1 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
4 with fast unaligned access.
5 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
7 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
9 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
10 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
12 2017-11-17 Richard Biener <rguenther@suse.de>
14 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
15 folding of references.
17 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
20 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
21 of replacing call to strncmp with corresponding call to strcmp when
24 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
26 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
27 option prefer-avx256 for skylake-avx512 configuration.
28 * config/i386/i386.c (ix86_option_override_internal): Ditto.
29 (get_builtin_code_for_version): Ditto.
31 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
32 Monk Chiang <sh.chiang04@gmail.com>
34 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
35 (FIXED_REGISTERS): Reserve more register numbers.
36 (CALL_USED_REGISTERS): Likewise.
37 (REG_ALLOC_ORDER): Likewise.
38 (REG_CLASS_CONTENTS): Likewise.
39 (REGISTER_NAMES): Likewise.
41 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
42 Kito Cheng <kito.cheng@gmail.com>
44 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
46 * config/nds32/iterators.md: Add vector mode iterators and attributes.
48 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
50 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
51 parameter list for vec_splats.
53 2017-11-16 Joseph Myers <joseph@codesourcery.com>
55 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
57 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
60 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
63 * doc/install.texi: Document incompatibility of Perl >=5.6.26
64 with the required version of automake 1.11.6.
66 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
68 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
70 (DU_C2_3_power9): Correct reservation combinations.
71 (FP_DIV_power9, VEC_DIV_power9): New.
72 (power9-alu): Split out rotate/shift...
73 (power9-rot): ...to here, correct dispatch resource.
74 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
76 (power9-fp): Correct latency.
77 (power9-sdiv): Add div/sqrt resource.
78 (power9-ddiv): Correct latency, add div/sqrt resource.
79 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
80 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
82 (power9-qpdiv, power9-qpmul): Adjust resource usage.
84 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
86 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
87 switch statement mapping KF built-ins to TF built-ins if we don't
88 have the proper ISA 3.0 assembler support.
90 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
92 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
93 (lower_emutls_function_body): Do not compute it.
95 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
97 * ipa-split.c (split_bb_info): Turn time to sreal.
98 (split_point): Likewise.
99 (dump_split_point): Likewise.
100 (fine_split_points): Likewise.
101 (execute_split_functions): Only zero split_bbs; turn time to sreals.
103 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
105 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
108 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
110 * predict.c (combine_predictions_for_bb): Preserve zero predicted
112 (expensive_function_p): Remove useless assert.
113 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
115 2017-11-16 Martin Sebor <msebor@redhat.com>
117 PR tree-optimization/82588
118 PR tree-optimization/82583
119 * tree-vrp.c (check_array_ref): Handle flexible array members,
120 string literals, and inner indices.
121 (search_for_addr_array): Add detail to diagnostics.
123 2017-11-16 Nathan Sidwell <nathan@acm.org>
127 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
128 (SET_DECL_ASSEMBLER_NAME): Forward to
129 overwrite_decl_assembler_name.
130 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
131 (overwrite_decl_assembler_name): Declare.
132 * tree.c (overwrite_decl_assembler_name): New.
133 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
134 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
135 (LANG_HOOKS_INITIALIZER): Add it.
136 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
137 * langhooks.c (lhd_set_decl_assembler_name): Use
138 SET_DECL_ASSEMBLER_NAME.
139 (lhd_overwrite_decl_assembler_name): Default implementation.
141 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
142 Jackson Woodruff <jackson.woodruff@arm.com>
144 PR tree-optimization/71026
145 * match.pd: Canonicalize constant multiplies in division.
147 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
149 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
150 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
151 Remove OPT_fomit_frame_pointer entry.
152 * common/config/arc/arc-common.c: Likewise.
153 * common/config/arm/arm-common.c: Likewise.
154 * common/config/avr/avr-common.c: Likewise.
155 * common/config/c6x/c6x-common.c: Likewise.
156 * common/config/cr16/cr16-common.c: Likewise.
157 * common/config/cris/cris-common.c: Likewise.
158 * common/config/epiphany/epiphany-common.c: Likewise.
159 * common/config/fr30/fr30-common.c: Likewise.
160 * common/config/frv/frv-common.c: Likewise.
161 * common/config/ia64/ia64-common.c: Likewise.
162 * common/config/iq2000/iq2000-common.c: Likewise.
163 * common/config/lm32/lm32-common.c: Likewise.
164 * common/config/m32r/m32r-common.c: Likewise.
165 * common/config/mcore/mcore-common.c: Likewise.
166 * common/config/microblaze/microblaze-common.c: Likewise.
167 * common/config/mips/mips-common.c: Likewise.
168 * common/config/mmix/mmix-common.c: Likewise.
169 * common/config/mn10300/mn10300-common.c: Likewise.
170 * common/config/nios2/nios2-common.c: Likewise.
171 * common/config/pa/pa-common.c: Likewise.
172 * common/config/pdp11/pdp11-common.c: Likewise.
173 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
174 * common/config/riscv/riscv-common.c: Likewise.
175 * common/config/rs6000/rs6000-common.c: Likewise.
176 * common/config/rx/rx-common.c: Likewise.
177 * common/config/s390/s390-common.c: Likewise.
178 * common/config/sh/sh-common.c: Likewise.
179 * common/config/sparc/sparc-common.c: Likewise.
180 * common/config/tilegx/tilegx-common.c: Likewise.
181 * common/config/tilepro/tilepro-common.c: Likewise.
182 * common/config/v850/v850-common.c: Likewise.
183 * common/config/visium/visium-common.c: Likewise.
184 * common/config/xstormy16/xstormy16-common.c: Likewise.
185 * common/config/xtensa/xtensa-common.c: Likewise.
186 * invoke.texi (-fomit-frame-pointer): Update documentation.
188 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
190 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
192 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
194 * cfg.c (scale_bbs_frequencies_int,
195 cale_bbs_frequencies_gcov_type): Remove.
196 * cfg.h (scale_bbs_frequencies_int,
197 cale_bbs_frequencies_gcov_type): Remove.
199 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
201 * tree-ssa-loop-manip.c
202 (scale_dominated_blocks_in_loop): Update to profile counts.
203 (tree_transform_and_unroll_loop): Likewise.
205 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
207 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
208 scale_bbs_frequencies_int.
210 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
212 * final.c (compute_alignments): Use counts rather than frequencies.
214 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
216 * cfgloopanal.c: Include sreal.h
217 (average_num_loop_insns): Use counts and sreal for accounting.
219 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
221 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
224 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
225 Kito Cheng <kito.cheng@gmail.com>
227 * config/nds32/constraints.md: Provide more constraints.
228 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
229 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
230 support constraints usage.
232 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
234 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
236 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
237 Kito Cheng <kito.cheng@gmail.com>
239 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
240 * config/nds32/nds32.opt: Refine the layout.
241 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
242 TARGET_EXT_STRING): Support new options.
243 * config/nds32/nds32.h: Likewise.
244 * config/nds32/nds32.md: Likewise.
245 * config/nds32/nds32-predicates.c: Likewise.
246 * config/nds32/constraints.md: Likewise.
247 * common/config/nds32/nds32-common.c: Likewise.
249 2017-11-16 Julia Koval <julia.koval@intel.com>
252 * config/i386/gfniintrin.h: Add sse check.
253 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
255 2017-11-16 Julia Koval <julia.koval@intel.com>
257 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
258 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
259 (ix86_handle_option): Handle -mavx512vbmi2.
260 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
261 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
262 * config/i386/i386-c.c (__AVX512VBMI2__): New.
263 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
264 (ix86_valid_target_attribute_inner_p): Ditto.
265 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
266 * config/i386/i386.opt (mavx512vbmi2): New option.
267 * doc/invoke.texi: Add new option.
269 [2017-11-16 Julia Koval <julia.koval@intel.com>
271 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
272 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
273 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
274 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
275 _mm512_gf2p8mul_epi8): New intrinsics.
276 * config/i386/i386-builtin-types.def
277 (V64QI_FTYPE_V64QI_V64QI): New type.
278 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
279 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
280 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
281 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
282 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
283 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
285 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
287 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
288 explicitly as a stream of bytes.
290 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
292 * config/rs6000/altivec.h (vec_xst_be): New #define.
293 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
294 and externalize from *altivec_vperm_<mode>_internal.
295 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
297 (XL_BE_V8HI): Likewise.
298 (XL_BE_V4SI): Likewise.
299 (XL_BE_V4SI): Likewise.
300 (XL_BE_V2DI): Likewise.
301 (XL_BE_V4SF): Likewise.
302 (XL_BE_V2DF): Likewise.
303 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
304 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
305 all array entries with these keys: VSX_BUILTIN_VEC_XL,
306 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
307 VSX_BUILTIN_VEC_XST_BE.
308 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
309 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
311 (altivec_init_builtins): Replace conditional calls to def_builtin
312 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
313 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
314 with unconditional calls. Remove calls to def_builtin for
315 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
316 __builtin_vec_xst_be.
317 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
318 to define_expand, and add alternate RTL generation for P8.
319 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
321 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
322 add alternate RTL generation for P8.
323 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
324 vsx_ld_elemrev_v16qi.
325 (vsx_st_elemrev_v8hi): Convert define_insn
326 to define_expand, and add alternate RTL generation for P8.
327 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
329 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
330 add alternate RTL generation for P8.
331 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
332 vsx_st_elemrev_v16qi.
334 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
337 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
338 TARGET_AVX512ER check.
339 (ix86_option_override_internal): Set MASK_VZEROUPPER if
340 neither -mzeroupper nor -mno-zeroupper is used and
341 TARGET_EMIT_VZEROUPPER is set.
342 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
343 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
345 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
347 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
348 folding of vector compares.
349 (fold_build_vec_cmp): New helper function.
350 (fold_compare_helper): New helper function.
351 (builtin_function_type): Add compare builtins to the list of functions
352 having unsigned arguments. Cosmetic updates to comment indentation.
353 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
354 the not+eq combination.
356 2017-11-15 Bin Cheng <bin.cheng@arm.com>
358 PR tree-optimization/82726
359 PR tree-optimization/70754
360 * tree-predcom.c (order_drefs_by_pos): New function.
361 (combine_chains): Move code setting has_max_use_after to...
362 (try_combine_chains): ...here. New parameter. Sort combined chains
363 according to position information.
364 (tree_predictive_commoning_loop): Update call to above function.
365 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
367 2017-11-15 Bin Cheng <bin.cheng@arm.com>
369 PR tree-optimization/82726
371 2017-01-23 Bin Cheng <bin.cheng@arm.com>
373 PR tree-optimization/70754
374 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
375 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
376 combined stmt before it if not NULL.
377 (combine_chains): Process refs reversely and compute dominance point
381 2017-02-23 Bin Cheng <bin.cheng@arm.com>
383 PR tree-optimization/79663
384 * tree-predcom.c (combine_chains): Process refs in reverse order
385 only for ZERO length chains, and add explaining comment.
387 2017-11-15 Tamar Christina <tamar.christina@arm.com>
389 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
390 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
391 * doc/invoke.texi (ARM Options): Add armv8.3-a.
393 2017-11-15 Tamar Christina <tamar.christina@arm.com>
395 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
397 2017-11-15 Martin Liska <mliska@suse.cz>
399 * tree-cfg.c (pass_warn_function_return::execute):
400 Compare warn_return_type for greater than zero.
402 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
406 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
407 to return true on Xeon and not on Xeon Phi.
408 (ix86_check_avx256_register): Changed to ...
409 (ix86_check_avx_upper_register): ... this. Add extra check for
410 VALID_AVX512F_REG_OR_XI_MODE.
411 (ix86_avx_u128_mode_needed): Changed
412 ix86_check_avx256_register to ix86_check_avx_upper_register.
413 (ix86_check_avx256_stores): Changed to ...
414 (ix86_check_avx_upper_stores): ... this. Changed
415 ix86_check_avx256_register to ix86_check_avx_upper_register.
416 (ix86_avx_u128_mode_after): Changed
417 avx_reg256_found to avx_upper_reg_found. Changed
418 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
419 (ix86_avx_u128_mode_entry): Changed
420 ix86_check_avx256_register to ix86_check_avx_upper_register.
421 (ix86_avx_u128_mode_exit): Ditto.
422 * config/i386/i386.h: (host_detect_local_cpu): New define.
424 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
426 * config/arm/xgene1.md (xgene1): Split into automatons
427 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
428 (xgene1_f_load): Adjust reservations and/or types.
429 (xgene1_f_store): Likewise.
430 (xgene1_load_pair): Likewise.
431 (xgene1_store_pair): Likewise.
432 (xgene1_fp_load1): Likewise.
433 (xgene1_load1): Likewise.
434 (xgene1_store1): Likewise.
435 (xgene1_move): Likewise.
436 (xgene1_alu): Likewise.
437 (xgene1_simd): Likewise.
438 (xgene1_bfm): Likewise.
439 (xgene1_neon_load1): Likewise.
440 (xgene1_neon_store1): Likewise.
441 (xgene1_neon_logic): Likewise.
442 (xgene1_neon_st1): Likewise.
443 (xgene1_neon_ld1r): Likewise.
444 (xgene1_alu_cond): Added.
445 (xgene1_shift_reg): Likwise.
446 (xgene1_bfx): Likewise.
447 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
449 2017-11-15 Jakub Jelinek <jakub@redhat.com>
452 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
454 (can_widen_mult_without_libcall): New function.
455 (expand_mul_overflow): If only checking unsigned mul overflow,
456 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
457 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
458 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
459 (expand_DIVMOD): Formatting fix.
460 * expmed.h (expand_mult): Add NO_LIBCALL argument.
461 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
462 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
464 PR tree-optimization/82977
465 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
466 constructed temporary to strlen_to_stridx.put.
468 2017-11-15 Martin Liska <mliska@suse.cz>
470 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
471 * configure: Regenerate.
473 2017-11-15 Martin Liska <mliska@suse.cz>
476 * config/sh/sh-mem.cc: Use proper probability for
479 2017-11-14 Jeff Law <law@redhat.com>
481 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
482 the red zone for stack_clash_protection_final_dynamic_probe targets
483 when the total dynamic stack size is zero bytes.
485 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
486 blocks is post order.
488 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
490 * dumpfile.h (TDF_COMPARE_DEBUG): New.
491 * final.c (rest_of_clean_state): Set it for the
492 -fcompare-debug dump.
493 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
494 class when TDF_COMPARE_DEBUG is set.
496 * dwarf2out.c (gen_producer_string): Discard
499 2017-11-15 Joseph Myers <joseph@codesourcery.com>
502 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
503 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
504 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
505 (__TGMATH_REAL_2_3): Remove macros.
506 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
507 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
509 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
511 (remquo): Define using __TGMATH_REAL_3.
513 2017-11-14 Jeff Law <law@redhat.com>
515 * vr-values.c: New file with contents extracted from tree-vrp.c.
516 * Makefile.in (OBJS): Add vr-values.o
517 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
518 (set_value_range, set_and_canonicalize_value_range): Likewise.
519 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
520 (value_range_constant_singleton, symbolic_range_p): Likewise.
521 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
522 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
523 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
524 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
525 (find_case_label_range, find_case_label_index): Likewise.
526 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
527 (range_int_cst_singleton_p, value_inside_range): Likewise.
528 (get_single_symbol): Likewise.
529 (switch_update): Move structure definition here.
530 (to_remove_edges, to_update_switch_stmts): Provide externs.
531 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
532 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
533 (vrp_val_is_min, set_value_range): Likewise.
534 (set_and_canonicalize_value_range, copy_value_range): Likewise.
535 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
536 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
537 (range_is_nonnull, range_int_cst_p): Likewwise.
538 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
539 (get_single_symbol, operand_less_p): Likewise
540 (compare_values_warnv, compare_values): Likewise.
541 (value_inside_range, value_range_constant_singleton): Likewise.
542 (zero_nonzero_bitgs_from_vr): Likewise.
543 (extract_range_from_binary_expr_1): Likewise.
544 (overflow_comparison_p): Likewise.
545 (to_remove_edges, to_update_switch_stmts): Likewise.
546 (find_case_label-index, find_case_label_range): Likewise.
547 (switch_update, set_value_range_to_nonnegative): Remove.
548 (set_value_range_to_truthvalue): Likewise.
549 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
550 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
551 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
552 (find_case_label_ranges, test_for_singularity): Likewise.
553 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
554 (x_vr_values): Move to its remaining use site.
556 2017-11-10 Jeff Law <law@redhat.com>
558 * vr-values.h (VR_INITIALIZER): Move #define here.
559 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
560 * Makefile.in (OBJS): Add tree-evrp.o
561 * tree-vrp.h (assert_info): Move structure definition here.
562 (set_value_range_to_varying): Prototype.
563 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
564 (infer_value_range, register_edge_assert_for): Likewise.
565 (stmt_interesting_for_vrp): Likewise.
566 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
567 (set_value_range_to_varying): No longer static.
568 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
569 (infer_value_range, register_edge_assert_for): Likewise.
571 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
573 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
574 generate the XXBRD instruction.
576 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
577 (rs6000_builtin_type_compatible): Treat _Float128 and long double
578 as being compatible if -mabi=ieeelongdouble.
579 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
580 to setup float128 built-ins with hardware support.
581 (BU_FLOAT128_HW_2): Likewise.
582 (BU_FLOAT128_HW_3): Likewise.
583 (BU_FLOAT128_HW_VSX_1): Likewise.
584 (BU_FLOAT128_HW_VSX_2): Likewise.
585 (scalar_extract_expq): Change float128 built-in functions to
586 accommodate having both KFmode and TFmode functions. Use the
587 KFmode variant as the default.
588 (scalar_extract_sigq): Likewise.
589 (scalar_test_neg_qp): Likewise.
590 (scalar_insert_exp_q): Likewise.
591 (scalar_insert_exp_qp): Likewise.
592 (scalar_test_data_class_qp): Likewise.
593 (sqrtf128_round_to_odd): Delete processing the round to odd
594 built-in functions as special built-in functions, and define them
595 as float128 built-ins. Use the KFmode variant as the default.
596 (truncf128_round_to_odd): Likewise.
597 (addf128_round_to_odd): Likewise.
598 (subf128_round_to_odd): Likewise.
599 (mulf128_round_to_odd): Likewise.
600 (divf128_round_to_odd): Likewise.
601 (fmaf128_round_to_odd): Likewise.
602 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
603 support for KFmode and TFmode xststdcqp calls.
604 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
605 point, switch the built-in handlers for the get/set float128
606 exponent, get float128 mantissa, float128 test built-ins, and the
607 float128 round to odd built-in functions. Eliminate creating the
608 float128 round to odd built-in functions as special built-ins.
609 (rs6000_init_builtins): Eliminate special creation of the float128
610 round to odd built-in functions.
611 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
612 function insns to support both TFmode and KFmode variants.
613 (xsxsigqp_<mode>): Likewise.
614 (xsiexpqpf_<mode>): Likewise.
615 (xsiexpqp_<mode>): Likewise.
616 (xststdcqp_<mode>): Likewise.
617 (xststdcnegqp_<mode>): Likewise.
618 (xststdcqp_<mode>): Likewise.
620 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
622 * tree-ssa-threadupdate.c (compute_path_counts): Remove
623 unused path_in_freq_ptr parameter.
624 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
626 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
628 * ipa-inline.c (edge_badness): Dump sreal frequency.
629 (compute_inlined_call_time): Match natural implementaiton ...
630 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
631 forgotten division by CGRAPH_FREQ_BASE.
633 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
635 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
636 Solaris 11. Update comment.
637 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
639 * config/sol2.h (STARTFILE_SPEC): Likewise.
640 * configure: Regenerate.
642 2017-11-14 Carl Love <cel@us.ibm.com>
644 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
645 le_ and be_ prefixes to swap* variables. Remove
646 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
649 2017-11-14 Jason Merrill <jason@redhat.com>
651 Support GTY((cache)) on hash_map.
652 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
653 (ggc_cache_remove): Override it instead of ggc_mx.
654 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
655 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
656 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
657 (simple_cache_map_traits): Override maybe_mx.
658 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
659 (hash_map): Friend gt_cleare_cache.
660 (gt_cleare_cache): New.
661 * tree.h (tree_cache_traits): New hash_map traits class.
662 (tree_cache_map): New typedef.
664 2017-11-14 Richard Biener <rguenther@suse.de>
666 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
667 paramter and handling.
668 (cleanup_control_flow_bb): Likewise.
669 (cleanup_control_flow_pre): New helper performing a DFS walk
670 to call cleanup_control_flow_bb in PRE order.
671 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
672 via cleanup_control_flow_pre.
674 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
676 * config/aarch64/aarch64-simd.md
677 (aarch64_simd_bsl<mode>_internal): Remove DImode.
678 (*aarch64_simd_bsl<mode>_alt): Likewise.
679 (aarch64_simd_bsldi_internal): New.
680 (aarch64_simd_bsldi_alt): Likewise.
682 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
684 * tracer.c (better_p): Do not compare frequencies.
685 * reg-stack.c (better_edge): Likewise.
686 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
689 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
691 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
692 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
693 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
694 * lto-streamer-in.c (input_function): Use update_max_bb_count.
695 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
696 * predict.c (maybe_hot_frequency_p): Inline to ...
697 (maybe_hot_count_p): ... here; rewrite to counts.
698 (counts_to_freqs): Rename to ...
699 (update_max_bb_count): ... this one.
700 (expensive_function_p): Use counts.
701 (estimate_bb_frequencies): Update.
702 (rebuild_frequencies): Update.
703 * predict.h (counts_to_freqs): Rename to ...
704 (update_max_bb_count): ... this one.
705 * profile.c (compute_branch_probabilities): Add debug info
706 * tree-inline.c (expand_call_inline): Update debug info.
707 (optimize_inline_calls): Use update_max_bb_count..
708 (tree_function_versioning): Use update_max_bb_count..
709 * value-prof.c (gimple_value_profile_transformations):
710 Do not use update_max_bb_count.
712 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
714 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
715 always use frequencies.
717 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
719 * bb-reorder.c: Remove frequencies from comments.
720 (better_edge_p): Use profile counts.
721 (find_traces): Dump profile counts.
722 (rotate_loop): Use profile counts.
723 (find_traces_1_round): Likewise.
724 (connect_better_edge_p): Use counts instead of probabilities for
726 (copy_bb_p): Drop early check for non-0 frequency.
727 (sanitize_hot_paths): Update comments.
729 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
731 * ipa-split.c (struct split_point): Add count.
732 (consider_split): Do not compute incoming frequency; compute incoming
733 count and store it to split_point.
734 (split_function): Set count of the call to split part correctly.
736 2017-11-13 Carl Love <cel@us.ibm.com>
738 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
740 2017-11-13 Tom Tromey <tom@tromey.com>
742 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
744 2017-11-13 Carl Love <cel@us.ibm.com>
746 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
747 Add support for builtins:
748 unsigned int vec_first_{,miss}_match_{,or_eos}index,
749 vector {un,}signed {char,int,short},
750 vector {un,}signed {char,int,short}) arguments.
751 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
752 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
753 Add BU_P9V_AV_2 expansions for the builtins.
754 * config/rs6000/altivec.h (vec_first_match_index,
755 vec_first_mismatch_index, vec_first_match_or_eos_index,
756 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
757 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
758 new extern declaration.
759 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
760 * config/rs6000/vsx.md (first_match_index_<mode>,
761 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
762 first_mismatch_or_eos_index_<mode>): Add define expand.
763 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
764 * doc/extend.texi: Update the built-in documenation file for the new
767 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
769 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
770 into the min/max operations for _Float<N> and _Float<N>X types.
772 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
775 * dwarf2out.c (do_eh_frame): New static variable.
776 (dwarf2out_begin_prologue): Set it.
777 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
779 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
781 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
783 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
785 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
788 * debug.h (dwarf2out_do_eh_frame): Declare.
789 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
790 (dwarf2out_do_frame): Use it.
791 (dwarf2out_do_cfi_asm): Likewise.
792 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
793 (dwarf2out_assembly_start): Likewise.
794 (dwarf2out_begin_prologue): Fix comment.
795 * toplev.c (compile_file): Always call dwarf2out_frame_finish
796 if the target needs either debug or unwind DWARF2 info.
797 * lto-opts.c (lto_write_options): Do not save -fexceptions,
798 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
799 -ftrapping-math, -ftrapv and -fwrapv.
801 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
803 * cgraph.c (cgraph_edge::sreal_frequency): New function.
804 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
805 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
806 (estimate_edge_size_and_time): Likewise.
807 (ipa_merge_fn_summary_after_inlining): Likewise.
808 * ipa-inline.c (cgraph_freq_base_rec): Remove.
809 (compute_uninlined_call_time): Use sreal_frequency.
810 (compute_inlined_call_time): Likewise.
811 (ipa_inline): Do not initialize cgraph_freq_base_rec.
812 * profile-count.c: Include sreal.h.
813 (profile_count::to_sreal_scale): New.
814 * profile-count.h: Forward declare sreal.
815 (profile_count::to_sreal_scale): Declare.
817 2017-11-13 Nathan Sidwell <nathan@acm.org>
819 * diagnostic.c (maybe_line_and_column): New.
820 (diagnostic_get_location_text): Use it.
821 (diagnostic_report_current_module): Likewise.
822 (test_diagnostic_get_location_text): Add tests.
824 2017-11-13 Luis Machado <luis.machado@linaro.org>
826 * doc/md.texi (Specifying processor pipeline description): Fix
827 incorrect latency for the div instruction example.
829 2017-11-13 Jakub Jelinek <jakub@redhat.com>
831 PR tree-optimization/78821
832 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
833 that bit_not_p is the same.
834 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
835 (split_group): Count precisely bit_not_p bits in each statement.
836 (invert_op): New function.
837 (imm_store_chain_info::output_merged_store): Use invert_op to
838 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
839 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
841 2017-11-13 Martin Liska <mliska@suse.cz>
843 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
844 (struct source_info): Likewise.
845 (add_branch_counts): Likewise.
846 (add_line_counts): Likewise.
847 (function_summary): Likewise.
848 (output_intermediate_line): Likewise.
849 (generate_results): Likewise.
851 2017-11-13 Martin Liska <mliska@suse.cz>
853 * gcov.c (struct block_info): Remove typedef for block_t.
854 (struct line_info): Likewise.
855 (line_info::has_block): Likewise.
856 (EXIT_BLOCK): Likewise.
859 (get_cycles_count): Likewise.
860 (process_file): Likewise.
861 (read_graph_file): Likewise.
862 (solve_flow_graph): Likewise.
863 (find_exception_blocks): Likewise.
864 (add_line_counts): Likewise.
865 (accumulate_line_info): Likewise.
866 (output_line_details): Likewise.
868 2017-11-13 Martin Liska <mliska@suse.cz>
870 * gcov.c (struct arc_info): Remove typedef for arc_t.
871 (struct line_info): Likewise.
872 (add_branch_counts): Likewise.
873 (output_branch_count): Likewise.
874 (function_info::~function_info): Likewise.
876 (output_intermediate_line): Likewise.
877 (read_graph_file): Likewise.
878 (solve_flow_graph): Likewise.
879 (find_exception_blocks): Likewise.
880 (add_line_counts): Likewise.
881 (accumulate_line_info): Likewise.
882 (output_line_details): Likewise.
883 (output_function_details): Likewise.
885 2017-11-13 Martin Liska <mliska@suse.cz>
887 * gcov.c (struct function_info): Remove typedef for function_t.
888 (struct source_info): Likewise.
889 (source_info::get_functions_at_location): Likewise.
890 (solve_flow_graph): Likewise.
891 (find_exception_blocks): Likewise.
892 (add_line_counts): Likewise.
893 (output_intermediate_file): Likewise.
894 (process_file): Likewise.
895 (generate_results): Likewise.
896 (release_structures): Likewise.
897 (read_graph_file): Likewise.
898 (read_count_file): Likewise.
899 (accumulate_line_counts): Likewise.
900 (output_lines): Likewise.
902 2017-11-13 Martin Liska <mliska@suse.cz>
904 * gcov.c (function_info::function_info): Remove num_counts
905 and add vector<gcov_type>.
906 (function_info::~function_info): Use the vector.
907 (process_file): Likewise.
908 (read_graph_file): Likewise.
909 (read_count_file): Likewise.
910 (solve_flow_graph): Likewise.
912 2017-11-13 Martin Liska <mliska@suse.cz>
914 * gcov.c (function_info::is_artificial): New function.
915 (process_file): Erase all artificial early.
916 (generate_results): Skip as all artificial are already
919 2017-11-13 Martin Liska <mliska@suse.cz>
921 * gcov.c (read_graph_file): Store to global vector of functions.
922 (read_count_file): Iterate the vector.
923 (process_file): Likewise.
924 (generate_results): Likewise.
925 (release_structures): Likewise.
927 2017-11-13 Jakub Jelinek <jakub@redhat.com>
929 PR tree-optimization/82954
930 * gimple-ssa-store-merging.c
931 (imm_store_chain_info::coalesce_immediate_stores): If
932 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
934 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
936 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
937 Upddate call to ENDIAN_LANE_N.
938 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
939 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
940 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
941 and use aarch64_endian_lane_rtx.
942 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
944 2017-11-12 Tom de Vries <tom@codesourcery.com>
946 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
948 2017-11-12 Tom de Vries <tom@codesourcery.com>
950 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
952 2017-11-12 Tom de Vries <tom@codesourcery.com>
954 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
956 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
957 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
958 * defaults.h (ASM_OUTPUT_LABELREF): Same.
960 2017-11-11 Martin Sebor <msebor@redhat.com>
963 * doc/extend.texi (attribute nonstring): Remove spurious argument.
966 * prefic.c (translate_name): Replace strncpy with memcpy to
967 avoid -Wstringop-truncation.
969 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
971 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
973 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
975 * predict.c (maybe_hot_frequency_p): Do not use cfun.
977 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
979 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
982 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
984 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
985 enable generating XXBRH if the value is in a vector register.
986 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
987 value is in a vector register.
988 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
989 register to register bswap64's instead of doing the GPR sequence
990 used on previous machines.
991 (bswapdi2_xxbrd): New insn.
992 (bswapdi2_reg): Disallow on ISA 3.0.
993 (register to register bswap64 splitter): Do not split the insn on
994 ISA 3.0 systems that use XXBRD.
996 2017-11-10 Martin Sebor <msebor@redhat.com>
999 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1001 (find_subframework_file): Same.
1003 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1005 * auto-profile.c (afdo_indirect_call): Drop frequency.
1006 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1007 (cgraph_node::create_edge): Drop frequency argument.
1008 (cgraph_node::create_indirect_edge): Drop frequency argument.
1009 (cgraph_edge::make_speculative): Drop frequency arguments.
1010 (cgraph_edge::resolve_speculation): Do not update frequencies
1011 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1012 (cgraph_node::dump): Check consistency in IPA mode.
1013 (cgraph_edge::maybe_hot_p): Use IPA counter.
1014 (cgraph_edge::verify_count_and_frequency): Rename to ...
1015 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1016 (cgraph_node::verify_node): Update.
1017 * cgraph.h (struct cgraph_edge): Drop frequency.
1018 (cgraph_edge::frequency): New function.
1019 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1021 (cgraph_edge::rebuild_edges): Likewise.
1022 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1023 (duplicate_thunk_for_node): Do not pass frequency.
1024 (cgraph_node::create_clone): Scale only counts.
1025 (cgraph_node::create_virtual_clone): Do not pass frequency.
1026 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1027 (cgraph_node::create_version_clone): Do not pass frequency.
1028 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1029 (cgraph_node::expand_thunk): Do not pass frequency.
1030 (cgraph_node::create_wrapper): Do not pass frequency.
1031 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1033 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1034 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1035 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1036 (gather_caller_stats): Use frequency function.
1037 (ipcp_cloning_candidate_p): Use frequency function.
1038 (ipcp_propagate_stage): Use frequency function.
1039 (get_info_about_necessary_edges): Use frequency function.
1040 (update_profiling_info): Update only IPA profile.
1041 (update_specialized_profile): Use frequency functoin.
1042 (perhaps_add_new_callers): Update only IPA profile.
1043 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1044 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1045 (dump_ipa_call_summary): Use frequency function.
1046 (estimate_edge_size_and_time): Use frequency function.
1047 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1048 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1049 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1050 (update_noncloned_counts): ... ths one; scale counts only.
1051 (clone_inlined_nodes): Do not scale frequency.
1052 (inline_call): Do not pass frequency.
1053 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1054 (compute_inlined_call_time): Use IPA profile.
1055 (want_inline_small_function_p): Use IPA profile.
1056 (want_inline_self_recursive_call_p): Use IPA profile.
1057 (edge_badness): Use IPA profile.
1058 (lookup_recursive_calls): Use IPA profile.
1059 (recursive_inlining): Do not pass frequency.
1060 (resolve_noninline_speculation): Do not update frequency.
1061 (inline_small_functions): Collect max of IPA profile.
1062 (dump_overall_stats): Dump IPA porfile.
1063 (dump_inline_stats): Dump IPA porfile.
1064 (ipa_inline): Collect IPA stats.
1065 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1066 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1067 (ipa_propagate_frequency): Use frequency function.
1068 (ipa_profile): Cleanup.
1069 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1070 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1071 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1072 (input_node): Do not stream frequency.
1073 (input_edge): Do not stream frequency.
1074 (merge_profile_summaries): Scale only IPA profiles.
1075 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1076 * predict.c (drop_profile): Do not recompute frequency.
1077 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1078 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1079 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1080 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1081 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1082 * tree-inline.c (copy_bb): Do not scale frequency.
1083 (expand_call_inline): Do not scale frequency.
1084 (tree_function_versioning): Do not scale frequency.
1085 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1087 2017-11-10 Julia Koval <julia.koval@intel.com>
1089 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1090 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1091 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1092 (_mm256_maskz_gf2p8affine_epi64_epi8)
1093 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1094 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1095 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1096 (__builtin_ia32_vgf2p8affineqb_v32qi)
1097 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1098 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1100 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1104 (arm_option_override): Refactor.
1105 (arm_option_reconfigure_globals): New.
1106 (arm_options_perform_arch_sanity_checks): New.
1107 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1109 (arm_options_perform_arch_sanity_checks): Likewise
1111 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1113 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1114 (power9-qpmul): New.
1115 * rs6000/rs6000.md ("type" attr): Add qmul.
1116 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1117 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1118 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1120 2017-11-10 Martin Sebor <msebor@redhat.com>
1123 * builtins.c (compute_objsize): Handle arrays that
1124 compute_builtin_object_size likes to fail for. Make extern.
1125 * builtins.h (compute_objsize): Declare.
1126 (check_strncpy_sizes): New function.
1127 (expand_builtin_strncpy): Call check_strncpy_sizes.
1128 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1129 -Wstringop-truncation.
1130 (gimple_fold_builtin_strncat): Same.
1131 * gimple.c (gimple_build_call_from_tree): Set call location.
1132 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1133 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1134 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1135 (handle_builtin_strlen): Use strlen_to_stridx.
1136 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1138 Use strlen_to_stridx.
1139 (pass_strlen::execute): Release strlen_to_stridx.
1140 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1141 (-Wstringop-truncation): Document new option.
1143 2017-11-10 Martin Liska <mliska@suse.cz>
1145 PR gcov-profile/82702
1146 * gcov.c (main): Handle intermediate files in a different
1148 (get_gcov_intermediate_filename): New function.
1149 (output_gcov_file): Remove support of intermediate files.
1150 (generate_results): Allocate intermediate file.
1151 (release_structures): Clean-up properly fn_end.
1152 (output_intermediate_file): Start iterating with line 1.
1154 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1156 PR tree-optimization/82929
1157 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1158 ops_swapped_p non-static data member.
1159 (store_immediate_info::store_immediate_info): Clear it.
1160 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1161 ops set ops_swapped_p.
1162 (count_multiple_uses): Handle ops_swapped_p.
1164 2017-11-10 Martin Liska <mliska@suse.cz>
1166 * coverage.c (coverage_init): Stream information about
1167 support of has_unexecuted_blocks.
1168 * doc/gcov.texi: Document that.
1169 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1170 * gcov.c (read_graph_file): Likewise.
1171 (output_line_beginning): Fix a small issue with
1174 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1176 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1177 reference of trivial component.
1179 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1182 * gimple-ssa-store-merging.c
1183 (pass_store_merging::terminate_all_aliasing_chains): For
1184 gimple_store_p stmts also call refs_output_dependent_p.
1186 PR rtl-optimization/82913
1187 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1190 2017-11-09 Jeff Law <law@redhat.com>
1192 * vr-values.h: New file with vr_values class.
1193 * tree-vrp.c: Include vr-values.h
1194 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1195 data objects into the vr_values class.
1196 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1197 (get_value_range): Make it a member function within vr_values class.
1198 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1199 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1200 (op_with_constant_singleton_value_range): Likewise.
1201 (extract_range_for_var_from_comparison_expr): Likewise.
1202 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1203 (extract_range_from_binary_expr): Likewise.
1204 (extract_range_from_unary_expr): Likewise.
1205 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1206 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1207 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1208 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1209 (compare_name_with_value, compare_names): Likewise.
1210 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1211 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1212 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1213 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1214 (extract_range_from_phi_node): Likewise.
1215 (simplify_truth_ops_using_ranges): Likewise.
1216 (simplify_div_or_mod_using_ranges): Likewise.
1217 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1218 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1219 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1220 (simplify_float_conversion_using_ranges): Likewise.
1221 (simplify_internal_call_using_ranges): Likewise.
1222 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1223 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1224 poitner via x_vr_values for calls into gimple folder.
1225 (vrp_initialize_lattice): Make this the vr_values ctor.
1226 (vrp_free_lattice): Make this the vr_values dtor.
1227 (set_vr_value): New function.
1228 (class vrp_prop): Add vr_values data member. Add various member
1229 functions as well as member functions that delegate to vr_values.
1230 (check_array_ref): Make a member function within vrp_prop class.
1231 (search_for_addr_array, vrp_initialize): Likewise.
1232 (vrp_finalize): Likewise. Revamp to avoid direct access to
1233 vr_value, values_propagated, etc.
1234 (check_array_bounds): Extract vrp_prop class instance pointer from
1235 walk info structure. Use it to call member functions.
1236 (check_all_array_refs): Make a member function within vrp_prop class.
1237 Smuggle class instance pointer via walk info structure.
1238 (x_vr_values): New local static.
1239 (vrp_valueize): Use x_vr_values to get class instance.
1240 (vr_valueize_1): Likewise.
1241 (class vrp_folder): Add vr_values data member. Add various member
1242 functions as well as member functions that delegate to vr_values.
1243 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1244 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1245 class instance from vr_values. Use it to call member functions.
1246 (vrp_dom_walker): Add vr_values data member.
1247 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1248 instance via x_vr_values.
1249 (identify_jump_threads): Accept vr_values as argument. Store
1250 it into the walker structure.
1251 (evrp_dom_walker): Add vr_values class data member. Add various
1253 (evrp_dom_walker::try_find_new_range): Use vr_values data
1254 member to access the memory allocator.
1255 (evrp_dom_walker::before_dom_children): Store vr_values class
1256 instance into the vrp_folder class.
1257 (evrp_dom_walker::push_value_range): Rework to avoid direct
1258 access to num_vr_values and vr_value.
1259 (evrp_dom_walker::pop_value_range): Likewise.
1260 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1261 Use vr_values to get to dump_all_value_ranges member function.
1262 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1263 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1264 Pass vr_values class instance down to identify_jump_threads.
1265 Remove call to vrp_free_lattice.
1266 (debug_all_value_ranges): Remove.
1268 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1269 (vrp_folder): Likewise.
1271 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1272 Get it from the existing bitmap instead.
1273 (vrp_intersect_ranges_1): Likewise.
1275 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1277 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1279 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1280 set bit_not_p to it.
1281 (imm_store_chain_info::coalesce_immediate_stores): Break group
1282 if bit_not_p is different.
1283 (count_multiple_uses, split_group,
1284 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1285 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1286 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1287 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1288 to store_immediate_info ctor.
1290 2017-11-09 Jim Wilson <jimw@sifive.com>
1292 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1293 (scan_prog_file): Likewise.
1295 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1297 * bb-reorder.c (max_entry_frequency): Remove.
1298 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1299 connect_traces, push_to_next_round_p): Remove prototypes.
1300 (find_traces_1_round): Use counts only.
1301 (push_to_next_round_p): Likewise.
1302 (find_traces): Likewise.
1303 (rotate_loop): Likewise.
1304 (find_traces_1_round): Likewise.
1305 (connect_traces): Likewise.
1306 (edge_order): Likewise.
1308 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1310 * config/arm/arm.c (output_return_instruction): Add comments to
1311 indicate requirement for cmse_nonsecure_entry return to account
1312 for the size of clearing instruction output here.
1313 (thumb_exit): Likewise.
1314 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1315 return in hardfloat mode.
1317 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1319 * config/rs6000/rs6000.c (machine_function): Add a bool,
1320 "toc_is_wrapped_separately".
1321 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1322 if it wasn't explicitly set or unset, we are optimizing for speed, and
1323 doing separate shrink-wrapping.
1324 (rs6000_get_separate_components): Enable the TOC component if
1325 saving the TOC register in the prologue.
1326 (rs6000_components_for_bb): Handle the TOC component.
1327 (rs6000_emit_prologue_components): Store the TOC register where needed.
1328 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1329 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1331 2017-11-09 Martin Jambor <mjambor@suse.cz>
1333 * ipa-param-manipulation.c: New file.
1334 * ipa-param-manipulation.h: Likewise.
1335 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1336 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1337 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1338 (ipa_parm_adjustment): Likewise.
1339 (ipa_parm_adjustment_vec): Likewise.
1340 (ipa_get_vector_of_formal_parms): Moved declaration to
1341 ipa-param-manipulation.h.
1342 (ipa_get_vector_of_formal_parm_types): Likewise.
1343 (ipa_modify_formal_parameters): Likewise.
1344 (ipa_modify_call_arguments): Likewise.
1345 (ipa_combine_adjustments): Likewise.
1346 (ipa_dump_param_adjustments): Likewise.
1347 (ipa_modify_expr): Likewise.
1348 (ipa_get_adjustment_candidate): Likewise.
1349 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1350 ipa-param-manipulation.c.
1351 (ipa_get_vector_of_formal_parm_types): Likewise.
1352 (ipa_modify_formal_parameters): Likewise.
1353 (ipa_modify_call_arguments): Likewise.
1354 (ipa_modify_expr): Likewise.
1355 (get_ssa_base_param): Likewise.
1356 (ipa_get_adjustment_candidate): Likewise.
1357 (index_in_adjustments_multiple_times_p): Likewise.
1358 (ipa_combine_adjustments): Likewise.
1359 (ipa_dump_param_adjustments): Likewise.
1360 * tree-sra.c: Also include ipa-param-manipulation.h
1361 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1364 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1365 Alan Hayward <alan.hayward@arm.com>
1366 David Sherwood <david.sherwood@arm.com>
1368 * doc/sourcebuild.texi (vect_masked_store): Document.
1370 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1371 Alan Hayward <alan.hayward@arm.com>
1372 David Sherwood <david.sherwood@arm.com>
1374 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1376 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1377 Alan Hayward <alan.hayward@arm.com>
1378 David Sherwood <david.sherwood@arm.com>
1380 * doc/sourcebuild.texi (vect_variable_length): Document.
1382 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1383 Alan Hayward <alan.hayward@arm.com>
1384 David Sherwood <david.sherwood@arm.com>
1386 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1388 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1389 Alan Hayward <alan.hayward@arm.com>
1390 David Sherwood <david.sherwood@arm.com>
1392 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1394 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1395 Alan Hayward <alan.hayward@arm.com>
1396 David Sherwood <david.sherwood@arm.com>
1398 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1399 previously undocumented selectors.
1400 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1402 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1404 * doc/rtl.texi (const_vector): Say that elements can be
1405 const_wide_ints too.
1406 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1407 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1408 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1409 * optabs.c (expand_vector_broadcast): Likewise.
1411 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1412 Alan Hayward <alan.hayward@arm.com>
1413 David Sherwood <david.sherwood@arm.com>
1415 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1416 scaled index even if the unscaled address was invalid.
1417 Don't increase the complexity of using a scale in that case.
1419 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1420 Alan Hayward <alan.hayward@arm.com>
1421 David Sherwood <david.sherwood@arm.com>
1423 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1424 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1425 UNITS_PER_WORD bytes.
1426 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1427 based on REGMODE_NATURAL_SIZE of the inner mode.
1428 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1429 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1430 * expmed.c (lowpart_bit_field_p): Divide the value up into
1431 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1432 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1433 whether something is likely to occupy more than one register.
1435 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1438 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1439 profile_count::adjust_for_ipa_scaling.
1440 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1441 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1443 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1445 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1447 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1448 (split_group): Add total_orig and total_new arguments, estimate the
1449 number of statements related to the store group without store merging
1450 and with store merging.
1451 (imm_store_chain_info::output_merged_store): Adjust split_group
1452 callers, punt if estimated number of statements with store merging
1453 is not smaller than estimated number of statements without it.
1455 (handled_load): Remove has_single_use checks.
1456 (pass_store_merging::process_store): Likewise.
1458 2017-11-09 Richard Biener <rguenther@suse.de>
1460 PR tree-optimization/82902
1461 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1463 2017-11-09 Martin Liska <mliska@suse.cz>
1466 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1469 2017-11-09 Martin Liska <mliska@suse.cz>
1471 PR tree-optimization/82669
1472 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1474 2017-11-09 Martin Liska <mliska@suse.cz>
1476 PR gcov-profile/48463
1477 * coverage.c (coverage_begin_function): Output also end locus
1478 of a function and information whether the function is
1480 * gcov-dump.c (tag_function): Parse and print the information.
1481 * gcov.c (INCLUDE_MAP): Add include.
1482 (INCLUDE_SET): Likewise.
1483 (struct line_info): Move earlier in the source file because
1484 of vector<line_info> in function_info structure.
1485 (line_info::line_info): Likewise.
1486 (line_info::has_block): Likewise.
1487 (struct source_info): Add new member index.
1488 (source_info::get_functions_at_location): New function.
1489 (function_info::group_line_p): New function.
1490 (output_intermediate_line): New function.
1491 (output_intermediate_file): Use the mentioned function.
1492 (struct function_start): New.
1493 (struct function_start_pair_hash): Likewise.
1494 (process_file): Add code that identifies group functions.
1495 Assign lines either to global or function scope.
1496 (generate_results): Skip artificial functions.
1497 (find_source): Assign index for each source file.
1498 (read_graph_file): Read new flag artificial and end_line.
1499 (add_line_counts): Assign it either to global of function scope.
1500 (accumulate_line_counts): Isolate core of the function to
1501 accumulate_line_info and call it for both function and global
1503 (accumulate_line_info): New function.
1504 (output_line_beginning): Fix GNU coding style.
1505 (print_source_line): New function.
1506 (output_line_details): Likewise.
1507 (output_function_details): Likewise.
1508 (output_lines): Iterate both source (global) scope and function
1510 (struct function_line_start_cmp): New class.
1511 * doc/gcov.texi: Reflect changes in documentation.
1513 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1516 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1517 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1518 and similarly for not instead of neg.
1520 2017-11-08 Andi Kleen <ak@linux.intel.com>
1522 * config/i386/i386.opt: Add -mforce-indirect-call.
1523 * config/i386/predicates.md: Check for flag_force_indirect_call.
1524 * doc/invoke.texi: Document -mforce-indirect-call
1526 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1528 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1530 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1531 config/riscv/riscv.c (predict.h): New include.
1532 (riscv_slow_unaligned_access_p): No longer static.
1533 (riscv_block_move_straight): Add require.
1534 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1536 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1539 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1540 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1541 nonimmediate_operand predicate for operand 1 instead of
1544 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1546 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1548 * config/aarch64/constraints.md (Uml): New constraint.
1549 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1552 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1554 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1555 of two vec_duplicates into a vec_concat.
1557 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1559 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1560 Simplify vec_merge of vec_duplicate and vec_concat.
1561 * config/aarch64/constraints.md (Utq): New constraint.
1562 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1565 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1567 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1568 Simplify vec_merge of vec_duplicate and const_vector.
1569 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1571 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1572 mode iterator. Update predicate on operand 1 to
1573 handle non-const_vec constants. Delete constraints.
1574 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1576 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1578 PR tree-optimization/78821
1579 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1581 (store_operand_info::store_operand_info): Initialize it to false.
1582 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1583 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1584 store in the group, and if chain_info is non-NULL, to ignore altogether
1586 (compatible_load_p): Fail if bit_not_p does not match.
1587 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1588 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1589 (pass_store_merging::process_store): Adjust
1590 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1591 call terminate_all_aliasing_chains newly when adding a store into
1592 a chain with non-NULL chain_info.
1594 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1596 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1598 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1600 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1602 (aarch64_layout_frame): Initialise emit_frame_chain.
1603 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1604 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1606 2017-11-08 Martin Liska <mliska@suse.cz>
1608 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1609 of gimple_call_internal_p.
1611 2017-11-07 Tom de Vries <tom@codesourcery.com>
1613 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1615 2017-11-07 Tom de Vries <tom@codesourcery.com>
1617 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1620 2017-11-08 Martin Liska <mliska@suse.cz>
1623 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1625 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1627 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1629 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1632 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1633 (*cmp<mode>_ccz_1): New insn with $k alternative.
1636 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1637 RTX_COMM_COMPARE as commutative as well.
1638 (ix86_binary_operator_ok): Formatting fix.
1639 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1640 *<code><mode>3<mask_name><round_saeonly_name>,
1641 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1642 *<s>mul<mode>3_highpart<mask_name>,
1643 *vec_widen_umult_even_v16si<mask_name>,
1644 *vec_widen_umult_even_v8si<mask_name>,
1645 *vec_widen_umult_even_v4si<mask_name>,
1646 *vec_widen_smult_even_v16si<mask_name>,
1647 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1648 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1649 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1650 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1651 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1652 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1653 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1654 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1655 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1656 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1657 ix86_binary_operator_ok. Formatting fixes.
1658 (*<plusminus_insn><mode>3<mask_name><round_name>,
1659 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1662 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1664 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1665 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1666 eq and ne if TARGET_ISEL.
1667 (cmp): New code_iterator.
1668 (UNS, UNSU_, UNSIK): New code_attrs.
1669 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1670 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1672 ("eq<mode>3"): ... this.
1673 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1675 ("ne<mode>3"): ... this.
1677 2017-11-07 Julia Koval <julia.koval@intel.com>
1680 * common/config/i386/i386-common.c
1681 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1682 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1683 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1684 * config/i386/i386.opt: Ditto.
1685 * config/i386/i386.c (ix86_target_string): Ditto.
1686 (ix86_option_override_internal): Ditto.
1687 (ix86_init_mpx_builtins): Move MPX to args2.
1688 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1689 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1690 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1691 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1692 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1693 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1694 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1696 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1699 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1700 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1701 (zero-extendsidi peephole2): Remove peephole.
1703 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1706 * stor-layout.c: Include attribs.h.
1707 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1708 explicit lookup of "aligned" attribute.
1710 2017-11-07 Andrew Waterman <andrew@sifive.com>
1712 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1713 (riscv_expand_block_move): Likewise.
1714 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1716 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1717 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1718 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1719 (riscv_adjust_block_mem): Likewise.
1720 (riscv_block_move_loop): Likewise.
1721 (riscv_expand_block_move): Likewise.
1722 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1724 2017-11-07 Michael Clark <michaeljclark@mac.com>
1726 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1727 (MUSL_DYNAMIC_LINKER): Likewise.
1729 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1731 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1732 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1735 2017-11-07 Richard Biener <rguenther@suse.de>
1737 * match.pd: Fix build.
1739 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1740 Jackson Woodruff <jackson.woodruff@arm.com>
1742 PR tree-optimization/71026
1743 * match.pd: Canonicalize negate in division.
1745 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1748 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1750 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1752 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1753 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1755 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1757 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1758 non-scalar integral types.
1759 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1760 (-(A-B), -(~A)): New transformations.
1762 2017-11-07 Tom de Vries <tom@codesourcery.com>
1764 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1765 semicolon after "do {} while (0)".
1766 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1767 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1768 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1769 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1770 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1772 2017-11-07 Tom de Vries <tom@codesourcery.com>
1774 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1775 after "do {} while (0)".
1776 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1777 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1778 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1779 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1780 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1782 2017-11-07 Tom de Vries <tom@codesourcery.com>
1785 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1787 (arm_rtx_costs_internal): Add missing semicolon after
1788 HANDLE_NARROW_SHIFT_ARITH call.
1790 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1792 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1793 disable isel if it was not set explicitly.
1795 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1797 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1799 (add_type_attribute) likewise.
1801 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1803 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1804 of struct ix86_frame.
1805 (ix86_initial_elimination_offset): Likewise.
1806 (ix86_expand_split_stack_prologue): Likewise.
1808 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1810 * tree-vrp.h (enum value_range_type): Update stale comment.
1812 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1813 Alan Hayward <alan.hayward@arm.com>
1814 David Sherwood <david.sherwood@arm.com>
1816 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1817 (aarch64_expand_vec_perm_const): Take the number of units too.
1818 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1819 (aarch64_expand_vec_perm_const): Likewise.
1820 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1821 (vec_perm<mode>): Update accordingly.
1823 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1824 Alan Hayward <alan.hayward@arm.com>
1825 David Sherwood <david.sherwood@arm.com>
1827 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1828 Take the number of units too.
1829 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1830 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1831 but check for a vector mode before rather than after the call.
1832 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1833 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1834 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1835 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1836 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1837 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1838 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1839 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1840 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1841 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1842 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1843 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1844 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1845 (aarch64_sqdmull2_n<mode>): Update accordingly.
1847 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1848 Alan Hayward <alan.hayward@arm.com>
1849 David Sherwood <david.sherwood@arm.com>
1851 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1852 the number of units too.
1853 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1854 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1855 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1856 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1857 (vec_store_lanesxi<mode>): Update accordingly.
1859 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1860 Alan Hayward <alan.hayward@arm.com>
1861 David Sherwood <david.sherwood@arm.com>
1863 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1864 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1865 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1866 of units rather than the mode.
1867 * config/aarch64/iterators.md (nunits): New mode attribute.
1868 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1869 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1870 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1871 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1872 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1873 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1874 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1875 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1876 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1877 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1878 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1879 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1880 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1881 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1882 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1883 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1884 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1885 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1886 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1887 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1888 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1889 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1890 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1891 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1892 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1893 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1894 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1895 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1896 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1897 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1898 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1899 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1900 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1901 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1902 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1903 (aarch64_simd_vec_setv2di): Likewise.
1905 2017-11-06 Carl Love <cel@us.ibm.com>
1907 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1909 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1910 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1912 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1914 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1915 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1916 Add power 8 macro expansions.
1917 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1918 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1919 power 8 instructions. (VSX_XXBR): Add iterator.
1921 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1923 * config/arm/arm.md (predicable_short_it): Change default to "no",
1924 improve documentation, remove uses that are identical to the default.
1925 (enabled_for_depr_it): Rename to enabled_for_short_it.
1926 * gcc/config/arm/arm-fixed.md (predicable_short_it):
1927 Remove default uses.
1928 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1929 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1930 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1931 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1933 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1936 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1937 float128 helper macros, which are no longer used after deleting
1938 the old 'q' built-in functions, and moving the round to odd
1939 built-in functions to being special built-in functions.
1940 (BU_FLOAT128_2): Likewise.
1941 (BU_FLOAT128_1_HW): Likewise.
1942 (BU_FLOAT128_2_HW): Likewise.
1943 (BU_FLOAT128_3_HW): Likewise.
1944 (FABSQ): Delete old 'q' built-in functions.
1945 (COPYSIGNQ): Likewise.
1946 (SQRTF128_ODD): Move round to odd built-in functions to be
1947 special built-in functions, so that we can handle
1948 -mabi=ieeelongdouble.
1949 (TRUNCF128_ODD): Likewise.
1950 (ADDF128_ODD): Likewise.
1951 (SUBF128_ODD): Likewise.
1952 (MULF128_ODD): Likewise.
1953 (DIVF128_ODD): Likewise.
1954 (FMAF128_ODD): Likewise.
1955 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1956 built-in names to 'f128'.
1957 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1958 old 'q' built-in functions, as the machine independent code for
1959 'f128' built-in functions handles this.
1960 (rs6000_expand_builtin): Add expansion for float128 round to odd
1961 functions, keying off on -mabi=ieeelongdouble of whether to use
1962 the KFmode or TFmode variant.
1963 (rs6000_init_builtins): Initialize the _Float128 round to odd
1965 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1966 _Float128 'q' built-in functions are now mapped into the new
1967 'f128' built-in functions.
1969 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1971 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1972 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1974 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1977 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1978 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1979 * toplev.c: Include "ipa-fnsummary.h".
1980 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1982 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1984 PR tree-optimization/82838
1985 * gimple-ssa-store-merging.c
1986 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1987 on a separate gimple_seq which is then appended to seq.
1989 2017-11-06 Jeff Law <law@redhat.com>
1992 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1993 (get_probe_interval): New functions.
1994 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1995 (ix86_adjust_stack_and_probe): Likewise.
1996 (output_adjust_stack_and_probe): Likewise.
1997 (ix86_emit_probe_stack_range): Likewise.
1998 (ix86_expand_prologue): Likewise.
2000 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2002 PR tree-optimization/82816
2003 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2004 if the modes of the two types are the same.
2005 (convert_plusminus_to_widen): Likewise.
2007 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2009 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2011 (usadv16qi): New define_expand.
2012 (usadv8hi): New define_expand.
2014 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2017 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2019 (inline_transform): Likewise.
2020 * predict.c (counts_to_freqs): Remove useless conditional.
2021 * profile-count.h (profile_count::apply_scale): Move sanity check.
2022 * tree-inline.c (copy_bb): Always scale.
2023 (copy_cfg_body): Likewise.
2025 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2028 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2030 (*cmp_ite0): Add enabled_for_depr_it attribute.
2031 (*cmp_ite1): Likewise.
2033 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2035 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2038 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2040 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2041 return the value by pointer.
2042 (extract_range_from_multiplicative_op_1): Update accordingly.
2043 Return as soon as an operation fails.
2045 2017-11-05 Tom de Vries <tom@codesourcery.com>
2048 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2049 (DEF_SANITIZER_BUILTIN): ... here.
2050 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2051 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2053 2017-11-05 Tom de Vries <tom@codesourcery.com>
2056 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2058 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2059 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2060 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2062 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2063 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2064 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2066 2017-11-05 Tom de Vries <tom@codesourcery.com>
2069 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2072 2017-11-04 Michael Clark <michaeljclark@mac.com>
2074 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2075 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2077 (*addsi3_extended): Likewise.
2078 (*addsi3_extended2): Likewise.
2079 (<optab>si3): Likewise.
2080 (<optab>di3): Likewise.
2081 (<optab><mode>3): Likewise.
2082 (<*optabe>si3_internal): Likewise.
2083 (zero_extendqi<SUPERQI:mode>2): Likewise.
2084 (*add<mode>hi3): Likewise.
2085 (*xor<mode>hi3): Likewise.
2086 (<optab>di3): Likewise.
2087 (*<optab>si3_extend): Likewise.
2088 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2089 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2090 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2092 2017-11-04 Andrew Waterman <andrew@sifive.com>
2094 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2095 TARGET_STRICT_ALIGN based upon -mtune argument.
2097 2017-11-04 Andrew Waterman <andrew@sifive.com>
2099 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2101 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2103 config/i386/i386.c (choose_basereg): Use optional scratch
2104 register and add assertion.
2105 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2106 needed, and don't allocate stack.
2107 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2108 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2109 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2111 2017-11-03 Jeff Law <law@redhat.com>
2113 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2114 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2115 to probe at the start of a noreturn function.
2117 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2119 PR tree-optimization/78821
2120 * gimple-ssa-store-merging.c: Update the file comment.
2121 (MAX_STORE_ALIAS_CHECKS): Define.
2122 (struct store_operand_info): New type.
2123 (store_operand_info::store_operand_info): New constructor.
2124 (struct store_immediate_info): Add rhs_code and ops data members.
2125 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2126 and op1r arguments to the ctor, initialize corresponding data members.
2127 (struct merged_store_group): Add load_align_base and load_align
2129 (merged_store_group::merged_store_group): Initialize them.
2130 (merged_store_group::do_merge): Update them.
2131 (merged_store_group::apply_stores): Pick the constant for
2132 encode_tree_to_bitpos from one of the two operands, or skip
2133 encode_tree_to_bitpos if neither operand is a constant.
2134 (class pass_store_merging): Add process_store method decl. Remove
2135 bool argument from terminate_all_aliasing_chains method decl.
2136 (pass_store_merging::terminate_all_aliasing_chains): Remove
2137 var_offset_p argument and corresponding handling.
2138 (stmts_may_clobber_ref_p): New function.
2139 (compatible_load_p): New function.
2140 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2141 if there is overlap and rhs_code is not INTEGER_CST. For
2142 non-overlapping stores terminate group if rhs is not mergeable.
2143 (get_alias_type_for_stmts): Change first argument from
2144 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2145 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2146 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2148 (get_location_for_stmts): Change first argument from
2149 auto_vec<gimple *> & to vec<gimple *> &.
2150 (struct split_store): Remove orig_stmts data member, add orig_stores.
2151 (split_store::split_store): Create orig_stores rather than orig_stmts.
2152 (find_constituent_stmts): Renamed to ...
2153 (find_constituent_stores): ... this. Change second argument from
2154 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2155 to info structures rather than the statements.
2156 (split_group): Rename ALLOW_UNALIGNED argument to
2157 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2158 it. Adjust find_constituent_stores caller.
2159 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2160 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2161 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2162 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2163 (mem_valid_for_store_merging): New function.
2164 (handled_load): New function.
2165 (pass_store_merging::process_store): New method.
2166 (pass_store_merging::execute): Use process_store method. Adjust
2167 terminate_all_aliasing_chains caller.
2169 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2171 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2172 Return true for more constants, symbols and label references.
2173 (aarch64_valid_floating_const): Remove unused function.
2175 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2178 * config/aarch64/aarch64.c (aarch64_layout_frame):
2179 Undo forcing of LR at bottom of frame.
2181 2017-11-03 Jeff Law <law@redhat.com>
2184 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2185 for int_registers_saved.
2187 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2188 extracted from tree-ssa-dom.c.
2189 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2190 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2191 (record_equivalences_from_incoming_edge): Add additional argument
2192 to single_pred_edge_ignoring_loop_edges call.
2193 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2194 (uncprop_dom_walker::before_dom_children): Add additional argument
2195 to single_pred_edge_ignoring_loop_edges call.
2196 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2197 single_pred_edge_ignoring_loop_edges rather than open coding.
2198 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2200 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2202 * match.pd (-(-A)): Rewrite.
2204 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2206 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2207 (rs6000_emit_int_cmove): New declaration.
2208 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2209 (rs6000_emit_sISEL): Delete.
2210 (rs6000_emit_int_cmove): Make non-static.
2211 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2212 instead of rs6000_emit_sISEL.
2214 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2216 * asan.c (create_cond_insert_point): Maintain profile.
2217 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2219 * basic-block.h (struct basic_block_def): Remove frequency.
2220 (EDGE_FREQUENCY): Use to_frequency
2221 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2223 (find_traces): Update to use to_frequency.
2224 (find_traces_1_round): Likewise; use only IPA counts.
2225 (bb_to_key): Likewise.
2226 (connect_traces): Use IPA counts only.
2227 (copy_bb_p): Update to use to_frequency.
2228 (fix_up_crossing_landing_pad): Likewise.
2229 (sanitize_hot_paths): Likewise.
2230 * bt-load.c (basic_block_freq): Likewise.
2231 * cfg.c (init_flow): Set count_max to uninitialized.
2232 (check_bb_profile): Remove frequencies; check counts.
2233 (dump_bb_info): Do not dump frequencies.
2234 (update_bb_profile_for_threading): Update counts only.
2235 (scale_bbs_frequencies_int): Likewise.
2236 (MAX_SAFE_MULTIPLIER): Remove.
2237 (scale_bbs_frequencies_gcov_type): Update counts only.
2238 (scale_bbs_frequencies_profile_count): Update counts only.
2239 (scale_bbs_frequencies): Update counts only.
2240 * cfg.h (struct control_flow_graph): Add count-max.
2241 (update_bb_profile_for_threading): Update prototype.
2242 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2243 (find_many_sub_basic_blocks): Likewise.
2244 * cfgcleanup.c (try_forward_edges): Likewise.
2245 (try_crossjump_to_edge): Likewise.
2246 * cfgexpand.c (expand_gimple_cond): Likewise.
2247 (expand_gimple_tailcall): Likewise.
2248 (construct_init_block): Likewise.
2249 (construct_exit_block): Likewise.
2250 * cfghooks.c (verify_flow_info): Check consistency of counts.
2251 (dump_bb_for_graph): Do not dump frequencies.
2252 (split_block_1): Do not update frequencies.
2253 (split_edge): Do not update frequencies.
2254 (make_forwarder_block): Do not update frequencies.
2255 (duplicate_block): Do not update frequencies.
2256 (account_profile_record): Do not update frequencies.
2257 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2258 for global heuristics.
2259 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2260 (expected_loop_iterations_unbounded): Use counts only.
2261 * cfgloopmanip.c (scale_loop_profile): Simplify.
2262 (create_empty_loop_on_edge): Simplify
2264 (duplicate_loop_to_header_edge): Simplify
2265 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2266 (update_br_prob_note): Take care of removing note when profile
2268 (relink_block_chain): Do not dump frequency.
2269 (rtl_account_profile_record): Use to_frequency.
2270 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2271 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2272 (cgraph_update_edges_for_call_stmt_node): Likewise.
2273 (cgraph_edge::verify_count_and_frequency): Update.
2274 (cgraph_node::verify_node): Temporarily disable frequency verification.
2275 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2276 to_cgraph_frequency.
2277 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2278 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2280 (cgraph_node::expand_thunk): Update profile.
2281 * except.c (dw2_build_landing_pads): Do not update frequency.
2282 * final.c (compute_alignments): Use to_frequency.
2283 (dump_basic_block_info): Do not dump frequency.
2284 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2285 (dump_gimple_bb_header): Do not dump frequency.
2286 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2288 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2289 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2290 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2291 (init_before_recovery): Do not update frequency.
2292 (sched_create_recovery_edges): Do not update frequency.
2293 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2294 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2295 (ipa_cp_c_finalize): Set max_count to uninitialized.
2296 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2297 (param_change_prob): Use counts.
2298 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2300 * ipa-split.c (consider_split): Use to_frequency.
2301 (split_function): Use to_frequency.
2302 * ira-build.c (loop_compare_func): Likewise.
2303 (mark_loops_for_removal): Likewise.
2304 (mark_all_loops_for_removal): Likewise.
2305 * loop-doloop.c (doloop_modify): Do not update frequency.
2306 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2308 * lto-streamer-in.c (input_function): Update count_max.
2309 * omp-expand.c (expand_omp_taskreg): Update count_max.
2310 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2311 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2312 (maybe_hot_count_p): Use ipa counts only.
2313 (maybe_hot_bb_p): Simplify.
2314 (maybe_hot_edge_p): Simplify.
2315 (probably_never_executed): Do not take frequency argument.
2316 (probably_never_executed_bb_p): Do not pass frequency.
2317 (probably_never_executed_edge_p): Likewise.
2318 (combine_predictions_for_bb): Check that profile is nonzero.
2319 (propagate_freq): Do not set frequency.
2320 (drop_profile): Simplify.
2321 (counts_to_freqs): Simplify.
2322 (expensive_function_p): Use to_frequency.
2323 (propagate_unlikely_bbs_forward): Simplify.
2324 (determine_unlikely_bbs): Simplify.
2325 (estimate_bb_frequencies): Add hack to silence graphite issues.
2326 (compute_function_frequency): Use ipa counts.
2327 (pass_profile::execute): Update.
2328 (rebuild_frequencies): Use counts only.
2329 (force_edge_cold): Use counts only.
2330 * profile-count.c (profile_count::dump): Dump new count types.
2331 (profile_count::differs_from_p): Check compatiblity.
2332 (profile_count::to_frequency): New function.
2333 (profile_count::to_cgraph_frequency): New function.
2334 * profile-count.h (struct function): Declare.
2335 (enum profile_quality): Add profile_guessed_local and
2336 profile_guessed_global0.
2337 (class profile_proability): Decrease number of bits to 29;
2338 update from_reg_br_prob_note and to_reg_br_prob_note.
2339 (class profile_count: Update comment; decrease number of bits
2340 to 61. Check compatibility.
2341 (profile_count::compatible_p): New private member function.
2342 (profile_count::ipa_p): New member function.
2343 (profile_count::operator<): Handle global zero correctly.
2344 (profile_count::operator>): Handle global zero correctly.
2345 (profile_count::operator<=): Handle global zero correctly.
2346 (profile_count::operator>=): Handle global zero correctly.
2347 (profile_count::nonzero_p): New member function.
2348 (profile_count::force_nonzero): New member function.
2349 (profile_count::max): New member function.
2350 (profile_count::apply_scale): Handle IPA scalling.
2351 (profile_count::guessed_local): New member function.
2352 (profile_count::global0): New member function.
2353 (profile_count::ipa): New member function.
2354 (profile_count::to_frequency): Declare.
2355 (profile_count::to_cgraph_frequency): Declare.
2356 * profile.c (OVERLAP_BASE): Delete.
2357 (compute_frequency_overlap): Delete.
2358 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2359 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2360 * sched-ebb.c (rank): Use counts only.
2361 * shrink-wrap.c (handle_simple_exit): Use counts only.
2362 (try_shrink_wrapping): Use counts only.
2363 (place_prologue_for_one_component): Use counts only.
2364 * tracer.c (find_best_predecessor): Use to_frequency.
2365 (find_trace): Use to_frequency.
2366 (tail_duplicate): Use to_frequency.
2367 * trans-mem.c (expand_transaction): Do not update frequency.
2368 * tree-call-cdce.c: Do not update frequency.
2369 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2370 (gimple_merge_blocks): Likewise.
2371 (gimple_split_edge): Likewise.
2372 (gimple_duplicate_sese_region): Likewise.
2373 (gimple_duplicate_sese_tail): Likewise.
2374 (move_sese_region_to_fn): Likewise.
2375 (gimple_account_profile_record): Likewise.
2376 (insert_cond_bb): Likewise.
2377 * tree-complex.c (expand_complex_div_wide): Likewise.
2378 * tree-eh.c (lower_resx): Update profile.
2379 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2381 (initialize_cfun): Do not initialize frequencies
2382 (freqs_to_counts): Delete.
2383 (copy_cfg_body): Ignore count parameter.
2384 (copy_body): Update.
2385 (expand_call_inline): Update count_max.
2386 (optimize_inline_calls): Update count_max.
2387 (tree_function_versioning): Update count_max.
2388 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2389 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2391 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2392 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2393 (try_peel_loop): Likewise.
2394 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2396 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2397 (tree_transform_and_unroll_loop): Do not use frequencies
2398 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2399 Use reliable prediction only.
2400 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2401 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2402 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2403 probability scaling.
2404 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2405 not update frequency
2406 (any_remaining_duplicated_blocks): Likewise.
2407 (update_profile): Likewise.
2408 (estimated_freqs_path): Delete.
2409 (freqs_to_counts_path): Delete.
2410 (clear_counts_path): Delete.
2411 (ssa_fix_duplicate_block_edges): Likewise.
2412 (duplicate_thread_path): Likewise.
2413 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2414 * tree-tailcall.c (decrease_profile): Do not update frequency.
2415 (eliminate_tail_call): Likewise.
2416 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2417 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2418 (optimize_mask_stores): Likewise.
2419 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2420 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2421 (ubsan_expand_ptr_ifn): Update profile.
2422 * value-prof.c (gimple_ic): Simplify.
2423 * value-prof.h (gimple_ic): Update prototype.
2424 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2425 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2427 (want_inline_self_recursive_call_p): Likewise.
2428 (resolve_noninline_speculation): Only cummulate defined counts.
2429 (inline_small_functions): Use nonzero_p.
2430 (ipa_inline): Do not access freed node.
2432 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2434 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2435 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2437 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2439 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2440 non-legitimate address.
2442 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2444 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2445 (*lt0_<mode>di, *lt0_<mode>si): New.
2447 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2449 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2450 TARGET_PAIRED_FLOAT.
2452 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2453 Jim Wilson <jim.wilson@linaro.org>
2455 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2456 * config/aarch64/aarch64-tune.md: Regenerated.
2457 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2458 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2460 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2462 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2463 (arc_expand_prologue): Restore blink for millicode.
2464 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2466 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2469 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2470 gen_vec_duplicate after forcing the scalar into a register.
2472 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2474 * combine (try_combine): Print the insns input to try_combine to the
2477 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2480 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2481 Remove second argument from aarch64_process_target_attr call.
2482 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2484 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2486 (aarch64_handle_attr_arch): Remove second argument.
2487 (aarch64_handle_attr_cpu): Ditto.
2488 (aarch64_handle_attr_tune): Ditto.
2489 (aarch64_handle_attr_isa_flags): Ditto.
2490 (aarch64_process_one_target_attr): Ditto.
2491 (aarch64_process_target_attr): Ditto.
2492 (aarch64_option_valid_attribute_p): Remove second argument.
2493 on aarch64_process_target_attr call.
2495 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2497 * diagnostic.c: Include "selftest-diagnostic.h".
2498 (selftest::assert_location_text): New function.
2499 (selftest::test_diagnostic_get_location_text): New function.
2500 (selftest::diagnostic_c_tests): Call it.
2502 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2504 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2505 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2506 (class selftest::test_diagnostic_context): Move to...
2507 * selftest-diagnostic.c: New file.
2508 * selftest-diagnostic.h: New file.
2510 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2512 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2513 offset range for FT32B.
2514 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2515 * config/ft32/ft32.md: Add TARGET_NOPM.
2516 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2517 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2519 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2521 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2523 2017-11-02 Jeff Law <law@redhat.com>
2525 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2526 virtual keyword on FINAL OVERRIDE members.
2528 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2529 virtuals together. Add virtual destructor.
2530 (substitute_and_fold_engine): Similarly.
2532 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2534 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2536 2017-11-02 Richard Biener <rguenther@suse.de>
2538 PR tree-optimization/82795
2539 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2541 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2543 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2545 (gcc_GAS_CHECK_FEATURE): Remove.
2546 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2547 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2548 * configure: Regenerate.
2550 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2552 * config/arc/arc.c (hwloop_optimize): Account for empty
2555 2017-11-02 Richard Biener <rguenther@suse.de>
2558 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2559 Truncate ARRAY_REF index and element size.
2561 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2563 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2565 2017-11-01 Jeff Law <law@redhat.com>
2567 * tree-ssa-ccp.c (ccp_folder): New class derived from
2568 substitute_and_fold_engine.
2569 (ccp_folder::get_value): New member function.
2570 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2571 (ccp_fold_stmt): Remove prototype.
2572 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2573 * tree-ssa-copy.c (copy_folder): New class derived from
2574 substitute_and_fold_engine.
2575 (copy_folder::get_value): Renamed from get_value.
2576 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2577 * tree-vrp.c (vrp_folder): New class derived from
2578 substitute_and_fold_engine.
2579 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2580 (vrp_folder::get_value): New member function.
2581 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2582 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2583 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2584 provide a class interface to folder/substitute routines.
2585 (ssa_prop_fold_stmt_fn): Remove typedef.
2586 (ssa_prop_get_value_fn): Likewise.
2587 (subsitute_and_fold): Remove prototype.
2588 (replace_uses_in): Likewise.
2589 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2590 Renamed from replace_uses_in. Call the virtual member function
2591 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2592 (substitute_and_fold_dom_walker): Remove initialization of
2593 data member entries for calbacks. Add substitute_and_fold_engine
2594 member and initialize it.
2595 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2596 member functions for get_value, replace_phi_args_in c
2597 replace_uses_in, and fold_stmt calls.
2598 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2599 substitute_and_fold. Remove assert. Update ctor call.
2601 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2602 (ssa_prop_visit_phi_fn): Likewise.
2603 (class ssa_propagation_engine): New class to provide an interface
2605 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2607 (ssa_prop_visit_phi): Likewise.
2608 (ssa_propagation_engine::simulate_stmt): Moved into class.
2609 Call visit_phi/visit_stmt from the class rather than via
2610 file scoped static variables.
2611 (ssa_propagation_engine::simulate_block): Moved into class.
2612 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2613 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2614 set file scoped statics for the visit_stmt/visit_phi callbacks.
2615 * tree-complex.c (complex_propagate): New class derived from
2616 ssa_propagation_engine.
2617 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2618 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2619 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2621 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2622 ssa_propagation_engine.
2623 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2624 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2625 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2626 * tree-ssa-copy.c (copy_prop): New class derived from
2627 ssa_propagation_engine.
2628 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2629 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2630 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2631 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2632 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2633 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2634 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2636 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2638 PR rtl-optimization/82778
2639 PR rtl-optimization/82597
2640 * compare-elim.c (struct comparison): Add in_a_setter field.
2641 (find_comparison_dom_walker::before_dom_children): Remove killed
2642 bitmap and df_simulate_find_defs call, instead walk the defs.
2643 Compute last_setter and initialize in_a_setter. Merge definitions
2644 with first initialization for a few variables.
2645 (try_validate_parallel): Use insn_invalid_p instead of
2646 recog_memoized. Return insn rather than just the pattern.
2647 (try_merge_compare): Fix up comment. Don't uselessly test if
2648 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2650 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2653 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2654 Alan Hayward <alan.hayward@arm.com>
2655 David Sherwood <david.sherwood@arm.com>
2657 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2658 aarch64_hard_regno_nregs to get the number of registers
2661 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2662 Alan Hayward <alan.hayward@arm.com>
2663 David Sherwood <david.sherwood@arm.com>
2665 * config/aarch64/constraints.md (Upl): Rename to...
2667 * config/aarch64/aarch64.md
2668 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2671 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2672 Alan Hayward <alan.hayward@arm.com>
2673 David Sherwood <david.sherwood@arm.com>
2675 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2676 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2679 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2680 Alan Hayward <alan.hayward@arm.com>
2681 David Sherwood <david.sherwood@arm.com>
2683 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2684 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2685 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2687 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2688 of a single element.
2689 * config/aarch64/iterators.md: Add a comment above the permute
2690 unspecs to say that they are generated directly by
2691 aarch64_expand_vec_perm_const.
2692 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2694 2017-11-01 Nathan Sidwell <nathan@acm.org>
2696 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2698 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2700 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2701 and medany code models, and describe what they do.
2703 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2705 Revert accidental duplicate:
2707 * combine.c (can_change_dest_mode): Reject changes in
2708 REGMODE_NATURAL_SIZE.
2710 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2712 PR rtl-optimization/64682
2713 PR rtl-optimization/69567
2714 PR rtl-optimization/69737
2715 PR rtl-optimization/82683
2716 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2717 register mentioned in the note, drop the note, unless it came from I3,
2718 in which case it should go to I3 again.
2720 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2722 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2723 and return false if not.
2724 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2726 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2728 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2729 range is known to be empty.
2731 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2732 Alan Hayward <alan.hayward@arm.com>
2733 David Sherwood <david.sherwood@arm.com>
2735 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2736 and CONST_VECTOR_NUNITS instead of computing the number of units from
2737 the byte sizes of the vector and element.
2738 (simplify_binary_operation_1): Likewise.
2739 (simplify_const_binary_operation): Likewise.
2740 (simplify_ternary_operation): Likewise.
2742 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2743 Alan Hayward <alan.hayward@arm.com>
2744 David Sherwood <david.sherwood@arm.com>
2746 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2747 (int_mem_offset): ...this new function.
2748 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2749 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2750 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2753 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2754 Alan Hayward <alan.hayward@arm.com>
2755 David Sherwood <david.sherwood@arm.com>
2757 * lower-subreg.c (interesting_mode_p): New function.
2758 (compute_costs, find_decomposable_subregs, decompose_register)
2759 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2760 (resolve_clobber, dump_choices): Use it.
2762 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2763 Alan Hayward <alan.hayward@arm.com>
2764 David Sherwood <david.sherwood@arm.com>
2766 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2768 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2769 Alan Hayward <alan.hayward@arm.com>
2770 David Sherwood <david.sherwood@arm.com>
2772 * alias.c (find_base_value, find_base_term): Only process integer
2773 truncations. Check the precision rather than the size.
2775 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2776 Alan Hayward <alan.hayward@arm.com>
2777 David Sherwood <david.sherwood@arm.com>
2779 * machmode.h (is_narrower_int_mode): New function
2780 * optabs.c (expand_float, expand_fix): Use it.
2781 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2783 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2784 Alan Hayward <alan.hayward@arm.com>
2785 David Sherwood <david.sherwood@arm.com>
2787 * rtl.h (narrower_subreg_mode): New function.
2788 * ira-color.c (update_costs_from_allocno): Use it.
2790 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2791 Alan Hayward <alan.hayward@arm.com>
2792 David Sherwood <david.sherwood@arm.com>
2794 * optabs-query.h (convert_optab_p): New function, split out from...
2795 (convert_optab_handler): ...here.
2796 (widening_optab_handler): Delete.
2797 (find_widening_optab_handler): Remove permit_non_widening parameter.
2798 (find_widening_optab_handler_and_mode): Likewise. Provide an
2799 override that operates on mode class wrappers.
2800 * optabs-query.c (widening_optab_handler): Delete.
2801 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2802 parameter. Assert that the two modes are the same class and that
2803 the "from" mode is narrower than the "to" mode. Use
2804 convert_optab_handler instead of widening_optab_handler.
2805 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2806 instead of widening_optab_handler.
2807 * expr.c (expand_expr_real_2): Update calls to
2808 find_widening_optab_handler.
2809 * optabs.c (expand_widen_pattern_expr): Likewise.
2810 (expand_binop_directly): Take the insn_code as a parameter.
2811 (expand_binop): Only call find_widening_optab_handler for
2812 conversion optabs; use optab_handler otherwise. Update calls
2813 to find_widening_optab_handler and expand_binop_directly.
2814 Use convert_optab_handler instead of widening_optab_handler.
2815 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2816 find_widening_optab_handler and use scalar_mode rather than
2818 (convert_plusminus_to_widen): Likewise.
2820 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2821 Alan Hayward <alan.hayward@arm.com>
2822 David Sherwood <david.sherwood@arm.com>
2824 * machmode.h (fixed_size_mode): New class.
2825 * rtl.h (get_pool_mode): Return fixed_size_mode.
2826 * gengtype.c (main): Add fixed_size_mode.
2827 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2828 (get_raw_arg_mode): Likewise.
2829 * doc/tm.texi: Regenerate.
2830 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2831 * targhooks.c (default_get_reg_raw_mode): Likewise.
2832 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2833 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2834 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2835 (msp430_get_raw_result_mode): Likewise.
2836 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2837 * dbxout.c (dbxout_parms): Require fixed-size modes.
2838 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2839 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2840 * omp-low.c (lower_oacc_reductions): Likewise.
2841 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2842 (simplify_subreg): Update accordingly.
2843 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2844 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2845 that aren't fixed-size.
2846 (get_pool_mode): Return a fixed_size_mode.
2847 (output_constant_pool_2): Take a fixed_size_mode.
2849 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2850 Alan Hayward <alan.hayward@arm.com>
2851 David Sherwood <david.sherwood@arm.com>
2853 * doc/rtl.texi (vec_series): Document.
2854 (const): Say that the operand can be a vec_series.
2855 * rtl.def (VEC_SERIES): New rtx code.
2856 * rtl.h (const_vec_series_p_1): Declare.
2857 (const_vec_series_p): New function.
2858 * emit-rtl.h (gen_const_vec_series): Declare.
2859 (gen_vec_series): Likewise.
2860 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2861 (gen_vec_series): Likewise.
2862 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2863 * simplify-rtx.c (simplify_unary_operation): Handle negations
2865 (simplify_binary_operation_series): New function.
2866 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2867 (test_vector_ops_series): New function.
2868 (test_vector_ops): Call it.
2869 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2870 gen_const_vec_series.
2871 (altivec_lvsr): Likewise.
2872 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2874 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2875 Alan Hayward <alan.hayward@arm.com>
2876 David Sherwood <david.sherwood@arm.com>
2878 * doc/rtl.texi (const): Update description of address constants.
2879 Say that vector constants are allowed too.
2880 * common.md (E, F): Use CONSTANT_P instead of checking for
2882 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2883 checking for CONST_VECTOR.
2884 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2886 * expr.c (expand_expr_real_2): Check for vector modes instead
2887 of checking for CONST_VECTOR.
2888 * rtl.h (const_vec_p): New function.
2889 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2890 (unwrap_const_vec_duplicate): Handle them here too.
2892 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2893 David Malcolm <dmalcolm@redhat.com>
2894 Alan Hayward <alan.hayward@arm.com>
2895 David Sherwood <david.sherwood@arm.com>
2897 * rtl.h (vec_duplicate_p): New function.
2898 * selftest-rtl.c (assert_rtx_eq_at): New function.
2899 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2900 (assert_rtx_eq_at): Declare.
2901 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2902 * selftest-run-tests.c (selftest::run_tests): Call it.
2903 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2904 (simplify_unary_operation_1): Recursively handle vector duplicates.
2905 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2907 (simplify_subreg): Handle subregs of vector duplicates.
2908 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2909 (selftest::simplify_rtx_c_tests): New functions.
2911 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2912 Alan Hayward <alan.hayward@arm.com>
2913 David Sherwood <david.sherwood@arm.com>
2915 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2916 (gen_vec_duplicate): Likewise.
2917 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2919 (gen_const_vector): ...here.
2920 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2921 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2922 whose elements are all equal.
2923 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2924 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2925 (simplify_relational_operation): Likewise.
2926 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2928 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2929 (aarch64_expand_vector_init): Likewise.
2930 * config/arm/arm.c (neon_vdup_constant): Likewise.
2931 (neon_expand_vector_init): Likewise.
2932 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2933 (arm_block_set_unaligned_vect): Likewise.
2934 (arm_block_set_aligned_vect): Likewise.
2935 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2936 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2937 (expand_vec_perm_even_odd_pack): Likewise.
2938 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2939 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2940 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2941 gen_const_vec_duplicate.
2942 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2943 * config/mips/mips.c (mips_gen_const_int_vector): Use
2944 gen_const_vec_duplicate.
2945 (mips_expand_vector_init): Use CONST0_RTX.
2946 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2947 (define_split): Use gen_const_vec_duplicate.
2948 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2949 (define_split): Use gen_const_vec_duplicate.
2950 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2951 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2952 * config/spu/spu.c (spu_const): Likewise.
2954 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2955 Alan Hayward <alan.hayward@arm.com>
2956 David Sherwood <david.sherwood@arm.com>
2958 * combine.c (can_change_dest_mode): Reject changes in
2959 REGMODE_NATURAL_SIZE.
2961 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2963 * configure.ac (--enable-libssp): New.
2964 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2965 trying to determine target-specific default. Adjust indentation.
2966 * configure: Regenerated.
2967 * doc/install.texi (Configuration): Expand --disable-libssp
2970 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2972 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2975 2017-10-31 Martin Jambor <mjambor@suse.cz>
2978 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2980 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2982 * auto-profile.c (autofdo_source_profile::read): Use
2983 UNKNOWN_LOCATION rather than 0.
2984 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2985 (warning_at): ...this overload.
2986 (warning_at_rich_loc_n): Rename to...
2987 (warning_n): ...this overload.
2988 (error_at_rich_loc): Rename to...
2989 (error_at): ...this overload.
2990 (pedwarn_at_rich_loc): Rename to...
2991 (pedwarn): ...this overload.
2992 (permerror_at_rich_loc): Rename to...
2993 (permerror): ...this overload.
2994 (inform_at_rich_loc): Rename to...
2995 (inform): ...this overload.
2996 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2997 (diagnostic_n_impl_richloc): Rename to...
2998 (diagnostic_n_impl): ...this rich_location *-based decl.
2999 (inform_at_rich_loc): Rename to...
3000 (inform): ...this, and add an assertion.
3001 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3002 (warning_at_rich_loc): Rename to...
3003 (warning_at): ...this, and add an assertion.
3004 (warning_at_rich_loc_n): Rename to...
3005 (warning_n): ...this, and add an assertion.
3006 (warning_n): Update location_t-based implementation for removal of
3007 location_t-based diagnostic_n_impl.
3008 (pedwarn_at_rich_loc): Rename to...
3009 (pedwarn): ...this, and add an assertion.
3010 (permerror_at_rich_loc): Rename to...
3011 (permerror): ...this, and add an assertion.
3012 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3013 (error_at_rich_loc): Rename to...
3014 (error_at): ...this, and add an assertion.
3015 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3016 (driver::do_spec_on_infiles): Likewise.
3017 * substring-locations.c (format_warning_va): Update for renaming
3018 of inform_at_rich_loc.
3020 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3022 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3023 _Float<N>X built-in functions so that the variant without the
3024 "__builtin_" prefix is only enabled for the GNU C and Objective C
3025 languages when they are in non-strict ANSI/ISO mode.
3026 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3027 * target.def (floatn_builtin_p): Add a target hook to control
3028 whether _Float<N> and _Float<N>X built-in functions without the
3029 "__builtin_" prefix are enabled, and return true for C and
3030 Objective C in the default hook. Include langhooks.h in
3032 * targhooks.h (default_floatn_builtin_p): Likewise.
3033 * targhooks.c (default_floatn_builtin_p): Likewise.
3034 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3035 floatn_builtin_p target hook.
3036 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3038 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3039 Eric Botcazou <ebotcazou@adacore.com>
3041 PR rtl-optimization/81803
3042 * lra-constraints.c (curr_insn_transform): Also reload the whole
3043 register for a strict subreg no wider than a word if this is for
3044 a WORD_REGISTER_OPERATIONS target.
3046 2017-10-31 Jason Merrill <jason@redhat.com>
3048 * gdbinit.in: Skip over inlines from timevar.h.
3050 2017-10-31 Martin Liska <mliska@suse.cz>
3052 * doc/gcov.texi: Document new option.
3053 * gcov.c (print_usage): Likewise print it.
3054 (process_args): Support the argument.
3055 (format_count): New function.
3056 (format_gcov): Use the function.
3058 2017-10-31 Martin Liska <mliska@suse.cz>
3060 * gcov.c (struct name_map): do not use typedef.
3061 Define operator== and operator<.
3062 (name_search): Remove.
3063 (name_sort): Remove.
3064 (main): Do not allocate names.
3065 (process_file): Add vertical space.
3066 (generate_results): Use std::find.
3067 (release_structures): Do not release memory.
3068 (find_source): Use std::find.
3070 2017-10-31 Martin Liska <mliska@suse.cz>
3072 * gcov.c (struct line_info): Remove it's typedef.
3073 (line_info::line_info): Add proper ctor.
3074 (line_info::has_block): Do not use a typedef.
3075 (struct source_info): Do not use typedef.
3076 (circuit): Likewise.
3077 (get_cycles_count): Likewise.
3078 (output_intermediate_file): Iterate via vector iterator.
3079 (add_line_counts): Use std::vector methods.
3080 (accumulate_line_counts): Likewise.
3081 (output_lines): Likewise.
3083 2017-10-31 Martin Liska <mliska@suse.cz>
3085 * gcov.c (struct source_info): Remove typedef.
3086 (source_info::source_info): Add proper ctor.
3087 (accumulate_line_counts): Use struct, not it's typedef.
3088 (output_gcov_file): Likewise.
3089 (output_lines): Likewise.
3090 (main): Do not allocate an array.
3091 (output_intermediate_file): Use size of vector container.
3092 (process_file): Resize the vector.
3093 (generate_results): Do not preallocate, use newly added vector
3095 (release_structures): Do not release sources.
3096 (find_source): Use vector methods.
3097 (add_line_counts): Do not use typedef.
3099 2017-10-31 Martin Liska <mliska@suse.cz>
3101 * doc/gcov.texi: Document that.
3102 * gcov.c (add_line_counts): Mark lines with a non-executed
3104 (output_line_beginning): Handle such lines.
3105 (output_lines): Pass new argument.
3106 (output_intermediate_file): Print it in intermediate format.
3108 2017-10-31 Martin Liska <mliska@suse.cz>
3110 * color-macros.h: New file.
3111 * diagnostic-color.c: Factor out color related to macros to
3113 * doc/gcov.texi: Document -k option.
3114 * gcov.c (INCLUDE_STRING): Include string.h.
3115 (print_usage): Add -k option.
3116 (process_args): Parse it.
3117 (pad_count_string): New function.
3118 (output_line_beginning): Likewise.
3119 (DEFAULT_LINE_START): New macro.
3120 (output_lines): Support color output.
3122 2017-10-31 Martin Liska <mliska@suse.cz>
3124 PR gcov-profile/82633
3125 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3126 their interaction with GCOV infrastructure.
3127 * configure.ac: Add -fkeep-{inline,static}-functions to
3129 * configure: Regenerate.
3131 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3134 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3136 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3139 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3140 into a register if it does not fit into an immediate offset field.
3142 2017-10-31 Olivier Hainque <hainque@adacore.com>
3144 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3146 2017-10-31 Julia Koval <julia.koval@intel.com>
3148 * config.gcc: Add gfniintrin.h.
3149 * config/i386/gfniintrin.h: New.
3150 * config/i386/i386-builtin-types.def
3151 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3152 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3153 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3154 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3155 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3156 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3157 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3158 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3159 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3160 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3161 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3162 * config/i386/immintrin.h: Include gfniintrin.h.
3163 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3165 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3167 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3169 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3171 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3172 (arm_ashldi3_1bit): Remove pattern.
3173 (ashrdi3): Remove shift by 1 expansion.
3174 (arm_ashrdi3_1bit): Remove pattern.
3175 (lshrdi3): Remove shift by 1 expansion.
3176 (arm_lshrdi3_1bit): Remove pattern.
3177 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3178 cost of ashldi3 by 1.
3179 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3180 (<shift>di3_neon): Likewise.
3182 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3184 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3185 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3186 and (*aarch64_simd_mov<VQ:mode>).
3187 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3188 pattern alternative.
3189 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3190 attributes to match pattern alternative.
3192 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3194 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3195 * config/rs6000/emmintrin.h: New file.
3196 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3198 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3200 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3201 * (movdi_vfp_cortexa8): Remove pattern.
3203 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3205 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3207 (Specific, alpha*-dec-osf5.1): Remove.
3208 (Specific, mips-sgi-irix5): Remove.
3209 (Specific, mips-sgi-irix6): Remove.
3211 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3214 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3215 arguments to clear_bit_region_be.
3217 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3219 * gimplify.c: Include memmodel.h.
3221 2017-10-30 Martin Jambor <mjambor@suse.cz>
3223 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3224 condition whether loop should be executed at all.
3226 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3228 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3229 gimple folding of vec_madd() intrinsics.
3230 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3231 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3232 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3234 2017-10-30 Richard Biener <rguenther@suse.de>
3236 PR tree-optimization/82762
3238 2017-10-23 Richard Biener <rguenther@suse.de>
3240 PR tree-optimization/82129
3242 2017-08-01 Richard Biener <rguenther@suse.de>
3244 PR tree-optimization/81181
3245 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3246 (compute_antic): ... end of iteration here.
3248 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3250 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3251 -std=iso9899:2017 and -std=gnu17.
3252 * doc/standards.texi (C Language): Document C17 support.
3253 * doc/cpp.texi (Overview): Mention -std=c17.
3254 (Standard Predefined Macros): Document C11 and C17 values of
3255 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3256 * doc/extend.texi (Inline): Do not list individual options for
3257 standards newer than C99.
3258 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3260 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3263 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3265 * asan.c (asan_finish_file): Align asan globals array by shadow
3268 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3271 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3272 (struct store_immediate_info): Add bitregion_start and bitregion_end
3274 (store_immediate_info::store_immediate_info): Add brs and bre
3275 arguments and initialize bitregion_{start,end} from those.
3276 (struct merged_store_group): Add bitregion_start, bitregion_end,
3277 align_base and mask fields. Drop unnecessary struct keyword from
3278 struct store_immediate_info. Add do_merge method.
3279 (clear_bit_region_be): Use memset instead of loop storing zeros.
3280 (merged_store_group::do_merge): New method.
3281 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3282 stores as long as the surrounding bitregions have no gaps.
3283 (merged_store_group::merge_overlapping): Use do_merge.
3284 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3285 is byte aligned, rather than requiring that start and width are
3286 byte aligned. Drop unnecessary struct keyword from
3287 struct store_immediate_info. Allocate and populate also mask array.
3288 Make start of the arrays relative to bitregion_start rather than
3289 start and size them according to bitregion_{end,start} difference.
3290 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3291 struct store_immediate_info.
3292 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3293 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3294 struct keyword from struct store_immediate_info.
3295 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3296 between stores as long as the surrounding bitregions have no gaps.
3298 (struct split_store): Add orig non-static data member.
3299 (split_store::split_store): Initialize orig to false.
3300 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3301 if there is exactly a single original stmt. Change stmts argument
3302 to pointer from reference, if NULL, don't push anything to it. Add
3303 first argument, use it to optimize skipping over orig stmts that
3304 are known to be before bitpos already. Simplify.
3305 (split_group): Return unsigned int count how many stores are or
3306 would be needed rather than a bool. Add allow_unaligned argument.
3307 Change split_stores argument from reference to pointer, if NULL,
3308 only do a dry run computing how many stores would be produced.
3309 Rewritten algorithm to use both alignment and misalign if
3310 !allow_unaligned and handle bitfield stores with gaps.
3311 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3312 from bitregion_start instead of start. Compute allow_unaligned
3313 here, if true, do 2 split_group dry runs to compute which one
3314 produces fewer stores and prefer aligned if equal. Punt if
3315 new count is bigger or equal than original before emitting any
3316 statements, rather than during that. Remove no longer needed
3317 new_ssa_names tracking. Replace num_stmts with
3318 split_stores.length (). Use 32-bit stack allocated entries
3319 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3320 if possible. Handle bitfields with gaps.
3321 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3322 Compute bitregion_{start,end} for the stores and construct
3323 store_immediate_info with that. Formatting fixes.
3325 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3328 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3329 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3331 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3333 * gimplify.c: Include tm_p.h.
3335 * common.opt (gcoff): Re-add as ignored option.
3336 (gcoff1, gcoff2, gcoff3): Likewise.
3338 * Makefile.in (OBJS): Delete sdbout.o.
3339 (GTFILES): Delete $(srcdir)/sdbout.c.
3340 * debug.h: Delete sdb_debug_hooks.
3341 * final.c: Delete sdbout.h include.
3342 (final_scan_insn): Delete SDB_DEBUG check.
3343 (rest_of_clean_state): Likewise.
3344 * output.h: Delete sdb_begin_function_line.
3347 * toplev.c: Delete sdbout.h include.
3348 (process_options): Delete SDB_DEBUG check.
3349 * tree-core.h (tree_type_common): Delete pointer field of
3351 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3352 TYPE_SYMTAB_POINTER.
3353 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3354 (TYPE_SYMTAB_IS_POINTER): Delete.
3355 (TYPE_SYMTAB_IS_DIE): Renumber.
3356 * xcoffout.c: Refer to former sdbout.c file.
3357 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3359 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3360 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3361 (Debugging Options): Delete -gcoff.
3362 (-gstabs): Delete SDB reference.
3364 (-gcoff@var{level}): Delete.
3365 * doc/passes.texi (Debugging information output): Delete SDB and
3366 sdbout.c references.
3367 * doc/tm.texi: Regenerate.
3368 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3369 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3371 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3372 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3373 (SDB_DEBUGGING_INFO): Delete.
3374 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3375 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3376 * target.def (output_source_filename): Delete COFF reference.
3378 * common.opt (gcoff): Delete.
3379 (gxcoff+): Update Negative chain.
3380 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3382 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3383 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3384 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3386 (expand_function_start): Change sdb reference to past tense.
3387 (expand_function_end): Change sdb reference to past tense.
3388 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3389 * opts.c (debug_type_names): Delete coff entry.
3390 (common_handle_option): Delete OPT_gcoff case.
3391 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3393 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3394 * config/cris/cris.h: Delete SDB reference in comment.
3395 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3396 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3397 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3398 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3400 (ix86_expand_prologue): Likewise.
3401 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3402 * config/ia64/ia64.h: Likewise.
3403 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3404 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3406 * config/mmix/mmix.h: Likewise.
3407 * config/nds32/nds32.c: Likewise.
3408 * config/stormy/storym16.h: Likewise.
3409 * config/visium/visium.h: Likewise.
3410 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3412 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3414 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3415 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3416 FRAME_POINTER_REGNUM point at high end of local var area.
3418 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3420 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3421 Move comment around. Do not reset best_edge for a copiable
3422 destination if the copy would cause a partition change.
3423 (better_edge_p): Remove redundant check.
3425 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3427 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3429 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3431 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3432 for math functions that have _Float<N> and _Float<N>X variants.
3433 (mathfn_built_in_2): Add support for math functions that have
3434 _Float<N> and _Float<N>X variants.
3435 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3436 (expand_builtin_mathfn_ternary): Add support for fma with
3437 _Float<N> and _Float<N>X variants.
3438 (expand_builtin): Likewise.
3439 (fold_builtin_3): Likewise.
3440 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3441 create math function _Float<N> and _Float<N>X variants as external
3443 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3444 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3445 the __builtin_ prefix and if not strict ansi, without the prefix.
3446 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3447 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3448 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3449 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3450 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3451 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3452 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3453 function signatures for fma _Float<N> and _Float<N>X variants.
3454 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3455 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3456 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3457 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3458 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3459 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3460 * gencfn-macros.c (print_case_cfn): Add support for math functions
3461 that have _Float<N> and _Float<N>X variants.
3462 (print_define_operator_list): Likewise.
3463 (fltfn_suffixes): Likewise.
3465 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3466 for math functions that have _Float<N> and _Float<N>X variants.
3467 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3468 and _Float<N>X variants.
3469 (COPYSIGN): Likewise.
3472 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3473 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3475 (integer_valued_read_call_p): Likewise.
3476 * fold-const-call.c (fold_const_call_ss): Likewise.
3477 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3478 _Float<N> and _Float<N>X variants.
3479 (fold_const_call_ssss): Add support for fma _Float<N> and
3480 _Float<N>X variants.
3481 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3482 support for copysign and fma _Float<N> and _Float<N>X variants.
3483 (backprop::process_builtin_call_use): Likewise.
3484 * tree-call-cdce.c (can_test_argument_range); Add support for
3485 sqrt _Float<N> and _Float<N>X variants.
3486 (edom_only_function): Likewise.
3487 (get_no_error_domain): Likewise.
3488 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3489 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3490 copysign _Float<N> and _Float<N>X variants.
3491 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3492 handled by machine independent code.
3493 (FMAF128): Likewise.
3494 * doc/cpp.texi (Common Predefined Macros): Document defining
3495 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3496 fma _Float<N> and _Float<N>X variants.
3498 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3501 * config/i386/i386-modes.def (CCFPU): Remove definition.
3502 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3503 (ix86_cc_modes_compatible): Ditto.
3504 (ix86_expand_carry_flag_compare): Ditto.
3505 (ix86_expand_int_movcc): Ditto.
3506 (ix86_expand_int_addcc): Ditto.
3507 (ix86_reverse_condition): Ditto.
3508 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3509 Return true/false for unordered/ordered fp comparisons.
3510 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3511 (ix86_prepare_fp_compare_args): Update for rename.
3512 (ix86_expand_fp_compare): Update for rename. Generate unordered
3513 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3514 (ix86_expand_sse_compare_and_jump): Ditto.
3515 * config/i386/predicates.md (fcmov_comparison_operator):
3516 Remove CCFPU mode handling.
3517 (ix86_comparison_operator): Ditto.
3518 (ix86_carry_flag_operator): Ditto.
3519 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3520 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3521 (*cmpu<mode>_cc_i387): Ditto.
3522 (FPCMP): Remove mode iterator.
3523 (unord): Remove mode attribute.
3524 (unord_subst): New define_subst transformation
3525 (unord): New define_subst attribute.
3527 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3528 (*cmpi<unord>xf_i387): Ditto.
3529 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3530 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3531 using unord_subst transformation.
3532 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3533 (round_saeonly): Also handle CCFP mode.
3534 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3535 Remove UNSPEC_SAHF unspec handling.
3537 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3539 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3541 2017-10-27 Jeff Law <law@redhat.com>
3543 * gimple-ssa-sprintf.c: Include domwalk.h.
3544 (class sprintf_dom_walker): New class, derived from dom_walker.
3545 (sprintf_dom_walker::before_dom_children): New function.
3546 (struct call_info): Moved into sprintf_dom_walker class
3547 (compute_formath_length, handle_gimple_call): Likewise.
3548 (sprintf_length::execute): Call the dominator walker rather
3549 than walking the statements.
3551 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3552 gimple statement locations.
3553 (check_array_bounds): Corresponding changes. Get the statement's
3554 location directly from wi->stmt.
3556 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3559 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3561 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3563 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3564 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3566 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3569 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3570 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3571 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3572 maybe_get_pool_constant.
3573 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3576 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3578 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3579 2.26 caveat. Update gas and gld versions.
3580 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3583 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3585 * cgraph.h (set_malloc_flag): Declare.
3586 * cgraph.c (set_malloc_flag_1): New function.
3587 (set_malloc_flag): Likewise.
3588 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3589 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3591 (read_ipa_call_summary): Add support for reading is_return_callee.
3592 (write_ipa_call_summary): Stream is_return_callee.
3593 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3594 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3595 ipa-prop.h, ipa-fnsummary.h.
3596 (pure_const_names): Change to static.
3597 (malloc_state_e): Define.
3598 (malloc_state_names): Define.
3599 (funct_state_d): Add field malloc_state.
3600 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3601 (check_retval_uses): New function.
3602 (malloc_candidate_p): Likewise.
3603 (analyze_function): Add support for malloc attribute.
3604 (pure_const_write_summary): Stream malloc_state.
3605 (pure_const_read_summary): Add support for reading malloc_state.
3606 (dump_malloc_lattice): New function.
3607 (propagate_malloc): New function.
3608 (warn_function_malloc): New function.
3609 (ipa_pure_const::execute): Call propagate_malloc and
3610 ipa_free_fn_summary.
3611 (pass_local_pure_const::execute): Add support for malloc attribute.
3612 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3613 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3615 2017-10-27 Martin Liska <mliska@suse.cz>
3617 PR gcov-profile/82457
3618 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3619 for fork-like functions to be properly instrumented.
3621 2017-10-27 Richard Biener <rguenther@suse.de>
3624 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3625 info when we redirected EH.
3627 2017-10-26 Michael Collison <michael.collison@arm.com>
3629 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3631 (<optab>_trunchf<GPI:mode>2: New pattern.
3632 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3633 * config/aarch64/iterators.md (wv): New mode attribute.
3634 (vf, VF): New mode attributes.
3635 (vgp, VGP): New mode attributes.
3636 (s): Update attribute with SImode and DImode prefixes.
3638 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3640 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3641 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3642 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3643 (nios2_option_overide): Initialize it. Don't allow R0-relative
3644 addressing with PIC.
3645 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3646 (nios2_symbolic_constant_p): Likewise.
3647 (nios2_legitimate_address_p): Likewise.
3648 (nios2_r0rel_section_name_p): New.
3649 (nios2_symbol_ref_in_r0rel_data_p): New.
3650 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3651 (r0rel_constant_p): New.
3652 (nios2_print_operand_address): Handle r0rel_constant_p.
3653 (nios2_cdx_narrow_form_p): Likewise.
3654 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3655 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3656 (Nios II Options): Document -mr0rel-sec.
3658 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3660 * config/nios2/nios2.c: Include xregex.h.
3661 (nios2_gprel_sec_regex): New.
3662 (nios2_option_overide): Initialize it. Don't allow GP-relative
3663 addressing with PIC.
3664 (nios2_small_section_name_p): Check for regex match.
3665 * config/nios2/nios2.opt (mgprel-sec=): New option.
3666 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3667 (Nios II Options): Document -mgprel-sec.
3669 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3671 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3673 2017-10-26 Tom de Vries <tom@codesourcery.com>
3675 PR tree-optimization/82707
3676 * gimple.c (gimple_copy): Fix unsharing of
3677 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3679 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3681 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3682 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3683 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3684 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3685 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3686 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3687 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3688 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3689 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3690 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3691 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3692 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3693 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3694 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3695 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3696 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3697 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3699 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3701 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3703 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3704 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3705 warning to rs6000.c. Remove the Undocumented flag, since it has
3707 (-mabi=ibmlongdouble): Likewise.
3708 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3709 already set, set the default format for long double.
3710 (rs6000_debug_reg_global): Print whether long double is IBM or
3712 (rs6000_option_override_internal): Rework setting long double
3713 format. Only warn if the user is changing the long double default
3714 and they did not use -Wno-psabi.
3715 * doc/invoke.texi (PowerPC options): Update the documentation for
3716 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3718 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3719 Alan Hayward <alan.hayward@arm.com>
3720 David Sherwood <david.sherwood@arm.com>
3722 * rtl.h (wider_subreg_mode): New function.
3723 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3724 rather than an unsigned int *.
3725 * ira-color.c (regno_max_ref_width): Replace with...
3726 (regno_max_ref_mode): ...this new variable.
3727 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3728 Use wider_subreg_mode.
3729 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3730 rather than an unsigned int *.
3731 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3732 (process_alt_operands): Likewise.
3733 (invariant_p): Likewise.
3734 * lra-spills.c (assign_mem_slot): Likewise.
3735 (add_pseudo_to_slot): Likewise.
3736 * lra.c (collect_non_operand_hard_regs): Likewise.
3737 (add_regs_to_insn_regno_info): Likewise.
3738 * reload1.c (regno_max_ref_width): Replace with...
3739 (regno_max_ref_mode): ...this new variable.
3740 (reload): Update accordingly. Update call to
3741 ira_sort_regnos_for_alter_reg.
3742 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3743 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3744 (scan_paradoxical_subregs): Likewise.
3746 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3748 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3749 (aarch64_frame): Add emit_frame_chain boolean.
3750 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3751 Move eh_return case to aarch64_layout_frame.
3752 (aarch64_layout_frame): Initialize emit_frame_chain.
3753 (aarch64_expand_prologue): Use emit_frame_chain.
3755 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3757 * config/aarch64/aarch64.c (aarch64_layout_frame):
3758 Ensure LR is always stored at the bottom of the callee-saves.
3759 Remove rarely used frame layout which saves callee-saves at top of
3760 frame, so the store of LR can be used as a valid probe in all cases.
3762 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3764 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3765 Improve unaligned TImode/TFmode base/offset split.
3767 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3768 Alan Hayward <alan.hayward@arm.com>
3769 David Sherwood <david.sherwood@arm.com>
3771 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3772 * combine.c (find_single_use_1): Likewise.
3773 (expand_field_assignment): Likewise.
3774 (move_deaths): Likewise.
3775 * lra-constraints.c (simplify_operand_subreg): Likewise.
3776 (curr_insn_transform): Likewise.
3777 * lra.c (collect_non_operand_hard_regs): Likewise.
3778 (add_regs_to_insn_regno_info): Likewise.
3779 * rtlanal.c (reg_referenced_p): Likewise.
3780 (covers_regno_no_parallel_p): Likewise.
3782 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3784 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3785 Don't print any bits outside the precision of the value.
3786 * wide-int.cc (test_printing): Add some new tests.
3788 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3790 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3791 supports -xbrace_comment option.
3792 * configure: Regenerate.
3793 * config.in: Regenerate.
3794 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3795 (ASM_CPU_SPEC): Use it.
3797 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3799 * target.def (static_rtx_alignment): New hook.
3800 * targhooks.h (default_static_rtx_alignment): Declare.
3801 * targhooks.c (default_static_rtx_alignment): New function.
3802 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3803 * doc/tm.texi: Regenerate.
3804 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3805 instead of targetm.constant_alignment. Remove call to
3807 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3808 (cris_preferred_mininum_alignment): New function, split out from...
3809 (cris_constant_alignment): ...here.
3810 (cris_static_rtx_alignment): New function.
3811 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3813 (ix86_constant_alignment): ...here.
3814 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3815 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3816 (mmix_static_rtx_alignment): New function.
3817 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3818 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3820 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3823 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3824 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3826 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3828 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3829 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3830 mark as a sign-extending load.
3831 (local_pic_load_u): Define.
3833 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3836 * fold-const.c (operand_equal_for_comparison_p): Also return true
3837 if ARG0 is a simple variant of ARG1 with narrower precision.
3838 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3840 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3842 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3844 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3845 scatter_static, scatter_per_elt.
3846 * x86-tune-costs.h: Add new cost entries.
3848 2017-10-25 Richard Biener <rguenther@suse.de>
3850 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3851 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3852 class pass_fre): Move to ...
3853 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3854 class pass_fre): ... here and adjust for statistics.
3856 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3859 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3860 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3861 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3862 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3865 2017-10-25 Richard Biener <rguenther@suse.de>
3867 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3868 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3869 eliminate_push_avail, eliminate_insert): Move inside...
3870 (class eliminate_dom_walker): ... this class in preparation
3872 (fini_eliminate): Remove by merging with ...
3873 (eliminate): ... this function. Adjust for class changes.
3874 (pass_pre::execute): Remove fini_eliminate call.
3875 (pass_fre::execute): Likewise.
3877 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3880 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3881 (VPERMI2, VPERMI2I): New mode iterators.
3882 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3883 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3885 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3886 mode iterator. Remove 3 old define_insn patterns.
3887 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3888 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3889 VPERMI2 mode iterator, remove the other two expanders.
3890 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3891 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3892 instructions, remove the other two patterns.
3893 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3894 mode iterator, remove the other two patterns.
3895 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3896 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3897 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3898 and adjust argument order accordingly.
3899 (ix86_expand_vec_perm): Adjust caller.
3900 (expand_vec_perm_1): Likewise.
3901 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3902 (expand_vec_perm_vpermt2_vpshub2): ... this.
3903 (ix86_expand_vec_perm_const_1): Adjust caller.
3904 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3907 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3908 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3909 (vec_shl_<mode>): Remove unused expander.
3910 (avx512bw_<shift_insn><mode>3): New define_insn.
3911 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3912 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3914 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3917 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3920 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3922 PR rtl-optimization/82396
3923 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3924 (autopref_multipass_init): Simplify initialization.
3925 (autopref_rank_data): Simplify sort order.
3926 * gcc/sched-int.h (autopref_multipass_data_): Remove
3927 multi_mem_insn_p, min_offset and max_offset.
3929 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3932 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3933 Check special value of flag_omit_frame_pointer.
3934 (aarch64_can_eliminate): Likewise.
3935 (aarch64_override_options_after_change_1): Simplify handling of
3936 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3938 2017-10-24 Richard Biener <rguenther@suse.de>
3940 PR tree-optimization/82697
3941 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3942 zero for conditional load and unconditional store.
3944 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3946 * doc/install.texi: Document bootstrap-cet.
3948 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3951 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3952 ENDBR instruction at function entrance if function is only
3955 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3958 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3959 patterns to better describe from which operation the CF is computed.
3960 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3961 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3962 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3963 is 0, use _0 suffixed expanders instead of emitting a comparison
3966 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3968 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3969 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3971 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3974 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3975 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3976 * loop-iv.c (iv_get_reaching_def): Likewise.
3977 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3978 variable is promoted and the partition contains undefined values.
3980 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3982 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3984 (nios2_address_cost): Define.
3985 (nios2_legitimize_address): Recognize (exp + constant) directly.
3986 (TARGET_ADDRESS_COST): Define.
3988 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3990 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3991 (nios2_symbolic_memory_operand_p): Declare.
3992 (nios2_split_large_constant): Declare.
3993 (nios2_split_symbolic_memory_operand): Declare.
3994 * config/nios2/nios2.c: Adjust includes.
3995 (nios2_symbolic_constant_allowed): New.
3996 (nios2_symbolic_constant_p): New.
3997 (nios2_plus_symbolic_constant_p): New.
3998 (nios2_valid_addr_expr_p): Recognize addresses involving
4000 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4001 (nios2_symbolic_memory_operand_p): New.
4002 (nios2_large_constant_p): New.
4003 (nios2_split_large_constant): New.
4004 (nios2_split_plus_large_constant): New.
4005 (nios2_split_symbolic_memory_operand): New.
4006 (nios2_legitimize_address): Code refactoring. Handle addresses
4007 involving symbolic constants.
4008 (nios2_emit_move_sequence): Likewise.
4009 (nios2_print_operand): Improve error output.
4010 (nios2_print_operand_address): Handle LO_SUM.
4011 (nios2_cdx_narrow_form_p): Likewise.
4012 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4013 operands involving symbolic constants.
4014 (movhi_internal, movsi_internal): Likewise.
4015 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4016 (extendhisi2, extendqi<mode>2): Likewise.
4018 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4020 * tree-pass.h (PROP_rtl_split_insns): Define.
4021 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4023 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4025 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4027 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4030 * target.def (const_not_ok_for_debug_p): Default to
4031 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4032 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4033 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4034 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4035 which targetm.const_not_ok_for_debug_p returned true.
4036 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4038 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4040 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4041 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4042 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4044 (ix86_const_not_ok_for_debug_p): New function.
4045 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4046 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4048 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4051 * system.h: Conditionally include "unique-ptr.h" if
4052 INCLUDE_UNIQUE_PTR is defined.
4053 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4054 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4056 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4058 * config/rl78/rl78.md: New define_expand "subdi3".
4060 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4063 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4064 DF_REF_INSN if DF_REF_INSN_INFO is false.
4066 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4068 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4069 xmm_move instead of sse_move.
4070 (sse_store_index): New function.
4071 (ix86_register_move_cost): Be more sensible about mismatch stall;
4072 model AVX moves correctly; make difference between sse->integer and
4074 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4075 moves; make difference between SSE and AVX.
4076 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4077 and zmm_move. Increase size of sse load and store tables;
4078 add unaligned load and store tables; add ssemmx_to_integer.
4079 * x86-tune-costs.h: Update all entries according to real
4080 move latencies from Agner Fog's manual and chip documentation.
4082 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4085 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4086 * config/i386/constraints.md (Wf): New constraint.
4087 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4088 (cmp<dwi>_doubleword): Removed.
4089 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4090 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4091 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4092 expand with cmp<dwi>_doubleword. For LTU and GEU use
4093 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4095 * common.opt (gcolumn-info): Enable by default.
4096 * doc/invoke.texi (gcolumn-info): Document new default.
4098 2017-10-23 Richard Biener <rguenther@suse.de>
4100 PR tree-optimization/82672
4101 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4102 Fold the stmt if we propagated into it.
4104 2017-10-23 Richard Biener <rguenther@suse.de>
4106 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4107 (bitmap_remove_expr_from_set): ... this. All callers call this
4108 for non-constant values.
4109 (bitmap_set_subtract): Rename to...
4110 (bitmap_set_subtract_expressions): ... this. Adjust and
4112 (bitmap_set_contains_value): Remove superfluous check.
4113 (bitmap_set_replace_value): Inline into single caller ...
4114 (bitmap_value_replace_in_set): ... here and simplify.
4115 (dependent_clean): Merge into ...
4116 (clean): ... this using an overload. Adjust.
4117 (prune_clobbered_mems): Adjust.
4118 (compute_antic_aux): Likewise.
4119 (compute_partial_antic_aux): Likewise.
4121 2017-10-23 Richard Biener <rguenther@suse.de>
4123 PR tree-optimization/82129
4125 2017-08-01 Richard Biener <rguenther@suse.de>
4127 PR tree-optimization/81181
4128 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4129 (compute_antic): ... end of iteration here.
4131 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4133 * target.def (starting_frame_offset): New hook.
4134 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4135 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4136 * doc/tm.texi.in: Regenerate.
4137 * hooks.h (hook_hwi_void_0): Declare.
4138 * hooks.c (hook_hwi_void_0): New function.
4139 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4140 STARTING_FRAME_OFFSET.
4141 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4142 * reload1.c (reload): Likewise.
4143 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4144 instead of STARTING_FRAME_OFFSET.
4145 * function.c (try_fit_stack_local): Likewise.
4146 (assign_stack_local_1): Likewise
4147 (instantiate_virtual_regs): Likewise.
4148 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4149 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4150 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4151 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4152 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4153 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4154 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4155 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4156 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4157 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4158 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4159 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4160 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4161 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4162 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4163 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4164 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4165 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4166 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4167 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4168 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4169 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4170 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4171 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4172 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4173 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4174 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4175 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4176 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4177 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4178 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4179 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4180 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4181 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4182 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4183 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4184 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4185 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4186 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4187 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4188 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4189 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4190 return a HOST_WIDE_INT.
4191 (avr_builtin_setjmp_frame_value): Use it instead of
4192 STARTING_FRAME_OFFSET.
4193 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4194 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4195 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4197 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4198 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4199 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4200 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4201 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4202 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4203 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4204 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4205 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4206 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4207 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4208 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4210 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4211 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4212 * config/mips/mips.c (mips_compute_frame_info): Refer to
4213 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4214 (mips_starting_frame_offset): New function.
4215 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4216 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4217 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4218 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4219 and return a HOST_WIDE_INT.
4220 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4221 (mmix_initial_elimination_offset): Refer to
4222 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4223 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4224 * config/pa/pa.c (pa_starting_frame_offset): New function.
4225 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4226 (pa_expand_prologue): Likewise.
4227 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4228 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4229 !FRAME_GROWS_DOWNWARD handling to...
4230 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4231 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4232 !FRAME_GROWS_DOWNWARD handling to...
4233 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4234 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4235 !FRAME_GROWS_DOWNWARD handling to...
4236 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4237 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4239 (rs6000_starting_frame_offset): New function.
4240 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4241 !FRAME_GROWS_DOWNWARD handling to...
4242 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4243 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4244 !FRAME_GROWS_DOWNWARD handling to...
4245 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4246 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4247 !FRAME_GROWS_DOWNWARD handling to...
4248 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4249 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4250 (rs6000_starting_frame_offset): New function.
4251 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4252 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4253 * config/vax/vax.c (vax_starting_frame_offset): New function.
4254 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4255 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4256 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4257 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4258 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4259 * system.h (STARTING_FRAME_OFFSET): Poison.
4261 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4263 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4264 SCALAR_TYPE_MODE instead of TYPE_MODE.
4266 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4267 Alan Hayward <alan.hayward@arm.com>
4268 David Sherwood <david.sherwood@arm.com>
4270 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4272 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4273 Alan Hayward <alan.hayward@arm.com>
4274 David Sherwood <david.sherwood@arm.com>
4276 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4278 2017-10-23 Richard Biener <rguenther@suse.de>
4280 PR tree-optimization/82129
4281 * tree-ssa-pre.c (bitmap_set_and): Remove.
4282 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4283 canonicalizing expressions in the set to those with lowest
4284 ID rather than taking that from the first edge.
4286 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4288 * combine.c (rtx_equal_for_field_assignment_p): Use
4289 byte_lowpart_offset.
4291 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4292 Alan Hayward <alan.hayward@arm.com>
4293 David Sherwood <david.sherwood@arm.com>
4295 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4296 to a SUBREG_PROMOTED_VAR.
4298 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4299 Alan Hayward <alan.hayward@arm.com>
4300 David Sherwood <david.sherwood@arm.com>
4302 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4303 (expand_debug_source_expr): Likewise.
4304 * combine.c (combine_simplify_rtx): Likewise.
4305 * cse.c (fold_rtx): Likewise.
4306 * optabs.c (expand_float): Likewise.
4307 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4308 (simplify_binary_operation_1): Likewise.
4310 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4311 Alan Hayward <alan.hayward@arm.com>
4312 David Sherwood <david.sherwood@arm.com>
4314 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4315 (record_promoted_value): Likewise.
4316 * expr.c (expand_expr_real_2): Likewise.
4317 * ree.c (update_reg_equal_equiv_notes): Likewise.
4318 (combine_set_extension): Likewise.
4319 * rtlanal.c (low_bitmask_len): Likewise.
4320 * simplify-rtx.c (neg_const_int): Likewise.
4321 (simplify_binary_operation_1): Likewise.
4323 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4324 Alan Hayward <alan.hayward@arm.com>
4325 David Sherwood <david.sherwood@arm.com>
4327 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4328 * regcprop.c (maybe_mode_change): Likewise.
4329 * reload1.c (alter_reg): Likewise.
4331 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4333 * inchash.h (inchash::hash::add_wide_int): New function.
4334 * lto-streamer-out.c (hash_tree): Use it.
4336 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4338 * inchash.h (inchash::hash::add_wide_int): Rename to...
4339 (inchash::hash::add_hwi): ...this.
4340 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4341 (polymorphic_call_target_hasher::hash): Likewise.
4342 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4343 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4344 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4345 * lto-streamer-out.c (hash_tree): Likewise.
4346 * optc-save-gen.awk: Likewise.
4347 * tree.c (add_expr): Likewise.
4349 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4352 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4353 for ordered inequality comparisons even with TARGET_IEEE_FP.
4355 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4358 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4359 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4360 Expand with cmp<dwi>_doubleword.
4362 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4364 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4365 List CET intrinsics.
4366 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4367 specific to -fcf-protection option.
4369 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4371 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4372 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4373 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4374 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4375 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4376 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4377 (extra_objs): Add cet.o for Linux/x86 targets.
4378 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4379 * config/i386/cet.c: New file.
4380 * config/i386/cetintrin.h: Likewise.
4381 * config/i386/t-cet: Likewise.
4382 * config/i386/cpuid.h (bit_SHSTK): New.
4383 (bit_IBT): Likewise.
4384 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4385 pass IBT and SHSTK bits.
4386 * config/i386/i386-builtin-types.def
4387 (VOID_FTYPE_UNSIGNED_PVOID): New.
4388 (VOID_FTYPE_UINT64_PVOID): Likewise.
4389 * config/i386/i386-builtin.def: Add CET intrinsics.
4390 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4391 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4392 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4393 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4395 * config/i386/i386.c (rest_of_insert_endbranch): New.
4396 (pass_data_insert_endbranch): Likewise.
4397 (pass_insert_endbranch): Likewise.
4398 (make_pass_insert_endbranch): Likewise.
4399 (ix86_notrack_prefixed_insn_p): Likewise.
4400 (ix86_target_string): Add -mibt, -mshstk flags.
4401 (ix86_option_override_internal): Add flag_cf_protection
4403 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4404 (ix86_print_operand): Add 'notrack' prefix output.
4405 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4406 (ix86_expand_builtin): Expand CET intrinsics.
4407 (x86_output_mi_thunk): Add 'endbranch' instruction.
4408 * config/i386/i386.h (TARGET_IBT): New.
4409 (TARGET_IBT_P): Likewise.
4410 (TARGET_SHSTK): Likewise.
4411 (TARGET_SHSTK_P): Likewise.
4412 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4413 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4414 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4415 (builtin_setjmp_setup): New pattern.
4416 (builtin_longjmp): Likewise.
4417 (rdssp<mode>): Likewise.
4418 (incssp<mode>): Likewise.
4419 (saveprevssp): Likewise.
4420 (rstorssp): Likewise.
4421 (wrss<mode>): Likewise.
4422 (wruss<mode>): Likewise.
4423 (setssbsy): Likewise.
4424 (clrssbsy): Likewise.
4425 (nop_endbr): Likewise.
4426 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4428 * config/i386/immintrin.h: Include <cetintrin.h>.
4429 * config/i386/linux-common.h
4430 (file_end_indicate_exec_stack_and_cet): New prototype.
4431 (TARGET_ASM_FILE_END): New.
4433 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4435 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4436 latencies instead of having separate table; make difference between
4437 integer and float costs.
4438 * i386.h (processor_costs): Remove scalar_stmt_cost,
4439 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4440 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4442 * x86-tune-costs.h: Remove entries which has been removed in
4443 procesor_costs from all tables; make cond_taken_branch_cost
4444 and cond_not_taken_branch_cost COST_N_INSNS based.
4446 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4448 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4450 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4452 * config/i386/i386.md (isa): Remove fma_avx512f.
4453 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4454 <avx512>_fmadd_<mode>_mask3<round_name>,
4455 <avx512>_fmsub_<mode>_mask<round_name>,
4456 <avx512>_fmsub_<mode>_mask3<round_name>,
4457 <avx512>_fnmadd_<mode>_mask<round_name>,
4458 <avx512>_fnmadd_<mode>_mask3<round_name>,
4459 <avx512>_fnmsub_<mode>_mask<round_name>,
4460 <avx512>_fnmsub_<mode>_mask3<round_name>,
4461 <avx512>_fmaddsub_<mode>_mask<round_name>,
4462 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4463 <avx512>_fmsubadd_<mode>_mask<round_name>,
4464 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4465 (*vec_widen_umult_even_v16si<mask_name>,
4466 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4467 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4469 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4471 * extend.texi: Add 'nocf_check' documentation.
4472 * gimple.texi: Add second parameter to
4473 gimple_build_call_from_tree.
4474 * invoke.texi: Add -fcf-protection documentation.
4475 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4477 2017-10-20 Richard Biener <rguenther@suse.de>
4479 PR tree-optimization/82473
4480 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4481 the largest input type.
4483 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4485 * c-attribs.c (handle_nocf_check_attribute): New function.
4486 (c_common_attribute_table): Add 'nocf_check' handling.
4487 * gimple-parser.c: Add second argument NULL to
4488 gimple_build_call_from_tree.
4489 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4490 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4492 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4493 * common.opt: Add fcf-protection flag.
4494 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4495 * flag-types.h: Add enum cf_protection_level.
4496 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4497 Add 'nocf_check' attribute propagation to gimple call.
4498 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4499 (gimple_build_call_from_tree): Update prototype.
4500 (gimple_call_nocf_check_p): New function.
4501 (gimple_call_set_nocf_check): Likewise.
4502 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4503 * ipa-icf.c: Add nocf_check attribute in statement hash.
4504 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4505 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4506 * toplev.c (process_options): Add flag_cf_protection handling.
4508 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4510 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4512 2017-10-20 Richard Biener <rguenther@suse.de>
4514 PR tree-optimization/82603
4515 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4516 remove false predicated stores.
4518 2017-10-20 Richard Biener <rguenther@suse.de>
4520 * graphite-isl-ast-to-gimple.c
4521 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4522 Remove return value and simplify, dump copied stmt after lhs
4524 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4525 Reduce dump verbosity.
4526 (gsi_insert_earliest): Likewise.
4527 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4528 * graphite.c (print_global_statistics): Adjust dumping.
4529 (print_graphite_scop_statistics): Likewise.
4530 (print_graphite_statistics): Do not dump loops here.
4531 (graphite_transform_loops): But here.
4533 2017-10-20 Nicolas Roche <roche@adacore.com>
4535 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4536 * configure: Regenerate.
4538 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4541 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4542 functions when optimizing replace GIMPLE_RETURN stmts with
4543 calls to __builtin_unreachable ().
4546 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4547 for -fsanitize=thread link of executables.
4548 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4549 link of executables.
4552 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4554 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4555 define_insns for logical vector shifts to use VI248_AVX512BW
4556 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4557 condition, useless isa and prefix attributes. Change the first
4558 2 of these define_insns to ...
4559 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4560 define_insn for avx512vl.
4561 (<shift_insn><mode>3): ... and this, new define_insn without
4562 masking for non-avx512vl.
4565 * config/i386/sse.md (*andnot<mode>3,
4566 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4567 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4568 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4569 not applied use empty suffix even for TARGET_AVX512VL.
4570 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4571 is applied, supply evex,evex or evex,evex,evex instead of just
4574 2017-10-20 Julia Koval <julia.koval@intel.com>
4576 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4577 (OPTION_MASK_ISA_GFNI_UNSET): New.
4578 (ix86_handle_option): Handle OPT_mgfni.
4579 * config/i386/cpuid.h (bit_GFNI): New.
4580 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4581 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4582 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4583 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4584 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4585 * config/i386/i386.opt: Add mgfni.
4587 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4589 * config/msp430/msp430.c (msp430_option_override): Disable
4590 -fdelete-null-pointer-checks.
4591 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4593 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4595 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4596 of x87 and SSE instructions.
4598 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4600 * asan.c (create_cond_insert_point): Do not update edge count.
4601 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4602 (afdo_propagate_circuit): Likewise.
4603 (afdo_calculate_branch_prob): Likewise.
4604 (afdo_annotate_cfg): Likewise.
4605 * basic-block.h (struct edge_def): Remove count.
4606 (edge_def::count): New accessor.
4607 * bb-reorder.c (rotate_loop): Update.
4608 (find_traces_1_round): Update.
4609 (connect_traces): Update.
4610 (sanitize_hot_paths): Update.
4611 * cfg.c (unchecked_make_edge): Update.
4612 (make_single_succ_edge): Update.
4613 (check_bb_profile): Update.
4614 (dump_edge_info): Update.
4615 (update_bb_profile_for_threading): Update.
4616 (scale_bbs_frequencies_int): Update.
4617 (scale_bbs_frequencies_gcov_type): Update.
4618 (scale_bbs_frequencies_profile_count): Update.
4619 (scale_bbs_frequencies): Update.
4620 * cfganal.c (connect_infinite_loops_to_exit): Update.
4621 * cfgbuild.c (compute_outgoing_frequencies): Update.
4622 (find_many_sub_basic_blocks): Update.
4623 * cfgcleanup.c (try_forward_edges): Update.
4624 (try_crossjump_to_edge): Update
4625 * cfgexpand.c (expand_gimple_cond): Update
4626 (expand_gimple_tailcall): Update
4627 (construct_exit_block): Update
4628 * cfghooks.c (verify_flow_info): Update
4629 (redirect_edge_succ_nodup): Update
4630 (split_edge): Update
4631 (make_forwarder_block): Update
4632 (duplicate_block): Update
4633 (account_profile_record): Update
4634 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4635 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4636 * cfgloopmanip.c (scale_loop_profile): Update.
4638 (lv_adjust_loop_entry_edge): Update.
4639 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4640 (force_nonfallthru_and_redirect): Update.
4641 (purge_dead_edges): Update.
4642 (rtl_flow_call_edges_add): Update.
4643 * cgraphunit.c (init_lowered_empty_function): Update.
4644 (cgraph_node::expand_thunk): Update.
4645 * gimple-pretty-print.c (dump_probability): Update.
4646 (dump_edge_probability): Update.
4647 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4648 * haifa-sched.c (sched_create_recovery_edges): Update.
4649 * hsa-gen.c (convert_switch_statements): Update.
4650 * ifcvt.c (dead_or_predicable): Update.
4651 * ipa-inline-transform.c (inline_transform): Update.
4652 * ipa-split.c (split_function): Update.
4653 * ipa-utils.c (ipa_merge_profiles): Update.
4654 * loop-doloop.c (add_test): Update.
4655 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4656 * lto-streamer-in.c (input_cfg): Update.
4657 (input_function): Update.
4658 * lto-streamer-out.c (output_cfg): Update.
4659 * modulo-sched.c (sms_schedule): Update.
4660 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4661 * predict.c (maybe_hot_edge_p): Update.
4662 (unlikely_executed_edge_p): Update.
4663 (probably_never_executed_edge_p): Update.
4664 (dump_prediction): Update.
4665 (drop_profile): Update.
4666 (propagate_unlikely_bbs_forward): Update.
4667 (determine_unlikely_bbs): Update.
4668 (force_edge_cold): Update.
4669 * profile.c (compute_branch_probabilities): Update.
4670 * reg-stack.c (better_edge): Update.
4671 * shrink-wrap.c (handle_simple_exit): Update.
4672 * tracer.c (better_p): Update.
4673 * trans-mem.c (expand_transaction): Update.
4674 (split_bb_make_tm_edge): Update.
4675 * tree-call-cdce.c: Update.
4676 * tree-cfg.c (gimple_find_sub_bbs): Update.
4677 (gimple_split_edge): Update.
4678 (gimple_duplicate_sese_region): Update.
4679 (gimple_duplicate_sese_tail): Update.
4680 (gimple_flow_call_edges_add): Update.
4681 (insert_cond_bb): Update.
4682 (execute_fixup_cfg): Update.
4683 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4684 * tree-complex.c (expand_complex_div_wide): Update.
4685 * tree-eh.c (lower_resx): Update.
4686 (unsplit_eh): Update.
4687 (cleanup_empty_eh_move_lp): Update.
4688 * tree-inline.c (copy_edges_for_bb): Update.
4689 (freqs_to_counts): Update.
4690 (copy_cfg_body): Update.
4691 * tree-ssa-dce.c (remove_dead_stmt): Update.
4692 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4693 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4694 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4695 (unloop_loops): Update.
4696 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4697 * tree-ssa-loop-split.c (connect_loops): Update.
4698 (split_loop): Update.
4699 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4700 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4701 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4702 * tree-ssa-reassoc.c (branch_fixup): Update.
4703 * tree-ssa-tail-merge.c (replace_block_by): Update.
4704 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4705 (compute_path_counts): Update.
4706 (update_profile): Update.
4707 (recompute_probabilities): Update.
4708 (update_joiner_offpath_counts): Update.
4709 (estimated_freqs_path): Update.
4710 (freqs_to_counts_path): Update.
4711 (clear_counts_path): Update.
4712 (ssa_fix_duplicate_block_edges): Update.
4713 (duplicate_thread_path): Update.
4714 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4715 (case_bit_test_cmp): Update.
4716 (collect_switch_conv_info): Update.
4717 (gen_inbound_check): Update.
4718 (do_jump_if_equal): Update.
4719 (emit_cmp_and_jump_insns): Update.
4720 * tree-tailcall.c (decrease_profile): Update.
4721 (eliminate_tail_call): Update.
4722 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4723 (vect_do_peeling): Update.
4724 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4725 * ubsan.c (ubsan_expand_null_ifn): Update.
4726 (ubsan_expand_ptr_ifn): Update.
4727 * value-prof.c (gimple_divmod_fixed_value): Update.
4728 (gimple_mod_pow2): Update.
4729 (gimple_mod_subtract): Update.
4730 (gimple_ic): Update.
4731 (gimple_stringop_fixed_value): Update.
4733 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4736 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4738 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4740 PR rtl-optimization/82395
4741 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4742 based on non_spilled_static_chain_regno_p.
4744 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4746 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4747 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4748 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4750 2017-10-19 Martin Sebor <msebor@redhat.com>
4752 PR tree-optimization/82596
4753 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4755 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4757 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4758 (get_mem_refs_of_builtin_call): Likewise.
4759 * builtins.c (expand_builtin_apply): Adjust call to
4760 allocate_dynamic_stack_space.
4761 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4762 the third argument to allocate_dynamic_stack_space, otherwise -1.
4763 (expand_builtin): Deal with all alloca variants.
4764 (is_inexpensive_builtin): Likewise.
4765 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4766 * calls.c (special_function_p): Deal with all alloca variants.
4767 (initialize_argument_information): Adjust call to
4768 allocate_dynamic_stack_space.
4769 (expand_call): Likewise.
4770 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4771 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4772 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4773 use it for the stack usage computation.
4774 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4775 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4776 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4777 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4778 (in_loop_p): Remove first argument and useless check.
4779 (pass_walloca::execute): Remove useless test and adjust call to above.
4780 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4781 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4782 (gimplify_call_expr): Deal with all alloca variants.
4783 * hsa-gen.c (gen_hsa_alloca): Likewise.
4784 (gen_hsa_insns_for_call): Likewise.
4785 * ipa-pure-const.c (special_builtin_state): Likewise.
4786 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4787 * tree-object-size.c (alloc_object_size): Likewise.
4788 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4789 (call_may_clobber_ref_p_1): Likewise.
4790 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4791 (ccp_fold_stmt): Likewise.
4792 (optimize_stack_restore): Likewise.
4793 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4794 (mark_all_reaching_defs_necessary_1): Likewise.
4795 (propagate_necessity): Likewise.
4796 (eliminate_unnecessary_stmts): Likewise.
4797 * tree.c (build_common_builtin_nodes): Build
4798 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4799 (build_alloca_call_expr): New function.
4800 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4801 (CASE_BUILT_IN_ALLOCA): Likewise.
4802 (build_alloca_call_expr): Declare.
4803 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4805 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4808 * dwarf2out.c (new_die_raw): New static inline function.
4809 (new_die): Use it to create the DIE.
4810 (add_AT_external_die_ref): Likewise.
4811 (clone_die): Likewise.
4812 (clone_as_declaration): Likewise.
4813 (dwarf2out_vms_debug_main_pointer): Likewise.
4814 (base_type_die): Likewise. Remove early return for corner cases.
4815 Do not call add_pubtype on the DIE here.
4816 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4817 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4818 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4819 native order exists for base types, attach the DIE manually and call
4820 add_pubtype on it. Do not equate a reverse order DIE to the type.
4822 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4824 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4825 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4827 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4829 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4831 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4832 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4833 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4834 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4835 * opts.c (sanitizer_opts): Add builtin.
4836 * ubsan.c (instrument_builtin): New function.
4837 (pass_ubsan::execute): Call it.
4838 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4839 * doc/invoke.texi: Document -fsanitize=builtin.
4841 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4842 builtins, store max (log2 (align), 0) into uchar field instead of
4843 align into uptr field.
4844 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4845 store uchar 0 field instead of uptr 0 field.
4846 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4847 instead of passing one address of struct with 2 locations pass
4848 two addresses of structs with 1 location each.
4849 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4850 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4851 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4852 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4853 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4854 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4855 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4856 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4858 2017-10-19 Martin Liska <mliska@suse.cz>
4861 * file-find.c (remove_prefix): Remove.
4862 * file-find.h (remove_prefix): Likewise.
4863 * gcc-ar.c: Remove smartness of lookup.
4865 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4867 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4868 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4869 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4871 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4874 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4875 (setcc + and to xor + setcc): New peephole2.
4877 2017-10-19 Tom de Vries <tom@codesourcery.com>
4879 * doc/sourcebuild.texi (Test Directives, Variants of
4880 dg-require-support): Add dg-require-stack-size.
4882 2017-10-19 Martin Liska <mliska@suse.cz>
4885 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4886 that have a large alignment.
4887 (gimplify_target_expr): Likewise.
4889 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4891 PR rtl-optimization/82602
4892 * ira.c (rtx_moveable_p): Return false for volatile asm.
4894 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4897 * config/i386/i386-modes.def (CCGZ): New CC mode.
4898 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4899 * config/i386/predicates.md (ix86_comparison_operator):
4901 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4902 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4903 with double-word subtraction.
4904 (put_condition_code): Handle CCGZmode.
4906 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4908 * wide-int.cc (debug (const wide_int &)): New.
4909 (debug (const wide_int *)): New.
4910 (debug (const widest_int &)): New.
4911 (debug (const widest_int *)): New.
4913 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4916 * lra-constraints.c (curr_insn_transform): Use non-input operand
4917 instead of output one for matched reload.
4919 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4921 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4922 (tree-ssa-loop-ivopts.h): New header file.
4923 (struct builtin_info): New fields.
4924 (classify_builtin_1): Compute and record base and offset parts for
4925 memset builtin partition by calling strip_offset.
4926 (offset_cmp, fuse_memset_builtins): New functions.
4927 (finalize_partitions): Fuse adjacent memset partitions by calling
4929 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4930 Expose the interface.
4931 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4933 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4935 PR tree-optimization/82574
4936 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4937 that data reference must be executed exactly once per iteration
4938 against the outermost loop in nest.
4939 (classify_partition): Update call to above function.
4941 2017-10-18 Richard Biener <rguenther@suse.de>
4943 PR tree-optimization/82591
4944 * graphite.c (graphite_transform_loops): Move code gen message
4946 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4947 Here. Handle scop_to_isl_ast failing.
4948 (scop_to_isl_ast): Limit the number of ISL operations.
4950 2017-10-18 Richard Biener <rguenther@suse.de>
4952 * graphite-isl-ast-to-gimple.c
4953 (translate_isl_ast_to_gimple::set_rename): Simplify.
4954 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4955 (graphite_copy_stmts_from_block): ... here.
4956 (copy_bb_and_scalar_dependences): Simplify.
4957 (add_parameters_to_ivs_params): Canonicalize.
4958 (generate_entry_out_of_ssa_copies): Simplify.
4959 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4960 by passing in ISL dimension.
4961 (parameter_index_in_region_1): Rename to ...
4962 (parameter_index_in_region): ... this.
4963 (extract_affine): Adjust assert, pass down parameter index.
4964 (add_param_constraints): Use range-info when available.
4965 (build_scop_context): Adjust.
4966 * sese.c (new_sese_info): Adjust.
4967 (free_sese_info): Likewise.
4968 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4969 Remove unused typedefs.
4970 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4972 2017-10-18 Martin Liska <mliska@suse.cz>
4974 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4976 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4978 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4979 (s390_sched_init): Do not reset s390_sched_state if we entered the
4980 current basic block via a fallthru edge and all others are unlikely.
4982 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4984 * config/s390/s390.c (NUM_SIDES): New variable.
4985 (LONGRUNNING_THRESHOLD): New variable.
4986 (LATENCY_FACTOR): New variable.
4987 (s390_sched_score): Decrease score for long-running instructions on
4989 (s390_sched_variable_issue): Perform bookkeeping for long-running
4992 2017-10-18 Richard Biener <rguenther@suse.de>
4994 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4995 Simplify with removal of the parameter rename map.
4996 (set_rename): Likewise.
4997 (should_copy_to_new_region): Likewise.
4998 (graphite_copy_stmts_from_block): Likewise.
4999 (copy_bb_and_scalar_dependences): Remove initialization of
5000 unused copied_bb_map.
5002 (copy_internal_parameters): Likewise.
5003 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5004 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5005 Use INTEGRAL_TYPE_P.
5006 (parameter_index_in_region_1): Rename to ...
5007 (assign_parameter_index_in_region): ... this. Assert we have
5008 a parameter we handle.
5009 (scan_tree_for_params): Adjust.
5010 * sese.h (parameter_rename_map_t): Remove.
5011 (struct sese_info_t): Remove unused parameter_rename_map and
5012 copied_bb_map members.
5013 * sese.c (new_sese_info): Adjust.
5014 (free_sese_info): Likewise.
5016 2017-10-18 Martin Liska <mliska@suse.cz>
5019 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5020 on an abnormal edge.
5022 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5024 * doc/invoke.texi (ffunction-sections and fdata-sections):
5027 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5029 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5030 the use statement can throw internally.
5032 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5034 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5035 any RTX present on the RHS of a SET.
5036 * compare-elim.c (try_eliminate_compare): Restore comment.
5038 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5040 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5041 may be also called on tcc_type nodes.
5042 * langhooks.c (lhd_tree_size): Likewise.
5044 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5046 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5047 format_warning_at_substring.
5048 (maybe_warn): Convert source_range * param to a location_t. Pass
5049 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5050 (format_directive): Remove code to extract source_ranges and
5051 source_range * in favor of just a location_t.
5052 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5054 * substring-locations.c (format_warning_va): Convert
5055 source_range * param to a location_t.
5056 (format_warning_at_substring): Likewise.
5057 * substring-locations.h (format_warning_va): Likewise.
5058 (format_warning_at_substring): Likewise.
5060 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5062 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5064 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5065 and scatter/gather ops.
5067 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5068 vec_gather_load and vec_scatter_store.
5069 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5070 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5071 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5072 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5073 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5074 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5076 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5078 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5079 Detect FTST instruction and handle its register pops. Only pop
5080 second operand if can_pop_second_op is true.
5081 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5082 set can_pop_second_op to false in the compare_for_stack_reg call.
5084 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5085 output_fp_compare for stack register operands.
5086 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5087 instructions here. Do not emit stack register pops here. Assert
5088 that FCOMPP pops next to top stack register. Rewrite function.
5090 2017-10-17 Nathan Sidwell <nathan@acm.org>
5093 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5094 use DECL_ASSEMBLER_NAME_RAW.
5097 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5100 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5101 Wilco Dijkstra <wilco.dijkstra@arm.com>
5103 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5104 converstion to Pmode from the buf_addr.
5106 2017-10-17 Richard Biener <rguenther@suse.de>
5108 * graphite-dependences.c (scop_get_reads_and_writes): Change
5109 output parameters to references.
5111 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5113 PR 71026/tree-optimization
5114 * fold-const.c (distribute_real_division): Removed.
5115 (fold_binary_loc): Remove calls to distribute_real_divison.
5117 2017-10-17 Richard Biener <rguenther@suse.de>
5119 * graphite-scop-detection.c
5120 (scop_detection::stmt_has_simple_data_refs_p): Always use
5121 the full nest as region.
5122 (try_generate_gimple_bb): Likewise.
5123 * sese.c (scalar_evolution_in_region): Simplify now that
5124 SCEV can handle instantiation in regions.
5125 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5126 in the non-loop part of a function if requested.
5128 2017-10-17 Richard Biener <rguenther@suse.de>
5130 PR tree-optimization/82563
5131 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5133 (graphite_regenerate_ast_isl): Call it.
5134 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5136 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5138 PR tree-optimization/82549
5139 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5140 Formatting fixes. Instead of calling make_bit_field_ref with negative
5143 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5145 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5146 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5147 _mm_maskz_reduce_ss): New.
5148 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5149 __builtin_ia32_reducess_mask): Ditto..
5150 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5151 * config/i386/sse.md (reduces<mode>): Renamed to ...
5152 (reduces<mode><mask_scalar_name>): ... this.
5153 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5155 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5156 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5158 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5160 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5161 * selftest-run-tests.c (selftest::run_tests): Call
5162 selftest::unique_ptr_tests_cc_tests.
5163 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5164 * unique-ptr-tests.cc: New file.
5166 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5169 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5171 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5172 (make_hard_regno_born, make_hard_regno_dead): Update
5173 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5175 2017-10-16 Jeff Law <law@redhat.com>
5177 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5179 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5181 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5183 2017-10-16 Olivier Hainque <hainque@adacore.com>
5185 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5186 with_cpu if we were configured for an e500v2 target cpu name.
5188 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5190 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5191 * doc/invoke.texi: Document +nodsp as a valid extension for
5194 2017-10-16 Martin Liska <mliska@suse.cz>
5196 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5197 (test_set_range): Likewise.
5198 (test_range_functions): Rename to ...
5199 (test_bit_in_range): ... this.
5200 (sbitmap_c_tests): Add new test.
5202 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5204 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5206 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5207 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5209 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5211 * config/aarch64/aarch64-builtins.c
5212 (aarch64_types_quadopu_lane_qualifiers): New.
5213 (TYPES_QUADOPU_LANE): New.
5214 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5215 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5216 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5217 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5218 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5219 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5220 (Vdottype, DOTPROD): New.
5221 (sur): Add SDOT and UDOT.
5223 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5225 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5226 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5227 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5229 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5230 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5231 Enable TARGET_DOTPROD.
5232 (cortex-a75.cortex-a55): Likewise.
5233 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5235 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5237 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5238 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5240 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5242 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5243 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5244 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5245 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5246 * config/arm/types.md (neon_dot, neon_dot_q): New.
5247 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5249 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5251 * config/arm/arm.h (TARGET_DOTPROD): New.
5252 * config/arm/arm.c (arm_arch_dotprod): New.
5253 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5254 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5255 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5256 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5257 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5258 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5259 * doc/invoke.texi (armv8.2-a): Document dotprod
5261 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5263 * i386.c (ix86_vec_cost): New function.
5264 (ix86_rtx_costs): Handle vector operations better.
5265 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5266 * x86-tune-costs.h: Add new costs to all tables.
5268 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5270 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5272 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5273 divsd, sqrtss and sqrtsd
5274 * x86-tune-costs.h: Add new entries to all costs.
5275 (znver1_cost): Fix to match real instruction latencies.
5277 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5278 Michael Collison <michael.collison@arm.com>
5280 * compare-elim.c: Include emit-rtl.h.
5281 (can_merge_compare_into_arith): New function.
5282 (try_validate_parallel): Likewise.
5283 (try_merge_compare): Likewise.
5284 (try_eliminate_compare): Call the above when no previous clobber
5286 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5289 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5293 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5294 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5298 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5299 to be any operand_equal_p operands. For & (B - 1) require
5300 B to be power of 2. Recognize
5301 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5303 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5306 * optabs.c (expand_memory_blockage): Fix call of
5307 targetm.have_memory_blockage.
5309 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5312 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5313 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5314 objects to extra_objs instead of overwriting it.
5316 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5318 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5319 Use any_fp_register_operand as operand[3] predicate. Simplify
5320 equality test for operands[2] and operands[4] memory location.
5321 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5322 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5323 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5324 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5325 any_fp_register_operand as operand[1] predicate. Simplify
5326 equality test for operands[0] and operands[3] memory location.
5327 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5328 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5329 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5331 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5333 * target-insns.def: Add memory_blockage.
5334 * optabs.c (expand_memory_blockage): New function.
5335 (expand_asm_memory_barrier): Rename ...
5336 (expand_asm_memory_blockage): ... to this.
5337 (expand_mem_thread_fence): Call expand_memory_blockage
5338 instead of expand_asm_memory_barrier.
5339 (expand_mem_singnal_fence): Ditto.
5340 (expand_atomic_load): Ditto.
5341 (expand_atomic_store): Ditto.
5342 * doc/md.texi (Standard Pattern Names For Generation):
5343 Document memory_blockage instruction pattern.
5345 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5347 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5348 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5349 * config/rl78/rl78.md: New define_expand "adddi3".
5351 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5353 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5356 2017-10-13 Jeff Law <law@redhat.com>
5358 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5360 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5363 * internal-fn.c (expand_mul_overflow): If both operands have
5364 the same highpart of -1 or 0 and the topmost bit of lowpart
5365 is different, overflow is if res <= 0 rather than res < 0.
5367 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5369 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5370 TARGET_P9_VECTOR code for unaligned_load case.
5372 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5374 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5376 2017-10-13 Nathan Sidwell <nathan@acm.org>
5378 * tree-core.h (tree_contains_struct): Make bool.
5379 * tree.c (tree_contains_struct): Likewise.
5380 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5381 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5382 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5383 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5384 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5386 2017-10-13 Richard Biener <rguenther@suse.de>
5388 * graphite-isl-ast-to-gimple.c
5389 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5390 parameters and dominance check.
5391 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5392 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5393 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5394 Do not update SSA form here or do intermediate IL verification.
5395 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5396 (graphite_initialize): Remove check on the number of loops in
5397 the function and inline into graphite_transform_loops.
5398 (graphite_finalize): Inline into graphite_transform_loops.
5399 (graphite_transform_loops): Perform SSA update and IL verification
5401 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5403 2017-10-13 Richard Biener <rguenther@suse.de>
5405 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5406 graphite_expression_type_precision): Avoid global constructor
5408 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5409 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5410 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5411 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5412 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5413 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5415 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5418 * config/i386/i386.h (ix86_red_zone_size): New.
5419 * config/i386/i386.md (push peephole2s): Replace
5420 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5422 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5423 Alan Hayward <alan.hayward@arm.com>
5424 David Sherwood <david.sherwood@arm.com>
5426 * combine.c (can_change_dest_mode): Reject changes in
5427 REGMODE_NATURAL_SIZE.
5429 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5430 Alan Hayward <alan.hayward@arm.com>
5431 David Sherwood <david.sherwood@arm.com>
5433 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5434 (expand_debug_source_expr): Likewise.
5435 * combine.c (combine_simplify_rtx): Likewise.
5436 * cse.c (fold_rtx): Likewise.
5437 * fwprop.c (canonicalize_address): Likewise.
5438 * targhooks.c (default_shift_truncation_mask): Likewise.
5440 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5441 Alan Hayward <alan.hayward@arm.com>
5442 David Sherwood <david.sherwood@arm.com>
5444 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5445 (widened_mode): Likewise.
5446 (expand_unop): Likewise.
5447 * ree.c (transform_ifelse): Likewise.
5448 (merge_def_and_ext): Likewise.
5449 (combine_reaching_defs): Likewise.
5450 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5452 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5453 Alan Hayward <alan.hayward@arm.com>
5454 David Sherwood <david.sherwood@arm.com>
5456 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5457 * combine.c (gen_lowpart_for_combine): Likewise.
5458 * dwarf2out.c (rtl_for_decl_location): Likewise.
5459 * final.c (alter_subreg): Likewise.
5460 * rtlhooks.c (gen_lowpart_general): Likewise.
5461 (gen_lowpart_if_possible): Likewise.
5463 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5464 Alan Hayward <alan.hayward@arm.com>
5465 David Sherwood <david.sherwood@arm.com>
5467 * calls.c (expand_call): Use subreg_lowpart_offset.
5468 * cse.c (cse_insn): Likewise.
5469 * regcprop.c (copy_value): Likewise.
5470 (copyprop_hardreg_forward_1): Likewise.
5472 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5475 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5476 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5477 =Q constraints to +Q and into insn condition add check
5478 that operands[0] and operands[1] are equal.
5479 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5480 =Q constraints to +Q and into insn condition add check
5481 that operands[0] is equal to either operands[1] or operands[2].
5484 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5485 instead of handling MINUS_EXPR twice (once for each argument),
5486 canonicalize operand order and handle just once, use rtype where
5487 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5490 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5491 any values of __C while still being pattern recognizable as a simple
5494 2017-10-13 Richard Biener <rguenther@suse.de>
5496 PR tree-optimization/82451
5498 2017-10-02 Richard Biener <rguenther@suse.de>
5500 PR tree-optimization/82355
5501 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5502 a mapping for the enclosing loop but avoid generating one for
5504 (copy_bb_and_scalar_dependences): Remove premature codegen
5505 error on PHIs in blocks duplicated into multiple places.
5506 * graphite-scop-detection.c
5507 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5508 in the region use it as loop and nest to analyze the DR in.
5509 (try_generate_gimple_bb): Likewise.
5510 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5511 (add_loop_constraints): For blocks in a loop not in the region
5512 create a dimension with a single iteration.
5513 * sese.h (gbb_loop_at_index): Remove assert.
5515 * cfgloop.c (loop_preheader_edge): For the loop tree root
5516 return the single successor of the entry block.
5517 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5518 Reset the SCEV hashtable and niters.
5519 * graphite-scop-detection.c
5520 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5521 assert that we only have POLYNOMIAL_CHREC that vary in loops
5522 contained in the region.
5523 (scop_detection::graphite_can_represent_expr): Adjust.
5524 (scop_detection::stmt_has_simple_data_refs_p): For loops
5525 not in the region set loop to NULL. The nest is now the
5526 entry edge to the region.
5527 (try_generate_gimple_bb): Likewise.
5528 * sese.c (scalar_evolution_in_region): Adjust for
5529 instantiate_scev change.
5530 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5531 Make nest parameter the edge into the region.
5532 (create_data_ref): Likewise.
5533 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5534 entry edge into a region and adjust instantiate_scev calls.
5535 (create_data_ref): Likewise.
5536 (graphite_find_data_references_in_stmt): Likewise.
5537 (find_data_references_in_stmt): Pass the loop preheader edge
5538 from the nest argument.
5539 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5540 parameter the edge into the region.
5541 (instantiate_parameters): Use the loop preheader edge as entry.
5542 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5544 (get_instantiated_value_entry): Make instantiate_below parameter
5545 the edge into the region.
5546 (instantiate_scev_name): Likewise. Adjust dominance checks,
5547 when we cannot use loop-based instantiation instantiate by
5548 walking use-def chains.
5549 (instantiate_scev_poly): Adjust.
5550 (instantiate_scev_binary): Likewise.
5551 (instantiate_scev_convert): Likewise.
5552 (instantiate_scev_not): Likewise.
5553 (instantiate_array_ref): Remove.
5554 (instantiate_scev_3): Likewise.
5555 (instantiate_scev_2): Likewise.
5556 (instantiate_scev_1): Likewise.
5557 (instantiate_scev_r): Do not blindly handle N-operand trees.
5558 Do not instantiate array-refs. Handle all constants and invariants.
5559 (instantiate_scev): Make instantiate_below parameter
5560 the edge into the region.
5561 (resolve_mixers): Use the loop preheader edge for the region
5562 parameter to instantiate_scev_r.
5563 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5565 2017-10-13 Richard Biener <rguenther@suse.de>
5567 PR tree-optimization/82525
5568 * graphite-isl-ast-to-gimple.c
5569 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5571 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5572 Fail code generation when we cannot represent the isl integer.
5573 (binary_op_to_tree): Elide modulo operations that are no-ops
5574 in the type we code generate. Remove now superfluous code
5577 2017-10-13 Richard Biener <rguenther@suse.de>
5579 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5580 (scop_detection::harmful_loop_in_region): Remove premature
5581 IV type restriction.
5582 (scop_detection::graphite_can_represent_scev): We can handle
5583 pointer IVs just fine.
5585 2017-10-13 Alan Modra <amodra@gmail.com>
5587 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5588 "Clobbers and Scratch Registers". Add paragraph on
5589 alternative to clobbers for scratch registers and OpenBLAS
5592 2017-10-13 Alan Modra <amodra@gmail.com>
5594 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5595 example of a memory input for a string of known length. Move
5596 commentary out of table. Add a number of new examples
5597 covering array memory inputs.
5599 2017-10-12 Martin Liska <mliska@suse.cz>
5601 PR tree-optimization/82493
5602 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5603 (test_range_functions): New function.
5604 (sbitmap_c_tests): Likewise.
5605 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5606 * selftest.h (sbitmap_c_tests): New function.
5608 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5610 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5612 * config/rs6000/amo.h: Fix spacing issue.
5614 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5617 * config/i386/i386.md (*ashl<mode>3_mask_1,
5618 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5619 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5622 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5624 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5625 (profile_probability): Set max_probability
5626 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5628 * profile-count.c (profile_probability::differs_from_p): Do not
5629 rely on max_probaiblity == 10000
5631 2017-10-12 Jeff Law <law@redhat.com>
5633 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5636 2017-10-12 Martin Sebor <msebor@redhat.com>
5640 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5641 (handle_alias_pairs): Call it.
5642 * common.opt (-Wattribute-alias): New option.
5643 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5644 * doc/invoke.texi (-Wattribute-alias): Document.
5646 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5649 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5651 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5653 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5654 (make_hard_regno_born, make_hard_regno_dead): Update
5655 bb_killed_pseudos and bb_gen_pseudos.
5657 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5659 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5661 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5663 * config/alpha/alpha.c (alpha_split_conditional_move):
5664 Use std::swap instead of manually swapping.
5665 (alpha_stdarg_optimize_hook): Ditto.
5666 (alpha_canonicalize_comparison): Ditto.
5668 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5670 * tree-loop-distribution.c (struct builtin_info): New struct.
5671 (struct partition): Refactor fields into struct builtin_info.
5672 (partition_free): Free struct builtin_info.
5673 (build_size_arg_loc, build_addr_arg_loc): Delete.
5674 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5675 information from struct builtin_info.
5676 (find_single_drs): New function refactored from classify_partition.
5677 Also moved builtin validity checks to this function.
5678 (compute_access_range, alloc_builtin): New functions.
5679 (classify_builtin_st, classify_builtin_ldst): New functions.
5680 (classify_partition): Refactor code into functions find_single_drs,
5681 classify_builtin_st and classify_builtin_ldst.
5682 (distribute_loop): Don't do runtime alias check when distributing
5684 (find_seed_stmts_for_distribution): New function.
5685 (pass_loop_distribution::execute): Refactor code finding seed
5686 stmts into above function. Support distribution for the innermost
5687 two-level loop nest. Adjust dump information.
5689 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5691 * tree-loop-distribution.c: Adjust the general comment.
5692 (NUM_PARTITION_THRESHOLD): New macro.
5693 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5694 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5695 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5696 in call to build_partition_graph.
5697 (finalize_partitions): New parameter. Make loop distribution more
5698 conservative by fusing more partitions.
5699 (distribute_loop): Don't do runtime alias check in case of loop nest
5701 (find_seed_stmts_for_distribution): New function.
5702 (prepare_perfect_loop_nest): New function.
5703 (pass_loop_distribution::execute): Refactor code finding seed stmts
5704 and loop nest into above functions. Support loop nest distribution.
5705 Adjust dump information accordingly.
5707 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5709 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5710 and set PTYPE_SEQUENTIAL for merged partition.
5712 2017-10-12 Richard Biener <rguenther@suse.de>
5714 PR tree-optimization/69728
5716 2017-09-19 Richard Biener <rguenther@suse.de>
5718 PR tree-optimization/69728
5719 * graphite-sese-to-poly.c (schedule_error): New global.
5720 (add_loop_schedule): Handle empty domain by failing the
5722 (build_original_schedule): Handle schedule_error.
5724 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5725 domain by returning an unchanged schedule.
5727 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5729 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5730 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5732 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5734 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5737 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5740 * expr.c (store_field): Don't optimize away bitsize == 0 store
5741 from CALL_EXPR with addressable return type.
5743 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5745 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5746 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5747 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5748 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5749 TARGET_ISEL instead of TARGET_ISEL<sel>.
5751 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5753 * config/rs6000/rs6000.c
5754 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5756 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5758 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5760 (reg_or_cint_operand): Fix comment.
5761 (reg_or_zero_operand): New predicate.
5762 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5763 * config/rs6000/rs6000.c (output_isel): Delete.
5764 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5765 instead of reg_or_cint_operand. Output instruction directly (not via
5767 (isel_unsigned_<mode>): Ditto.
5768 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5769 gpc_reg_operand. Add an instruction alternative for this. Output
5770 instruction directly.
5771 (*isel_reversed_unsigned_<mode>): Ditto.
5773 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5775 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5776 (TARGET_CANONICALIZE_COMPARISON): Define.
5778 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5781 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5782 Check whether the dest is REG before adding REG_EQUIV note.
5784 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5787 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5789 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5790 (make_hard_regno_born, make_hard_regno_dead): Update
5791 bb_killed_pseudos and bb_gen_pseudos.
5793 2017-10-11 Nathan Sidwell <nathan@acm.org>
5795 * incpath.h (enum incpath_kind): Name enum, prefix values.
5796 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5797 * incpath.c (heads, tails): Use INC_MAX.
5798 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5799 (merge_include_chains, split_quote_chain,
5800 register_include_chains): Update incpath_kind names.
5801 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5802 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5804 (add_framework_path, darwin_register_objc_includes): Likewise.
5805 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5807 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5809 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5810 Do not use float_operator operator predicate.
5811 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5812 * config/i386/predicates.md (float_operator): Remove predicate.
5814 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5816 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5817 (*jccxf_i387): Ditto.
5818 (*jcc<mode>_i387): Ditto.
5819 (*jccu<mode>_i387): Ditto.
5820 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5821 (*jcc_*_i387 splitters): Remove.
5822 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5823 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5824 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5827 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5829 * profile-count.h (slow_safe_scale_64bit): New function.
5830 (safe_scale_64bit): New inline.
5831 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5832 * profile-count.c: Include wide-int.h
5833 (slow_safe_scale_64bit): New.
5835 2017-10-11 Nathan Sidwell <nathan@acm.org>
5837 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5838 HAS_DECL_ASSEMBLER_NAME_P.
5839 * gimple-expr.c (gimple_decl_printable_name: Check
5840 HAS_DECL_ASSEMBLER_NAME_P too.
5841 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5842 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5843 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5844 * passes.c (rest_of_decl_compilation): Check
5845 HAS_DECL_ASSEMBLER_NAME_P too.
5846 * recog.c (verify_changes): Likewise.
5847 * tree-pretty-print.c (dump_decl_name): Likewise.
5848 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5850 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5851 (SET_DECL_ASSEMBLER_NAME): Use it.
5852 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5853 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5854 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5856 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5858 * config.gcc (i386, x86_64): Add extra objects.
5859 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5860 (ix86_min_insn_size): Declare.
5861 (ix86_issue_rate): Declare.
5862 (ix86_adjust_cost): Declare.
5863 (ia32_multipass_dfa_lookahead): Declare.
5864 (ix86_macro_fusion_p): Declare.
5865 (ix86_macro_fusion_pair_p): Declare.
5866 (ix86_bd_has_dispatch): Declare.
5867 (ix86_bd_do_dispatch): Declare.
5868 (ix86_core2i7_init_hooks): Declare.
5869 (ix86_atom_sched_reorder): Declare.
5870 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5871 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5872 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5873 (rip_relative_addr_p): Rename to ...
5874 (ix86_rip_relative_addr_p): ... this one; export.
5875 (memory_address_length): Update.
5876 (ix86_issue_rate): Move to x86-tune-sched.c.
5877 (ix86_flags_dependent): Move to x86-tune-sched.c.
5878 (ix86_agi_dependent): Move to x86-tune-sched.c.
5879 (exact_dependency_1): Move to x86-tune-sched.c.
5880 (exact_store_load_dependency): Move to x86-tune-sched.c.
5881 (ix86_adjust_cost): Move to x86-tune-sched.c.
5882 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5883 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5884 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5885 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5886 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5887 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5888 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5889 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5890 (min_insn_size): Rename to ...
5891 (ix86_min_insn_size): ... this one; export.
5892 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5893 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5894 (core2i7_first_cycle_multipass_backtrack): Move to
5895 x86-tune-sched-core.c.
5896 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5897 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5898 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5899 (ix86_avoid_jump_mispredicts): Update.
5900 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5901 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5902 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5903 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5904 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5905 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5906 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5907 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5908 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5909 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5910 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5911 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5912 (BIG): Move to ix86-tune-sched-bd.c.
5913 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5914 (enum insn_path): Move to ix86-tune-sched-bd.c.
5915 (get_mem_group): Move to ix86-tune-sched-bd.c.
5916 (is_cmp): Move to ix86-tune-sched-bd.c.
5917 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5918 (is_branch): Move to ix86-tune-sched-bd.c.
5919 (is_prefetch): Move to ix86-tune-sched-bd.c.
5920 (init_window): Move to ix86-tune-sched-bd.c.
5921 (allocate_window): Move to ix86-tune-sched-bd.c.
5922 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5923 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5924 (process_end_window): Move to ix86-tune-sched-bd.c.
5925 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5926 (find_constant): Move to ix86-tune-sched-bd.c.
5927 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5928 (has_immediate): Move to ix86-tune-sched-bd.c.
5929 (get_insn_path): Move to ix86-tune-sched-bd.c.
5930 (get_insn_group): Move to ix86-tune-sched-bd.c.
5931 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5932 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5933 (add_insn_window): Move to ix86-tune-sched-bd.c.
5934 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5935 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5936 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5937 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5938 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5939 (do_dispatch): Move to ix86-tune-sched-bd.c.
5940 (has_dispatch): Move to ix86-tune-sched-bd.c.
5941 * config/i386/t-i386: Add new object files.
5942 * config/i386/x86-tune-costs.h: New file.
5943 * config/i386/x86-tune-sched-atom.c: New file.
5944 * config/i386/x86-tune-sched-bd.c: New file.
5945 * config/i386/x86-tune-sched-core.c: New file.
5946 * config/i386/x86-tune-sched.c: New file.
5948 2017-10-11 Liu Hao <lh_mouse@126.com>
5950 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5951 the generic version below instead.
5952 (should_colorize): Recognize Windows consoles as terminals
5954 * pretty-print.c [__MINGW32__] (write_all): New function.
5955 [__MINGW32__] (find_esc_head): Likewise.
5956 [__MINGW32__] (find_esc_terminator): Likewise.
5957 [__MINGW32__] (eat_esc_sequence): Likewise.
5958 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5960 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5963 2017-10-11 Richard Biener <rguenther@suse.de>
5965 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5966 Properly call analyze_scalar_evolution with the loop of the stmt.
5968 2017-10-11 Richard Biener <rguenther@suse.de>
5970 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5971 * tree-core.h (tree_base): Add chrec_var union member.
5972 * tree.h (CHREC_VAR): Remove.
5973 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5974 * tree-chrec.h (build_polynomial_chrec): Adjust.
5975 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5976 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5978 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5980 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5981 * match.pd: ... here.
5982 ((T) X == (T) Y): Relax condition.
5984 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5986 PR tree-optimization/82472
5987 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5989 (break_alias_scc_partitions): Update postorder number.
5991 2017-10-11 Martin Liska <mliska@suse.cz>
5994 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5996 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5998 2017-10-11 Martin Liska <mliska@suse.cz>
6000 * print-rtl.c (print_insn): Move declaration of idbuf
6001 to same scope as name.
6003 2017-10-11 Martin Liska <mliska@suse.cz>
6008 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6011 2017-10-11 Martin Liska <mliska@suse.cz>
6014 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6017 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6019 * config.gcc (default_gnu_indirect_function): Default to yes for
6020 arm*-*-linux* with glibc.
6022 2017-10-11 Richard Biener <rguenther@suse.de>
6024 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6025 default-defs and types we do not want to analyze.
6026 (interpret_loop_phi): Replace unreachable code with an assert.
6027 (compute_scalar_evolution_in_loop): Remove and inline ...
6028 (analyze_scalar_evolution_1): ... here, replacing condition with
6029 what makes the intent clearer. Remove handling of cases
6030 get_scalar_evolution now handles.
6032 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6034 PR rtl-optimization/81434
6035 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6036 comment for main loop. In sched_group_found if, also add checks for
6037 pass and min_cost_group.
6039 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6041 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6042 (rs6000_insn_cost): New function.
6043 * config/rs6000/rs6000.md (cost): New attribute.
6045 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6046 H.J. Lu <hongjiu.lu@intel.com>
6050 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6051 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6052 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6053 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6054 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6055 OPTION_MASK_ISA_MMX and some other bit set require both
6056 mmx and the ISAs without the mmx bit.
6057 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6058 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6059 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6060 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6061 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6062 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6063 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6064 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6065 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6066 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6067 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6068 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6069 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6070 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6071 Add OPTION_MASK_ISA_MMX.
6073 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6075 * config.gcc (armv7*-*-freebsd*): New target.
6076 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6078 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6080 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6081 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6084 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6086 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6087 a template parameter.
6088 (WIDE_INT_REF_FOR): Update accordingly.
6089 * tree.h (wi::int_traits <const_tree>): Delete.
6090 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6091 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6092 (wi::tree_to_wide_ref): New typedef.
6093 (wi::to_wide): New function.
6094 * calls.c (get_size_range): Use wi::to_wide when operating on
6096 * cgraph.c (cgraph_node::create_thunk): Likewise.
6097 * config/i386/i386.c (ix86_data_alignment): Likewise.
6098 (ix86_local_alignment): Likewise.
6099 * dbxout.c (stabstr_O): Likewise.
6100 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6101 * expr.c (const_vector_from_tree): Likewise.
6102 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6103 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6104 (fold_negate_expr_1, int_const_binop_1, const_binop)
6105 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6106 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6107 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6108 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6109 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6110 (fold_not_const, round_up_loc): Likewise.
6111 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6112 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6113 (alloca_call_type): Likewise.
6114 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6115 * godump.c (go_output_typedef): Likewise.
6116 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6117 * internal-fn.c (get_min_precision): Likewise.
6118 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6119 * ipa-polymorphic-call.c
6120 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6121 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6122 (ipa_modify_call_arguments): Likewise.
6123 * match.pd: Likewise.
6124 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6125 * print-tree.c (print_node_brief, print_node): Likewise.
6126 * stmt.c (expand_case): Likewise.
6127 * stor-layout.c (layout_type): Likewise.
6128 * tree-affine.c (tree_to_aff_combination): Likewise.
6129 * tree-cfg.c (group_case_labels_stmt): Likewise.
6130 * tree-data-ref.c (dr_analyze_indices): Likewise.
6131 (prune_runtime_alias_test_list): Likewise.
6132 * tree-dump.c (dequeue_and_dump): Likewise.
6133 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6134 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6135 * tree-pretty-print.c (dump_generic_node): Likewise.
6136 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6137 (simple_iv_with_niters): Likewise.
6138 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6139 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6140 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6141 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6142 (refine_value_range_using_guard, number_of_iterations_ne_max)
6143 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6144 (get_cst_init_from_scev, record_nonwrapping_iv)
6145 (scev_var_range_cant_overflow): Likewise.
6146 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6147 * tree-ssa-pre.c (compute_avail): Likewise.
6148 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6149 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6150 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6151 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6152 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6153 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6154 (dump_case_nodes, try_switch_expansion): Likewise.
6155 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6156 (vect_do_peeling): Likewise.
6157 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6158 * tree-vect-stmts.c (vectorizable_load): Likewise.
6159 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6160 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6161 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6162 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6163 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6164 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6165 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6166 (evrp_dom_walker::before_dom_children): Likewise.
6167 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6168 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6169 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6170 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6171 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6172 (get_range_pos_neg): Likewise.
6173 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6174 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6175 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6176 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6177 * config/avr/avr.c (avr_fold_builtin): Likewise.
6178 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6179 * config/msp430/msp430.c (msp430_attr): Likewise.
6180 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6181 * config/powerpcspe/powerpcspe-c.c
6182 (altivec_resolve_overloaded_builtin): Likewise.
6183 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6184 (rs6000_expand_ternop_builtin): Likewise.
6185 * config/rs6000/rs6000-c.c
6186 (altivec_resolve_overloaded_builtin): Likewise.
6187 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6188 (rs6000_expand_ternop_builtin): Likewise.
6189 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6191 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6193 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6194 when copying loop nest with only one inner loop.
6196 2017-10-10 Richard Biener <rguenther@suse.de>
6198 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6199 blocks if SCEV is active.
6200 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6202 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6203 (scev_initialize): Assert we are not yet initialized.
6205 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6207 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6208 inner loop's exit stmt by making it always exit the loop, otherwise
6209 we would generate an infinite empty loop.
6211 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6213 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6214 renaming variables in new preheader if it's deleted.
6216 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6218 * tree-loop-distribution.c (struct partition): Remove unused field
6219 loops of the structure.
6220 (partition_alloc, partition_free): Ditto.
6221 (build_rdg_partition_for_vertex): Ditto.
6223 2017-10-09 Jeff Law <law@redhat.com>
6225 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6226 return type to match prototype and documentation.
6228 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6230 * config/rs6000/rs6000.c (processor_costs): Move to ...
6231 * config/rs6000/rs6000.h: ... here.
6232 (rs6000_cost): Declare.
6234 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6236 * except.c (setjmp_fn): New global variable.
6237 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6238 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6239 if DONT_USE_BUILTIN_SETJMP is defined.
6241 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6243 * target.def (insn_cost): New hook.
6244 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6245 * doc/tm.texi: Regenerate.
6246 * rtlanal.c (insn_cost): Use the new hook.
6248 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6250 * combine.c (combine_validate_cost): Compute the new insn_cost,
6251 not just pattern_cost.
6252 (try_combine): Adjust comment.
6254 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6256 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6258 * combine.c (uid_insn_cost): Adjust comment.
6259 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6261 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6262 * dse.c (find_shift_sequence): Ditto.
6263 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6264 (bb_valid_for_noce_process_p): Use pattern_cost.
6265 * rtl.h (insn_rtx_cost): Delete.
6266 (pattern_cost): New prototype.
6267 (insn_cost): New prototype.
6268 * rtlanal.c (insn_rtx_cost): Rename to...
6269 (pattern_cost): ... this.
6272 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6274 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6275 (*jcc<mode>_0_r_i387): Ditto.
6276 (*jccxf_r_i387): Ditto.
6277 (*jcc<mode>_r_i387): Ditto.
6278 (*jccu<mode>_r_i387): Ditto.
6279 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6280 (*jcc): Rename from *jcc_1.
6282 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6284 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6285 deferred rescans after the lvx/stvx recombination pre-pass.
6287 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6289 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6290 memory operation instruction support.
6291 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6292 (rs6000-ibm-aix[789]*): Likewise.
6293 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6294 Document new functions.
6296 2017-10-09 Richard Biener <rguenther@suse.de>
6298 PR tree-optimization/82397
6299 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6300 equality only for semantically equal trees.
6302 2017-10-09 Richard Biener <rguenther@suse.de>
6304 PR tree-optimization/82449
6305 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6306 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6307 allow constant addresses.
6308 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6311 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6313 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6316 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6319 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6322 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6325 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6327 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6330 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6331 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6333 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6335 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6336 (WI_BINARY_PREDICATE_RESULT): Likewise.
6337 (wi::binary_traits::operator_result): New type.
6338 (wi::binary_traits::predicate_result): Likewise.
6339 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6340 (generic_wide_int::operator==, generic_wide_int::operator!=)
6341 (generic_wide_int::operator&, generic_wide_int::and_not)
6342 (generic_wide_int::operator|, generic_wide_int::or_not)
6343 (generic_wide_int::operator^, generic_wide_int::operator+
6344 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6346 (operator~, unary operator-, operator==, operator!=, operator&)
6347 (operator|, operator^, operator+, binary operator-, operator*): New
6349 * expr.c (get_inner_reference): Use wi::bit_and_not.
6350 * fold-const.c (fold_binary_loc): Likewise.
6351 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6352 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6353 (bit_value_binop): Likewise.
6354 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6355 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6356 (extract_range_from_binary_expr_1): Likewise.
6357 (masked_increment): Likewise.
6358 (simplify_bit_ops_using_ranges): Likewise.
6360 2017-10-09 Martin Jambor <mjambor@suse.cz>
6363 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6364 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6365 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6367 (hsa_fixup_mov_insn_type): New function.
6368 (hsa_op_with_type::get_in_type): Use it.
6369 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6370 immediates in an assert.
6371 (hsa_op_with_type::extend_int_to_32bit): New method.
6372 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6373 types. Convert to dest type if necessary.
6374 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6375 (reg_for_gimple_ssa): Pass false as min32int to
6376 hsa_type_for_scalar_tree_type.
6377 (gen_hsa_addr): Fixup type when creating addresable temporary.
6378 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6379 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6380 necessary. Call hsa_fixup_mov_insn_type.
6381 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6382 extend operands and convert to dest type if necessary.
6383 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6384 to dest type if necessary.
6385 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6386 if conversion nt necessary and size matches.
6387 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6388 to dest type if necessary.
6389 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6390 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6392 (gen_hsa_clrsb): Likewise.
6393 (gen_hsa_ffs): Likewise.
6394 (gen_hsa_divmod): Extend operands and convert to dest type if
6396 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6398 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6400 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6401 Remove empty default arguments. Use a brace block as output
6403 (conditional return): Ditto.
6405 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6406 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6408 (group_ending_nop): Ditto.
6409 (doloop_end): Ditto.
6410 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6411 (splitters for those): Ditto.
6413 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6415 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6416 a conditional jump (and the compare for it) so that pc_rtx is the
6418 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6419 for the deleted and renamed ctr<mode>_internal[234] patterns.
6420 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6421 Delete second conditional return pattern.
6422 (ctr<mode>_internal2): Delete this second bdnz pattern.
6423 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6424 (ctr<mode>_internal4): Delete this second bdz pattern.
6426 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6428 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6429 (always_initialized_rtx_for_ssa_name_p): New predicate.
6430 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6431 (finish_out_of_ssa): Free new field of SA.
6432 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6433 * tree-ssa-coalesce.c: Include tree-ssa.h.
6434 (get_parm_default_def_partitions): Remove extern keyword.
6435 (get_undefined_value_partitions): New function.
6436 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6437 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6439 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6441 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6443 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6445 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6447 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6448 for targets that preffer 128bit.
6450 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6452 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6454 2017-10-08 Olivier Hainque <hainque@adacore.com>
6456 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6457 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6459 (thumb_set_return_address): Likewise.
6461 2017-10-08 Olivier Hainque <hainque@adacore.com>
6463 * common/config/arm/arm-common.c (arm_except_unwind_info):
6464 Handle DWARF2_UNWIND_INFO.
6466 2017-10-07 Michael Collison <michael.collison@arm.com>
6468 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6471 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6473 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6474 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6475 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6476 defined, force the creation of a new block for a dispatch label.
6478 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6480 * invoke.texi (Wsuggest-attribute=cold): Document.
6481 * common.opt (Wsuggest-attribute=cold): New
6482 * ipa-pure-const.c (warn_function_cold): New function.
6483 * predict.c (compute_function_frequency): Use it.
6484 * predict.h (warn_function_cold): Declare.
6486 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6488 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6491 2017-10-06 Martin Liska <mliska@suse.cz>
6493 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6494 keyword for member functions.
6495 (struct sanopt_tree_couple): New struct.
6496 (struct sanopt_tree_couple_hash): New function.
6497 (struct sanopt_ctx): Add new hash_map.
6498 (has_dominating_ubsan_ptr_check): New function.
6499 (record_ubsan_ptr_check_stmt): Likewise.
6500 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6501 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6502 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6504 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6507 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6508 aarch64_simd_valid_immediate on CONST_VECTORs.
6509 (aarch64_reg_or_bic_imm): Likewise.
6511 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6513 PR rtl-optimization/82396
6514 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6516 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6518 * graphite-dependences.c (scop_get_reads): Move code to...
6519 (scop_get_must_writes): Move code to...
6520 (scop_get_may_writes): Move code to...
6521 (scop_get_reads_and_writes): ... here.
6522 (scop_get_dependences): Call scop_get_reads_and_writes.
6524 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6526 PR tree-optimization/82434
6527 * fold-const.h (can_native_encode_type_p,
6528 can_native_encode_string_p): Remove.
6529 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6530 don't encode anything, just return what would be otherwise returned.
6531 (native_encode_fixed, native_encode_complex, native_encode_vector):
6533 (native_encode_string): Likewise. Inline by hand
6534 can_native_encode_string_p.
6535 (can_native_encode_type_p): Remove.
6536 (can_native_encode_string_p): Remove.
6537 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6538 STRING_CSTs using can_native_encode_string_p, test all
6539 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6540 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6541 argument from native_encode_expr.
6542 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6543 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6546 2017-10-06 Richard Biener <rguenther@suse.de>
6548 PR tree-optimization/82397
6549 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6550 operand_equal_p but rely on data_ref_compare_tree for detecting
6552 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6553 to match up with dr_group_sort_cmp.
6555 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6558 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6560 * config/s390/s390-builtin-types.def: Regenerate.
6562 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6565 * config/s390/s390-builtin-types.def: Regenerate.
6566 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6567 Change flag from B_VXE to B_VX.
6568 (s390_vec_min_dbl): Remove B_VXE flag.
6570 2017-10-06 Richard Biener <rguenther@suse.de>
6572 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6573 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6574 translate_isl_ast_to_gimple::is_valid_rename,
6575 translate_isl_ast_to_gimple::get_rename,
6576 translate_isl_ast_to_gimple::get_def_bb_for_const,
6577 translate_isl_ast_to_gimple::get_new_name,
6578 translate_isl_ast_to_gimple::collect_all_ssa_names,
6579 translate_isl_ast_to_gimple::copy_loop_phi_args,
6580 translate_isl_ast_to_gimple::collect_all_ssa_names,
6581 translate_isl_ast_to_gimple::copy_loop_phi_args,
6582 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6583 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6584 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6585 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6586 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6587 translate_isl_ast_to_gimple::copy_cond_phi_args,
6588 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6589 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6590 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6591 translate_isl_ast_to_gimple::rename_uses,
6592 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6593 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6594 (set_rename_for_each_def): Likewise.
6595 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6596 here. Handle rewriting SCEV analyzable uses here.
6597 (copy_bb_and_scalar_dependences): Generate code for PHI
6599 (graphite_regenerate_ast_isl): Adjust.
6600 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6601 (add_write, add_read): New functions.
6602 (build_cross_bb_scalars_def): Use it and simplify.
6603 (build_cross_bb_scalars_use): Likewise.
6604 (graphite_find_cross_bb_scalar_vars): Inline into...
6605 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6606 simulating out-of-SSA. Compute liveout and add dependencies.
6607 (build_scops): Force an empty entry block.
6608 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6610 (sese_build_liveouts): Declare.
6611 (sese_trivially_empty_bb_p): Likewise.
6612 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6613 compute liveout and debug_liveout.
6614 (sese_bad_liveouts_use): Remove.
6615 (sese_reset_debug_liveouts_bb): Likewise.
6616 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6617 (sese_build_liveouts): Build liveout and debug_liveout and store
6619 (new_sese_info): Adjust.
6620 (free_sese_info): Likewise.
6621 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6622 do not build liveout here.
6623 (move_sese_in_condition): Adjust region entry.
6624 (scev_analyzable_p): Match up with chrec_apply requirements.
6625 (sese_trivially_empty_bb_p): New.
6626 * tree-into-ssa.c (get_reaching_def): Properly support generating
6627 default-defs for incremental rewrite of anonymous names.
6629 2017-10-06 Richard Biener <rguenther@suse.de>
6631 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6632 precision do not perform modulo reduction.
6634 2017-10-06 Richard Biener <rguenther@suse.de>
6636 PR tree-optimization/82436
6637 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6638 conservatively choose the vectorization factor when checking
6639 whether we can perform the required load permutation.
6640 (vect_transform_slp_perm_load): Assert when we may not fail.
6642 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6644 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6645 message for incompatible -msdata=* and -mcall-* options.
6647 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6649 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6650 rate for post-reload scheduling.
6652 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6654 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6656 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6658 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6659 to improve monte carlo in scimark.
6661 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6663 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6664 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6665 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6666 pentium4_cost, nocona_cost): Set reassociation width to 1.
6667 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6668 width to 2 for fp operations and 1 otherwise.
6669 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6671 (atom_cost): Set reassociation width to 2.
6672 (slm_cost, generic_cost): Set fp reassociation width
6673 to 2 and 1 otherwise.
6674 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6675 (core_cost): Set fp reassociation width to 4 and vector to 2.
6676 (ix86_reassociation_width): Rewrite using cost table; special case
6677 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6678 and TARGET_AVX128_OPTIMAL.
6679 * config/i386/i386.h (processor_costs): Add
6680 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6681 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6682 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6683 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6684 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6685 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6687 2017-10-05 Nathan Sidwell <nathan@acm.org>
6689 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6691 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6693 * config/arm/arm.c (arm_test_fpu_data): New.
6694 (arm_run_selftests): Call arm_test_fpu_data.
6696 2017-10-04 Nathan Sidwell <nathan@acm.org>
6698 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6700 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6702 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6704 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6705 check type for aarch64_simd_valid_immediate.
6706 (aarch64_output_simd_mov_immediate): Update prototype.
6707 (aarch64_simd_valid_immediate): Update prototype.
6708 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6709 support for ORR-immediate.
6710 (and<mode>3): modified pattern to add support for BIC-immediate.
6711 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6712 now checks for valid immediate for BIC and ORR based on new enum
6714 (aarch64_output_simd_mov_immediate): Function now used to output
6715 BIC/ORR imm as well based on new enum argument.
6716 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6718 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6719 (aarch64_reg_or_bic_imm): Likewise.
6721 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6723 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6724 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6726 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6730 PR rtl-optimization/82396
6731 * haifa-sched.c (autopref_multipass_init): Simplify
6733 (autopref_rank_data): Simplify sort order.
6734 * sched-int.h (autopref_multipass_data_): Remove
6735 multi_mem_insn_p, min_offset and max_offset.
6737 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6739 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6741 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6743 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6746 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6748 * doc/sourcebuild.texi: Document vect_long_mult.
6750 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6752 PR tree-optimization/82413
6753 * fold-const.c (build_range_check): Use widest_int when comparing
6754 the maximum ETYPE value with HIGH.
6756 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6758 PR rtl-optimization/82396
6759 * haifa-sched.c (autopref_multipass_init): Simplify
6761 (autopref_rank_data): Simplify sort order.
6762 * sched-int.h (autopref_multipass_data_): Remove
6763 multi_mem_insn_p, min_offset and max_offset.
6765 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6767 PR tree-optimization/82381
6768 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6769 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6772 PR tree-optimization/82374
6773 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6774 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6775 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6776 current_function_decl to the new decl.
6778 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6780 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6781 helper macro for IEEE float128 hardware built-in functions.
6782 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6784 (TRUNCF128_ODD): Likewise.
6785 (ADDF128_ODD): Likewise.
6786 (SUBF128_ODD): Likewise.
6787 (MULF128_ODD): Likewise.
6788 (DIVF128_ODD): Likewise.
6789 (FMAF128_ODD): Likewise.
6790 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6791 UNSPEC_TRUNC_ROUND_TO_ODD.
6792 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6793 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6794 floating point round to odd instructions.
6795 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6796 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6797 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6798 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6799 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6800 (trunc<mode>sf2_hw): Change the truncate with round to odd
6801 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6802 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6803 to odd hardware instructions.
6804 (sub<mode>3_odd): Likewise.
6805 (mul<mode>3_odd): Likewise.
6806 (div<mode>3_odd): Likewise.
6807 (sqrt<mode>2_odd): Likewise.
6808 (fma<mode>4_odd): Likewise.
6809 (fms<mode>4_odd): Likewise.
6810 (nfma<mode>4_odd): Likewise.
6811 (nfms<mode>4_odd): Likewise.
6812 (trunc<mode>df2_odd): Change the truncate with round to odd
6813 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6815 * doc/extend.texi (PowerPC built-in functions): Update documentation
6816 for existing IEEE float128-bit built-in functions. Add built-in
6817 functions that generate the IEEE 128-bit floating point round to
6820 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6822 PR rtl-optimization/77729
6823 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6824 to (X&(C1&~C2))|C2 transformations.
6826 2017-10-03 Martin Jambor <mjambor@suse.cz>
6828 PR tree-optimization/82363
6829 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6830 mismatch, mark lacc written regardless of racc.
6832 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6834 PR tree-optimization/82381
6835 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6836 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6837 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6838 fallthrough into reassoc_stmt_dominates_stmt_p.
6841 * combine.c (combine_instructions): Don't combine in unreachable
6844 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6847 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6848 function to not use the have_cpu variable. Do not set cpu_index,
6849 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6851 (rs6000_valid_attribute_p): Remove duplicate initializations of
6852 old_optimize and func_optimize.
6853 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6854 (rs6000_activate_target_options): Make global.
6855 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6858 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6860 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6861 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6862 Set *poffset to 0 and *psize and *pmax_size to -1 if
6863 *poffset + *psize overflows in HOST_WIDE_INT.
6865 PR tree-optimization/82387
6866 PR tree-optimization/82388
6867 PR tree-optimization/82389
6868 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6869 instead of live_bytes non-NULL.
6871 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6874 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6877 2017-10-02 Richard Biener <rguenther@suse.de>
6879 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6880 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6881 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6883 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6885 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6886 requested precision matches the type's.
6887 * calls.c (alloc_max_size): Calculate the new candidate size as
6888 a widest_int and use wi::to_widest when comparing it with the
6889 current candidate size.
6890 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6891 zero rather than integer_zero_node.
6892 * match.pd: Check for a no-op conversion before using wi::add
6893 rather than after. Use tree_to_uhwi when summing small shift
6894 counts into an unsigned int.
6896 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6897 Alan Hayward <alan.hayward@arm.com>
6898 David Sherwood <david.sherwood@arm.com>
6901 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6902 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6903 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6904 POINTER_AND_FP_REGS.
6906 2017-10-02 Richard Biener <rguenther@suse.de>
6908 PR tree-optimization/82355
6909 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6910 a mapping for the enclosing loop but avoid generating one for
6912 (copy_bb_and_scalar_dependences): Remove premature codegen
6913 error on PHIs in blocks duplicated into multiple places.
6914 * graphite-scop-detection.c
6915 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6916 in the region use it as loop and nest to analyze the DR in.
6917 (try_generate_gimple_bb): Likewise.
6918 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6919 (add_loop_constraints): For blocks in a loop not in the region
6920 create a dimension with a single iteration.
6921 * sese.h (gbb_loop_at_index): Remove assert.
6923 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6925 * omp-expand.c (adjust_context_scope): New function.
6926 (expand_parallel_call): Call adjust_context_scope.
6928 2017-10-01 Jeff Law <law@redhat.com>
6930 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6931 dom_opt_dom_walker class with direct access to private members.
6932 Add comments. Call test_for_singularity.
6933 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6934 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6935 m_dummy_cond anymore.
6936 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6938 (pass_dominator:execute): Build the dummy_cond here and pass it
6939 to the dom_opt_dom_walker ctor.
6940 (test_for_singularity): New function.
6942 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6943 Maya Rashish <coypu@sdf.org>
6945 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6946 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6947 (alpha*-*-netbsd*) Use nbsd_tm_file.
6948 (arm*-*-netbsdelf*) Likewise.
6949 (i[34567]86-*-netbsdelf*) Likewise.
6950 (x86_64-*-netbsd*) Likewise.
6951 (mips*-*-netbsd*) Likewise.
6952 (powerpc-*-netbsd*) Likewise.
6953 (sh*-*-netbsd*) Likewise.
6954 (sparc-*-netbsdelf*) Likewise.
6955 (sparc64-*-netbsd*) Likewise.
6956 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6958 (vax-*-netbsdelf*) Likewise.
6959 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6960 (UINT_FAST8_TYPE) Likewise.
6961 (INT_FAST16_TYPE) Check CHAR_FAST16.
6962 (UINT_FAST16_TYPE) Likewise.
6964 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6967 * config/i386/i386.md
6968 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6969 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6970 *divmodsi4_zext_2): New define_insn_and_split.
6971 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6972 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6973 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6974 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6975 New define_insn_and_split.
6976 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6977 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6978 operands[1] having DImode when mode is SImode.
6980 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6981 always SImode for DIV and MOD in REG_EQUAL notes.
6983 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6986 * match.pd: Fix handling of NaNs in pattern.
6988 2017-09-29 Jeff Law <law@redhat.com>
6990 * sbitmap.c (bitmap_bit_in_range_p): New function.
6991 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6992 * tree-ssa-dse.c (live_bytes_read): New function.
6993 (dse_classify_store): Ignore reads of dead bytes.
6995 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6996 typos and whitespace errors.
6997 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6998 * config/s390/s390.c (s390_emit_prologue): Likewise.
7000 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7003 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7004 with a symbol for LRA.
7006 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7008 PR rtl-optimization/82338
7009 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7011 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7013 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7014 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7015 (qsort_chk): Declare.
7016 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7017 (qsort_chk): New function.
7019 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7021 PR tree-optimization/82337
7022 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7023 phi definition if the PHI result appears in an abnormal PHI.
7024 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7025 basis appears in an abnormal PHI.
7027 2017-09-29 Richard Biener <rguenther@suse.de>
7029 * graphite-isl-ast-to-gimple.c
7030 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7031 (binary_op_to_tree): Use it.
7032 (get_rename_from_scev): Likewise.
7033 (copy_loop_phi_nodes): Likewise.
7034 (copy_bb_and_scalar_dependences): Likewise.
7035 (translate_pending_phi_nodes): Likewise.
7037 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7040 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7041 for movabsq $(i32 << shift), r64.
7043 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7045 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7046 index when encoding %esp as %rsp to avoid 0x67 prefix.
7048 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7050 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7051 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7053 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7055 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7056 Extensions with more than 16 double VFP registers.
7057 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7058 to_clear_mask and all code related to it. Replace the remaining
7059 entry by a sbitmap and adapt code accordingly.
7061 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7063 * brig-builtins.def: Change pure attributes to const.
7065 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7067 * config.gcc (default_gnu_indirect_function): Default to yes for
7068 sparc*-*-linux* with glibc.
7070 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7072 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7073 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7074 when creating .init_array and .fini_array sections with priority
7077 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7080 * config/aarch64/aarch64.c
7081 (aarch64_builtin_support_vector_misalignment): Always return false
7082 when misalignment is unknown.
7084 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7086 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7087 this function to return false if the definition used by the swap
7088 instruction is artificial, or if the memory address from which the
7089 constant value is loaded is not represented by a base address held
7090 in a register or if the base address register is a frame or stack
7091 pointer. Additionally, return false if the base address of the
7092 loaded constant is a SYMBOL_REF but is not considered to be a
7094 (replace_swapped_load_constant): New function.
7095 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7096 loaded constant vector with a load of a swapped constant vector.
7098 2017-09-27 Carl Love <cel@us.ibm.com>
7100 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7101 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7102 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7105 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7107 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7108 first, always call autopref_rank_data otherwise.
7110 2017-09-27 Richard Biener <rguenther@suse.de>
7112 * graphite-scop-detection.c (find_scop_parameters): Move
7113 loop bound handling ...
7114 (gather_bbs::before_dom_children): ... here, avoiding the need
7115 to build scop_info->loop_nest.
7116 (record_loop_in_sese): Remove.
7117 * sese.h (sese_info_t::loop_nest): Remove.
7118 * sese.c (new_sese_info): Do not allocate loop_nest.
7119 (free_sese_info): Do not free loop_nest.
7121 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7124 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7125 lhs from calls if the lhs has addressable type.
7127 2017-09-27 Richard Biener <rguenther@suse.de>
7129 * graphite.h (scop::max_alias_set): New member.
7130 * graphite-scop-detection.c: Remove references to non-existing
7131 --param in comments.
7132 (build_alias_sets): Record the maximum alias set used for drs.
7133 (build_scops): Support zero as unlimited for
7134 --param graphite-max-arrays-per-scop.
7135 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7137 (build_poly_sr_1): ... here. Compute alias set based on the
7138 maximum alias set used for drs rather than
7139 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7141 2017-09-27 Richard Biener <rguenther@suse.de>
7143 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7144 --param loop-block-tile-size=0 to disable tiling.
7146 2017-09-27 Richard Biener <rguenther@suse.de>
7148 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7149 (graphite-max-nb-scop-params): Document special value zero.
7150 * domwalk.h (dom_walker::STOP): New symbolical constant.
7151 (dom_walker::dom_walker): Add optional parameter for bb to
7153 (dom_walker::~dom_walker): Declare.
7154 (dom_walker::before_dom_children): Document STOP return value.
7155 (dom_walker::m_user_bb_to_rpo): New member.
7156 (dom_walker::m_bb_to_rpo): Likewise.
7157 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7158 mapping here if not provided by the user.
7159 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7160 provided by the user.
7161 (dom_walker::STOP): Define.
7162 (dom_walker::walk): Do not compute bb to RPO mapping here.
7163 Support STOP return value from before_dom_children to stop
7165 * graphite-optimize-isl.c (optimize_isl): If the schedule
7166 is the same still generate code if -fgraphite-identity
7167 or -floop-parallelize-all are given.
7168 * graphite-scop-detection.c: Include cfganal.h.
7169 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7171 (gather_bbs::before_dom_children): Return STOP for BBs
7173 (build_scops): Compute bb to RPO mapping and pass it to
7174 the domwalk. Treat --param graphite-max-nb-scop-params=0
7175 as not limiting the number of params.
7176 * graphite.c (graphite_initialize): Remove limit on the
7177 number of basic-blocks in a function.
7178 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7179 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7180 default value of 10.
7182 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7184 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7185 Adjust code to eliminate needing to do the shift right 32-bits
7186 operation after XSCVDPSPN.
7188 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7190 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7191 ((X / Y) != 0 -> X >= Y): Likewise.
7193 2017-09-26 Carl Love <cel@us.ibm.com>
7195 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7196 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7197 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7198 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7199 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7200 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7201 definitions and overloading.
7202 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7203 statement for P9V_BUILTIN_XST_LEN_R.
7204 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7205 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7206 define_expand and define_insn for the instructions and builtins.
7207 * doc/extend.texi: Update the built-in documentation file for the new
7209 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7210 define_insn for the instructions
7212 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7215 * gcc/config/netbsd-protos.h: New file.
7216 * gcc/config/netbsd.c: New file.
7217 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7218 * gcc/config/t-netbsd: New file.
7219 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7220 (tmake_file) Add t-netbsd.
7221 (extra_objs) Add netbsd.o.
7223 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7227 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7229 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7231 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7232 sign extension from a vector register to a GPR by doing a 32-bit
7233 direct move and then an EXTSW.
7234 (extendsi<mode>2 splitter): Likewise.
7235 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7236 right or vector extract after doing XSCVDPSPN. Use
7237 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7239 (movdi_from_sf_zero_ext): Likewise.
7240 (reload_gpr_from_vsxsf): Likewise.
7241 (p8_mfvsrd_4_disf): Delete, no longer used.
7242 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7243 then needing to move the SFmode to a GPR to use the XSCVDPSP
7244 instruction instead of FRSP and XSCVDPSPN.
7245 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7246 it is adjacent to the other XSCVSPDP insns.
7247 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7248 SFmode to be in traditional Altivec registers.
7249 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7250 (vsx_xscvspdpn): Likewise.
7251 (vsx_xscvspdpn_scalar): Likewise.
7253 2017-09-26 Martin Jambor <mjambor@suse.cz>
7255 * tree-sra.c (compare_access_positions): Put integral types first,
7256 stabilize sorting of integral types, remove conditions putting
7257 non-full-precision integers last.
7258 (sort_and_splice_var_accesses): Disable scalarization if a
7259 non-integert would be represented by a non-full-precision integer.
7261 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7263 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7264 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7265 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7266 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7267 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7268 conditionals inside the function instead of around it. Call
7269 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7270 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7272 2017-09-26 Richard Biener <rguenther@suse.de>
7274 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7276 (scop_detection::build_scop_breadth): ... this. Removed.
7277 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7278 (scop_detection::harmful_stmt_in_bb): Likewise.
7279 (scop_detection::graphite_can_represent_stmt): Likewise.
7280 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7281 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7282 (scop_detection::can_represent_loop_1): ... this. Removed.
7283 (scop_detection::harmful_loop_in_region): Simplify after inlining
7284 the above and remove more quadraticness.
7285 (build_scops): Adjust.
7286 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7289 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7292 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7293 REGNO (base) == SP_REG if base is a REG.
7296 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7297 if it is different SSA_NAME.
7298 (optimize_range_tests_cmp_bitwise): New function.
7299 (optimize_range_tests): Call it.
7301 2017-09-26 Richard Biener <rguenther@suse.de>
7303 PR tree-optimization/82321
7304 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7305 for the def being inside the loop.
7307 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7309 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7311 * config/s390/s390-builtins.def: Fix constraint on op4.
7313 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7315 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7316 independent expanders.
7317 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7318 ("vec_ordered", "vec_unordered"): New expanders.
7320 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7322 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7325 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7327 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7328 vec_unpacks_lo_v16qi.
7329 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7331 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7333 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7334 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7335 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7337 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7339 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7341 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7343 ("*vec_slb<mode>"): New insn pattern.
7344 ("vec_shr_<mode>"): New expander.
7345 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7346 and force the shift count operand to V16QImode.
7347 ("vec_srb<mode>"): Set shift count mode to V16QI.
7349 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7351 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7352 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7353 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7355 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7358 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7360 2017-09-26 Richard Biener <rguenther@suse.de>
7362 PR tree-optimization/82320
7363 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7366 2017-09-25 Jeff Law <law@redhat.com>
7368 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7369 prototype for new argument.
7370 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7371 mostly extracted from rs6000_emit_allocate_stack.
7372 (rs6000_emit_probe_stack_range_stack_clash): New function.
7373 (rs6000_emit_allocate_stack): Call
7374 rs6000_emit_probe_stack_range_stack_clash as needed.
7375 (rs6000_emit_probe_stack_range): Add additional argument
7376 to call to gen_probe_stack_range{si,di}.
7377 (output_probe_stack_range): New.
7378 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7379 (output_probe_stack_range_stack_clash): New.
7380 (rs6000_emit_prologue): Emit notes into dump file as requested.
7381 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7382 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7383 Add additional operand and pass it to output_probe_stack_range.
7385 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7387 PR tree-optimization/82163
7388 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7389 (checking_verify_loop_closed_ssa): New parameter.
7390 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7391 (check_loop_closed_ssa_stmt): Delete.
7392 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7393 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7394 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7397 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7399 * brig-builtins.def: Treat HSAIL barrier builtins as
7400 setjmp/longjump style functions.
7402 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7404 * target.def (constant_alignment): New hook.
7405 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7406 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7407 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7408 * doc/tm.texi: Regenerate.
7409 * targhooks.h (default_constant_alignment): Declare.
7410 (constant_alignment_word_strings): Likewise.
7411 * targhooks.c (default_constant_alignment): New function.
7412 (constant_alignment_word_strings): Likewise.
7413 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7414 instead of CONSTANT_ALIGNMENT.
7415 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7416 (force_const_mem): Likewise.
7417 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7418 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7419 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7420 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7421 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7423 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7424 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7425 constant_alignment_word_strings.
7426 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7427 (CONSTANT_ALIGNMENT): Likewise.
7428 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7429 (arm_constant_alignment): New function.
7430 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7431 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7432 constant_alignment_word_strings.
7433 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7434 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7435 constant_alignment_word_strings.
7436 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7437 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7438 (cris_constant_alignment): New function.
7439 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7440 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7441 (epiphany_constant_alignment): New function.
7442 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7443 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7444 constant_alignment_word_strings.
7445 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7446 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7447 constant_alignment_word_strings.
7448 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7449 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7450 constant_alignment_word_strings.
7451 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7452 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7453 * config/i386/i386.c (ix86_constant_alignment): Make static.
7454 Use the same interface as the target hook.
7455 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7456 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7457 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7458 constant_alignment_word_strings.
7459 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7460 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7461 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7462 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7463 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7464 constant_alignment_word_strings.
7465 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7466 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7467 constant_alignment_word_strings.
7468 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7469 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7470 constant_alignment_word_strings.
7471 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7472 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7474 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7475 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7476 * config/mips/mips.c (mips_constant_alignment): New function.
7477 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7478 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7479 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7480 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7481 (mmix_constant_alignment): Make static. Use the same interface
7483 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7484 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7485 constant_alignment_word_strings.
7486 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7487 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7488 constant_alignment_word_strings.
7489 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7490 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7491 constant_alignment_word_strings.
7492 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7493 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7494 (rs6000_constant_alignment): New function.
7495 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7496 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7497 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7498 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7499 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7500 (rs6000_constant_alignment): New function.
7501 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7502 * config/s390/s390.c (s390_constant_alignment): New function.
7503 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7504 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7505 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7506 constant_alignment_word_strings.
7507 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7508 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7509 (sparc_constant_alignment): New function.
7510 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7511 * config/spu/spu.c (spu_constant_alignment): New function.
7512 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7513 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7514 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7515 constant_alignment_word_strings.
7516 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7517 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7518 constant_alignment_word_strings.
7519 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7520 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7521 constant_alignment_word_strings.
7522 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7523 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7524 (visium_constant_alignment): New function.
7525 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7526 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7527 (xtensa_constant_alignment): New function.
7528 * system.h (CONSTANT_ALIGNMENT): Poison.
7530 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7532 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7533 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7534 (rs6000_builtin_valid_without_lhs): New helper function.
7535 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7536 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7538 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7540 * target.h (vec_perm_indices): Use unsigned short rather than
7542 (auto_vec_perm_indices): Likewise.
7543 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7544 Use unsigned int rather than unsigned char.
7545 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7547 2017-09-25 Richard Biener <rguenther@suse.de>
7549 * cfgloop.h (sort_sibling_loops): Declare.
7550 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7551 (sort_sibling_loops): New function sorting the sibling loop list
7553 * graphite.c (graphite_transform_loops): Sort sibling loops.
7555 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7557 * target.def (vec_perm_const_ok): Change sel parameter to
7559 * optabs-query.c (can_vec_perm_p): Update accordingly.
7560 * doc/tm.texi: Regenerate.
7561 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7562 auto_vec_perm_indices and remove separate nelt field.
7563 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7564 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7565 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7566 (aarch64_expand_vec_perm_const): Update accordingly.
7567 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7568 to vec_perm_indices.
7569 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7570 auto_vec_perm_indices and remove separate nelt field.
7571 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7572 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7573 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7575 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7576 to vec_perm_indices.
7577 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7578 sel to vec_perm_indices.
7579 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7580 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7581 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7583 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7586 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7589 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7590 on the FUNCTION_DECL function context if it has a DIE that is a
7593 2017-09-25 Richard Biener <rguenther@suse.de>
7595 PR tree-optimization/82285
7596 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7599 2017-09-25 Tom de Vries <tom@codesourcery.com>
7603 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7606 2017-09-25 Richard Biener <rguenther@suse.de>
7608 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7609 ISL errors other than isl_error_quota happen. Dump if the
7610 schedule is the same.
7611 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7612 errors instead of aborting inside ISL.
7614 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7617 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7618 of libgcc_eh for m64.
7619 * config/i386/darwin64.h: Likewise.
7621 2017-09-25 Richard Biener <rguenther@suse.de>
7624 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7625 attribute for incomplete types nor twice for complete ones.
7627 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7630 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7631 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7634 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7637 * config/i386/i386.opt (mprefer-avx256): Use
7638 ix86_target_flags variable.
7639 * config/i386/i386.c (ix86_target_string): Move
7640 -mprefer-avx256 to flag2_opts.
7642 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7645 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7646 and x != -1 | y != -1 into (x & y) != -1.
7648 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7650 * config.gcc: Add new case statement to set
7651 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7652 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7653 s390x-*-linux* case statements. Added aarch64 to the list of
7654 supported architectures.
7656 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7658 PR tree-optimization/82289
7659 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7660 STMT_VINFO_RELEVANT_P.
7662 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7663 Alan Hayward <alan.hayward@arm.com>
7664 David Sherwood <david.sherwood@arm.com>
7666 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7667 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7668 (extract_range_from_binary_expr_1): Don't call
7669 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7671 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7672 Alan Hayward <alan.hayward@arm.com>
7673 David Sherwood <david.sherwood@arm.com>
7675 * target.def (preferred_vector_alignment): New hook.
7676 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7678 * doc/tm.texi: Regenerate.
7679 * targhooks.h (default_preferred_vector_alignment): Declare.
7680 * targhooks.c (default_preferred_vector_alignment): New function.
7681 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7683 (DR_TARGET_ALIGNMENT): New macro.
7684 (aligned_access_p): Update commentary.
7685 (vect_known_alignment_in_bytes): New function.
7686 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7688 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7689 Calculate the misalignment based on the target alignment rather than
7691 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7692 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7693 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7694 the target alignment, rather than masking the element misalignment
7695 with the number of elements in a vector. Also use the target
7696 alignment when calculating the maximum number of peels.
7697 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7698 instead of TYPE_ALIGN_UNIT.
7699 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7700 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7701 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7702 (vect_create_data_ref_ptr): Likewise.
7703 (vect_setup_realignment): Realign by ANDing with
7704 -DR_TARGET_MISALIGNMENT.
7705 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7706 the number of peels based on DR_TARGET_ALIGNMENT.
7707 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7708 with the guaranteed alignment boundary when deciding whether
7710 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7711 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7712 (ensure_base_align): Remove stmt_info parameter. Get the
7713 target base alignment from DR_TARGET_ALIGNMENT.
7714 (vectorizable_store): Update call accordingly. Interpret
7715 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7717 (vectorizable_load): Likewise.
7719 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7720 Alan Hayward <alan.hayward@arm.com>
7721 David Sherwood <david.sherwood@arm.com>
7723 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7724 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7725 (vect_enhance_data_refs_alignment): Likewise.
7727 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7729 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7730 error. Only quit immediately if parsing is complete.
7731 (BEGIN): Initialize fatal_err and parse_done.
7732 (begin fpu, end fpu): Check number of arguments.
7733 (begin arch, end arch): Likewise.
7734 (begin cpu, end cpu): Likewise.
7735 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7736 (optalias): Likewise.
7738 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7740 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7741 * config/arm/arm-isa.h: Delete. Move definitions to ...
7742 * arm-cpus.in: ... here. Use new feature and fgroup values.
7743 * config/arm/arm.c (arm_option_override): Use lower case for feature
7745 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7746 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7747 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7749 (print_isa_bits_for): New function.
7750 (gen_isa): New function.
7751 (gen_comm_data): Use print_isa_bits_for.
7752 (define feature): New keyword.
7753 (define fgroup): New keyword.
7754 * config/arm/t-arm (TM_H): Remove.
7755 (GTM_H): Add arm-isa.h.
7756 (arm-isa.h): Add rule to generate file.
7757 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7758 case for feature bit names.
7760 2017-09-22 Richard Biener <rguenther@suse.de>
7762 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7764 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7765 print of no dependency loops ...
7766 * graphite.c (graphite_transform_loops): ... here.
7767 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7769 (same_close_phi_node, remove_duplicate_close_phi,
7770 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7771 (canonicalize_loop_closed_ssa): ... here and simplify.
7772 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7773 (optimize_isl): Use dump_printf_loc to tell when we stopped
7774 optimizing because of an ISL timeout.
7776 2017-09-22 Richard Biener <rguenther@suse.de>
7778 PR tree-optimization/82291
7779 * tree-if-conv.c (predicate_mem_writes): Make sure to
7780 remove writes in blocks predicated with false.
7782 2017-09-22 Richard Biener <rguenther@suse.de>
7784 * sese.c: Include cfganal.h.
7785 (if_region_set_false_region): Remove.
7786 (create_if_region_on_edge): Likewise.
7787 (move_sese_in_condition): Re-implement without destroying
7790 2017-09-22 Richard Biener <rguenther@suse.de>
7792 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7793 Verify both BBs contain loop PHI nodes before dispatching to
7795 (graphite_regenerate_ast_isl): Do not recompute dominators,
7796 do not verify three times. Restructure for clarity.
7797 * graphite-scop-detection.c (same_close_phi_node,
7798 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7799 defined_in_loop_p, canonicalize_loop_closed_ssa,
7800 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7801 checking and SSA rewrite, move to ...
7802 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7803 (graphite_initialize): Do not pass in ctx, do not reset the
7804 SCEV cache, compute only dominators.
7805 (graphite_transform_loops): Allocate ISL ctx after
7806 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7807 Maintain post-dominators only around build_scops.
7808 * sese.c (if_region_set_false_region): Make static. Free
7809 and recompute dominators.
7810 (move_sese_in_condition): Assert we don't get called with
7811 post-dominators computed.
7812 * sese.h (if_region_set_false_region): Remove.
7814 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7816 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7817 mode attribute for TARGET_AVX512VL.
7819 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7821 * config/i386/i386.opt (mprefer-avx256): New option.
7822 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7824 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7825 for TARGET_PREFER_AVX256.
7826 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7828 2017-09-21 Jeff Law <law@redhat.com>
7830 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7831 Fix dump output if the only stack space is for pushed registers.
7833 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7835 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7838 2017-09-21 Martin Sebor <msebor@redhat.com>
7841 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7842 code (in C++) or code that triggers warnings.
7844 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7846 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7848 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7850 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7851 * sched-rgn.c: Ditto.
7852 * sel-sched-ir.c: Ditto.
7854 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7856 * toplev.h (set_random_seed): Adjust return type.
7857 * toplev.c (init_local_tick): Move eager initialization of random_seed
7858 to get_random_seed. Adjust comment.
7859 (init_random_seed): Inline to get_random_seed, delete.
7860 (get_random_seed): Initialize random_seed lazily.
7861 (set_random_seed): Do not return previous value.
7862 (print_switch_value): Do not call get_random_seed.
7864 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7866 * cgraph.c (delete_function_version): New, broken out from...
7867 (cgraph_node::delete_function_version): ...here. Rename to
7868 cgraph_node::delete_function_version_by_decl. Update all uses.
7869 (cgraph_node::remove): Call delete_function_version.
7871 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7874 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7875 VAR_DECLs to which addressable non-volatile parameters are mapped
7876 and for id->retvar after the return value assignment. Clear
7877 id->retval and id->retbnd after inlining.
7879 2017-09-21 Richard Biener <rguenther@suse.de>
7881 PR tree-optimization/82276
7882 PR tree-optimization/82244
7883 * tree-vrp.c (build_assert_expr_for): Set
7884 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7886 (remove_range_assertions): Revert earlier change.
7888 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7891 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7893 2017-09-21 Richard Biener <rguenther@suse.de>
7895 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7896 Restore valid IL after code generation errors.
7897 * graphite.c (graphite_transform_loops): Diagnose code
7898 generation issues as MSG_MISSED_OPTIMIZATION and continue
7899 with processing SCOPs.
7901 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7902 Alan Hayward <alan.hayward@arm.com>
7903 David Sherwood <david.sherwood@arm.com>
7905 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7906 rather than choosing between plus_constant and gen_rtx_<CODE>.
7907 * expr.c (emit_push_insn): Likewise.
7908 (expand_expr_real_2): Likewise.
7910 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7911 Alan Hayward <alan.hayward@arm.com>
7912 David Sherwood <david.sherwood@arm.com>
7914 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7916 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7917 Alan Hayward <alan.hayward@arm.com>
7918 David Sherwood <david.sherwood@arm.com>
7920 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7921 calling tree_to_uhwi.
7923 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7924 Alan Hayward <alan.hayward@arm.com>
7925 David Sherwood <david.sherwood@arm.com>
7927 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7928 INTEGER_CST rather than a negative test for ADDR_EXPR.
7930 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7931 Alan Hayward <alan.hayward@arm.com>
7932 David Sherwood <david.sherwood@arm.com>
7934 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7935 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7937 2017-09-21 Richard Biener <rguenther@suse.de>
7939 PR tree-optimization/71351
7940 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7941 graphite_create_new_loop_guard): Remove, fold remaining parts
7943 (translate_isl_ast_node_for): ... here and simplify.
7945 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7948 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7949 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7950 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7951 alternative always use QI mode, for -Os imov (=R,R) alternative
7952 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7955 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7956 Jeff Law <law@redhat.com>
7958 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7959 (allocate_stack_space): New function, partially extracted from
7961 (s390_emit_prologue): Track offset to most recent stack probe.
7962 Code to allocate space moved into allocate_stack_space.
7963 Dump actions when no stack is allocated.
7964 (s390_prologue_plus_offset): New function.
7965 (s390_emit_stack_probe): Likewise.
7967 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7969 * common.opt (Wa, Wl, Wp, g, gz=): Add
7971 (gno-column-info): Remove.
7972 (gcolumn-info): Drop RejectNegative.
7974 (gno-record-gcc-switches): Remove.
7975 (grecord-gcc-switches): Drop RejectNegative.
7976 (gno-split-dwarf): Remove.
7977 (gsplit-dwarf): Drop RejectNegative.
7978 (gno-strict-dwarf): Remove.
7979 (gstrict-dwarf): Drop RejectNegative.
7980 * config/darwin.opt (gfull, gused): Add RejectNegative.
7981 * dwarf2out.c (gen_producer_string): Drop
7982 gno-record-gcc-switches handler.
7983 * optc-gen.awk: Add g to prefixes with negative forms.
7984 * opts-common.c (remapping_prefix_p): New.
7985 (find_opt): Check it.
7986 (generate_canonical_option): Test g prefix.
7987 (option_map): Add -gno- mapping.
7988 (add_misspelling_candidates): Check remapping_prefix_p.
7990 2017-09-20 Jeff Law <law@redhat.com>
7992 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7993 thinko in stack clash protection support.
7995 * explow.c (compute_stack_clash_protection_loop_data): Use
7996 CONST_INT_P instead of explicit test. Verify object is a
7997 CONST_INT_P before looking at INTVAL.
7998 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7999 instead of explicit test.
8001 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8004 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8005 address instead of to r1 and r11.
8007 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8009 * config.gcc: Support "knm".
8010 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8011 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8013 * config/i386/i386.c (m_KNM): Define.
8014 (processor_target_table): Add "knm".
8016 (ix86_option_override_internal): Add "knm".
8017 (ix86_issue_rate): Add PROCESSOR_KNM.
8018 (ix86_adjust_cost): Ditto.
8019 (ia32_multipass_dfa_lookahead): Ditto.
8020 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8021 (fold_builtin_cpu): Add M_INTEL_KNM.
8022 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8023 (processor_type): Add PROCESSOR_KNM.
8024 * config/i386/x86-tune.def: Add m_KNM.
8025 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8027 2017-09-20 Richard Biener <rguenther@suse.de>
8029 PR tree-optimization/80213
8030 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8031 are allowed in empty BBs as well.
8032 (canonicalize_loop_closed_ssa): Also look for other complex
8034 (scop_detection::get_sese): Include the loop-closed PHI block
8036 (scop_detection::merge_sese): Remove code adding extra blocks.
8037 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8038 (build_scops): Assert the final returned scop is invalid.
8040 2017-09-20 Richard Biener <rguenther@suse.de>
8042 PR tree-optimization/82264
8043 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8045 (vn_phi_lookup): Likewise.
8046 (vn_phi_insert): Likewise.
8048 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8050 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8051 that fits into uhwi or shwi, add DW_AT_const_value regardless
8052 of early_dwarf without going through RTL, using add_AT_unsigned
8055 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8056 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8057 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8058 (DEBUG_MACRO_SECTION): Likewise.
8059 (DEBUG_LTO_MACRO_SECTION): Likewise.
8060 (DEBUG_STR_DWO_SECTION): Likewise.
8061 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8062 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8063 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8064 (DEBUG_LTO_LINE_STR_SECTION): Define.
8065 (init_sections_and_labels): Initialize debug_line_str_section
8066 variable. Initialize debug_loc_section for -gdwarf-5 to
8067 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8069 2017-09-20 Richard Biener <rguenther@suse.de>
8071 * graphite-sese-to-poly.c (extract_affine): Properly handle
8072 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8074 2017-09-20 Richard Biener <rguenther@suse.de>
8076 PR tree-optimization/81373
8077 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8078 Force SESE live-out defs to be handled even if they are
8081 2017-09-19 Jeff Law <law@redhat.com>
8083 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8084 nothing for stack adjustments with REG_STACK_CHECK.
8085 * sched-deps.c (parse_add_or_inc): Reject insns with
8086 REG_STACK_CHECK from dependency breaking.
8087 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8088 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8089 * reg-notes.def (STACK_CHECK): New note.
8091 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8092 (ix86_expand_prologue): Dump stack clash info as needed.
8093 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8095 * function.c (dump_stack_clash_frame_info): New function.
8096 * function.h (dump_stack_clash_frame_info): Prototype.
8097 (enum stack_clash_probes): New enum.
8099 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8100 flag_stack_clash_protection.
8101 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8102 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8103 (arm_frame_pointer_required): Likewise.
8104 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8105 (ia64_expand_prologue): Likewise.
8106 * config/mips/mips.c (mips_expand_prologue): Likewise.
8107 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8108 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8109 (sparc_flat_expand_prologue): Likewise.
8110 * config/spu/spu.c (spu_expand_prologue): Likewise.
8112 * explow.c: Include "params.h".
8113 (anti_adjust_stack_and_probe_stack_clash): New function.
8114 (get_stack_check_protect): Likewise.
8115 (compute_stack_clash_protection_loop_data): Likewise.
8116 (emit_stack_clash_protection_loop_start): Likewise.
8117 (emit_stack_clash_protection_loop_end): Likewise.
8118 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8119 Use anti_adjust_stack_and_probe_stack_clash.
8120 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8121 (emit_stack_clash_protection_loop_start): Likewise.
8122 (emit_stack_clash_protection_loop_end): Likewise.
8123 * rtl.h (get_stack_check_protect): Prototype.
8124 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8125 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8126 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8128 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8130 * doc/tm.texi: Rebuilt.
8131 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8132 get_stack_check_protect.
8133 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8134 * config/arm/arm.c (arm_expand_prologue): Likewise.
8135 (arm_frame_pointer_required): Likewise.
8136 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8137 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8138 * config/mips/mips.c (mips_expand_prologue): Likewise.
8139 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8140 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8141 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8142 (sparc_flat_expand_prologue): Likewise.
8144 * common.opt (-fstack-clash-protection): New option.
8145 * flag-types.h (enum stack_check_type): Note difference between
8146 -fstack-check= and -fstack-clash-protection.
8147 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8148 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8149 * toplev.c (process_options): Issue warnings/errors for cases
8150 not handled with -fstack-clash-protection.
8151 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8152 (-fstack-check): Note additional problem with -fstack-check=generic.
8153 Note that -fstack-check is primarily for Ada and refer users
8154 to -fstack-clash-protection for stack-clash-protection.
8155 Document new params for stack clash protection.
8157 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8159 * config/i386/i386.c (ix86_split_long_move): Do not handle
8160 address used for LEA in a special way.
8162 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8164 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8166 2017-09-19 Martin Sebor <msebor@redhat.com>
8169 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8170 of incompatible types.
8172 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8174 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8175 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8176 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8177 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8179 2017-09-19 Richard Biener <rguenther@suse.de>
8181 PR tree-optimization/82244
8182 * tree-vrp.c (remove_range_assertions): Do not propagate
8183 a constant to abnormals but replace the assert with a copy.
8185 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8187 PR rtl-optimization/57878
8188 PR rtl-optimization/68988
8189 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8190 avoidance test involving non_reload_pseudos. Move frequency test
8191 below the general fragmentation avoidance test.
8193 2017-09-19 Richard Biener <rguenther@suse.de>
8195 PR tree-optimization/69728
8196 * graphite-sese-to-poly.c (schedule_error): New global.
8197 (add_loop_schedule): Handle empty domain by failing the
8199 (build_original_schedule): Handle schedule_error.
8201 2017-09-19 Richard Biener <rguenther@suse.de>
8203 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8204 Do not iterate to sibling loops but only to siblings of inner
8207 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8210 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8211 that the registers are different.
8213 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8215 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8216 to processor_model and "amdfam17h" to arch_names_table.
8217 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8219 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8222 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8224 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8225 Alan Hayward <alan.hayward@arm.com>
8226 David Sherwood <david.sherwood@arm.com>
8228 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8230 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8232 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8233 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8234 vect_schedule_slp_instance.
8235 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8236 Update call to vect_slp_analyze_node_operations. Simplify return
8238 (vect_slp_analyze_bb_1): Update call accordingly.
8239 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8240 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8241 (vect_schedule_slp): Update call accordingly.
8243 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8244 Alan Hayward <alan.hayward@arm.com>
8245 David Sherwood <david.sherwood@arm.com>
8247 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8248 with types that aren't in fact scalar.
8250 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8252 * tree-vect-slp.c (vect_record_max_nunits): New function,
8254 (vect_build_slp_tree_1): ...here.
8255 (vect_build_slp_tree_2): Call it for phis too.
8257 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8259 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8260 to vect_get_vec_def_for_operand when getting the mask operand.
8262 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8263 Alan Hayward <alan.hayward@arm.com>
8264 David Sherwood <david.sherwood@arm.com>
8266 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8269 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8270 Alan Hayward <alan.hayward@arm.com>
8271 David Sherwood <david.sherwood@arm.com>
8273 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8274 calculation for vector booleans.
8276 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8277 Alan Hayward <alan.hayward@arm.com>
8278 David Sherwood <david.sherwood@arm.com>
8280 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8282 (vect_transform_stmt): ...here.
8283 (vect_analyze_stmt): Use it instead of calling
8284 vectorizable_live_operation directly.
8286 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8288 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8289 non-SIMT targets in acc vector loops.
8291 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8293 * configure.ac: Add arc and check if assembler supports gdwarf2.
8294 * configure: Regenerate.
8296 2017-09-18 Richard Biener <rguenther@suse.de>
8298 PR tree-optimization/82220
8299 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8300 epilogue niters from the min_profitable_iters compute.
8302 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8305 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8306 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8307 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8309 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8312 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8313 switching to a new text section.
8315 2017-09-18 Richard Biener <rguenther@suse.de>
8317 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8319 (build_alias_set): Reject aliases with no access function.
8321 2017-09-18 Richard Biener <rguenther@suse.de>
8323 PR tree-optimization/79622
8324 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8326 (build_cross_bb_scalars_use): Likewise.
8328 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8330 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8332 2017-09-18 Alan Modra <amodra@gmail.com>
8335 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8336 stack_pointer_rtx for count 0. Update comments. Break up
8337 large rtl expression.
8339 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8341 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8342 Increase to 20 bytes.
8343 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8344 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8345 or avx version of the stub.
8347 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8350 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8351 compute the minimum stack alignment. Also update preferred stack
8352 boundary for leaf functions.
8354 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8356 PR tree-optimization/82228
8357 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8360 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8362 * common/config/nds32/nds32-common.c
8363 (nds32_option_optimization_table): Refine formatting.
8364 (nds32_option_optimization_table): Use -fsched-pressure and
8365 -fomit-frame-pointer for specific optimization level.
8367 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8369 * config/nds32/nds32.c: Refine formatting and comments.
8370 * config/nds32/nds32.h: Likewise.
8371 * config/nds32/nds32.md: Likewise.
8372 * config/nds32/nds32-cost.c: Likewise.
8373 * config/nds32/nds32-isr.c: Likewise.
8374 * config/nds32/nds32-md-auxiliary.c: Likewise.
8375 * config/nds32/nds32-multiple.md: Likewise.
8376 * config/nds32/nds32-predicates.c: Likewise.
8378 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8379 Jakub Jelinek <jakub@redhat.com>
8381 Add support for -std=c++2a.
8382 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8384 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8386 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8389 * doc/extend.texi (Common Function Attributes): Add
8390 references to ARM, AArch64, and S/390 specific attributes.
8391 (Function Specific Option Pragmas): Add AArch64 and S/390
8392 to list of back ends that support the target pragma.
8394 2017-09-15 Nathan Sidwell <nathan@acm.org>
8396 * doc/standards.texi: Fix C++17 description. Update URLs for
8399 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8401 * common.opt (Wcast-align=strict): New warning option.
8402 * doc/invoke.texi: Document -Wcast-align=strict.
8404 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8406 * cgraph.h (cgraph_thunk_info): Add comments.
8407 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8408 assert for VIRTUAL_* arguments stricter.
8410 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8412 PR tree-optimization/71026
8413 * match.pd: Move RDIV patterns from fold-const.c
8414 * fold-const.c (distribute_real_division): Removed.
8415 (fold_binary_loc): Remove calls to distribute_real_divison.
8417 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8419 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8420 c++1z and gnu++1z as deprecated. Change other references to
8421 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8422 Change -Wc++1z-compat to -Wc++17-compat.
8423 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8424 * dwarf2out.c (highest_c_language): Handle C++17.
8425 (gen_compile_unit_die): Likewise.
8427 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8429 PR rtl-optimization/82192
8430 * combine.c (make_extraction): Don't look through non-paradoxical
8431 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8434 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8435 Alan Hayward <alan.hayward@arm.com>
8436 David Sherwood <david.sherwood@arm.com>
8438 * target.def (function_arg_offset): New hook.
8439 * targhooks.h (default_function_arg_offset): Declare.
8440 * targhooks.c (default_function_arg_offset): New function.
8441 * function.c (locate_and_pad_parm): Use
8442 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8443 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8444 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8445 * doc/tm.texi: Regenerate.
8446 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8447 * config/spu/spu.c (spu_function_arg_offset): New function.
8448 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8449 * system.h (FUNCTION_ARG_OFFSET): Poison.
8451 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8452 Alan Hayard <alan.hayward@arm.com>
8453 David Sherwood <david.sherwood@arm.com>
8455 * target.def (truly_noop_truncation): New hook.
8456 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8457 than TRULY_NOOP_TRUNCATION.
8458 * hooks.h (hook_bool_uint_uint_true): Declare.
8459 * hooks.c (hook_bool_uint_uint_true): New function.
8460 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8461 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8462 * doc/tm.texi: Regenerate.
8463 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8464 rather than TRULY_NOOP_TRUNCATION in comments.
8465 (simplify_comparison): Likewise.
8466 (record_truncated_value): Likewise.
8467 * expmed.c (extract_bit_field_1): Likewise.
8468 (extract_split_bit_field): Likewise.
8469 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8470 instead of TRULY_NOOP_TRUNCATION.
8471 * function.c (assign_parm_setup_block): Likewise.
8472 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8473 * rtlhooks.c: Include target.h.
8474 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8475 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8476 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8477 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8478 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8479 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8480 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8481 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8482 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8483 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8484 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8485 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8486 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8487 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8488 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8489 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8490 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8491 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8492 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8493 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8494 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8495 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8496 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8497 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8498 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8499 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8500 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8501 rather than TRULY_NOOP_TRUNCATION in comments.
8502 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8503 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8504 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8505 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8506 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8507 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8508 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8509 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8510 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8511 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8512 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8513 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8514 rather than TRULY_NOOP_TRUNCATION in comments.
8515 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8516 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8517 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8518 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8519 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8520 TRULY_NOOP_TRUNCATION condition.
8521 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8522 (TRULY_NOOP_TRUNCATION): Delete.
8523 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8524 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8525 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8526 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8527 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8528 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8529 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8530 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8531 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8532 rather than TRULY_NOOP_TRUNCATION in comments.
8533 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8534 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8535 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8536 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8537 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8538 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8540 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8543 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8544 (*cmp_ior): Likewise.
8545 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8546 (*ior_scc_scc_cmp): Likewise.
8547 (*and_scc_scc): Likewise.
8548 (*and_scc_scc_cmp): Likewise.
8550 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8551 Alan Hayard <alan.hayward@arm.com>
8552 David Sherwood <david.sherwood@arm.com>
8554 * target.def (can_change_mode_class): New hook.
8555 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8556 (hard_regno_nregs): Likewise.
8557 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8558 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8559 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8560 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8561 (LOAD_EXTEND_OP): Update accordingly.
8562 * doc/tm.texi: Regenerate.
8563 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8564 CANNOT_CHANGE_MODE_CLASS.
8565 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8566 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8567 * combine.c (simplify_set): Update accordingly.
8568 * emit-rtl.c (validate_subreg): Likewise.
8569 * recog.c (general_operand): Likewise.
8570 * regcprop.c (mode_change_ok): Likewise.
8571 * reload1.c (choose_reload_regs): Likewise.
8572 (inherit_piecemeal_p): Likewise.
8573 * rtlanal.c (simplify_subreg_regno): Likewise.
8574 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8575 instead of CANNOT_CHANGE_MODE_CLASS.
8576 (reload_cse_simplify_operands): Likewise.
8577 * reload.c (push_reload): Use targetm.can_change_mode_class
8578 instead of CANNOT_CHANGE_MODE_CLASS.
8579 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8580 REG_CANNOT_CHANGE_MODE_P.
8581 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8582 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8583 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8584 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8585 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8586 (arm_can_change_mode_class): New function.
8587 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8588 than CANNOT_CHANGE_MODE_CLASS in comments.
8589 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8590 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8591 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8592 (ix86_can_change_mode_class): ...this new function, inverting the
8593 sense of the return value.
8594 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8595 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8596 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8597 (ia64_can_change_mode_class): New function.
8598 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8599 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8600 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8601 (m32c_can_change_mode_class): ...this new function, inverting the
8602 sense of the return value.
8603 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8604 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8605 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8606 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8607 (mips_can_change_mode_class): ...this new function, inverting the
8608 sense of the return value.
8609 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8610 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8611 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8612 (msp430_can_change_mode_class): New function.
8613 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8614 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8615 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8616 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8617 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8618 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8619 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8620 (pa_cannot_change_mode_class): Replace with...
8621 (pa_can_change_mode_class): ...this new function, inverting the
8622 sense of the return value.
8623 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8624 than CANNOT_CHANGE_MODE_CLASS in comments.
8625 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8626 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8627 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8628 (pdp11_cannot_change_mode_class): Replace with...
8629 (pdp11_can_change_mode_class): ...this new function, inverting the
8630 sense of the return value.
8631 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8632 * config/powerpcspe/powerpcspe-protos.h
8633 (rs6000_cannot_change_mode_class_ptr): Delete.
8634 * config/powerpcspe/powerpcspe.c
8635 (rs6000_cannot_change_mode_class_ptr): Delete.
8636 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8637 (rs6000_option_override_internal): Assign to
8638 targetm.can_change_mode_class instead of
8639 rs6000_cannot_change_mode_class_ptr.
8640 (rs6000_cannot_change_mode_class): Replace with...
8641 (rs6000_can_change_mode_class): ...this new function, inverting the
8642 sense of the return value.
8643 (rs6000_debug_cannot_change_mode_class): Replace with...
8644 (rs6000_debug_can_change_mode_class): ...this new function.
8645 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8646 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8647 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8648 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8649 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8651 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8652 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8653 (rs6000_option_override_internal): Assign to
8654 targetm.can_change_mode_class instead of
8655 rs6000_cannot_change_mode_class_ptr.
8656 (rs6000_cannot_change_mode_class): Replace with...
8657 (rs6000_can_change_mode_class): ...this new function, inverting the
8658 sense of the return value.
8659 (rs6000_debug_cannot_change_mode_class): Replace with...
8660 (rs6000_debug_can_change_mode_class): ...this new function.
8661 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8662 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8663 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8664 (s390_can_change_mode_class): ...this new function, inverting the
8665 sense of the return value.
8666 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8667 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8668 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8669 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8670 (sh_cannot_change_mode_class): Replace with...
8671 (sh_can_change_mode_class): ...this new function, inverting the
8672 sense of the return value.
8673 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8674 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8675 (sparc_can_change_mode_class): New function.
8676 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8677 * config/spu/spu.c (spu_can_change_mode_class): New function.
8678 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8679 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8680 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8681 (visium_can_change_mode_class): New function.
8682 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8684 2017-09-15 Richard Biener <rguenther@suse.de>
8686 PR tree-optimization/82217
8687 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8688 but not undefined case.
8690 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8693 * postreload.c (reload_cse_simplify_operands): Skip
8694 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8696 2017-09-15 Richard Biener <rguenther@suse.de>
8698 PR tree-optimization/68823
8699 * graphite-scop-detection.c (build_alias_set): If we have a
8700 possible dependence check whether we can handle them by just
8701 looking at the DRs DR_ACCESS_FNs.
8702 (build_scops): If build_alias_set fails, fail the SCOP.
8704 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8706 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8707 to support float128 built-in functions that require the ISA 3.0
8709 (BU_FLOAT128_3_HW): Likewise.
8710 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8712 (FMAF128): Likewise.
8714 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8715 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8716 floating point instructions.
8717 (rs6000_invalid_builtin): Likewise.
8718 (rs6000_builtin_mask_names): Likewise.
8719 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8720 (RS6000_BTM_FLOAT128_HW): Likewise.
8721 (RS6000_BTM_COMMON): Likewise.
8722 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8724 * doc/extend.texi (RS/6000 built-in functions): Document the
8725 IEEE 128-bit floating point square root and fused multiply-add
8728 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8730 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8731 reg (r2) isn't in the set of registers defined in the prologue.
8733 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8734 Alan Hayward <alan.hayward@arm.com>
8735 David Sherwood <david.sherwood@arm.com>
8737 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8738 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8739 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8740 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8741 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8743 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8744 max_vectorization_factor.
8745 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8747 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8748 Alan Hayward <alan.hayward@arm.com>
8749 David Sherwood <david.sherwood@arm.com>
8751 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8752 (vect_worthwhile_without_simd_p): Declare.
8753 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8754 (vectorizable_reduction): Use it.
8755 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8756 (vectorizable_operation): Likewise.
8758 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8759 Alan Hayward <alan.hayward@arm.com>
8760 David Sherwood <david.sherwood@arm.com>
8762 * tree-vectorizer.h (vect_get_num_copies): New function.
8763 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8764 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8765 (vectorizable_induction): Likewise.
8766 (vectorizable_live_operation): Likewise.
8767 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8768 (vectorizable_bswap): Likewise.
8769 (vectorizable_call): Likewise.
8770 (vectorizable_conversion): Likewise.
8771 (vectorizable_assignment): Likewise.
8772 (vectorizable_shift): Likewise.
8773 (vectorizable_operation): Likewise.
8774 (vectorizable_store): Likewise.
8775 (vectorizable_load): Likewise.
8776 (vectorizable_condition): Likewise.
8777 (vectorizable_comparison): Likewise.
8778 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8780 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8781 Alan Hayward <alan.hayward@arm.com>
8782 David Sherwood <david.sherwood@arm.com>
8784 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8785 of vect_init_vector.
8787 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8788 Alan Hayward <alan.hayward@arm.com>
8789 David Sherwood <david.sherwood@arm.com>
8791 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8792 an inline wrapper that provides a location.
8793 (gimple_build_vector): Likewise.
8794 * gimple-fold.c (gimple_build_vector_from_val): New function.
8795 (gimple_build_vector): Likewise.
8796 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8797 functions to build the initial value. Always return a gimple value.
8798 (get_initial_defs_for_reduction): Likewise. Only compute
8800 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8801 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8802 (vectorizable_induction): Use gimple_build_vector rather than
8805 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8806 Alan Hayward <alan.hayward@arm.com>
8807 David Sherwood <david.sherwood@arm.com>
8809 * target.h (vec_perm_indices): New typedef.
8810 (auto_vec_perm_indices): Likewise.
8811 * optabs-query.h: Include target.h
8812 (can_vec_perm_p): Take a vec_perm_indices *.
8813 * optabs-query.c (can_vec_perm_p): Likewise.
8814 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8815 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8816 * tree-vect-generic.c (lower_vec_perm): Likewise.
8817 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8818 (vect_grouped_load_supported): Likewise.
8819 (vect_shift_permute_load_chain): Likewise.
8820 (vect_permute_store_chain): Use auto_vec_perm_indices.
8821 (vect_permute_load_chain): Likewise.
8822 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8823 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8824 Update uses of can_vec_perm_p.
8825 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8826 mode with a number of elements. Take a vec_perm_indices *.
8827 (vect_create_epilog_for_reduction): Update accordingly.
8828 Use auto_vec_perm_indices.
8829 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8830 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8831 (vect_transform_slp_perm_load): Likewise.
8832 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8833 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8834 (vect_gen_perm_mask_checked): Likewise.
8835 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8836 (vect_gen_perm_mask_checked): Likewise.
8837 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8838 (vectorizable_store): Likewise.
8839 (vectorizable_load): Likewise.
8840 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8841 (vectorizable_bswap): Likewise.
8843 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8844 Alan Hayward <alan.hayward@arm.com>
8845 David Sherwood <david.sherwood@arm.com>
8847 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8848 * tree.c (build_vector): Likewise.
8849 (build_vector_from_ctor): Update accordingly.
8850 (build_vector_from_val): Likewise.
8851 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8852 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8853 * tree-vect-generic.c (add_rshift): Likewise.
8854 (expand_vector_divmod): Likewise.
8855 (optimize_vector_constructor): Likewise.
8856 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8857 (vect_transform_slp_perm_load): Likewise.
8858 (vect_schedule_slp_instance): Likewise.
8859 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8860 (vectorizable_call): Likewise.
8861 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8862 * expmed.c (make_tree): Likewise.
8863 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8864 a vector passed to build_vector.
8865 (fold_convert_const): Likewise.
8866 (exact_inverse): Likewise.
8867 (fold_ternary_loc): Likewise.
8868 (fold_relational_const): Likewise.
8869 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8870 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8871 (const_unop): Likewise. Store the reduction accumulator in a
8872 variable rather than an array.
8873 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8874 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8875 the new vector, rather than constructing it after the input arrays.
8876 (native_interpret_vector): Use auto_vec<tree> when building
8877 a vector passed to build_vector. Add elements in order.
8878 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8879 auto_vec<tree> when building a vector passed to build_vector.
8880 (vect_create_epilog_for_reduction): Likewise.
8881 (vectorizable_induction): Likewise.
8882 (get_initial_def_for_reduction): Likewise. Fix indentation of
8884 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8886 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8887 passed to build_vector.
8889 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8890 Alan Hayward <alan.hayward@arm.com>
8891 David Sherwood <david.sherwood@arm.com>
8893 * tree-core.h (tree_base::u): Add an "nelts" field.
8894 (tree_vector): Use VECTOR_CST_NELTS as the length.
8895 * tree.c (tree_size): Likewise.
8896 (make_vector): Initialize VECTOR_CST_NELTS.
8897 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8898 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8899 TYPE_VECTOR_SUBPARTS.
8900 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8901 as the number of units, setting it from VECTOR_CST_NELTS.
8902 (const_vector_from_tree): Likewise.
8903 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8904 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8905 (fold_negate_expr_1): Likewise.
8906 (fold_convert_const): Likewise.
8907 (const_binop): Likewise. Differentiate the number of output and
8909 (const_unop): Likewise.
8910 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8911 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8912 in cases that did the opposite.
8914 2017-09-14 Richard Biener <rguenther@suse.de>
8916 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8919 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8921 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8923 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8926 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8927 if and where to split a bb, except for splitting before debug insn
8928 sequences followed by non-label real insn. Delete debug insns
8929 in between basic blocks.
8931 * combine.c (make_compound_operation_int): Formatting fixes.
8933 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8934 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8935 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8936 * config/sol2.h (LINK_EH_SPEC): Likewise.
8937 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8938 * config/s390/linux.h (LINK_SPEC): Likewise.
8939 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8940 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8941 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8942 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8943 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8944 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8945 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8946 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8948 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8950 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8952 (ENDFILE_LINUX_SPEC): Likewise.
8953 (LINK_EH_SPEC): Likewise.
8954 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8955 (LINK_OS_LINUX_SPEC32): Likewise.
8956 (LINK_OS_LINUX_SPEC64): Likewise.
8957 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8958 (LINK_OS_LINUX_SPEC): Likewise.
8960 2017-09-13 Martin Liska <mliska@suse.cz>
8963 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8964 CASE_HIGH is NULL_TREE.
8966 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8967 Alan Hayward <alan.hayward@arm.com>
8968 David Sherwood <david.sherwood@arm.com>
8970 * target.def (secondary_memory_needed): New hook.
8971 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8972 instead of SECONDARY_MEMORY_NEEDED.
8973 (secondary_memory_needed_mode): Likewise.
8974 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8975 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8976 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8977 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8978 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8979 * doc/tm.texi: Regenerate.
8980 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8981 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8982 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8983 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8984 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8985 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8986 mode argument first and change the reg_class arguments to reg_class_t.
8987 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8988 Make static. Update the call to inline_secondary_memory_needed.
8989 (ix86_register_move_cost): Update the call to
8990 inline_secondary_memory_needed.
8991 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8992 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8994 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8995 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8997 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8998 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8999 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9000 and match hook interface. Add comment from mips.h.
9001 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9002 * config/mmix/mmix.md (truncdfsf2): Refer to
9003 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9005 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9006 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9007 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9008 (pa_secondary_memory_needed): New function.
9009 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9010 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9011 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9012 (pdp11_secondary_memory_needed): Make static and match hook interface.
9013 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9014 * config/powerpcspe/powerpcspe-protos.h
9015 (rs6000_secondary_memory_needed_ptr): Delete.
9016 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9018 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9019 (rs6000_option_override_internal): Assign to
9020 targetm.secondary_memory_needed rather than
9021 rs6000_secondary_memory_needed_ptr.
9022 (rs6000_secondary_memory_needed): Match hook interface.
9023 (rs6000_debug_secondary_memory_needed): Likewise.
9024 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9025 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9026 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9027 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9028 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9029 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9031 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9032 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9033 (rs6000_option_override_internal): Assign to
9034 targetm.secondary_memory_needed rather than
9035 rs6000_secondary_memory_needed_ptr.
9036 (rs6000_secondary_memory_needed): Match hook interface.
9037 (rs6000_debug_secondary_memory_needed): Likewise.
9038 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9039 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9040 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9041 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9042 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9043 (sparc_secondary_memory_needed): New function.
9044 * lra-constraints.c (check_and_process_move): Refer to
9045 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9047 (curr_insn_transform): Likewise.
9048 (process_alt_operands): Use targetm.secondary_memory_needed
9049 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9050 (check_secondary_memory_needed_p): Likewise.
9051 (choose_split_class): Likewise.
9052 * reload.c: Unconditionally include code that was previously
9053 conditional on SECONDARY_MEMORY_NEEDED.
9054 (push_secondary_reload): Use targetm.secondary_memory_needed
9055 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9056 (push_reload): Likewise.
9057 * reload1.c: Unconditionally include code that was previously
9058 conditional on SECONDARY_MEMORY_NEEDED.
9059 (choose_reload_regs): Use targetm.secondary_memory_needed
9060 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9061 (gen_reload): Likewise.
9062 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9064 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9065 Alan Hayward <alan.hayward@arm.com>
9066 David Sherwood <david.sherwood@arm.com>
9068 * target.def (secondary_memory_needed_mode): New hook:
9069 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9070 * targhooks.h (default_secondary_memory_needed_mode): New function.
9071 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9072 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9073 * doc/tm.texi: Regenerate.
9074 * lra-constraints.c (check_and_process_move): Use
9075 targetm.secondary_memory_needed_mode instead of
9076 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9077 (curr_insn_transform): Likewise.
9078 * reload.c (get_secondary_mem): Likewise.
9079 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9080 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9082 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9083 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9084 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9085 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9086 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9088 * config/powerpcspe/powerpcspe-protos.h
9089 (rs6000_secondary_memory_needed_mode): Delete.
9090 * config/powerpcspe/powerpcspe.c
9091 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9092 (rs6000_secondary_memory_needed_mode): Make static.
9093 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9094 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9096 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9098 (rs6000_secondary_memory_needed_mode): Make static.
9099 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9100 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9101 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9102 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9103 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9105 (sparc_secondary_memory_needed_mode): New function.
9106 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9108 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9110 * config/aarch64/constraints.md (Umq): New constraint.
9111 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9113 (mov<mode>): Update condition.
9115 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9117 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9118 when bitposition is the same.
9120 2017-09-13 Richard Biener <rguenther@suse.de>
9122 * dwarf2out.c (output_die_symbol): Remove.
9123 (output_die): Do not output a DIEs symbol.
9125 2017-09-13 Richard Biener <rguenther@suse.de>
9128 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9129 default-def to avoid breaking iterator update with the weird
9130 interaction with cgraph_update_edges_for_call_stmt_node.
9132 2017-09-13 Richard Biener <rguenther@suse.de>
9134 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9135 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9136 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9137 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9139 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9141 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9142 Disable pc relative literal load irrespective of
9143 TARGET_FIX_ERR_A53_84341 for default.
9145 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9147 * config/sparc/sparc.c (output_return): Output the source location of
9148 the insn in the delay slot, if any.
9149 (output_sibcall): Likewise.
9151 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9154 * common.opt (-static-pie): New alias.
9155 (shared): Negate static-pie.
9156 (-no-pie): Update help text.
9158 (static-pie): New option.
9159 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9160 -static-pie support.
9161 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9162 (LINK_EH_SPEC): Likewise.
9163 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9164 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9165 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9166 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9167 (init_gcc_specs): Likewise.
9168 (init_spec): Likewise.
9169 (display_help): Update help message for -pie.
9170 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9173 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9175 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9176 (movdi_aarch64): Likewise.
9177 (movti_aarch64): Likewise.
9179 2017-09-12 Simon Wright <simon@pushface.org>
9182 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9183 calculation of the minor version, always output as 0.
9185 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9188 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9189 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9190 on it early, rather than manual conversion late. For
9191 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9192 instead of performing manual conversion.
9194 2017-09-12 Carl Love <cel@us.ibm.com>
9196 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9197 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9199 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9200 VMULOSW): Add definitions.
9201 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9202 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9203 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9204 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9205 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9207 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9209 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9211 (movti_aarch64): Likewise.
9212 (movdf_aarch64): Likewise.
9213 (movtf_aarch64): Likewise.
9214 (load_pairdi): Likewise.
9215 (store_pairdi): Likewise.
9216 (load_pairdf): Likewise.
9217 (store_pairdf): Likewise.
9218 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9219 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9220 (ldr_got_small_<mode>): Likewise.
9221 (ldr_got_small_28k_<mode>): Likewise.
9222 (ldr_got_tiny): Likewise.
9223 * config/aarch64/iterators.md (ldst_sz): New.
9224 (ldpstp_sz): Likewise.
9225 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9227 (thunderx_load): Split load_8 to load_16.
9228 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9230 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9231 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9232 (xgene1_store_pair): Split store_8 to store_16.
9233 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9234 (falkor_st_0_st_sd): Split store_8 to store_16.
9236 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9238 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9239 and store1/2/3/4 to store_4/8/12/16.
9240 * config/aarch64/aarch64.md: Update for rename.
9241 * config/arm/arm.md: Likewise.: Likewise.
9242 * config/arm/arm.c: Likewise.
9243 * config/arm/thumb1.md: Likewise.
9244 * config/arm/thumb2.md: Likewise.
9245 * config/arm/vfp.md: Likewise.
9246 * config/arm/arm-generic.md: Likewise.
9247 * config/arm/arm1020e.md: Likewise.
9248 * config/arm/arm1026ejs.md: Likewise.
9249 * config/arm/arm1136jfs.md: Likewise.
9250 * config/arm/arm926ejs.md: Likewise.
9251 * config/arm/cortex-a15.md: Likewise.
9252 * config/arm/cortex-a17.md: Likewise.
9253 * config/arm/cortex-a5.md: Likewise.
9254 * config/arm/cortex-a53.md: Likewise.
9255 * config/arm/cortex-a57.md: Likewise.
9256 * config/arm/cortex-a7.md: Likewise.
9257 * config/arm/cortex-a8.md: Likewise.
9258 * config/arm/cortex-a9.md: Likewise.
9259 * config/arm/cortex-m4.md: Likewise.
9260 * config/arm/cortex-m7.md: Likewise.
9261 * config/arm/cortex-r4.md: Likewise.
9262 * config/arm/exynos-m1.md: Likewise.
9263 * config/arm/fa526.md: Likewise.
9264 * config/arm/fa606te.md: Likewise.
9265 * config/arm/fa626te.md: Likewise.
9266 * config/arm/fa726te.md: Likewise.
9267 * config/arm/fmp626.md: Likewise.
9268 * config/arm/iwmmxt.md: Likewise.
9269 * config/arm/ldmstm.md: Likewise.
9270 * config/arm/marvell-pj4.md: Likewise.
9271 * config/arm/xgene1.md: Likewise.
9272 * config/aarch64/thunderx.md: Likewise.
9273 * config/aarch64/thunderx2t99.md: Likewise.
9274 * config/aarch64/falkor.md: Likewise.
9276 2017-09-12 Martin Liska <mliska@suse.cz>
9278 * attribs.c (private_lookup_attribute): New function.
9279 * attribs.h (private_lookup_attribute): Declared here.
9280 (lookup_attribute): Called from this place.
9282 2017-09-12 Richard Biener <rguenther@suse.de>
9284 PR tree-optimization/82157
9285 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9286 stmts with side-effects.
9288 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9289 Alan Hayward <alan.hayward@arm.com>
9290 David Sherwood <david.sherwood@arm.com>
9292 * target.def (hard_regno_nregs): New hook.
9293 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9294 * targhooks.h (default_hard_regno_nregs): Declare.
9295 * targhooks.c (default_hard_regno_nregs): New function.
9296 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9297 (TARGET_HARD_REGNO_NREGS): ...this hook.
9298 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9299 (CLASS_MAX_NREGS): Likewise.
9300 * doc/tm.texi: Regenerate.
9301 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9302 instead of HARD_REGNO_NREGS.
9303 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9304 HARD_REGNO_NREGS in the comment.
9305 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9306 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9307 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9308 Return an unsigned int.
9309 (TARGET_HARD_REGNO_NREGS): Redefine.
9310 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9311 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9312 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9313 (arc_hard_regno_nregs): New function.
9314 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9315 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9316 (arm_hard_regno_nregs): New function.
9317 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9318 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9319 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9320 (TARGET_HARD_REGNO_NREGS): Redefine.
9321 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9322 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9323 (HARD_REGNO_NREGS): Delete.
9324 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9325 (cr16_hard_regno_nregs): New function.
9326 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9327 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9328 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9329 (cris_hard_regno_nregs): New function.
9330 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9331 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9332 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9333 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9334 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9335 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9336 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9337 (frv_hard_regno_nregs): Make static. Take and return an
9339 (frv_class_max_nregs): Remove outdated copy of documentation.
9340 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9341 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9342 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9343 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9344 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9345 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9346 (TARGET_HARD_REGNO_NREGS): Redefine.
9347 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9348 (CLASS_MAX_NREGS): Update comment.
9349 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9350 (ia64_hard_regno_nregs): New function.
9351 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9352 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9353 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9354 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9355 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9357 (m32c_hard_regno_nregs): Likewise. Make static.
9358 (TARGET_HARD_REGNO_NREGS): Redefine.
9359 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9360 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9361 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9362 (m68k_hard_regno_nregs): New function.
9363 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9364 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9365 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9366 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9367 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9368 Take and return an unsigned int.
9369 (TARGET_HARD_REGNO_NREGS): Redefine.
9370 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9371 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9372 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9373 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9374 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9375 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9376 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9377 (msp430_hard_regno_nregs): Make static. Take and return an
9379 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9380 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9381 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9382 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9383 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9384 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9385 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9386 (TARGET_HARD_REGNO_NREGS): Redefine.
9387 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9388 (PA_HARD_REGNO_NREGS): ...this.
9389 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9390 (PA_HARD_REGNO_NREGS): ...this.
9391 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9392 (pa_hard_regno_nregs): New function.
9393 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9394 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9395 (pdp11_hard_regno_nregs): New function.
9396 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9397 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9398 (rs6000_hard_regno_nregs_hook): New function.
9399 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9400 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9401 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9402 Take and return an unsigned int. Move earlier in file.
9403 (TARGET_HARD_REGNO_NREGS): Redefine.
9404 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9405 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9406 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9407 (rl78_hard_regno_nregs): Make static. Take and return an
9409 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9410 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9411 (rs6000_hard_regno_nregs_hook): New function.
9412 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9413 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9414 (TARGET_HARD_REGNO_NREGS): Redefine.
9415 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9416 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9417 instead of HARD_REGNO_NREGS.
9418 (s390_hard_regno_nregs): New function.
9419 (s390_hard_regno_mode_ok): Add comment from s390.h.
9420 (TARGET_HARD_REGNO_NREGS): Redefine.
9421 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9422 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9423 (sh_hard_regno_nregs): New function.
9424 (sh_pass_in_reg_p): Use it.
9425 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9426 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9427 (sparc_hard_regno_nregs): New function.
9428 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9429 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9430 (spu_function_arg_advance): Use it, supplying a valid register number.
9431 (TARGET_HARD_REGNO_NREGS): Redefine.
9432 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9433 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9434 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9435 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9436 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9437 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9438 (CLASS_MAX_NREGS): Remove copy of old documentation.
9439 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9440 (visium_hard_regno_nregs): New function.
9441 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9442 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9443 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9444 xtensa_hard_regno_nregs): New function.
9445 * system.h (HARD_REGNO_NREGS): Poison.
9447 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9449 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9450 hard_regno_nregs instead of HARD_REGNO_NREGS.
9451 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9452 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9453 (c6x_expand_epilogue): Likewise.
9454 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9455 (frv_read_iacc_argument): Likewise.
9456 * config/sh/sh.c: Include regs.h.
9457 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9458 (regs_used): Likewise.
9459 (output_stack_adjust): Likewise.
9460 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9461 * expmed.c: Include regs.h.
9462 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9463 * ree.c: Include regs.h.
9464 (combine_reaching_defs): Use hard_regno_nregs instead of
9466 (add_removable_extension): Likewise.
9468 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9470 * regs.h (hard_regno_nregs): Turn into a function.
9471 (end_hard_regno): Update accordingly.
9472 * caller-save.c (setup_save_areas): Likewise.
9473 (save_call_clobbered_regs): Likewise.
9474 (replace_reg_with_saved_mem): Likewise.
9475 (insert_restore): Likewise.
9476 (insert_save): Likewise.
9477 * combine.c (can_change_dest_mode): Likewise.
9478 (move_deaths): Likewise.
9479 (distribute_notes): Likewise.
9480 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9481 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9482 (rs6000_split_multireg_move): Likewise.
9483 (rs6000_register_move_cost): Likewise.
9484 (rs6000_memory_move_cost): Likewise.
9485 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9486 (rs6000_split_multireg_move): Likewise.
9487 (rs6000_register_move_cost): Likewise.
9488 (rs6000_memory_move_cost): Likewise.
9489 * cselib.c (cselib_reset_table): Likewise.
9490 (cselib_lookup_1): Likewise.
9491 * emit-rtl.c (set_mode_and_regno): Likewise.
9492 * function.c (aggregate_value_p): Likewise.
9493 * ira-color.c (setup_profitable_hard_regs): Likewise.
9494 (check_hard_reg_p): Likewise.
9495 (calculate_saved_nregs): Likewise.
9496 (assign_hard_reg): Likewise.
9497 (improve_allocation): Likewise.
9498 (calculate_spill_cost): Likewise.
9499 * ira-emit.c (modify_move_list): Likewise.
9500 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9501 (ira_hard_reg_in_set_p): Likewise.
9502 * ira.c (setup_reg_mode_hard_regset): Likewise.
9503 (clarify_prohibited_class_mode_regs): Likewise.
9504 (check_allocation): Likewise.
9505 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9506 (lra_setup_reg_renumber): Likewise.
9507 (setup_try_hard_regno_pseudos): Likewise.
9508 (spill_for): Likewise.
9509 (assign_hard_regno): Likewise.
9510 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9511 * lra-constraints.c (in_class_p): Likewise.
9512 (lra_constraint_offset): Likewise.
9513 (simplify_operand_subreg): Likewise.
9514 (lra_constraints): Likewise.
9515 (split_reg): Likewise.
9516 (split_if_necessary): Likewise.
9517 (invariant_p): Likewise.
9518 (inherit_in_ebb): Likewise.
9519 * lra-lives.c (process_bb_lives): Likewise.
9520 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9521 (get_hard_regs): Likewise.
9522 (do_remat): Likewise.
9523 * lra-spills.c (assign_spill_hard_regs): Likewise.
9524 * mode-switching.c (create_pre_exit): Likewise.
9525 * postreload.c (reload_combine_recognize_pattern): Likewise.
9526 * recog.c (peep2_find_free_register): Likewise.
9527 * regcprop.c (kill_value_regno): Likewise.
9528 (set_value_regno): Likewise.
9529 (copy_value): Likewise.
9530 (maybe_mode_change): Likewise.
9531 (find_oldest_value_reg): Likewise.
9532 (copyprop_hardreg_forward_1): Likewise.
9533 * regrename.c (check_new_reg_p): Likewise.
9534 (regrename_do_replace): Likewise.
9535 * reload.c (push_reload): Likewise.
9536 (combine_reloads): Likewise.
9537 (find_dummy_reload): Likewise.
9538 (operands_match_p): Likewise.
9539 (find_reloads): Likewise.
9540 (find_equiv_reg): Likewise.
9541 (reload_adjust_reg_for_mode): Likewise.
9542 * reload1.c (count_pseudo): Likewise.
9543 (count_spilled_pseudo): Likewise.
9544 (find_reg): Likewise.
9545 (clear_reload_reg_in_use): Likewise.
9546 (free_for_value_p): Likewise.
9547 (allocate_reload_reg): Likewise.
9548 (choose_reload_regs): Likewise.
9549 (reload_adjust_reg_for_temp): Likewise.
9550 (emit_reload_insns): Likewise.
9551 (delete_output_reload): Likewise.
9552 * rtlanal.c (subreg_get_info): Likewise.
9553 * sched-deps.c (sched_analyze_reg): Likewise.
9554 * sel-sched.c (init_regs_for_mode): Likewise.
9555 (mark_unavailable_hard_regs): Likewise.
9556 (choose_best_reg_1): Likewise.
9557 (verify_target_availability): Likewise.
9558 * valtrack.c (dead_debug_insert_temp): Likewise.
9559 * var-tracking.c (track_loc_p): Likewise.
9560 (emit_note_insn_var_location): Likewise.
9561 * varasm.c (make_decl_rtl): Likewise.
9562 * reginfo.c (choose_hard_reg_mode): Likewise.
9563 (init_reg_modes_target): Refer directly to
9564 this_target_regs->x_hard_regno_nregs.
9566 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9568 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9569 instead of hard_regno_nregs.
9571 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9573 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9574 end_hard_regno instead of hard_regno_nregs.
9575 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9576 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9577 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9578 * ira-color.c (improve_allocation): Likewise.
9579 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9580 * lra-lives.c (mark_regno_live): Likewise.
9581 (mark_regno_dead): Likewise.
9582 * lra-remat.c (operand_to_remat): Likewise.
9583 * lra.c (collect_non_operand_hard_regs): Likewise.
9584 * postreload.c (reload_combine_note_store): Likewise.
9585 (move2add_valid_value_p): Likewise.
9586 * reload.c (regno_clobbered_p): Likewise.
9588 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9590 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9592 * config/v850/v850.c (v850_reorg): Likewise.
9593 * reload.c (refers_to_regno_for_reload_p): Likewise.
9594 (find_equiv_reg): Likewise.
9595 * reload1.c (reload_reg_reaches_end_p): Likewise.
9597 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9599 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9601 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9602 * config/arm/arm.c (output_move_neon): Likewise.
9603 (arm_attr_length_move_neon): Likewise.
9604 (neon_split_vcombine): Likewise.
9605 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9606 (c6x_mark_reg_written): Likewise.
9607 (c6x_dwarf_register_span): Likewise.
9608 * config/i386/i386.c (ix86_save_reg): Likewise.
9609 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9610 (rws_access_reg): Likewise.
9611 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9612 * mode-switching.c (create_pre_exit): Likewise.
9613 * ree.c (combine_reaching_defs): Likewise.
9614 (add_removable_extension): Likewise.
9615 * regcprop.c (find_oldest_value_reg): Likewise.
9616 (copyprop_hardreg_forward_1): Likewise.
9617 * reload.c (reload_inner_reg_of_subreg): Likewise.
9618 (push_reload): Likewise.
9619 (combine_reloads): Likewise.
9620 (find_dummy_reload): Likewise.
9621 (reload_adjust_reg_for_mode): Likewise.
9622 * reload1.c (find_reload_regs): Likewise.
9623 (forget_old_reloads_1): Likewise.
9624 (reload_reg_free_for_value_p): Likewise.
9625 (reload_adjust_reg_for_temp): Likewise.
9626 (emit_reload_insns): Likewise.
9627 (delete_output_reload): Likewise.
9628 * sel-sched.c (choose_best_reg_1): Likewise.
9629 (choose_best_pseudo_reg): Likewise.
9631 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9632 Alan Hayward <alan.hayward@arm.com>
9633 David Sherwood <david.sherwood@arm.com>
9635 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9636 * target.def (slow_unaligned_access): New hook.
9637 * targhooks.h (default_slow_unaligned_access): Declare.
9638 * targhooks.c (default_slow_unaligned_access): New function.
9639 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9640 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9641 * doc/tm.texi: Regenerate.
9642 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9643 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9644 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9646 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9647 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9649 (rs6000_slow_unaligned_access): New function.
9650 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9651 (expand_block_compare): Likewise.
9652 (expand_strn_compare): Likewise.
9653 (rs6000_rtx_costs): Likewise.
9654 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9655 (riscv_slow_unaligned_access): Likewise.
9656 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9657 (riscv_slow_unaligned_access_p): ...this and make static.
9658 (riscv_option_override): Update accordingly.
9659 (riscv_slow_unaligned_access): New function.
9660 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9661 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9662 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9663 (rs6000_slow_unaligned_access): New function.
9664 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9665 (rs6000_rtx_costs): Likewise.
9666 * config/rs6000/rs6000-string.c (expand_block_compare)
9667 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9668 of SLOW_UNALIGNED_ACCESS.
9669 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9670 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9671 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9672 of SLOW_UNALIGNED_ACCESS.
9673 * expmed.c (simple_mem_bitfield_p): Likewise.
9674 * expr.c (alignment_for_piecewise_move): Likewise.
9675 (emit_group_load_1): Likewise.
9676 (emit_group_store): Likewise.
9677 (copy_blkmode_from_reg): Likewise.
9678 (emit_push_insn): Likewise.
9679 (expand_assignment): Likewise.
9680 (store_field): Likewise.
9681 (expand_expr_real_1): Likewise.
9682 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9683 * lra-constraints.c (simplify_operand_subreg): Likewise.
9684 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9685 * gimple-ssa-store-merging.c: Likewise in block comment at start
9687 * tree-ssa-strlen.c: Include target.h.
9688 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9689 of SLOW_UNALIGNED_ACCESS.
9690 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9692 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9694 PR rtl-optimization/82185
9695 * expmed.c (emit_store_flag_int): Only test tem if it has been
9698 2017-09-12 Richard Biener <rguenther@suse.de>
9701 * match.pd ((FTYPE) N CMP CST): Fix typo.
9703 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9705 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9707 (mips_near_type_p): Add 'short_call' attribute as a synonym
9709 * doc/extend.texi (short_call): Document new function attribute.
9711 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9714 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9715 assertion check that in the condition.
9716 (get_atomic_generic_size): Likewise. Before testing if parameter
9717 has pointer type, if it has array type, call for C++
9718 default_conversion to perform array-to-pointer conversion.
9720 2017-09-12 Richard Biener <rguenther@suse.de>
9722 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9723 for operations we cannot scalarize.
9725 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9727 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9728 vectors heap vectors. Clean up comments.
9729 Make visited_bbs a reference.
9730 (profitable_jump_thread_path): Make GC
9731 vectors heap vectors. Clean up comments.
9733 (convert_and_register_jump_thread_path): Make GC vectors heap
9735 (check_subpath_and_update_thread_path): Same. Clean up comments.
9736 Make visited_bbs a reference.
9737 (handle_phi): Abstract common code to to
9738 register_jump_thread_path_if_profitable.
9739 Rename VAR_BB to DEF_BB.
9741 Make GC vectors heap vectors.
9742 Make visited_bbs a reference.
9743 (handle_assignment): Same.
9744 (register_jump_thread_path_if_profitable): New.
9745 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9747 Make GC vectors heap vectors. Clean up comments.
9748 Make visited_bbs a reference.
9749 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9750 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9753 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9756 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9757 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9759 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9761 Revert r251800 and r251799.
9763 2017-09-11 Martin Jambor <mjambor@suse.cz>
9766 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9767 arguments in advance.
9768 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9769 use it to find predecessor edges.
9770 (naive_outof_ssa): Collect vector of predecessors.
9772 2017-09-08 Jason Merrill <jason@redhat.com>
9774 PR c++/70029 - ICE with ref-qualifier and -flto
9775 * langhooks.h (struct lang_hooks_for_types): Add
9776 copy_lang_qualifiers.
9777 * attribs.c (build_type_attribute_qual_variant): Use it.
9778 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9780 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9781 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9783 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9786 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9787 (*mulsi3_sp64): New instruction.
9788 (mulsi3): New expander.
9790 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9792 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9794 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9796 * sancov.c: Include memmodel.h.
9798 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9801 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9804 2017-09-07 Carl Love <cel@us.ibm.com>
9806 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9807 the sldi instruction.
9809 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9811 * sancov.c: Include tm_p.h.
9813 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9816 * output.h (switch_to_other_text_partition): New declaration.
9817 * varasm.c (switch_to_other_text_partition): New function.
9818 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9819 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9820 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9821 to the other text partition before emitting LCL label and switch back
9822 after emitting the word after it.
9824 2017-09-07 Richard Biener <rguenther@suse.de>
9826 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9827 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9828 critical edges here, after loop init.
9829 (pass_data_pre): Remove PROP_no_crit_edges flags.
9830 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9831 for valueization of call args to avoid leaking VN_TOP.
9832 (visit_use): Assert we do not visit default defs.
9833 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9834 Use error_mark_node to more easily detect leaking VN_TOP.
9835 All default-defs are varying, not VN_TOP. Mark them visited.
9836 (run_scc_vn): Make code match comment.
9838 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9840 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9841 OPTION_MASK_FLOAT128_KEYWORD.
9842 (POWERPC_MASKS): Likewise.
9843 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9844 support for the -mfloat128-type option, and make -mfloat128
9845 default on PowerPC Linux systems. Define or undefine
9846 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9847 Define __float128 to be __ieee128 if IEEE 128-bit support is
9848 enabled, or undefine it.
9849 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9850 Delete defining __FLOAT128_TYPE__.
9851 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9852 -mfloat128-type option and make -mfloat128 default on PowerPC
9854 (TARGET_FLOAT128_TYPE): Likewise.
9855 (-mfloat128-type): Likewise.
9856 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9857 Delete the -mfloat128-type option and make -mfloat128 default on
9858 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9859 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9860 128-bit floating point is enabled. Change tests from using
9861 -mfloat128-type to -mfloat128.
9862 (rs6000_mangle_type): Use the correct mangling for the __float128
9863 type even if normal long double is restricted to 64-bits.
9864 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9866 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9867 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9868 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9870 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9871 documentation for -mfloat128.
9873 2017-09-06 Olivier Hainque <hainque@adacore.com>
9875 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9877 2017-09-06 Wish Wu <wishwu007@gmail.com>
9878 Jakub Jelinek <jakub@redhat.com>
9880 * asan.c (initialize_sanitizer_builtins): Add
9881 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9882 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9883 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9884 BT_FN_VOID_UINT64_PTR variables.
9885 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9886 (BT_FN_VOID_UINT16_UINT16): Likewise.
9887 (BT_FN_VOID_UINT32_UINT32): Likewise.
9888 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9889 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9890 (BT_FN_VOID_UINT64_PTR): Likewise.
9891 * common.opt (flag_sanitize_coverage): New variable.
9892 (fsanitize-coverage=trace-pc): Remove.
9893 (fsanitize-coverage=): Add.
9894 * flag-types.h (enum sanitize_coverage_code): New enum.
9895 * fold-const.c (fold_range_test): Disable non-short-circuit
9896 optimization if flag_sanitize_coverage.
9897 (fold_truth_andor): Likewise.
9898 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9899 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9900 (coverage_sanitizer_opts): New array.
9901 (get_closest_sanitizer_option): Add OPTS argument, handle also
9902 OPT_fsanitize_coverage_.
9903 (parse_sanitizer_options): Adjusted to also handle
9904 OPT_fsanitize_coverage_.
9905 (common_handle_option): Add OPT_fsanitize_coverage_.
9906 * sancov.c (instrument_comparison, instrument_switch): New function.
9907 (sancov_pass): Add trace-cmp support.
9908 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9909 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9910 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9911 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9912 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9913 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9914 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9915 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9916 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9917 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9919 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9921 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9922 error. Only quit immediately if parsing is complete.
9923 (BEGIN): Initialize fatal_err and parse_done.
9924 (begin fpu, end fpu): Check number of arguments.
9925 (begin arch, end arch): Likewise.
9926 (begin cpu, end cpu): Likewise.
9927 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9928 (optalias): Likewise.
9930 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9932 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9933 * config/arm/arm-isa.h: Delete. Move definitions to ...
9934 * arm-cpus.in: ... here. Use new feature and fgroup values.
9935 * config/arm/arm.c (arm_option_override): Use lower case for feature
9937 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9938 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9939 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9941 (print_isa_bits_for): New function.
9942 (gen_isa): New function.
9943 (gen_comm_data): Use print_isa_bits_for.
9944 (define feature): New keyword.
9945 (define fgroup): New keyword.
9946 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9947 (arm-isa.h): Add rule to generate file.
9948 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9949 case for feature bit names.
9951 2017-09-06 Richard Biener <rguenther@suse.de>
9953 * tree-ssa-pre.c (NECESSARY): Remove.
9954 (create_expression_by_pieces): Do not touch pass-local flags.
9955 (insert_into_preds_of_block): Likewise.
9956 (do_pre_regular_insertion): Likewise.
9957 (eliminate_insert): Likewise.
9958 (eliminate_dom_walker::before_dom_children): Likewise.
9959 (fini_eliminate): Do not look at inserted_exprs.
9960 (mark_operand_necessary): Remove.
9961 (remove_dead_inserted_code): Replace with simple work-list
9962 algorithm based on inserted_exprs and SSA uses.
9963 (pass_pre::execute): Re-order fini_eliminate and
9964 remove_dead_inserted_code.
9966 2017-09-06 Olivier Hainque <hainque@adacore.com>
9968 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9969 for VxWorks 7. Adjust surrounding comments.
9971 2017-09-06 Richard Biener <rguenther@suse.de>
9973 * gimple-ssa-strength-reduction.c
9974 (find_candidates_dom_walker::before_dom_children): Also allow
9977 2017-09-06 Richard Biener <rguenther@suse.de>
9979 PR tree-optimization/82108
9980 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9981 for gap in the non-permutation SLP case.
9983 2017-09-06 Martin Jambor <mjambor@suse.cz>
9985 PR tree-optimization/82078
9986 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9987 add_access_to_work_queue...
9988 (build_accesses_from_assign): ...here.
9989 (propagate_all_subaccesses): Make sure racc is the group
9990 representative, if there is one.
9992 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9995 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9998 2017-09-06 Richard Biener <rguenther@suse.de>
10000 * gimple-ssa-strength-reduction.c
10001 (find_candidates_dom_walker::before_doom_children): Use a
10002 type and not a mode check.
10004 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10007 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10009 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10011 2017-09-05 Jeff Law <law@redhat.com>
10013 PR tree-optimization/64910
10014 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10015 cases where we have 3 or more operands.
10017 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10019 PR middle-end/81768
10020 * omp-low.c (lower_omp_for): Recompute tree invariant if
10021 gimple_omp_for_initial/final is ADDR_EXPR.
10023 PR middle-end/81768
10024 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10025 into gimple val before gimplification fo the COND_EXPR.
10027 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10029 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10030 REGION_COPY argument.
10031 (thread_through_all_blocks): Remove unused argument to
10032 duplicate_thread_path.
10034 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10035 Alan Hayward <alan.hayward@arm.com>
10036 David Sherwood <david.sherwood@arm.com>
10038 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10039 Take a scalar_mode rather than a machine_mode.
10040 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10041 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10042 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10043 (aarch64_gen_adjusted_ldpstp): Likewise.
10044 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10046 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10047 Alan Hayward <alan.hayward@arm.com>
10048 David Sherwood <david.sherwood@arm.com>
10050 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10051 Take a scalar_int_mode instead of a machine_mode.
10052 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10053 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10054 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10055 (aarch64_simd_attr_length_rglist): Delete.
10056 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10057 a scalar_int_mode instead of a machine_mode.
10058 (aarch64_add_offset): Likewise.
10059 (aarch64_internal_mov_immediate): Likewise
10060 (aarch64_add_constant_internal): Likewise.
10061 (aarch64_add_constant): Likewise.
10062 (aarch64_movw_imm): Likewise.
10063 (aarch64_rtx_arith_op_extract_p): Likewise.
10064 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10065 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10066 Remove assert that the mode isn't a vector.
10067 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10068 (aarch64_expand_mov_immediate): Update calls after above changes.
10069 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10070 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10071 (aarch64_move_imm): Likewise.
10072 (aarch64_can_const_movi_rtx_p): Likewise.
10073 (aarch64_strip_extend): Likewise.
10074 (aarch64_extr_rtx_p): Likewise.
10075 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10076 a CONST_INT when the mode parameter is VOIDmode.
10077 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10079 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10081 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10082 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10083 (bitwise_type_for_mode): Update accordingly.
10085 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10087 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10088 * stor-layout.c (mode_for_size_tree): Likewise.
10089 (mode_for_array): Update accordingly.
10090 (layout_decl): Likewise.
10091 (compute_record_mode): Likewise. Only set the mode once.
10093 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10095 * target.def (get_mask_mode): Change return type to opt_mode.
10097 * doc/tm.texi: Regenerate.
10098 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10099 * targhooks.c (default_get_mask_mode): Likewise.
10100 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10101 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10102 targetm.get_mask_mode.
10103 * tree.c (build_truth_vector_type): Likewise.
10105 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10107 * machmode.h (mode_for_vector): Return an opt_mode.
10108 * stor-layout.c (mode_for_vector): Likewise.
10109 (mode_for_int_vector): Update accordingly.
10110 (layout_type): Likewise.
10111 * config/i386/i386.c (emit_memmov): Likewise.
10112 (ix86_expand_set_or_movmem): Likewise.
10113 (ix86_expand_vector_init): Likewise.
10114 (ix86_get_mask_mode): Likewise.
10115 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10117 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10118 * expmed.c (extract_bit_field_1): Likewise.
10119 * expr.c (expand_expr_real_2): Likewise.
10120 * optabs-query.c (can_vec_perm_p): Likewise.
10121 (can_vec_mask_load_store_p): Likewise.
10122 * optabs.c (expand_vec_perm): Likewise.
10123 * targhooks.c (default_get_mask_mode): Likewise.
10124 * tree-vect-stmts.c (vectorizable_store): Likewise.
10125 (vectorizable_load): Likewise.
10126 (get_vectype_for_scalar_type_and_size): Likewise.
10128 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10130 * machmode.h (mode_for_int_vector): New function.
10131 * stor-layout.c (mode_for_int_vector): Likewise.
10132 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10133 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10134 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10135 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10136 (s390_expand_vcond): Likewise.
10138 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10140 * machmode.h (opt_machine_mode): New type.
10141 (opt_mode<T>): Allow construction from anything that can be
10143 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10144 (mode_for_size): Return an opt_machine_mode.
10145 * stor-layout.c (mode_for_size): Likewise.
10146 (mode_for_size_tree): Update call accordingly.
10147 (bitwise_mode_for_mode): Likewise.
10148 (make_fract_type): Likewise.
10149 (make_accum_type): Likewise.
10150 * caller-save.c (replace_reg_with_saved_mem): Update call
10152 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10153 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10154 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10155 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10156 * expmed.c (extract_bit_field_1): Likewise.
10157 * reload.c (get_secondary_mem): Likewise.
10158 * varasm.c (assemble_integer): Likewise.
10159 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10162 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10164 * machmode.h (decimal_float_mode_for_size): New function.
10165 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10166 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10167 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10168 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10169 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10170 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10172 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10174 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10175 (get_builtin_sync_mode): Likewise.
10176 (expand_ifn_atomic_compare_exchange): Likewise.
10177 (expand_builtin_atomic_clear): Likewise.
10178 (expand_builtin_atomic_test_and_set): Likewise.
10179 (fold_builtin_atomic_always_lock_free): Likewise.
10180 * calls.c (compute_argument_addresses): Likewise.
10181 (emit_library_call_value_1): Likewise.
10182 (store_one_arg): Likewise.
10183 * combine.c (combine_instructions): Likewise.
10184 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10185 * config/arm/arm.c (arm_function_value): Likewise.
10186 (aapcs_allocate_return_reg): Likewise.
10187 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10188 * config/i386/i386.c (construct_container): Likewise.
10189 (ix86_gimplify_va_arg): Likewise.
10190 (ix86_expand_sse_cmp): Likewise.
10191 (emit_memmov): Likewise.
10192 (emit_memset): Likewise.
10193 (expand_small_movmem_or_setmem): Likewise.
10194 (ix86_expand_pextr): Likewise.
10195 (ix86_expand_pinsr): Likewise.
10196 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10197 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10199 * config/mips/mips.c (mips_function_value_1) Likewise.
10200 (mips_block_move_straight): Likewise.
10201 (mips_expand_ins_as_unaligned_store): Likewise.
10202 * config/powerpcspe/powerpcspe.c
10203 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10204 (rs6000_darwin64_record_arg_flush): Likewise.
10205 * config/rs6000/rs6000.c
10206 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10207 (rs6000_darwin64_record_arg_flush): Likewise.
10208 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10209 (sparc_function_value_1): Likewise.
10210 * config/spu/spu.c (adjust_operand): Likewise.
10211 (spu_emit_branch_or_set): Likewise.
10212 (arith_immediate_p): Likewise.
10213 * emit-rtl.c (gen_lowpart_common): Likewise.
10214 * expr.c (expand_expr_real_1): Likewise.
10215 * function.c (assign_parm_setup_block): Likewise.
10216 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10217 * reload1.c (alter_reg): Likewise.
10218 * stor-layout.c (mode_for_vector): Likewise.
10219 (layout_type): Likewise.
10221 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10223 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10224 (spu_convert_move): Likewise.
10225 * lower-subreg.c (resolve_simple_move): Likewise.
10227 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10230 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10231 define_insn to a define_expand.
10232 (altivec_vsum2sws_direct): New define_insn.
10233 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10235 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10237 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10240 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10245 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10246 realign stack if stack alignment needed is less than incoming
10249 2017-09-05 Marek Polacek <polacek@redhat.com>
10252 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10255 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10257 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10259 2017-09-05 Richard Biener <rguenther@suse.de>
10261 PR tree-optimization/82084
10262 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10264 2017-09-05 Richard Biener <rguenther@suse.de>
10266 PR tree-optimization/82102
10267 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10269 2017-09-05 Martin Liska <mliska@suse.cz>
10271 PR tree-optimization/82032
10272 * tree-cfg.c (generate_range_test): New function.
10273 * tree-cfg.h (generate_range_test): Declared here.
10274 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10275 (cleanup_control_expr_graph): Use it.
10276 * tree-switch-conversion.c (try_switch_expansion): Remove
10278 (emit_case_nodes): Use generate_range_test.
10280 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10283 * config/i386/i386.md (*<btsc><mode>_mask): Add
10284 TARGET_USE_BT to insn constraint.
10285 (*btr<mode>_mask): Ditto.
10287 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10289 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10290 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10292 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10295 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10296 TARGET_NEON and TARGET_IWMMXT.
10297 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10298 TARGET_NEON and TARGET_IWMMXT.
10299 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10301 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10303 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10304 (ix86_rewrite_tls_address): Ditto.
10305 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10306 (ix86_rewrite_tls_address_1): Ditto.
10307 (ix86_rewrite_tls_address): Ditto.
10308 * config/i386/predicates.md (tls_address_pattern): New predicate.
10309 * config/i386/i386.md (TLS address splitter): New splitter.
10311 2017-09-04 Richard Biener <rguenther@suse.de>
10313 PR tree-optimization/82084
10314 * fold-const.h (can_native_encode_string_p): Declare.
10315 * fold-const.c (can_native_encode_string_p): Factor out from ...
10316 (native_encode_string): ... here.
10317 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10318 vectorizing stores from constants we later cannot handle.
10320 2017-09-04 Marek Polacek <polacek@redhat.com>
10323 * doc/invoke.texi: Update -Wtautological-compare documentation.
10325 2017-09-04 Jeff Law <law@redhat.com>
10327 PR tree-optimization/64910
10328 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10329 swap the first and last operand if the last is a constant.
10331 2017-09-04 Marek Polacek <polacek@redhat.com>
10334 * convert.c (do_narrow): When sanitizing signed integer overflows,
10335 bail out for signed types.
10336 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10338 2017-09-04 Richard Biener <rguenther@suse.de>
10340 PR tree-optimization/82060
10341 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10342 Move devirtualization after stmt folding and before EH/AB/noreturn
10343 cleanup to get the stmt refs canonicalized. Use a bool instead
10344 of gimple_modified_p since that doesn't work for NOPs. Schedule
10345 NOPs generated by folding for removal.
10347 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10348 Alan Hayward <alan.hayward@arm.com>
10349 David Sherwood <david.sherwood@arm.com>
10351 * coretypes.h (pad_direction): New enum.
10352 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10353 (FUNCTION_ARG_PADDING): Likewise.
10354 * target.def (function_arg_padding): New hook.
10355 * targhooks.h (default_function_arg_padding): Declare.
10356 * targhooks.c (default_function_arg_padding): New function.
10357 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10358 (TARGET_FUNCTION_ARG_PADDING): ...this.
10359 * doc/tm.texi: Regenerate.
10360 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10361 instead of direction.
10362 (compute_argument_addresses): Likewise.
10363 (load_register_parameters): Likewise.
10364 (emit_library_call_value_1): Likewise.
10365 (store_one_arg): Use targetm.calls.function_arg_padding instead
10366 of FUNCTION_ARG_PADDING.
10367 (must_pass_in_stack_var_size_or_pad): Likewise.
10368 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10369 (emit_group_store): Likewise.
10370 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10371 instead of FUNCTION_ARG_PADDING.
10372 (emit_push_insn): Likewise, and propagate enum change throughout
10374 * function.h (direction): Delete.
10375 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10377 * function.c (assign_parm_find_stack_rtl): Likewise.
10378 (assign_parm_setup_block_p): Likewise.
10379 (assign_parm_setup_block): Likewise.
10380 (gimplify_parameters): Likewise.
10381 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10382 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10384 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10385 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10386 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10387 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10388 (aarch64_function_arg_padding): ...this new function.
10389 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10390 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10391 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10392 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10393 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10394 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10395 (arm_pad_arg_upward): Replace with...
10396 (arm_function_arg_padding): ...this new function.
10397 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10399 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10400 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10401 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10402 (ia64_hpux_function_arg_padding): Replace with...
10403 (ia64_function_arg_padding): ...this new function. Use pad_direction
10404 instead of direction. Check for TARGET_HPUX.
10405 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10406 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10407 (iq2000_function_arg_padding): New function.
10408 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10409 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10410 (mips_function_arg_padding): ...this new function.
10411 (mips_pad_reg_upward): Update accordingly.
10412 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10413 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10414 targetm.calls.function_arg_padding.
10415 (FUNCTION_ARG_PADDING): Delete.
10416 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10417 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10418 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10419 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10420 (nios2_block_reg_padding): Return pad_direction instead of direction.
10421 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10422 instead of direction.
10423 (nios2_function_arg_padding): Likewise. Make static.
10424 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10425 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10426 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10427 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10428 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10429 (pa_function_arg_padding): Make static. Return pad_direction instead
10431 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10432 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10433 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10434 instead of direction. Use targetm.calls.function_arg_padding.
10435 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10436 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10437 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10438 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10439 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10441 (function_arg_padding): Rename to...
10442 (rs6000_function_arg_padding): ...this. Make static. Return
10443 pad_direction instead of direction.
10444 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10445 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10446 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10447 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10448 instead of direction. Use targetm.calls.function_arg_padding.
10449 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10450 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10451 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10452 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10453 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10454 (function_arg_padding): Rename to...
10455 (rs6000_function_arg_padding): ...this. Make static. Return
10456 pad_direction instead of direction.
10457 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10458 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10459 * config/s390/s390.c (s390_function_arg_padding): New function.
10460 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10461 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10462 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10463 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10464 (function_arg_padding): Rename to...
10465 (sparc_function_arg_padding): ...this. Make static. Return
10466 pad_direction instead of direction.
10467 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10468 * config/spu/spu.c (spu_function_arg_padding): New function.
10469 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10470 * system.h (FUNCTION_ARG_PADDING): Poison.
10472 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10473 Alan Hayward <alan.hayward@arm.com>
10474 David Sherwood <david.sherwood@arm.com>
10476 * target.def (modes_tieable_p): New hook.
10477 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10478 (TARGET_MODES_TIEABLE_P): ...this.
10479 * doc/tm.texi.in: Regenerate.
10480 * hooks.h (hook_bool_mode_mode_true): Declare.
10481 * hooks.c (hook_bool_mode_mode_true): New function.
10482 * combine.c (subst): Use targetm.modes_tieable_p instead of
10484 * dse.c (find_shift_sequence): Likewise.
10485 * expmed.c (extract_low_bits): Likewise.
10486 * lower-subreg.c: Include target.h.
10487 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10489 * rtlanal.c (rtx_cost): Likewise.
10490 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10491 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10492 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10493 (TARGET_MODES_TIEABLE_P): Redefine.
10494 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10495 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10496 (TARGET_MODES_TIEABLE_P): Redefine.
10497 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10498 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10499 (arc_modes_tieable_p): New function.
10500 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10501 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10502 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10503 (arm_modes_tieable_p): Make static.
10504 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10505 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10506 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10507 (TARGET_MODES_TIEABLE_P): Redefine.
10508 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10509 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10510 (TARGET_MODES_TIEABLE_P): Redefine.
10511 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10512 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10513 (cr16_modes_tieable_p): New function.
10514 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10515 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10516 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10517 (TRULY_NOOP_TRUNCATION): Update comment.
10518 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10519 (TRULY_NOOP_TRUNCATION): Update comment.
10520 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10521 (frv_modes_tieable_p): New function.
10522 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10523 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10524 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10525 (TARGET_MODES_TIEABLE_P): Redefine.
10526 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10527 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10528 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10529 (TARGET_MODES_TIEABLE_P): Redefine.
10530 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10531 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10532 (ia64_modes_tieable_p): New function.
10533 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10534 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10535 (iq2000_modes_tieable_p): New function.
10536 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10537 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10538 (lm32_modes_tieable_p): New function.
10539 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10540 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10541 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10542 (TARGET_MODES_TIEABLE_P): Redefine.
10543 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10544 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10545 (m32r_modes_tieable_p): New function.
10546 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10547 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10548 (m68k_modes_tieable_p): New function.
10549 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10550 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10551 (mcore_modes_tieable_p): New function.
10552 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10553 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10555 (TARGET_MODES_TIEABLE_P): Redefine.
10556 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10557 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10558 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10559 (TARGET_MODES_TIEABLE_P): Redefine.
10560 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10561 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10562 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10563 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10564 (mn10300_modes_tieable_p): ...this and make static.
10565 (TARGET_MODES_TIEABLE_P): Redefine.
10566 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10567 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10568 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10569 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10570 (msp430_modes_tieable_p): Make static.
10571 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10572 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10573 (TARGET_MODES_TIEABLE_P): Redefine.
10574 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10575 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10576 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10577 (TARGET_MODES_TIEABLE_P): Redefine.
10578 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10579 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10580 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10581 (TARGET_MODES_TIEABLE_P): Redefine.
10582 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10583 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10584 (pdp11_modes_tieable_p): New function.
10585 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10586 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10587 (rs6000_modes_tieable_p): New function.
10588 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10589 * config/powerpcspe/powerpcspe.md: Update comment.
10590 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10591 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10592 (TARGET_MODES_TIEABLE_P): Redefine.
10593 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10594 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10595 (rl78_modes_tieable_p): New function.
10596 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10597 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10598 (rs6000_modes_tieable_p): New function.
10599 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10600 * config/rs6000/rs6000.md: Update comment.
10601 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10602 * config/rx/rx.c (rx_modes_tieable_p): New function.
10603 (TARGET_MODES_TIEABLE_P): Redefine.
10604 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10605 * config/s390/s390.c (s390_modes_tieable_p): New function.
10606 (TARGET_MODES_TIEABLE_P): Redefine.
10607 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10608 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10609 (sh_modes_tieable_p): New function.
10610 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10611 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10612 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10613 (sparc_modes_tieable_p): Make static.
10614 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10615 * config/spu/spu.c (spu_modes_tieable_p): New function.
10616 (TARGET_MODES_TIEABLE_P): Redefine.
10617 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10618 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10619 (TARGET_MODES_TIEABLE_P): Redefine.
10620 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10621 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10622 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10623 * config/v850/v850.c (v850_modes_tieable_p): New function.
10624 (TARGET_MODES_TIEABLE_P): Redefine.
10625 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10626 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10627 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10628 (visium_modes_tieable_p): New function.
10629 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10630 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10631 (xtensa_modes_tieable_p): New function.
10632 * system.h (MODES_TIEABLE_P): Poison.
10634 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10635 Alan Hayward <alan.hayward@arm.com>
10636 David Sherwood <david.sherwood@arm.com>
10638 * target.def (hard_regno_mode_ok): New hook.
10639 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10640 (TARGET_HARD_REGNO_MODE_OK): ...this.
10641 * doc/tm.texi.in: Regenerate.
10642 * hooks.h (hook_bool_uint_mode_true): Declare.
10643 * hooks.c (hook_bool_uint_mode_true): New function.
10644 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10645 HARD_REGNO_MODE_OK.
10646 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10647 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10648 instead of HARD_REGNO_MODE_OK.
10649 * caller-save.c: Include target.h.
10650 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10651 HARD_REGNO_MODE_OK.
10652 * combine.c (can_combine_p): Likewise.
10653 (combinable_i3pat): Likewise.
10654 (can_change_dest_mode): Likewise.
10655 * expr.c (init_expr_target): Likewise.
10656 (convert_move): Likewise.
10657 (convert_modes): Likewise.
10658 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10659 (setup_prohibited_mode_move_regs): Likewise.
10660 * ira.h (target_ira): Likewise.
10661 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10662 * lra-constraints.c (process_alt_operands): Likewise.
10663 (split_reg): Likewise.
10664 * recog.c (peep2_find_free_register): Likewise.
10665 * ree.c (combine_reaching_defs): Likewise.
10666 * regcprop.c (maybe_mode_change): Likewise.
10667 * reginfo.c (init_reg_sets_1): Likewise.
10668 (choose_hard_reg_mode): Likewise.
10669 (simplifiable_subregs): Likewise.
10670 * regrename.c (check_new_reg_p): Likewise.
10671 * reload.c (find_valid_class): Likewise.
10672 (find_valid_class_1): Likewise.
10673 (reload_inner_reg_of_subreg): Likewise.
10674 (push_reload): Likewise.
10675 (combine_reloads): Likewise.
10676 (find_dummy_reload): Likewise.
10677 (find_reloads): Likewise.
10678 * reload1.c (find_reg): Likewise.
10679 (set_reload_reg): Likewise.
10680 (allocate_reload_reg): Likewise.
10681 (choose_reload_regs): Likewise.
10682 (reload_adjust_reg_for_temp): Likewise.
10683 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10684 (simplify_subreg_regno): Likewise.
10685 * sel-sched.c (init_regs_for_mode): Likewise.
10686 * varasm.c (make_decl_rtl): Likewise.
10687 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10688 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10689 HARD_REGNO_MODE_OK.
10690 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10691 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10692 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10693 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10694 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10695 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10696 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10697 (arc_mode_class): Delete.
10698 (HARD_REGNO_MODE_OK): Delete.
10699 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10700 (arc_hard_regno_mode_ok): Rename old array to...
10701 (arc_hard_regno_mode_ok_modes): ...this.
10702 (arc_conditional_register_usage): Update accordingly.
10703 (arc_mode_class): Make static.
10704 (arc_hard_regno_mode_ok): New function.
10705 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10706 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10707 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10708 (arm_hard_regno_mode_ok): Make static.
10709 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10710 HARD_REGNO_MODE_OK.
10711 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10712 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10713 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10715 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10716 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10717 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10718 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10719 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10720 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10721 * config/bfin/predicates.md (valid_reg_operand): Use
10722 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10723 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10724 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10725 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10726 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10727 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10728 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10729 (cr16_hard_regno_mode_ok): Make static and return a bool.
10730 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10731 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10732 (cris_hard_regno_mode_ok): New function.
10733 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10734 (epiphany_mode_class): Delete.
10735 (HARD_REGNO_MODE_OK): Delete.
10736 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10737 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10738 (hard_regno_mode_ok): Rename to...
10739 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10740 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10741 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10742 HARD_REGNO_MODE_OK.
10743 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10744 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10745 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10746 (frv_hard_regno_mode_ok): Make static and return a bool.
10747 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10748 HARD_REGNO_MODE_OK.
10749 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10750 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10751 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10752 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10754 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10755 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10756 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10757 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10759 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10760 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10761 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10762 (ia64_hard_regno_mode_ok): New function.
10763 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10764 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10765 (iq2000_hard_regno_mode_ok): New function.
10766 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10767 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10768 (lm32_hard_regno_mode_ok): New function.
10769 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10770 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10771 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10772 instead of HARD_REGNO_MODE_OK.
10773 (m32c_hard_regno_ok): Rename to...
10774 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10775 (m32c_cannot_change_mode_class): Update accordingly.
10776 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10777 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10778 (m32r_mode_class): Delete.
10779 (HARD_REGNO_MODE_OK): Delete.
10780 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10781 (m32r_hard_regno_mode_ok): Rename to...
10782 (m32r_hard_regno_modes): ...this.
10783 (m32r_mode_class): Make static.
10784 (m32r_hard_regno_mode_ok): New function.
10785 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10786 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10787 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10788 (m68k_hard_regno_mode_ok): Make static.
10789 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10790 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10791 (mcore_hard_regno_mode_ok): New function.
10792 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10793 (HARD_REGNO_MODE_OK): Delete.
10794 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10796 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10797 (microblaze_hard_regno_mode_ok): New function.
10798 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10799 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10800 (mips_hard_regno_mode_ok): Delete.
10801 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10802 (mips_hard_regno_mode_ok_p): ...this and make static.
10803 (mips_hard_regno_mode_ok_p): Rename to...
10804 (mips_hard_regno_mode_ok_uncached): ...this.
10805 (mips_hard_regno_mode_ok): New function.
10806 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10807 of HARD_REGNO_MODE_OK.
10808 (mips_option_override): Update after above name changes.
10809 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10810 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10811 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10812 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10813 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10814 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10815 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10816 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10817 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10818 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10819 (msp430_hard_regno_mode_ok): Make static and return a bool.
10820 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10821 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10822 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10824 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10825 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10826 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10827 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10828 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10829 (PA_HARD_REGNO_MODE_OK): ...this
10830 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10831 (PA_HARD_REGNO_MODE_OK): ...this.
10832 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10833 (pa_hard_regno_mode_ok): New function.
10834 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10835 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10836 (pdp11_hard_regno_mode_ok): New function.
10837 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10838 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10840 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10842 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10843 (rs6000_hard_regno_mode_ok): Rename to...
10844 (rs6000_hard_regno_mode_ok_uncached): ...this.
10845 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10846 (rs6000_hard_regno_mode_ok): New function.
10847 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10848 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10849 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10850 (riscv_hard_regno_mode_ok): ...this and make static.
10851 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10852 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10853 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10854 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10855 (rl78_hard_regno_mode_ok): Make static and return bool.
10856 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10857 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10859 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10860 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10861 (rs6000_hard_regno_mode_ok): Rename to...
10862 (rs6000_hard_regno_mode_ok_uncached): ...this.
10863 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10864 (rs6000_hard_regno_mode_ok): New function.
10865 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10866 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10867 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10868 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10869 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10870 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10871 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10872 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10873 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10874 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10875 (sh_hard_regno_mode_ok): Make static.
10876 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10877 instead of HARD_REGNO_MODE_OK.
10878 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10879 (sparc_mode_class): Delete.
10880 (HARD_REGNO_MODE_OK): Delete.
10881 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10882 (hard_regno_mode_classes): Make static.
10883 (sparc_mode_class): Likewise.
10884 (sparc_hard_regno_mode_ok): New function.
10885 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10886 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10887 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10889 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10890 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10891 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10892 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10893 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10894 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10895 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10896 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10897 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10898 (visium_hard_regno_mode_ok): New function.
10899 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10900 instead of HARD_REGNO_MODE_OK.
10901 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10902 (HARD_REGNO_MODE_OK): Delete.
10903 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10904 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10905 (xtensa_option_override): Update accordingly.
10906 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10907 (xtensa_hard_regno_mode_ok): New function.
10908 * system.h (HARD_REGNO_MODE_OK): Poison.
10910 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10911 Alan Hayward <alan.hayward@arm.com>
10912 David Sherwood <david.sherwood@arm.com>
10914 * target.def (hard_regno_call_part_clobbered): New hook.
10915 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10916 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10917 * doc/tm.texi: Regenerate.
10918 * hooks.h (hook_bool_uint_mode_false): Declare.
10919 * hooks.c (hook_bool_uint_mode_false): New function.
10920 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10921 * cselib.c (cselib_process_insn): Use
10922 targetm.hard_regno_call_part_clobbered instead of
10923 HARD_REGNO_CALL_PART_CLOBBERED.
10924 * ira-conflicts.c (ira_build_conflicts): Likewise.
10925 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10926 * lra-constraints.c (need_for_call_save_p): Likewise.
10927 * lra-lives.c: Include target.h.
10928 (check_pseudos_live_through_calls): Use
10929 targetm.hard_regno_call_part_clobbered instead of
10930 HARD_REGNO_CALL_PART_CLOBBERED.
10931 * regcprop.c: Include target.h.
10932 (copyprop_hardreg_forward_1): Use
10933 targetm.hard_regno_call_part_clobbered instead of
10934 HARD_REGNO_CALL_PART_CLOBBERED.
10935 * reginfo.c (choose_hard_reg_mode): Likewise.
10936 * regrename.c (check_new_reg_p): Likewise.
10937 * reload.c (find_equiv_reg): Likewise.
10938 * reload1.c (emit_reload_insns): Likewise.
10939 * sched-deps.c (deps_analyze_insn): Likewise.
10940 * sel-sched.c (init_regs_for_mode): Likewise.
10941 (mark_unavailable_hard_regs): Likewise.
10942 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10943 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10944 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10946 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10947 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10948 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10950 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10952 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10953 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10954 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10956 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10957 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10958 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10960 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10961 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10963 * config/powerpcspe/powerpcspe.c
10964 (rs6000_hard_regno_call_part_clobbered): New function.
10965 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10966 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10967 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10969 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10970 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10971 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10973 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10974 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10975 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10977 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10978 Alan Hayward <alan.hayward@arm.com>
10979 David Sherwood <david.sherwood@arm.com>
10981 * rtl.h (subreg_memory_offset): Declare.
10982 * emit-rtl.c (subreg_memory_offset): New function.
10983 * expmed.c (store_bit_field_1): Use it.
10984 * expr.c (undefined_operand_subword_p): Likewise.
10985 * simplify-rtx.c (simplify_subreg): Likewise.
10987 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10989 PR rtl-optimization/57448
10992 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10993 using atomic_load pattern.
10994 (expand_atomic_store): Likewise.
10996 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10999 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11000 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11001 handling. Use replace_call_with_value with NULL instead of
11002 gsi_replace, unlink_stmt_vdef and release_defs.
11004 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11007 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11009 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11012 * rtl.h (emit_library_call_value_1): Declare.
11013 (emit_library_call): Replace declaration with a series of overloads.
11014 Remove the parameter count argument.
11015 (emit_library_call_value): Likewise.
11016 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11017 with an "rtx_mode_t *".
11018 (emit_library_call_value): Delete.
11019 (emit_library_call): Likewise.
11020 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11021 (asan_emit_allocas_unpoison): Likewise.
11022 * builtins.c (expand_builtin_powi): Likewise.
11023 (expand_asan_emit_allocas_unpoison): Likewise.
11024 * cfgexpand.c (expand_main_function): Likewise.
11025 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11026 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11027 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11028 * config/arm/arm.c (arm_trampoline_init): Likewise.
11029 (arm_call_tls_get_addr): Likewise.
11030 (arm_expand_divmod_libfunc): Likewise.
11031 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11032 (smulsi3_highpart): Likewise.
11033 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11034 (c6x_expand_compare): Likewise.
11035 (c6x_expand_movmem): Likewise.
11036 * config/frv/frv.c (frv_trampoline_init): Likewise.
11037 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11038 (ix86_expand_divmod_libfunc): Likewise.
11039 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11040 (ia64_expand_compare): Likewise.
11041 (ia64_profile_hook): Likewise.
11042 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11043 (nonlocal_goto): Likewise.
11044 (restore_stack_nonlocal): Likewise.
11045 * config/m32r/m32r.c (block_move_call): Likewise.
11046 (m32r_trampoline_init): Likewise.
11047 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11048 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11049 (m68k_call_m68k_read_tp): Likewise.
11050 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11051 (microblaze_expand_divide): Likewise.
11052 * config/mips/mips.h (mips_args): Likewise.
11053 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11054 (MIPS_ICACHE_SYNC): Likewise.
11055 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11056 (nios2_trampoline_init): Likewise.
11057 * config/pa/pa.c (hppa_tls_call): Likewise.
11058 (pa_trampoline_init): Likewise.
11059 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11060 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11061 (expand_strn_compare): Likewise.
11062 (rs6000_generate_compare): Likewise.
11063 (rs6000_expand_float128_convert): Likewise.
11064 (output_profile_hook): Likewise.
11065 (rs6000_trampoline_init): Likewise.
11066 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11067 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11068 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11069 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11070 (rs6000_generate_compare): Likewise.
11071 (rs6000_expand_float128_convert): Likewise.
11072 (output_profile_hook): Likewise.
11073 (rs6000_trampoline_init): Likewise.
11074 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11075 * config/sh/sh.c (sh_trampoline_init): Likewise.
11076 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11077 (sparc_emit_float_lib_cmp): Likewise.
11078 (sparc32_initialize_trampoline): Likewise.
11079 (sparc64_initialize_trampoline): Likewise.
11080 (sparc_profile_hook): Likewise.
11081 * config/spu/spu.c (ea_load_store): Likewise.
11082 * config/spu/spu.md (floatunssidf2): Likewise.
11083 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11084 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11085 * config/visium/visium.c (expand_block_move_4): Likewise.
11086 (expand_block_move_2): Likewise.
11087 (expand_block_move_1): Likewise.
11088 (expand_block_set_4): Likewise.
11089 (expand_block_set_2): Likewise.
11090 (expand_block_set_1): Likewise.
11091 (visium_trampoline_init): Likewise.
11092 (visium_profile_hook): Likewise.
11093 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11094 (xtensa_setup_frame_addresses): Likewise.
11095 (xtensa_trampoline_init): Likewise.
11096 * except.c (sjlj_emit_function_enter): Likewise.
11097 (sjlj_emit_function_exit): Likewise.
11098 * explow.c (allocate_dynamic_stack_space): Likewise.
11099 (probe_stack_range): Likewise.
11100 * expr.c (convert_mode_scalar): Likewise.
11101 * optabs.c (expand_binop): Likewise.
11102 (expand_twoval_binop_libfunc): Likewise.
11103 (expand_unop): Likewise.
11104 (prepare_cmp_insn): Likewise.
11105 (prepare_float_lib_cmp): Likewise.
11106 (expand_float): Likewise.
11107 (expand_fix): Likewise.
11108 (expand_fixed_convert): Likewise.
11109 (maybe_emit_sync_lock_test_and_set): Likewise.
11110 (expand_atomic_compare_and_swap): Likewise.
11111 (expand_mem_thread_fence): Likewise.
11112 (expand_atomic_fetch_op): Likewise.
11114 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11116 * doc/generic.texi (OpenACC): Adjust URL.
11117 * doc/invoke.texi (C Dialect Options): Ditto.
11119 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11121 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11122 predicate for operand 1. Add (m,<S>) constraint.
11123 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11124 Prevent memory operand 1 with register operand 2.
11126 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11128 PR rtl-optimization/82024
11129 * combine.c (try_combine): If the combination result is a PARALLEL,
11130 and we only need to retain the SET in there that would be placed
11131 at I2, check that we can place that at I3 instead, before doing so.
11133 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11136 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11138 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11139 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11142 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11143 Jeff Law <law@redhat.com>
11145 * varasm.c (bss_initializer_p): Do not put constants into .bss
11146 (categorize_decl_for_section): Handle bss_initializer_p returning
11147 false when DECL_INITIAL is NULL.
11149 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11152 * config/s390/s390.c (s390_can_inline_p): New function.
11154 2017-09-01 Jeff Law <law@redhat.com>
11156 PR tree-optimization/82052
11157 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11158 Always initialize the returned slot after a hash table miss
11159 when INSERT is true.
11161 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11163 * config/s390/s390.md (mem_signal_fence): Remove.
11164 * doc/md.texi (mem_signal_fence): Remove.
11165 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11167 * target-insns.def (mem_signal_fence): Remove.
11169 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11172 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11175 * asan.c (create_odr_indicator): Strip name encoding from assembler
11176 name before appending it after __odr_asan_.
11178 2017-09-01 Martin Liska <mliska@suse.cz>
11180 PR tree-optimization/82059
11181 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11182 frequency only when an edge is redirected.
11184 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11186 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11187 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11188 (arc_conditional_register_usage): Remove ARC600 lp_count
11190 (arc_file_start): Emit Tag_ARC_CPU_variation.
11191 (arc_can_use_doloop_p): New conditions to use ZOLs.
11192 (hwloop_fail): New function.
11193 (hwloop_optimize): Likewise.
11194 (hwloop_pattern_reg): Likewise.
11195 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11196 (arc_reorg_loops): New function, calls reorg_loops.
11197 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11198 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11200 (arc_loop_hazard): Remove function, functionality moved into
11202 (arc_hazard): Remove arc_loop_hazard call.
11203 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11204 into hwloop_optimize.
11205 (arc_label_align): Remove ZOL handling.
11206 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11207 * config/arc/arc.md (doloop_begin): Remove pattern.
11208 (doloop_begin_i): Likewise.
11209 (doloop_end_i): Likewise.
11210 (doloop_fallback): Likewise.
11211 (doloop_fallback_m): Likewise.
11212 (doloop_end): Reimplement expand.
11213 (arc_lp): New pattern for LP instruction.
11214 (loop_end): New pattern.
11215 (loop_fail): Likewise.
11216 (decrement_and_branch_until_zero): Likewise.
11217 * config/arc/arc.opt (mlpc-width): New option.
11218 * doc/invoke.texi (mlpc-width): Document option.
11220 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11222 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11223 (arc_ccfsm_advance): Fix checking for delay slots.
11224 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11226 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11228 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11230 (movhi_insn): Update store instruction constraint which are saving
11231 6-bit short immediates.
11232 (movsi_insn): Consider also short scaled load operations.
11233 (zero_extendhisi2_i): Use Usd constraint instead of T.
11234 (extendhisi2_i): Add q constraint.
11235 (arc_clzsi2): Add type and length attributes.
11236 (arc_ctzsi2): Likewise.
11237 * config/arc/constraints.md (Usc): Update constraint, the
11238 assembler can parse two relocations for a single instruction.
11240 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11242 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11243 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11245 2017-08-31 Olivier Hainque <hainque@adacore.com>
11247 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11248 match as powerpc-wrs-vxworks*.
11250 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11252 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11253 register constraint for by-element operand.
11254 (aarch64_mls_elt_merge<mode>): Likewise.
11256 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11258 * config/arc/arc.c (arc_can_follow_jump): Check for short
11261 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11263 * config.gcc: Use g.opt for arc.
11264 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11265 functionality moved to ...
11266 (legitimate_scaled_address_p): New function, ...here.
11267 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11268 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11269 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11271 (arc_override_options): Handle G option.
11272 (arc_output_pic_addr_const): Correct function definition.
11273 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11274 (arc_decl_anon_ns_mem_p): Delete.
11275 (arc_in_small_data_p): Overhaul this function to take into
11276 consideration the value given via G option.
11277 (arc_rewrite_small_data_1): Renamed and corrected old
11278 arc_rewrite_small_data function.
11279 (arc_rewrite_small_data): New function.
11280 (small_data_pattern): Don't use pic_offset_table_rtx.
11281 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11282 * config/arc/simdext.md (movmisalignv2hi): Use
11283 prepare_move_operands function.
11285 (movmisalign*): Likewise.
11286 * doc/invoke.texi (ARC options): Document -G option.
11288 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11290 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11292 * config/arc/arc.c (arc_print_operand): Output scalled address for
11293 sdata whenever is possible.
11294 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11295 load/stores are available.
11296 (compact_sda_memory_operand): Check for the alignment required by
11297 code density instructions.
11298 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11300 * config/arc/constraints.md (Usd): Update constraint.
11301 (Us0): New constraint.
11302 (Usc): Update constraint.
11304 2017-08-31 Richard Biener <rguenther@suse.de>
11306 PR middle-end/82054
11307 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11308 function only once.
11310 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11312 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11313 Resize type_signature.
11315 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11316 Alan Hayward <alan.hayward@arm.com>
11317 David Sherwood <david.sherwood@arm.com>
11319 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11320 subregs whose inner modes can be stored in GPRs.
11321 (aarch64_classify_index): Likewise.
11323 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11324 Alan Hayward <alan.hayward@arm.com>
11325 David Sherwood <david.sherwood@arm.com>
11327 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11328 (V_INT_EQUIV): ...this.
11329 (v_cmp_result): Rename to...
11330 (v_int_equiv): ...this.
11331 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11332 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11333 (copysign<mode>3): Likewise.
11334 (aarch64_simd_bsl<mode>_internal): Likewise.
11335 (aarch64_simd_bsl<mode>): Likewise.
11336 (vec_cmp<mode><mode>): Likewise.
11337 (vcond<mode><mode>): Likewise.
11338 (vcond<v_cmp_mixed><mode>): Likewise.
11339 (vcondu<mode><v_cmp_mixed>): Likewise.
11340 (aarch64_cm<optab><mode>): Likewise.
11341 (aarch64_cmtst<mode>): Likewise.
11342 (aarch64_fac<optab><mode>): Likewise.
11343 (vec_perm_const<mode>): Likewise.
11344 (vcond_mask_<mode><v_cmp_result>): Rename to...
11345 (vcond_mask_<mode><v_int_equiv>): ...this.
11346 (vec_cmp<mode><v_cmp_result>): Rename to...
11347 (vec_cmp<mode><v_int_equiv>): ...this.
11349 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11350 Alan Hayward <alan.hayward@arm.com>
11351 David Sherwood <david.sherwood@arm.com>
11353 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11355 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11356 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11357 (UNSPEC_LD4_DREG): New unspecs.
11358 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11359 (aarch64_ld2<mode>_dreg_be): Replace with...
11360 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11362 (aarch64_ld3<mode>_dreg_le)
11363 (aarch64_ld3<mode>_dreg_be): Replace with...
11364 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11366 (aarch64_ld4<mode>_dreg_le)
11367 (aarch64_ld4<mode>_dreg_be): Replace with...
11368 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11371 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11373 PR tree-optimization/81987
11374 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11375 insert an initializer in a location not dominated by the stride
11378 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11380 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11381 on the entire header of the finally block in the fallthru case.
11383 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11385 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11387 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11389 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11390 rs6000_emit_move_from_cr and call renamed function.
11391 (rs6000_emit_prologue): Call renamed functions.
11392 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11393 movesi_from_cr, remove volatile CRs.
11395 2017-08-30 Jon Beniston <jon@beniston.com>
11396 Richard Biener <rguenther@suse.de>
11398 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11399 of VECTOR_MODE_P check.
11400 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11401 element vector types.
11403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11405 * df.h (df_read_modify_subreg_p): Remove in favor of...
11406 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11407 const_rtx instead of an rtx.
11408 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11409 * df-problems.c (df_word_lr_mark_ref): Likewise.
11410 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11411 (mark_pseudo_reg_dead): Likewise.
11412 (mark_ref_dead): Likewise.
11413 * reginfo.c (init_subregs_of_mode): Likewise.
11414 * sched-deps.c (sched_analyze_1): Likewise.
11415 * df-scan.c (df_def_record_1): Likewise.
11416 (df_uses_record): Likewise.
11417 (df_read_modify_subreg_p): Remove in favor of...
11418 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11419 const_rtx instead of an rtx.
11421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11422 Alan Hayward <alan.hayward@arm.com>
11423 David Sherwood <david.sherwood@arm.com>
11425 * rtl.h (partial_subreg_p): New function.
11426 * caller-save.c (save_call_clobbered_regs): Use it.
11427 * calls.c (expand_call): Likewise.
11428 * combine.c (combinable_i3pat): Likewise.
11429 (simplify_set): Likewise.
11430 (make_extraction): Likewise.
11431 (make_compound_operation_int): Likewise.
11432 (gen_lowpart_or_truncate): Likewise.
11433 (force_to_mode): Likewise.
11434 (make_field_assignment): Likewise.
11435 (reg_truncated_to_mode): Likewise.
11436 (record_truncated_value): Likewise.
11437 (move_deaths): Likewise.
11438 * cse.c (record_jump_cond): Likewise.
11439 (cse_insn): Likewise.
11440 * cselib.c (cselib_lookup_1): Likewise.
11441 * expmed.c (extract_bit_field_using_extv): Likewise.
11442 * function.c (assign_parm_setup_reg): Likewise.
11443 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11444 * ira-build.c (create_insn_allocnos): Likewise.
11445 * lra-coalesce.c (merge_pseudos): Likewise.
11446 * lra-constraints.c (match_reload): Likewise.
11447 (simplify_operand_subreg): Likewise.
11448 (curr_insn_transform): Likewise.
11449 * lra-lives.c (process_bb_lives): Likewise.
11450 * lra.c (new_insn_reg): Likewise.
11451 (lra_substitute_pseudo): Likewise.
11452 * regcprop.c (mode_change_ok): Likewise.
11453 (maybe_mode_change): Likewise.
11454 (copyprop_hardreg_forward_1): Likewise.
11455 * reload.c (push_reload): Likewise.
11456 (find_reloads): Likewise.
11457 (find_reloads_subreg_address): Likewise.
11458 * reload1.c (alter_reg): Likewise.
11459 (eliminate_regs_1): Likewise.
11460 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11462 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11464 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11465 back to if statements, including unpack.
11467 2017-08-30 Martin Liska <mliska@suse.cz>
11469 PR inline-asm/82001
11470 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11472 (func_checker::compare_asm_inputs_outputs): ... this function.
11473 (func_checker::compare_gimple_asm): Use the function to compare
11474 also ASM constrains.
11475 * ipa-icf-gimple.h: Rename the function.
11477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11478 Alan Hayward <alan.hayward@arm.com>
11479 David Sherwood <david.sherwood@arm.com>
11481 * coretypes.h (complex_mode): New type.
11482 * gdbhooks.py (build_pretty_printer): Handle it.
11483 * machmode.h (complex_mode): New class.
11484 (complex_mode::includes_p): New function.
11485 (is_complex_int_mode): Likewise.
11486 (is_complex_float_mode): Likewise.
11487 * genmodes.c (get_mode_class): Handle complex mode classes.
11488 * function.c (expand_function_end): Use is_complex_int_mode.
11490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11491 Alan Hayward <alan.hayward@arm.com>
11492 David Sherwood <david.sherwood@arm.com>
11494 * coretypes.h (scalar_mode_pod): New typedef.
11495 * gdbhooks.py (build_pretty_printer): Handle it.
11496 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11497 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11498 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11499 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11500 as_a <scalar_mode>.
11502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11503 Alan Hayward <alan.hayward@arm.com>
11504 David Sherwood <david.sherwood@arm.com>
11506 * machmode.h (mode_for_vector): Take a scalar_mode instead
11508 * stor-layout.c (mode_for_vector): Likewise.
11509 * explow.c (promote_mode): Use as_a <scalar_mode>.
11510 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11513 Alan Hayward <alan.hayward@arm.com>
11514 David Sherwood <david.sherwood@arm.com>
11516 * target.def (preferred_simd_mode): Take a scalar_mode
11517 instead of a machine_mode.
11518 * targhooks.h (default_preferred_simd_mode): Likewise.
11519 * targhooks.c (default_preferred_simd_mode): Likewise.
11520 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11521 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11522 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11523 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11524 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11525 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11526 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11527 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11529 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11530 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11531 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11532 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11533 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11534 * doc/tm.texi: Regenerate.
11535 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11539 Alan Hayward <alan.hayward@arm.com>
11540 David Sherwood <david.sherwood@arm.com>
11542 * target.def (scalar_mode_supported_p): Take a scalar_mode
11543 instead of a machine_mode.
11544 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11545 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11546 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11547 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11548 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11549 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11550 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11551 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11552 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11553 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11554 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11555 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11556 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11557 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11559 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11560 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11561 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11562 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11563 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11565 * doc/tm.texi: Regenerate.
11567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11568 Alan Hayward <alan.hayward@arm.com>
11569 David Sherwood <david.sherwood@arm.com>
11571 * coretypes.h (opt_scalar_mode): New typedef.
11572 * gdbhooks.py (build_pretty_printers): Handle it.
11573 * machmode.h (mode_iterator::get_2xwider): Add overload for
11575 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11577 * expr.c (convert_mode_scalar): Likewise.
11578 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11579 * optabs.c (expand_float): Likewise.
11580 (expand_fix): Likewise.
11581 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11583 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11584 Alan Hayward <alan.hayward@arm.com>
11585 David Sherwood <david.sherwood@arm.com>
11587 * optabs.c (expand_float): Explicitly check for scalars before
11588 using a branching expansion.
11589 (expand_fix): Likewise.
11591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11592 Alan Hayward <alan.hayward@arm.com>
11593 David Sherwood <david.sherwood@arm.com>
11595 * expr.c (convert_mode): Split scalar handling out into...
11596 (convert_mode_scalar): ...this new function. Treat the modes
11599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11600 Alan Hayward <alan.hayward@arm.com>
11601 David Sherwood <david.sherwood@arm.com>
11603 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11605 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11608 Alan Hayward <alan.hayward@arm.com>
11609 David Sherwood <david.sherwood@arm.com>
11611 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11612 rather than a machine_mode.
11613 (fixed_from_string): Likewise.
11614 (fixed_convert): Likewise.
11615 (fixed_convert_from_int): Likewise.
11616 (fixed_convert_from_real): Likewise.
11617 (real_convert_from_fixed): Likewise.
11618 * fixed-value.c (fixed_from_double_int): Likewise.
11619 (fixed_from_string): Likewise.
11620 (fixed_convert): Likewise.
11621 (fixed_convert_from_int): Likewise.
11622 (fixed_convert_from_real): Likewise.
11623 (real_convert_from_fixed): Likewise.
11624 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11627 Alan Hayward <alan.hayward@arm.com>
11628 David Sherwood <david.sherwood@arm.com>
11630 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11631 of separate mode class checks. Do not allow vector modes here.
11632 (immed_wide_int_const): Use as_a <scalar_mode>.
11633 * explow.c (trunc_int_for_mode): Likewise.
11634 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11635 (wi::shwi): Likewise.
11636 (wi::min_value): Likewise.
11637 (wi::max_value): Likewise.
11638 * dwarf2out.c (loc_descriptor): Likewise.
11639 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11640 for CONST_WIDE_INT.
11642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11643 Alan Hayward <alan.hayward@arm.com>
11644 David Sherwood <david.sherwood@arm.com>
11646 * tree.h (SCALAR_TYPE_MODE): New macro.
11647 * expr.c (expand_expr_addr_expr_1): Use it.
11648 (expand_expr_real_2): Likewise.
11649 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11650 (fold_convert_const_fixed_from_int): Likewise.
11651 (fold_convert_const_fixed_from_real): Likewise.
11652 (native_encode_fixed): Likewise
11653 (native_encode_complex): Likewise
11654 (native_encode_vector): Likewise.
11655 (native_interpret_fixed): Likewise.
11656 (native_interpret_real): Likewise.
11657 (native_interpret_complex): Likewise.
11658 (native_interpret_vector): Likewise.
11659 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11660 (simd_clone_adjust_argument_types): Likewise.
11661 (simd_clone_init_simd_arrays): Likewise.
11662 (simd_clone_adjust): Likewise.
11663 * stor-layout.c (layout_type): Likewise.
11664 * tree.c (build_minus_one_cst): Likewise.
11665 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11666 * tree-inline.c (estimate_move_cost): Likewise.
11667 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11668 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11669 (vectorizable_reduction): Likewise.
11670 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11671 (vect_recog_mixed_size_cond_pattern): Likewise.
11672 (check_bool_pattern): Likewise.
11673 (adjust_bool_pattern): Likewise.
11674 (search_type_for_mask_1): Likewise.
11675 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11676 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11677 (vectorizable_load): Likewise.
11678 (vectorizable_store): Likewise.
11679 * ubsan.c (ubsan_encode_value): Likewise.
11680 * varasm.c (output_constant): Likewise.
11682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11683 Alan Hayward <alan.hayward@arm.com>
11684 David Sherwood <david.sherwood@arm.com>
11686 * coretypes.h (scalar_mode): New class.
11687 * machmode.h (scalar_mode): Likewise.
11688 (scalar_mode::includes_p): New function.
11689 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11690 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11691 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11692 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11693 * expmed.c (store_bit_field_1): Likewise.
11694 (extract_bit_field_1): Likewise.
11695 * expr.c (write_complex_part): Likewise.
11696 (read_complex_part): Likewise.
11697 (emit_move_complex_push): Likewise.
11698 (expand_expr_real_2): Likewise.
11699 * function.c (assign_parm_setup_reg): Likewise.
11700 (assign_parms_unsplit_complex): Likewise.
11701 * optabs.c (expand_binop): Likewise.
11702 * rtlanal.c (subreg_get_info): Likewise.
11703 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11704 * varasm.c (output_constant_pool_2): Likewise.
11706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11707 Alan Hayward <alan.hayward@arm.com>
11708 David Sherwood <david.sherwood@arm.com>
11710 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11712 (expmed_mult_highpart_optab): Likewise.
11713 (expmed_mult_highpart): Likewise.
11715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11716 Alan Hayward <alan.hayward@arm.com>
11717 David Sherwood <david.sherwood@arm.com>
11719 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11720 instead of a machine_mode.
11721 (builtin_memset_read_str): Likewise.
11722 * builtins.c (c_readstr): Likewise.
11723 (builtin_memcpy_read_str): Likewise.
11724 (builtin_strncpy_read_str): Likewise.
11725 (builtin_memset_read_str): Likewise.
11726 (builtin_memset_gen_str): Likewise.
11727 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11728 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11729 instead of a machine_mode.
11730 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11732 (make_extraction): Likewise.
11733 (try_widen_shift_mode): Take and return scalar_int_modes instead
11735 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11736 a scalar_int_mode instead of a machine_mode.
11737 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11738 (avr_addr_space_pointer_mode): Likewise.
11739 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11740 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11741 (msp430_unwind_word_mode): Likewise.
11742 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11743 (spu_addr_space_pointer_mode): Likewise.
11744 (spu_addr_space_address_mode): Likewise.
11745 (spu_libgcc_cmp_return_mode): Likewise.
11746 (spu_libgcc_shift_count_mode): Likewise.
11747 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11748 (rl78_addr_space_pointer_mode): Likewise.
11749 (fl78_unwind_word_mode): Likewise.
11750 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11752 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11753 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11754 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11755 (mips_valid_pointer_mode): Likewise.
11756 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11757 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11758 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11760 (ft32_addr_space_address_mode): Likewise.
11761 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11762 scalar_int_mode instead of a machine_mode.
11763 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11765 (m32c_addr_space_address_mode): Likewise.
11766 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11767 (rs6000_eh_return_filter_mode): Likewise.
11768 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11769 (rs6000_eh_return_filter_mode): Likewise.
11770 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11771 (s390_libgcc_shift_count_mode): Likewise.
11772 (s390_unwind_word_mode): Likewise.
11773 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11775 * target.def (mode_rep_extended): Likewise.
11776 (valid_pointer_mode): Likewise.
11777 (addr_space.valid_pointer_mode): Likewise.
11778 (eh_return_filter_mode): Return a scalar_int_mode rather than
11780 (libgcc_cmp_return_mode): Likewise.
11781 (libgcc_shift_count_mode): Likewise.
11782 (unwind_word_mode): Likewise.
11783 (addr_space.pointer_mode): Likewise.
11784 (addr_space.address_mode): Likewise.
11785 * doc/tm.texi: Regenerate.
11786 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11788 (do_jump): Use scalar_int_mode for local variables.
11789 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11790 rather than a machine_mode.
11791 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11792 (scompare_loc_descriptor_wide): Likewise.
11793 (scompare_loc_descriptor_narrow): Likewise.
11794 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11796 * except.c (sjlj_emit_dispatch_table): Likewise.
11797 (expand_builtin_eh_copy_values): Likewise.
11798 * explow.c (convert_memory_address_addr_space_1): Likewise.
11799 Take a scalar_int_mode rather than a machine_mode.
11800 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11801 than a machine_mode.
11802 (memory_address_addr_space): Use scalar_int_mode for local variables.
11803 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11804 rather than a machine_mode.
11805 * expmed.c (mask_rtx): Likewise.
11806 (init_expmed_one_conv): Likewise.
11807 (expand_mult_highpart_adjust): Likewise.
11808 (extract_high_half): Likewise.
11809 (expmed_mult_highpart_optab): Likewise.
11810 (expmed_mult_highpart): Likewise.
11811 (expand_smod_pow2): Likewise.
11812 (expand_sdiv_pow2): Likewise.
11813 (emit_store_flag_int): Likewise.
11814 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11816 (extract_low_bits): Likewise.
11817 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11819 * expr.c (pieces_addr::adjust): Likewise.
11820 (can_store_by_pieces): Likewise.
11821 (store_by_pieces): Likewise.
11822 (clear_by_pieces_1): Likewise.
11823 (expand_expr_addr_expr_1): Likewise.
11824 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11825 (expand_expr_real_1): Likewise.
11826 (try_casesi): Likewise.
11827 * final.c (shorten_branches): Likewise.
11828 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11829 type of "mode" to machine_mode.
11830 * internal-fn.c (expand_arith_overflow_result_store): Take a
11831 scalar_int_mode rather than a machine_mode.
11832 (expand_mul_overflow): Use scalar_int_mode for local variables.
11833 * loop-doloop.c (doloop_modify): Likewise.
11834 (doloop_optimize): Likewise.
11835 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11836 than a machine_mode.
11837 (expand_doubleword_shift_condmove): Likewise.
11838 (expand_doubleword_shift): Likewise.
11839 (expand_doubleword_clz): Likewise.
11840 (expand_doubleword_popcount): Likewise.
11841 (expand_doubleword_parity): Likewise.
11842 (expand_absneg_bit): Use scalar_int_mode for local variables.
11843 (prepare_float_lib_cmp): Likewise.
11844 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11845 rather than a machine_mode.
11846 (convert_memory_address_addr_space): Likewise.
11847 (get_mode_bounds): Likewise.
11848 (get_address_mode): Return a scalar_int_mode rather than a
11850 * rtlanal.c (get_address_mode): Likewise.
11851 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11852 than a machine_mode.
11853 * targhooks.c (default_mode_rep_extended): Likewise.
11854 (default_valid_pointer_mode): Likewise.
11855 (default_addr_space_valid_pointer_mode): Likewise.
11856 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11857 than a machine_mode.
11858 (default_libgcc_cmp_return_mode): Likewise.
11859 (default_libgcc_shift_count_mode): Likewise.
11860 (default_unwind_word_mode): Likewise.
11861 (default_addr_space_pointer_mode): Likewise.
11862 (default_addr_space_address_mode): Likewise.
11863 * targhooks.h (default_eh_return_filter_mode): Likewise.
11864 (default_libgcc_cmp_return_mode): Likewise.
11865 (default_libgcc_shift_count_mode): Likewise.
11866 (default_unwind_word_mode): Likewise.
11867 (default_addr_space_pointer_mode): Likewise.
11868 (default_addr_space_address_mode): Likewise.
11869 (default_mode_rep_extended): Take a scalar_int_mode rather than
11871 (default_valid_pointer_mode): Likewise.
11872 (default_addr_space_valid_pointer_mode): Likewise.
11873 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11875 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11876 rather than a machine_mode.
11877 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11878 for local variables.
11879 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11880 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11881 than a machine_mode.
11883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11884 Alan Hayward <alan.hayward@arm.com>
11885 David Sherwood <david.sherwood@arm.com>
11887 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11888 the mode argument to scalar_int_mode.
11889 (do_jump_by_parts_zero_rtx): Likewise.
11890 (do_jump_by_parts_equality_rtx): Likewise.
11891 (do_jump_by_parts_greater): Take a mode argument.
11892 (do_jump_by_parts_equality): Likewise.
11893 (do_jump_1): Update calls accordingly.
11895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11896 Alan Hayward <alan.hayward@arm.com>
11897 David Sherwood <david.sherwood@arm.com>
11899 * is-a.h (safe_dyn_cast): New function.
11900 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11901 (jump_table_for_label): Likewise.
11902 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11903 instead of an rtx_insn *.
11904 (shorten_branches): Use dyn_cast instead of LABEL_P and
11905 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11906 rtx_jump_table_data::get_data_mode.
11907 (final_scan_insn): Likewise.
11909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11910 Alan Hayward <alan.hayward@arm.com>
11911 David Sherwood <david.sherwood@arm.com>
11913 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11914 trying to combine a full-register integer set with a subreg
11917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11918 Alan Hayward <alan.hayward@arm.com>
11919 David Sherwood <david.sherwood@arm.com>
11921 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11922 that is always either address_mode or pointer_mode.
11924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11925 Alan Hayward <alan.hayward@arm.com>
11926 David Sherwood <david.sherwood@arm.com>
11928 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11929 when the two are known to be equal.
11931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11932 Alan Hayward <alan.hayward@arm.com>
11933 David Sherwood <david.sherwood@arm.com>
11935 * simplify-rtx.c (simplify_const_unary_operation): Use
11936 is_a <scalar_int_mode> instead of checking for a nonzero
11937 precision. Forcibly convert op_mode to a scalar_int_mode
11938 in that case. More clearly differentiate the operand and
11939 result modes and use the former when deciding what the value
11940 of a count-bits operation should be. Use is_int_mode instead
11941 of checking for a MODE_INT. Remove redundant check for whether
11942 this mode has a zero precision.
11944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11945 Alan Hayward <alan.hayward@arm.com>
11946 David Sherwood <david.sherwood@arm.com>
11948 * optabs.c (widen_leading): Change the type of the mode argument
11949 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11950 (widen_bswap): Likewise.
11951 (expand_parity): Likewise.
11952 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11953 (expand_ffs): Likewise.
11954 (epand_unop): Check for scalar integer modes before calling the
11957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11958 Alan Hayward <alan.hayward@arm.com>
11959 David Sherwood <david.sherwood@arm.com>
11961 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11963 (expand_expr_real_1): Update call accordingly.
11965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11966 Alan Hayward <alan.hayward@arm.com>
11967 David Sherwood <david.sherwood@arm.com>
11969 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11970 value_mode arguments. Use scalar_int_mode internally.
11971 (store_bit_field_1): Rename the new integer mode from imode
11972 to op0_mode and use it instead of GET_MODE (op0). Update calls
11973 to store_split_bit_field, store_bit_field_using_insv and
11974 store_fixed_bit_field.
11975 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11976 Use scalar_int_mode internally. Use a bit count rather than a mode
11977 when calculating the largest bit size for get_best_mode.
11978 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11979 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11980 Remove assertion that OP0 has a scalar integer mode.
11981 (store_split_bit_field): Add op0_mode and value_mode arguments.
11982 Update calls to extract_fixed_bit_field.
11983 (extract_bit_field_using_extv): Add an op0_mode argument.
11984 Use scalar_int_mode internally.
11985 (extract_bit_field_1): Rename the new integer mode from imode to
11986 op0_mode and use it instead of GET_MODE (op0). Update calls to
11987 extract_split_bit_field, extract_bit_field_using_extv and
11988 extract_fixed_bit_field.
11989 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11990 to extract_split_bit_field and extract_fixed_bit_field_1.
11991 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11992 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11993 on the target mode.
11994 (extract_split_bit_field): Add an op0_mode argument. Update call
11995 to extract_fixed_bit_field.
11997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11998 Alan Hayward <alan.hayward@arm.com>
11999 David Sherwood <david.sherwood@arm.com>
12001 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12002 * explow.c (hard_function_value): Likewise.
12003 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12004 convert_to_mode call outside the loop.
12005 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12006 for the mode iterator. Require the mode specified by max_pieces
12008 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12010 (copy_blkmode_to_reg): Likewise.
12011 (set_storage_via_setmem): Likewise.
12012 * optabs.c (prepare_cmp_insn): Likewise.
12013 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12014 * stor-layout.c (finish_bitfield_representative): Likewise.
12016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12017 Alan Hayward <alan.hayward@arm.com>
12018 David Sherwood <david.sherwood@arm.com>
12020 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12021 * expr.c (convert_move): Use them.
12022 (convert_modes): Likewise.
12023 (store_expr_with_bounds): Likewise.
12025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12026 Alan Hayward <alan.hayward@arm.com>
12027 David Sherwood <david.sherwood@arm.com>
12029 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12030 parameter for the mode of "x". Remove the "known_x", "known_mode"
12031 and "known_ret" arguments. Change the type of the mode argument
12032 to scalar_int_mode.
12033 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12034 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12035 (reg_num_sign_bit_copies_for_combine): Likewise.
12036 * rtlanal.c (nonzero_bits1): Likewise.
12037 (num_sign_bit_copies1): Likewise.
12038 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12039 (reg_num_sign_bit_copies_general): Likewise.
12040 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12041 (reg_nonzero_bits_general): Likewise.
12043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12044 Alan Hayward <alan.hayward@arm.com>
12045 David Sherwood <david.sherwood@arm.com>
12047 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12048 than in subroutines. Return 1 for non-integer modes.
12049 (cached_num_sign_bit_copies): Change the type of the mode parameter
12050 to scalar_int_mode.
12051 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12052 classes. Handle CONST_INT_P first and then check whether X also
12053 has a scalar integer mode. Check the same thing for inner registers
12054 of a SUBREG and for values that are being extended or truncated.
12056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12057 Alan Hayward <alan.hayward@arm.com>
12058 David Sherwood <david.sherwood@arm.com>
12060 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12061 in subroutines. Return the mode mask for non-integer modes.
12062 (cached_nonzero_bits): Change the type of the mode parameter
12063 to scalar_int_mode.
12064 (nonzero_bits1): Likewise. Remove early exit for other mode
12065 classes. Handle CONST_INT_P first and then check whether X
12066 also has a scalar integer mode.
12068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12069 Alan Hayward <alan.hayward@arm.com>
12070 David Sherwood <david.sherwood@arm.com>
12072 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12073 Return a scalar_int_mode and assert that the size is greater than
12075 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12076 (op_by_pieces_d::op_by_pieces_d): Likewise.
12077 (op_by_pieces_d::run): Likewise.
12078 (can_store_by_pieces): Likewise.
12080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12081 Alan Hayward <alan.hayward@arm.com>
12082 David Sherwood <david.sherwood@arm.com>
12084 * combine.c (extract_left_shift): Add a mode argument and update
12086 (make_compound_operation_int): Change the type of the mode parameter
12087 to scalar_int_mode and update the call to extract_left_shift.
12089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12090 Alan Hayward <alan.hayward@arm.com>
12091 David Sherwood <david.sherwood@arm.com>
12093 * combine.c (simplify_and_const_int): Change the type of the mode
12094 parameter to scalar_int_mode.
12095 (simplify_and_const_int_1): Likewise. Update recursive call.
12097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12098 Alan Hayward <alan.hayward@arm.com>
12099 David Sherwood <david.sherwood@arm.com>
12101 * combine.c (simplify_compare_const): Check that the mode is a
12102 scalar_int_mode (rather than VOIDmode) before testing its
12104 (simplify_comparison): Move COMPARISON_P handling out of the
12105 loop and restrict the latter part of the loop to scalar_int_modes.
12106 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12107 and when considering SUBREG_REGs. Use is_int_mode instead of
12108 checking GET_MODE_CLASS against MODE_INT.
12110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12111 Alan Hayward <alan.hayward@arm.com>
12112 David Sherwood <david.sherwood@arm.com>
12114 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12115 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12116 shift_unit_mode and for modes involved in scalar shifts.
12118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12119 Alan Hayward <alan.hayward@arm.com>
12120 David Sherwood <david.sherwood@arm.com>
12122 * combine.c (force_int_to_mode): New function, split out from...
12123 (force_to_mode): ...here. Keep xmode up-to-date and use it
12124 instead of GET_MODE (x).
12126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12127 Alan Hayward <alan.hayward@arm.com>
12128 David Sherwood <david.sherwood@arm.com>
12130 * optabs-query.h (extraction_insn::struct_mode): Change type to
12131 opt_scalar_int_mode and update comment.
12132 (extraction_insn::field_mode): Change type to scalar_int_mode.
12133 (extraction_insn::pos_mode): Likewise.
12134 * combine.c (make_extraction): Update accordingly.
12135 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12136 (get_optab_extraction_insn): Likewise.
12137 * recog.c (simplify_while_replacing): Likewise.
12138 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12139 parameter to opt_scalar_int_mode.
12141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12142 Alan Hayward <alan.hayward@arm.com>
12143 David Sherwood <david.sherwood@arm.com>
12145 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12146 to a scalar_int_mode instead of a machine_mode.
12147 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12148 (get_best_mode): Return a boolean and use a pointer argument to store
12149 the selected mode. Replace the limit mode parameter with a bit limit.
12150 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12151 for the values returned by bit_field_mode_iterator::next_mode.
12152 (store_bit_field): Update call to get_best_mode.
12153 (store_fixed_bit_field): Likewise.
12154 (extract_fixed_bit_field): Likewise.
12155 * expr.c (optimize_bitfield_assignment_op): Likewise.
12156 * fold-const.c (optimize_bit_field_compare): Likewise.
12157 (fold_truth_andor_1): Likewise.
12158 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12159 Update for new type of m_mode.
12160 (get_best_mode): As above.
12162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12163 Alan Hayward <alan.hayward@arm.com>
12164 David Sherwood <david.sherwood@arm.com>
12166 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12167 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12168 (store_bit_field): Check is_a <scalar_int_mode> before calling
12169 strict_volatile_bitfield_p.
12170 (extract_bit_field): Likewise.
12172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12173 Alan Hayward <alan.hayward@arm.com>
12174 David Sherwood <david.sherwood@arm.com>
12176 * target.def (cstore_mode): Return a scalar_int_mode.
12177 * doc/tm.texi: Regenerate.
12178 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12179 * targhooks.h (default_cstore_mode): Likewise.
12180 * targhooks.c (default_cstore_mode): Likewise, using a forced
12182 * expmed.c (emit_cstore): Expect the target of the cstore to be
12185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12186 Alan Hayward <alan.hayward@arm.com>
12187 David Sherwood <david.sherwood@arm.com>
12189 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12191 (niter_desc): Likewise mode.
12192 (iv_analyze): Add a mode parameter.
12194 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12195 and change its type to scalar_int_mode.
12196 * loop-iv.c: Update commentary at head of file.
12197 (iv_constant): Pass the mode paraeter before the rtx it describes
12198 and change its type to scalar_int_mode. Remove VOIDmode handling.
12199 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12200 (iv_extend): Likewise.
12201 (shorten_into_mode): Likewise.
12202 (iv_add): Use scalar_int_mode.
12203 (iv_mult): Likewise.
12204 (iv_shift): Likewise.
12205 (canonicalize_iv_subregs): Likewise.
12206 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12207 it describes and change its mode to scalar_int_mode. Also change
12208 the type of the returned inner_mode to scalar_int_mode.
12209 (get_biv_step): Likewise, turning outer_mode from a pointer
12210 into a direct parameter. Update call to get_biv_step_1.
12211 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12212 iv_constant and get_biv_step.
12213 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12214 and change its type to scalar_int_mode. Don't initialise iv->mode
12215 to VOIDmode and remove later checks for its still being VOIDmode.
12216 Update calls to iv_analyze_op and iv_analyze_expr. Check
12217 is_a <scalar_int_mode> when changing the mode under consideration.
12218 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12219 Update call to iv_analyze_expr.
12220 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12221 inner register is not also a scalar_int_mode. Update call to
12223 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12224 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12225 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12226 separate mode class checks. Update calls to iv_analyze. Remove
12227 fix-up of VOIDmodes after iv_analyze_biv.
12228 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12229 don't have a scalar_int_mode. Update call to biv_p.
12231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12232 Alan Hayward <alan.hayward@arm.com>
12233 David Sherwood <david.sherwood@arm.com>
12235 * cfgexpand.c (convert_debug_memory_address): Use
12236 as_a <scalar_int_mode>.
12237 * combine.c (expand_compound_operation): Likewise.
12238 (make_extraction): Likewise.
12239 (change_zero_ext): Likewise.
12240 (simplify_comparison): Likewise.
12241 * cse.c (cse_insn): Likewise.
12242 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12243 (mem_loc_descriptor): Likewise.
12244 (loc_descriptor): Likewise.
12245 * expmed.c (init_expmed_one_mode): Likewise.
12246 (synth_mult): Likewise.
12247 (emit_store_flag_1): Likewise.
12248 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12249 of a comparison with size.
12250 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12251 (reduce_to_bit_field_precision): Likewise.
12252 * function.c (expand_function_end): Likewise.
12253 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12254 * loop-doloop.c (doloop_modify): Likewise.
12255 * optabs.c (expand_binop): Likewise.
12256 (expand_unop): Likewise.
12257 (expand_copysign_absneg): Likewise.
12258 (prepare_cmp_insn): Likewise.
12259 (maybe_legitimize_operand): Likewise.
12260 * recog.c (const_scalar_int_operand): Likewise.
12261 * rtlanal.c (get_address_mode): Likewise.
12262 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12263 (simplify_cond_clz_ctz): Likewise.
12264 * tree-nested.c (get_nl_goto_field): Likewise.
12265 * tree.c (build_vector_type_for_mode): Likewise.
12266 * var-tracking.c (use_narrower_mode): Likewise.
12268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12269 Alan Hayward <alan.hayward@arm.com>
12270 David Sherwood <david.sherwood@arm.com>
12272 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12273 * builtins.c (expand_builtin_signbit): Use it.
12274 * cfgexpand.c (expand_debug_expr): Likewise.
12275 * dojump.c (do_jump): Likewise.
12276 (do_compare_and_jump): Likewise.
12277 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12278 * expmed.c (make_tree): Likewise.
12279 * expr.c (expand_expr_real_2): Likewise.
12280 (expand_expr_real_1): Likewise.
12281 (try_casesi): Likewise.
12282 * fold-const-call.c (fold_const_call_ss): Likewise.
12283 * fold-const.c (unextend): Likewise.
12284 (extract_muldiv_1): Likewise.
12285 (fold_single_bit_test): Likewise.
12286 (native_encode_int): Likewise.
12287 (native_encode_string): Likewise.
12288 (native_interpret_int): Likewise.
12289 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12290 * internal-fn.c (expand_addsub_overflow): Likewise.
12291 (expand_neg_overflow): Likewise.
12292 (expand_mul_overflow): Likewise.
12293 (expand_arith_overflow): Likewise.
12294 * match.pd: Likewise.
12295 * stor-layout.c (layout_type): Likewise.
12296 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12297 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12298 * tree-ssanames.c (get_range_info): Likewise.
12299 * tree-switch-conversion.c (array_value_type) Likewise.
12300 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12301 (vect_recog_divmod_pattern): Likewise.
12302 (vect_recog_mixed_size_cond_pattern): Likewise.
12303 * tree-vrp.c (extract_range_basic): Likewise.
12304 (simplify_float_conversion_using_ranges): Likewise.
12305 * tree.c (int_fits_type_p): Likewise.
12306 * ubsan.c (instrument_bool_enum_load): Likewise.
12307 * varasm.c (mergeable_string_section): Likewise.
12308 (narrowing_initializer_constant_valid_p): Likewise.
12309 (output_constant): Likewise.
12311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12312 Alan Hayward <alan.hayward@arm.com>
12313 David Sherwood <david.sherwood@arm.com>
12315 * machmode.h (NARROWEST_INT_MODE): New macro.
12316 * expr.c (alignment_for_piecewise_move): Use it instead of
12317 GET_CLASS_NARROWEST_MODE (MODE_INT).
12318 (push_block): Likewise.
12319 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12321 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12324 Alan Hayward <alan.hayward@arm.com>
12325 David Sherwood <david.sherwood@arm.com>
12327 * postreload.c (move2add_valid_value_p): Change the type of the
12328 mode parameter to scalar_int_mode.
12329 (move2add_use_add2_insn): Add a mode parameter and use it instead
12331 (move2add_use_add3_insn): Likewise.
12332 (reload_cse_move2add): Update accordingly.
12334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12335 Alan Hayward <alan.hayward@arm.com>
12336 David Sherwood <david.sherwood@arm.com>
12338 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12340 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12341 * optabs.c (expand_unop): Likewise.
12343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12344 Alan Hayward <alan.hayward@arm.com>
12345 David Sherwood <david.sherwood@arm.com>
12347 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12348 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12349 (popcount_loc_descriptor): Likewise.
12350 (bswap_loc_descriptor): Likewise.
12351 (rotate_loc_descriptor): Likewise.
12352 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12353 calling the functions above.
12355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12356 Alan Hayward <alan.hayward@arm.com>
12357 David Sherwood <david.sherwood@arm.com>
12359 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12361 (try_combine): Likewise.
12362 (simplify_if_then_else): Likewise.
12363 * cse.c (cse_insn): Likewise.
12364 * dwarf2out.c (mem_loc_descriptor): Likewise.
12365 * emit-rtl.c (gen_lowpart_common): Likewise.
12366 * simplify-rtx.c (simplify_truncation): Likewise.
12367 (simplify_binary_operation_1): Likewise.
12368 (simplify_const_relational_operation): Likewise.
12369 (simplify_ternary_operation): Likewise.
12370 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12373 Alan Hayward <alan.hayward@arm.com>
12374 David Sherwood <david.sherwood@arm.com>
12376 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12377 * reload.c (push_reload): Likewise.
12378 (find_reloads): Likewise.
12380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12381 Alan Hayward <alan.hayward@arm.com>
12382 David Sherwood <david.sherwood@arm.com>
12384 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12385 (make_compound_operation_int): Likewise.
12386 (change_zero_ext): Likewise.
12387 * expr.c (convert_move): Likewise.
12388 (convert_modes): Likewise.
12389 * fwprop.c (forward_propagate_subreg): Likewise.
12390 * loop-iv.c (get_biv_step_1): Likewise.
12391 * optabs.c (widen_operand): Likewise.
12392 * postreload.c (move2add_valid_value_p): Likewise.
12393 * recog.c (simplify_while_replacing): Likewise.
12394 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12395 (simplify_binary_operation_1): Likewise. Remove redundant
12396 mode equality check.
12398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12399 Alan Hayward <alan.hayward@arm.com>
12400 David Sherwood <david.sherwood@arm.com>
12402 * combine.c (combine_simplify_rtx): Add checks for
12403 is_a <scalar_int_mode>.
12404 (simplify_if_then_else): Likewise.
12405 (make_field_assignment): Likewise.
12406 (simplify_comparison): Likewise.
12407 * ifcvt.c (noce_try_bitop): Likewise.
12408 * loop-invariant.c (canonicalize_address_mult): Likewise.
12409 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12412 Alan Hayward <alan.hayward@arm.com>
12413 David Sherwood <david.sherwood@arm.com>
12415 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12416 is_a <scalar_int_mode> instead of != BLKmode.
12418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12419 Alan Hayward <alan.hayward@arm.com>
12420 David Sherwood <david.sherwood@arm.com>
12422 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12423 instead of != VOIDmode.
12424 * combine.c (if_then_else_cond): Likewise.
12425 (change_zero_ext): Likewise.
12426 * dwarf2out.c (mem_loc_descriptor): Likewise.
12427 (loc_descriptor): Likewise.
12428 * rtlanal.c (canonicalize_condition): Likewise.
12429 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12432 Alan Hayward <alan.hayward@arm.com>
12433 David Sherwood <david.sherwood@arm.com>
12435 * simplify-rtx.c (simplify_binary_operation_1): Use
12436 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12439 Alan Hayward <alan.hayward@arm.com>
12440 David Sherwood <david.sherwood@arm.com>
12442 * wide-int.h (int_traits<unsigned char>) New class.
12443 (int_traits<unsigned short>) Likewise.
12444 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12445 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12447 * combine.c (set_nonzero_bits_and_sign_copies): Use
12448 is_a <scalar_int_mode>.
12449 (find_split_point): Likewise.
12450 (combine_simplify_rtx): Likewise.
12451 (simplify_logical): Likewise.
12452 (expand_compound_operation): Likewise.
12453 (expand_field_assignment): Likewise.
12454 (make_compound_operation): Likewise.
12455 (extended_count): Likewise.
12456 (change_zero_ext): Likewise.
12457 (simplify_comparison): Likewise.
12458 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12459 (ucompare_loc_descriptor): Likewise.
12460 (minmax_loc_descriptor): Likewise.
12461 (mem_loc_descriptor): Likewise.
12462 (loc_descriptor): Likewise.
12463 * expmed.c (init_expmed_one_mode): Likewise.
12464 * lra-constraints.c (lra_constraint_offset): Likewise.
12465 * optabs.c (prepare_libcall_arg): Likewise.
12466 * postreload.c (move2add_note_store): Likewise.
12467 * reload.c (operands_match_p): Likewise.
12468 * rtl.h (load_extend_op): Likewise.
12469 * rtlhooks.c (gen_lowpart_general): Likewise.
12470 * simplify-rtx.c (simplify_truncation): Likewise.
12471 (simplify_unary_operation_1): Likewise.
12472 (simplify_binary_operation_1): Likewise.
12473 (simplify_const_binary_operation): Likewise.
12474 (simplify_const_relational_operation): Likewise.
12475 (simplify_subreg): Likewise.
12476 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12477 * var-tracking.c (adjust_mems): Likewise.
12478 (prepare_call_arguments): Likewise.
12480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12481 Alan Hayward <alan.hayward@arm.com>
12482 David Sherwood <david.sherwood@arm.com>
12484 * machmode.h (is_int_mode): New fuction.
12485 * combine.c (find_split_point): Use it.
12486 (combine_simplify_rtx): Likewise.
12487 (simplify_if_then_else): Likewise.
12488 (simplify_set): Likewise.
12489 (simplify_shift_const_1): Likewise.
12490 (simplify_comparison): Likewise.
12491 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12492 * cse.c (notreg_cost): Likewise.
12493 (cse_insn): Likewise.
12494 * cselib.c (cselib_lookup_1): Likewise.
12495 * dojump.c (do_jump_1): Likewise.
12496 (do_compare_rtx_and_jump): Likewise.
12497 * dse.c (get_call_args): Likewise.
12498 * dwarf2out.c (rtl_for_decl_init): Likewise.
12499 (native_encode_initializer): Likewise.
12500 * expmed.c (emit_store_flag_1): Likewise.
12501 (emit_store_flag): Likewise.
12502 * expr.c (convert_modes): Likewise.
12503 (store_field): Likewise.
12504 (expand_expr_real_1): Likewise.
12505 * fold-const.c (fold_read_from_constant_string): Likewise.
12506 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12507 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12508 * optabs.c (expand_binop): Likewise.
12509 (expand_unop): Likewise.
12510 (expand_abs_nojump): Likewise.
12511 (expand_one_cmpl_abs_nojump): Likewise.
12512 * simplify-rtx.c (mode_signbit_p): Likewise.
12513 (val_signbit_p): Likewise.
12514 (val_signbit_known_set_p): Likewise.
12515 (val_signbit_known_clear_p): Likewise.
12516 (simplify_relational_operation_1): Likewise.
12517 * tree.c (vector_type_mode): Likewise.
12519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12520 Alan Hayward <alan.hayward@arm.com>
12521 David Sherwood <david.sherwood@arm.com>
12523 * machmode.h (smallest_mode_for_size): Fix formatting.
12524 (smallest_int_mode_for_size): New function.
12525 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12526 instead of smallest_mode_for_size.
12527 * combine.c (make_extraction): Likewise.
12528 * config/arc/arc.c (arc_expand_movmem): Likewise.
12529 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12530 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12531 * config/s390/s390.c (s390_expand_insv): Likewise.
12532 * config/sparc/sparc.c (assign_int_registers): Likewise.
12533 * config/spu/spu.c (spu_function_value): Likewise.
12534 (spu_function_arg): Likewise.
12535 * coverage.c (get_gcov_type): Likewise.
12536 (get_gcov_unsigned_t): Likewise.
12537 * dse.c (find_shift_sequence): Likewise.
12538 * expmed.c (store_bit_field_1): Likewise.
12539 * expr.c (convert_move): Likewise.
12540 (store_field): Likewise.
12541 * internal-fn.c (expand_arith_overflow): Likewise.
12542 * optabs-query.c (get_best_extraction_insn): Likewise.
12543 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12544 * stor-layout.c (layout_type): Likewise.
12545 (initialize_sizetypes): Likewise.
12546 * targhooks.c (default_get_mask_mode): Likewise.
12547 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12550 Alan Hayward <alan.hayward@arm.com>
12551 David Sherwood <david.sherwood@arm.com>
12553 * machmode.h (opt_mode::else_blk): New function.
12554 (int_mode_for_mode): Declare.
12555 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12556 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12558 * cfgexpand.c (expand_debug_expr): Likewise.
12559 * combine.c (gen_lowpart_or_truncate): Likewise.
12560 (gen_lowpart_for_combine): Likewise.
12561 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12562 * config/avr/avr.c (avr_to_int_mode): Likewise.
12563 (avr_out_plus_1): Likewise.
12564 (avr_out_plus): Likewise.
12565 (avr_out_round): Likewise.
12566 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12567 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12568 (s390_expand_vcond): Likewise.
12569 * config/spu/spu.c (spu_split_immediate): Likewise.
12570 (spu_expand_mov): Likewise.
12571 * dse.c (get_stored_val): Likewise.
12572 * expmed.c (store_bit_field_1): Likewise.
12573 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12575 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12576 (extract_low_bits): Likewise.
12577 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12578 handling rather than repeating the check.
12579 (emit_group_store): Likewise.
12580 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12581 * optabs.c (expand_absneg_bit): Likewise.
12582 (expand_copysign_absneg): Likewise.
12583 (expand_copysign_bit): Likewise.
12584 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12585 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12586 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12587 * var-tracking.c (prepare_call_arguments): Likewise.
12588 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12589 int_mode_for_mode instead of mode_for_size.
12590 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12593 Alan Hayward <alan.hayward@arm.com>
12594 David Sherwood <david.sherwood@arm.com>
12596 * machmode.h (int_mode_for_size): New function.
12597 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12598 instead of mode_for_size.
12599 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12601 * combine.c (expand_field_assignment): Use int_mode_for_size
12602 instead of mode_for_size.
12603 (make_extraction): Likewise.
12604 (simplify_shift_const_1): Likewise.
12605 (simplify_comparison): Likewise.
12606 * dojump.c (do_jump): Likewise.
12607 * dwarf2out.c (mem_loc_descriptor): Likewise.
12608 * emit-rtl.c (init_derived_machine_modes): Likewise.
12609 * expmed.c (flip_storage_order): Likewise.
12610 (convert_extracted_bit_field): Likewise.
12611 * expr.c (copy_blkmode_from_reg): Likewise.
12612 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12613 * internal-fn.c (expand_mul_overflow): Likewise.
12614 * lower-subreg.c (simple_move): Likewise.
12615 * optabs-libfuncs.c (init_optabs): Likewise.
12616 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12617 * tree.c (vector_type_mode): Likewise.
12618 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12619 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12620 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12621 * tree-vect-stmts.c (vectorizable_load): Likewise.
12622 (vectorizable_store): Likewise.
12624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12625 Alan Hayward <alan.hayward@arm.com>
12626 David Sherwood <david.sherwood@arm.com>
12628 * coretypes.h (pod_mode): New type.
12629 (scalar_int_mode_pod): New typedef.
12630 * machmode.h (pod_mode): New class.
12631 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12632 * genmodes.c (emit_mode_int_n): Update accordingly.
12633 * lower-subreg.h (target_lower_subreg): Change type to
12634 scalar_int_mode_pod.
12635 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12636 scalar_int_mode_pod.
12638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12639 Alan Hayward <alan.hayward@arm.com>
12640 David Sherwood <david.sherwood@arm.com>
12642 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12643 machine_mode to scalar_int_mode.
12644 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12645 (rs6000_option_override_internal): Remove cast to int.
12646 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12647 machine_mode to scalar_int_mode.
12648 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12649 (rs6000_option_override_internal): Remove cast to int.
12650 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12651 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12653 * config/s390/s390.c (s390_expand_builtin): Likewise.
12654 * coretypes.h (scalar_int_mode): New type.
12655 (opt_scalar_int_mode): New typedef.
12656 * machmode.h (scalar_int_mode): New class.
12657 (scalar_int_mode::includes_p): New function.
12658 (byte_mode): Change type to scalar_int_mode.
12659 (word_mode): Likewise.
12660 (ptr_mode): Likewise.
12661 * emit-rtl.c (byte_mode): Likewise.
12662 (word_mode): Likewise.
12663 (ptr_mode): Likewise.
12664 (init_derived_machine_modes): Update accordingly.
12665 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12666 and MODE_PARTIAL_INT.
12667 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12668 opt_scalar_int_mode.
12670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12671 Alan Hayward <alan.hayward@arm.com>
12672 David Sherwood <david.sherwood@arm.com>
12674 * target.def (libgcc_floating_mode_supported_p): Take a
12676 * doc/tm.texi: Regenerate.
12677 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12679 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12680 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12684 Alan Hayward <alan.hayward@arm.com>
12685 David Sherwood <david.sherwood@arm.com>
12687 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12688 * doc/tm.texi: Regenerate.
12689 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12690 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12691 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12692 * targhooks.h (default_floatn_mode): Likewise.
12693 * targhooks.c (default_floatn_mode): Likewise.
12694 * tree.c (build_common_tree_nodes): Update accordingly.
12696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12697 Alan Hayward <alan.hayward@arm.com>
12698 David Sherwood <david.sherwood@arm.com>
12700 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12701 (mode_iterator::iterate_p): Likewise.
12702 (mode_iterator::get_wider): Likewise.
12703 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12706 Alan Hayward <alan.hayward@arm.com>
12707 David Sherwood <david.sherwood@arm.com>
12709 * coretypes.h (opt_scalar_float_mode): New typedef.
12710 * machmode.h (float_mode_for_size): New function.
12711 * emit-rtl.c (double_mode): Delete.
12712 (init_emit_once): Use float_mode_for_size.
12713 * stor-layout.c (layout_type): Likewise.
12714 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12717 Alan Hayward <alan.hayward@arm.com>
12718 David Sherwood <david.sherwood@arm.com>
12720 * output.h (assemble_real): Take a scalar_float_mode.
12721 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12722 * config/arm/arm.md (consttable_4): Likewise.
12723 (consttable_8): Likewise.
12724 (consttable_16): Likewise.
12725 * config/mips/mips.md (consttable_float): Likewise.
12726 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12727 * varasm.c (assemble_real): Take a scalar_float_mode.
12728 (output_constant_pool_2): Update accordingly.
12729 (output_constant): Likewise.
12731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12732 Alan Hayward <alan.hayward@arm.com>
12733 David Sherwood <david.sherwood@arm.com>
12735 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12736 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12737 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12738 (native_encode_real): Likewise.
12739 (native_interpret_real): Likewise.
12740 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12741 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12744 Alan Hayward <alan.hayward@arm.com>
12745 David Sherwood <david.sherwood@arm.com>
12747 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12748 <scalar_float_mode>. Simplify.
12749 (gen_extend_conv_libfunc): Likewise.
12751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12752 Alan Hayward <alan.hayward@arm.com>
12753 David Sherwood <david.sherwood@arm.com>
12755 * coretypes.h (scalar_float_mode): New type.
12756 * machmode.h (mode_traits::from_int): Use machine_mode if
12757 USE_ENUM_MODES is defined.
12758 (is_a): New function.
12760 (dyn_cast): Likewise.
12761 (scalar_float_mode): New class.
12762 (scalar_float_mode::includes_p): New function.
12763 (is_float_mode): Likewise.
12764 * gdbhooks.py (MachineModePrinter): New class.
12765 (build_pretty_printer): Use it for scalar_float_mode.
12766 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12767 (format_helper::format_helper): Turn into a template.
12768 * genmodes.c (get_mode_class): New function.
12769 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12770 or machine_mode if none.
12771 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12772 as_a <scalar_float_mode>.
12773 * dwarf2out.c (mem_loc_descriptor): Likewise.
12774 (insert_float): Likewise.
12775 (add_const_value_attribute): Likewise.
12776 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12777 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12778 (expand_unop): Update accordingly.
12779 (expand_abs_nojump): Likewise.
12780 (expand_copysign_absneg): Take a scalar_float_mode.
12781 (expand_copysign_bit): Likewise.
12782 (expand_copysign): Update accordingly.
12784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12785 Alan Hayward <alan.hayward@arm.com>
12786 David Sherwood <david.sherwood@arm.com>
12788 * coretypes.h (opt_mode): New class.
12789 * machmode.h (opt_mode): Likewise.
12790 (opt_mode::else_void): New function.
12791 (opt_mode::require): Likewise.
12792 (opt_mode::exists): Likewise.
12793 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12794 (GET_MODE_2XWIDER_MODE): Likewise.
12795 (mode_iterator::get_wider): Update accordingly.
12796 (mode_iterator::get_2xwider): Likewise.
12797 (mode_iterator::get_known_wider): Likewise, turning into a template.
12798 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12799 forcing a wider mode to exist.
12800 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12801 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12802 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12803 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12804 * lower-subreg.c (init_lower_subreg): Likewise.
12805 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12806 on the final iteration.
12807 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12808 a wider mode exists before asking for a move pattern.
12809 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12810 forcing a wider mode to exist.
12811 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12812 returning false if no such mode exists.
12813 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12814 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12815 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12816 Avoid checking for a MODE_INT if we already know the mode is not a
12818 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12819 forcing a wider mode to exist.
12820 (expmed_mult_highpart_optab): Likewise.
12821 (expmed_mult_highpart): Likewise.
12822 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12824 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12825 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12826 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12827 * internal-fn.c (expand_mul_overflow): Update use of
12828 GET_MODE_2XWIDER_MODE.
12829 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12830 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12831 GET_MODE_WIDER_MODE.
12832 (convert_plusminus_to_widen): Likewise.
12833 * tree-switch-conversion.c (array_value_type): Likewise.
12834 * var-tracking.c (emit_note_insn_var_location): Likewise.
12835 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12836 Return false inside rather than outside the loop if no wider mode
12838 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12839 and GET_MODE_2XWIDER_MODE
12840 (can_compare_p): Use else_void.
12841 * gdbhooks.py (OptMachineModePrinter): New class.
12842 (build_pretty_printer): Use it for opt_mode.
12844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12845 Alan Hayward <alan.hayward@arm.com>
12846 David Sherwood <david.sherwood@arm.com>
12848 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12849 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12852 Alan Hayward <alan.hayward@arm.com>
12853 David Sherwood <david.sherwood@arm.com>
12855 * machmode.h (mode_traits): New structure.
12856 (get_narrowest_mode): New function.
12857 (mode_iterator::start): Likewise.
12858 (mode_iterator::iterate_p): Likewise.
12859 (mode_iterator::get_wider): Likewise.
12860 (mode_iterator::get_known_wider): Likewise.
12861 (mode_iterator::get_2xwider): Likewise.
12862 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12863 (FOR_EACH_MODE): Likewise.
12864 (FOR_EACH_MODE_FROM): Likewise.
12865 (FOR_EACH_MODE_UNTIL): Likewise.
12866 (FOR_EACH_WIDER_MODE): Likewise.
12867 (FOR_EACH_2XWIDER_MODE): Likewise.
12868 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12869 * combine.c (simplify_comparison): Likewise
12870 * config/i386/i386.c (type_natural_mode): Likewise.
12871 * cse.c (cse_insn): Likewise.
12872 * dse.c (find_shift_sequence): Likewise.
12873 * emit-rtl.c (init_derived_machine_modes): Likewise.
12874 (init_emit_once): Likewise.
12875 * explow.c (hard_function_value): Likewise.
12876 * expmed.c (extract_fixed_bit_field_1): Likewise.
12877 (extract_bit_field_1): Likewise.
12878 (expand_divmod): Likewise.
12879 (emit_store_flag_1): Likewise.
12880 * expr.c (init_expr_target): Likewise.
12881 (convert_move): Likewise.
12882 (alignment_for_piecewise_move): Likewise.
12883 (widest_int_mode_for_size): Likewise.
12884 (emit_block_move_via_movmem): Likewise.
12885 (copy_blkmode_to_reg): Likewise.
12886 (set_storage_via_setmem): Likewise.
12887 (compress_float_constant): Likewise.
12888 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12889 * optabs-query.c (get_best_extraction_insn): Likewise.
12890 * optabs.c (expand_binop): Likewise.
12891 (expand_twoval_unop): Likewise.
12892 (expand_twoval_binop): Likewise.
12893 (widen_leading): Likewise.
12894 (widen_bswap): Likewise.
12895 (expand_parity): Likewise.
12896 (expand_unop): Likewise.
12897 (prepare_cmp_insn): Likewise.
12898 (prepare_float_lib_cmp): Likewise.
12899 (expand_float): Likewise.
12900 (expand_fix): Likewise.
12901 (expand_sfix_optab): Likewise.
12902 * postreload.c (move2add_use_add2_insn): Likewise.
12903 * reg-stack.c (reg_to_stack): Likewise.
12904 * reginfo.c (choose_hard_reg_mode): Likewise.
12905 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12906 * stor-layout.c (mode_for_size): Likewise.
12907 (smallest_mode_for_size): Likewise.
12908 (mode_for_vector): Likewise.
12909 (finish_bitfield_representative): Likewise.
12910 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12911 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12912 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12913 * var-tracking.c (prepare_call_arguments): Likewise.
12915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12916 Alan Hayward <alan.hayward@arm.com>
12917 David Sherwood <david.sherwood@arm.com>
12919 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12920 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12921 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12922 * machmode.h (mode_size): Move earlier in file.
12923 (mode_precision): Likewise.
12924 (mode_inner): Likewise.
12925 (mode_nunits): Likewise.
12926 (mode_unit_size): Likewise.
12927 (unit_unit_precision): Likewise.
12928 (mode_wider): Likewise.
12929 (mode_2xwider): Likewise.
12930 (machine_mode): New class.
12931 (mode_to_bytes): New function.
12932 (mode_to_bits): Likewise.
12933 (mode_to_precision): Likewise.
12934 (mode_to_inner): Likewise.
12935 (mode_to_unit_size): Likewise.
12936 (mode_to_unit_precision): Likewise.
12937 (mode_to_nunits): Likewise.
12938 (GET_MODE_SIZE): Use mode_to_bytes.
12939 (GET_MODE_BITSIZE): Use mode_to_bits.
12940 (GET_MODE_PRECISION): Use mode_to_precision.
12941 (GET_MODE_INNER): Use mode_to_inner.
12942 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12943 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12944 (GET_MODE_NUNITS): Use mode_to_nunits.
12945 * system.h (ALWAYS_INLINE): New macro.
12946 * config/powerpcspe/powerpcspe-c.c
12947 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12948 int for arg1_mode and arg2_mode.
12950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12951 Alan Hayward <alan.hayward@arm.com>
12952 David Sherwood <david.sherwood@arm.com>
12954 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12955 Prefix mode names with E_ in case statements.
12956 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12957 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12958 (aarch64_split_simd_move): Likewise.
12959 (aarch64_gen_storewb_pair): Likewise.
12960 (aarch64_gen_loadwb_pair): Likewise.
12961 (aarch64_gen_store_pair): Likewise.
12962 (aarch64_gen_load_pair): Likewise.
12963 (aarch64_get_condition_code_1): Likewise.
12964 (aarch64_constant_pool_reload_icode): Likewise.
12965 (get_rsqrte_type): Likewise.
12966 (get_rsqrts_type): Likewise.
12967 (get_recpe_type): Likewise.
12968 (get_recps_type): Likewise.
12969 (aarch64_gimplify_va_arg_expr): Likewise.
12970 (aarch64_simd_container_mode): Likewise.
12971 (aarch64_emit_load_exclusive): Likewise.
12972 (aarch64_emit_store_exclusive): Likewise.
12973 (aarch64_expand_compare_and_swap): Likewise.
12974 (aarch64_gen_atomic_cas): Likewise.
12975 (aarch64_emit_bic): Likewise.
12976 (aarch64_emit_atomic_swap): Likewise.
12977 (aarch64_emit_atomic_load_op): Likewise.
12978 (aarch64_evpc_trn): Likewise.
12979 (aarch64_evpc_uzp): Likewise.
12980 (aarch64_evpc_zip): Likewise.
12981 (aarch64_evpc_ext): Likewise.
12982 (aarch64_evpc_rev): Likewise.
12983 (aarch64_evpc_dup): Likewise.
12984 (aarch64_gen_ccmp_first): Likewise.
12985 (aarch64_gen_ccmp_next): Likewise.
12986 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12987 (alpha_emit_xfloating_libcall): Likewise.
12988 (emit_insxl): Likewise.
12989 (alpha_arg_type): Likewise.
12990 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12991 (arc_preferred_simd_mode): Likewise.
12992 (arc_secondary_reload): Likewise.
12993 (get_arc_condition_code): Likewise.
12994 (arc_print_operand): Likewise.
12995 (arc_legitimate_constant_p): Likewise.
12996 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12997 * config/arc/arc.md (casesi_load): Likewise.
12998 (casesi_compact_jump): Likewise.
12999 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13000 (cc_use_register): Likewise.
13001 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13002 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13003 (arm_init_iwmmxt_builtins): Likewise.
13004 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13005 (neon_expand_vector_init): Likewise.
13006 (arm_attr_length_move_neon): Likewise.
13007 (maybe_get_arm_condition_code): Likewise.
13008 (arm_emit_vector_const): Likewise.
13009 (arm_preferred_simd_mode): Likewise.
13010 (arm_output_iwmmxt_tinsr): Likewise.
13011 (thumb1_output_casesi): Likewise.
13012 (thumb2_output_casesi): Likewise.
13013 (arm_emit_load_exclusive): Likewise.
13014 (arm_emit_store_exclusive): Likewise.
13015 (arm_expand_compare_and_swap): Likewise.
13016 (arm_evpc_neon_vuzp): Likewise.
13017 (arm_evpc_neon_vzip): Likewise.
13018 (arm_evpc_neon_vrev): Likewise.
13019 (arm_evpc_neon_vtrn): Likewise.
13020 (arm_evpc_neon_vext): Likewise.
13021 (arm_validize_comparison): Likewise.
13022 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13023 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13024 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13025 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13026 (c6x_preferred_simd_mode): Likewise.
13027 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13028 (epiphany_rtx_costs): Likewise.
13029 * config/epiphany/predicates.md (proper_comparison_operator):
13031 * config/frv/frv.c (condexec_memory_operand): Likewise.
13032 (frv_emit_move): Likewise.
13033 (output_move_single): Likewise.
13034 (output_condmove_single): Likewise.
13035 (frv_hard_regno_mode_ok): Likewise.
13036 (frv_matching_accg_mode): Likewise.
13037 * config/h8300/h8300.c (split_adds_subs): Likewise.
13038 (h8300_rtx_costs): Likewise.
13039 (h8300_print_operand): Likewise.
13040 (compute_mov_length): Likewise.
13041 (output_logical_op): Likewise.
13042 (compute_logical_op_length): Likewise.
13043 (compute_logical_op_cc): Likewise.
13044 (h8300_shift_needs_scratch_p): Likewise.
13045 (output_a_shift): Likewise.
13046 (compute_a_shift_length): Likewise.
13047 (compute_a_shift_cc): Likewise.
13048 (expand_a_rotate): Likewise.
13049 (output_a_rotate): Likewise.
13050 * config/i386/i386.c (classify_argument): Likewise.
13051 (function_arg_advance_32): Likewise.
13052 (function_arg_32): Likewise.
13053 (function_arg_64): Likewise.
13054 (function_value_64): Likewise.
13055 (ix86_gimplify_va_arg): Likewise.
13056 (ix86_legitimate_constant_p): Likewise.
13057 (put_condition_code): Likewise.
13058 (split_double_mode): Likewise.
13059 (ix86_avx256_split_vector_move_misalign): Likewise.
13060 (ix86_expand_vector_logical_operator): Likewise.
13061 (ix86_split_idivmod): Likewise.
13062 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13063 (ix86_build_const_vector): Likewise.
13064 (ix86_build_signbit_mask): Likewise.
13065 (ix86_match_ccmode): Likewise.
13066 (ix86_cc_modes_compatible): Likewise.
13067 (ix86_expand_branch): Likewise.
13068 (ix86_expand_sse_cmp): Likewise.
13069 (ix86_expand_sse_movcc): Likewise.
13070 (ix86_expand_int_sse_cmp): Likewise.
13071 (ix86_expand_vec_perm_vpermi2): Likewise.
13072 (ix86_expand_vec_perm): Likewise.
13073 (ix86_expand_sse_unpack): Likewise.
13074 (ix86_expand_int_addcc): Likewise.
13075 (ix86_split_to_parts): Likewise.
13076 (ix86_vectorize_builtin_gather): Likewise.
13077 (ix86_vectorize_builtin_scatter): Likewise.
13078 (avx_vpermilp_parallel): Likewise.
13079 (inline_memory_move_cost): Likewise.
13080 (ix86_tieable_integer_mode_p): Likewise.
13081 (x86_maybe_negate_const_int): Likewise.
13082 (ix86_expand_vector_init_duplicate): Likewise.
13083 (ix86_expand_vector_init_one_nonzero): Likewise.
13084 (ix86_expand_vector_init_one_var): Likewise.
13085 (ix86_expand_vector_init_concat): Likewise.
13086 (ix86_expand_vector_init_interleave): Likewise.
13087 (ix86_expand_vector_init_general): Likewise.
13088 (ix86_expand_vector_set): Likewise.
13089 (ix86_expand_vector_extract): Likewise.
13090 (emit_reduc_half): Likewise.
13091 (ix86_emit_i387_round): Likewise.
13092 (ix86_mangle_type): Likewise.
13093 (ix86_expand_round_sse4): Likewise.
13094 (expand_vec_perm_blend): Likewise.
13095 (canonicalize_vector_int_perm): Likewise.
13096 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13097 (expand_vec_perm_1): Likewise.
13098 (expand_vec_perm_interleave3): Likewise.
13099 (expand_vec_perm_even_odd_pack): Likewise.
13100 (expand_vec_perm_even_odd_1): Likewise.
13101 (expand_vec_perm_broadcast_1): Likewise.
13102 (ix86_vectorize_vec_perm_const_ok): Likewise.
13103 (ix86_expand_vecop_qihi): Likewise.
13104 (ix86_expand_mul_widen_hilo): Likewise.
13105 (ix86_expand_sse2_abs): Likewise.
13106 (ix86_expand_pextr): Likewise.
13107 (ix86_expand_pinsr): Likewise.
13108 (ix86_preferred_simd_mode): Likewise.
13109 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13110 * config/i386/sse.md (*andnot<mode>3): Likewise.
13111 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13112 (*<code><mode>3): Likewise.
13113 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13114 (ia64_expand_atomic_op): Likewise.
13115 (ia64_arg_type): Likewise.
13116 (ia64_mode_to_int): Likewise.
13117 (ia64_scalar_mode_supported_p): Likewise.
13118 (ia64_vector_mode_supported_p): Likewise.
13119 (expand_vec_perm_broadcast): Likewise.
13120 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13121 (iq2000_function_arg_advance): Likewise.
13122 (iq2000_function_arg): Likewise.
13123 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13124 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13125 (m68k_libcall_value): Likewise.
13126 (m68k_function_value): Likewise.
13127 (sched_attr_op_type): Likewise.
13128 * config/mcore/mcore.c (mcore_output_move): Likewise.
13129 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13131 (microblaze_function_arg): Likewise.
13132 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13133 (mips_print_operand): Likewise.
13134 (mips_mode_ok_for_mov_fmt_p): Likewise.
13135 (mips_vector_mode_supported_p): Likewise.
13136 (mips_preferred_simd_mode): Likewise.
13137 (mips_expand_vpc_loongson_even_odd): Likewise.
13138 (mips_expand_vec_unpack): Likewise.
13139 (mips_expand_vi_broadcast): Likewise.
13140 (mips_expand_vector_init): Likewise.
13141 (mips_expand_vec_reduc): Likewise.
13142 (mips_expand_msa_cmp): Likewise.
13143 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13144 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13145 (cc_flags_for_mode): Likewise.
13146 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13147 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13148 (nds32_output_casesi_pc_relative): Likewise.
13149 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13150 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13151 (nvptx_gen_unpack): Likewise.
13152 (nvptx_gen_pack): Likewise.
13153 (nvptx_gen_shuffle): Likewise.
13154 (nvptx_gen_wcast): Likewise.
13155 (nvptx_preferred_simd_mode): Likewise.
13156 * config/pa/pa.c (pa_secondary_reload): Likewise.
13157 * config/pa/predicates.md (base14_operand): Likewise.
13158 * config/powerpcspe/powerpcspe-c.c
13159 (altivec_resolve_overloaded_builtin): Likewise.
13160 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13162 (rs6000_preferred_simd_mode): Likewise.
13163 (output_vec_const_move): Likewise.
13164 (rs6000_expand_vector_extract): Likewise.
13165 (rs6000_split_vec_extract_var): Likewise.
13166 (reg_offset_addressing_ok_p): Likewise.
13167 (rs6000_legitimate_offset_address_p): Likewise.
13168 (rs6000_legitimize_address): Likewise.
13169 (rs6000_emit_set_const): Likewise.
13170 (rs6000_const_vec): Likewise.
13171 (rs6000_emit_move): Likewise.
13172 (spe_build_register_parallel): Likewise.
13173 (rs6000_darwin64_record_arg_recurse): Likewise.
13174 (swap_selector_for_mode): Likewise.
13175 (spe_init_builtins): Likewise.
13176 (paired_init_builtins): Likewise.
13177 (altivec_init_builtins): Likewise.
13178 (do_load_for_compare): Likewise.
13179 (rs6000_generate_compare): Likewise.
13180 (rs6000_expand_float128_convert): Likewise.
13181 (emit_load_locked): Likewise.
13182 (emit_store_conditional): Likewise.
13183 (rs6000_output_function_epilogue): Likewise.
13184 (rs6000_handle_altivec_attribute): Likewise.
13185 (rs6000_function_value): Likewise.
13186 (emit_fusion_gpr_load): Likewise.
13187 (emit_fusion_p9_load): Likewise.
13188 (emit_fusion_p9_store): Likewise.
13189 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13190 (fusion_gpr_mem_load): Likewise.
13191 (fusion_addis_mem_combo_load): Likewise.
13192 (fusion_addis_mem_combo_store): Likewise.
13193 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13194 (fusion_gpr_mem_load): Likewise.
13195 (fusion_addis_mem_combo_load): Likewise.
13196 (fusion_addis_mem_combo_store): Likewise.
13197 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13199 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13200 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13201 (rs6000_preferred_simd_mode): Likewise.
13202 (output_vec_const_move): Likewise.
13203 (rs6000_expand_vector_extract): Likewise.
13204 (rs6000_split_vec_extract_var): Likewise.
13205 (reg_offset_addressing_ok_p): Likewise.
13206 (rs6000_legitimate_offset_address_p): Likewise.
13207 (rs6000_legitimize_address): Likewise.
13208 (rs6000_emit_set_const): Likewise.
13209 (rs6000_const_vec): Likewise.
13210 (rs6000_emit_move): Likewise.
13211 (rs6000_darwin64_record_arg_recurse): Likewise.
13212 (swap_selector_for_mode): Likewise.
13213 (paired_init_builtins): Likewise.
13214 (altivec_init_builtins): Likewise.
13215 (rs6000_expand_float128_convert): Likewise.
13216 (emit_load_locked): Likewise.
13217 (emit_store_conditional): Likewise.
13218 (rs6000_output_function_epilogue): Likewise.
13219 (rs6000_handle_altivec_attribute): Likewise.
13220 (rs6000_function_value): Likewise.
13221 (emit_fusion_gpr_load): Likewise.
13222 (emit_fusion_p9_load): Likewise.
13223 (emit_fusion_p9_store): Likewise.
13224 * config/rx/rx.c (rx_gen_move_template): Likewise.
13225 (flags_from_mode): Likewise.
13226 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13227 (s390_slb_comparison): Likewise.
13228 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13229 (s390_vector_mode_supported_p): Likewise.
13230 (s390_cc_modes_compatible): Likewise.
13231 (s390_match_ccmode_set): Likewise.
13232 (s390_canonicalize_comparison): Likewise.
13233 (s390_emit_compare_and_swap): Likewise.
13234 (s390_branch_condition_mask): Likewise.
13235 (s390_rtx_costs): Likewise.
13236 (s390_secondary_reload): Likewise.
13237 (__SECONDARY_RELOAD_CASE): Likewise.
13238 (s390_expand_cs): Likewise.
13239 (s390_preferred_simd_mode): Likewise.
13240 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13241 * config/sh/sh.c (sh_print_operand): Likewise.
13242 (dump_table): Likewise.
13243 (sh_secondary_reload): Likewise.
13244 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13245 * config/sh/sh.md (casesi_worker_1): Likewise.
13246 (casesi_worker_2): Likewise.
13247 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13248 (fcc_comparison_operator): Likewise.
13249 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13250 (emit_soft_tfmode_cvt): Likewise.
13251 (sparc_preferred_simd_mode): Likewise.
13252 (output_cbranch): Likewise.
13253 (sparc_print_operand): Likewise.
13254 (sparc_expand_vec_perm_bmask): Likewise.
13255 (vector_init_bshuffle): Likewise.
13256 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13257 (spu_vector_mode_supported_p): Likewise.
13258 (spu_expand_insv): Likewise.
13259 (spu_emit_branch_or_set): Likewise.
13260 (spu_handle_vector_attribute): Likewise.
13261 (spu_builtin_splats): Likewise.
13262 (spu_builtin_extract): Likewise.
13263 (spu_builtin_promote): Likewise.
13264 (spu_expand_sign_extend): Likewise.
13265 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13266 (tilegx_simd_int): Likewise.
13267 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13268 (tilepro_simd_int): Likewise.
13269 * config/v850/v850.c (const_double_split): Likewise.
13270 (v850_print_operand): Likewise.
13271 (ep_memory_offset): Likewise.
13272 * config/vax/vax.c (vax_rtx_costs): Likewise.
13273 (vax_output_int_move): Likewise.
13274 (vax_output_int_add): Likewise.
13275 (vax_output_int_subtract): Likewise.
13276 * config/visium/predicates.md (visium_branch_operator): Likewise.
13277 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13278 (visium_print_operand_address): Likewise.
13279 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13280 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13281 (xtensa_expand_conditional_branch): Likewise.
13282 (xtensa_copy_incoming_a7): Likewise.
13283 (xtensa_output_literal): Likewise.
13284 * dfp.c (decimal_real_maxval): Likewise.
13285 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13288 Alan Hayward <alan.hayward@arm.com>
13289 David Sherwood <david.sherwood@arm.com>
13291 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13292 (mode_nunits_inline): Likewise.
13293 (mode_inner_inline): Likewise.
13294 (mode_unit_size_inline): Likewise.
13295 (mode_unit_precision_inline): Likewise.
13296 (emit_insn_modes_h): Likewise. Also emit a #define of the
13298 (emit_mode_wider): Add an E_ prefix to mode names.
13299 (emit_mode_complex): Likewise.
13300 (emit_mode_inner): Likewise.
13301 (emit_mode_adjustments): Likewise.
13302 (emit_mode_int_n): Likewise.
13303 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13304 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13305 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13306 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13307 (CRC32_BUILTIN, ENTRY): Likewise.
13308 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13309 (aarch64_pop_regs): Likewise.
13310 (aarch64_process_components): Likewise.
13311 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13312 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13313 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13314 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13315 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13316 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13318 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13319 (pcmpistr): Likewise.
13320 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13321 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13322 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13324 * config/rl78/rl78.c (mduc_regs): Likewise.
13325 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13326 (htm_expand_builtin): Likewise.
13327 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13328 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13329 * config/xtensa/xtensa.c (print_operand): Likewise.
13330 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13331 (NUM_MODE_VECTOR_INT): Likewise.
13332 * genoutput.c (null_operand): Likewise.
13333 (output_operand_data): Likewise.
13334 * genrecog.c (print_parameter_value): Likewise.
13335 * lra.c (debug_operand_data): Likewise.
13337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13339 * dwarf2out.c (scompare_loc_descriptor_wide)
13340 (scompare_loc_descriptor_narrow): New functions, split out from...
13341 (scompare_loc_descriptor): ...here.
13342 * expmed.c (emit_store_flag_int): New function, split out from...
13343 (emit_store_flag): ...here.
13345 2017-08-30 Richard Biener <rguenther@suse.de>
13347 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13348 (dwarf2out_early_finish): Move setting of AT_pubnames from
13349 early debug output to early finish.
13351 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13353 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13354 and -mdata-region to the assembler.
13356 2017-08-30 Richard Biener <rguenther@suse.de>
13358 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13360 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13361 (dwarf2out_early_global_decl): Only generate a DIE for the
13362 abstract origin if it doesn't already exist or is a declaration DIE.
13363 (resolve_addr): Do not add the linkage name twice when
13364 generating a stub DIE for the DW_TAG_GNU_call_site target.
13366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13368 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13369 Use machine_mode rather than int for arg1_mode.
13371 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13374 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13375 that the second argument of the built-in functions to unpack
13376 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13377 switch statement instead a lot of if statements.
13378 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13379 Allow 64-bit values to be in Altivec registers as well as
13380 traditional floating point registers.
13381 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13383 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13385 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13386 MAX_REGS_PER_ADDRESS == 1.
13388 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13390 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13391 * config/i386/i386.c (ix86_option_override_internal): Simplify
13392 setting of opts->x_flag_entry.
13394 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13395 Jakub Jelinek <jakub@redhat.com>
13396 Richard Biener <rguenther@suse.de>
13398 PR tree-optimization/81503
13399 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13400 folded constant fits in the target type; reorder tests for clarity.
13402 2017-08-29 Martin Liska <mliska@suse.cz>
13404 * passes.def: Include pass_lower_switch.
13405 * stmt.c (dump_case_nodes): Remove and move to
13406 tree-switch-conversion.
13407 (case_values_threshold): Likewise.
13408 (expand_switch_as_decision_tree_p): Likewise.
13409 (emit_case_decision_tree): Likewise.
13410 (expand_case): Likewise.
13411 (balance_case_nodes): Likewise.
13412 (node_has_low_bound): Likewise.
13413 (node_has_high_bound): Likewise.
13414 (node_is_bounded): Likewise.
13415 (emit_case_nodes): Likewise.
13416 (struct simple_case_node): New struct.
13417 (add_case_node): Remove.
13418 (emit_case_dispatch_table): Use vector instead of case_list.
13419 (reset_out_edges_aux): Remove.
13420 (compute_cases_per_edge): Likewise.
13421 (expand_case): Build list of simple_case_node.
13422 (expand_sjlj_dispatch_table): Use it.
13423 * tree-switch-conversion.c (struct case_node): Moved from
13424 stmt.c and adjusted.
13425 (emit_case_nodes): Likewise.
13426 (node_has_low_bound): Likewise.
13427 (node_has_high_bound): Likewise.
13428 (node_is_bounded): Likewise.
13429 (case_values_threshold): Likewise.
13430 (reset_out_edges_aux): Likewise.
13431 (compute_cases_per_edge): Likewise.
13432 (add_case_node): Likewise.
13433 (dump_case_nodes): Likewise.
13434 (balance_case_nodes): Likewise.
13435 (expand_switch_as_decision_tree_p): Likewise.
13436 (emit_jump): Likewise.
13437 (emit_case_decision_tree): Likewise.
13438 (try_switch_expansion): Likewise.
13439 (do_jump_if_equal): Likewise.
13440 (emit_cmp_and_jump_insns): Likewise.
13441 (fix_phi_operands_for_edge): New function.
13442 (record_phi_operand_mapping): Likewise.
13443 (class pass_lower_switch): New pass.
13444 (pass_lower_switch::execute): New function.
13445 (make_pass_lower_switch): Likewise.
13446 (conditional_probability):
13447 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13448 * tree-pass.h: Add make_pass_lower_switch.
13450 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13453 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13456 2017-08-29 Richard Biener <rguenther@suse.de>
13458 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13459 we do not add a DW_AT_inline attribute twice.
13460 (gen_subprogram_die): Remove code setting DW_AT_inline on
13461 DECL_ABSTRACT_P nodes.
13463 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13465 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13466 calls to internal functions.
13467 (gimplify_modify_expr): Likewise.
13468 * tree-call-cdce.c (use_internal_fn): Likewise.
13469 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13470 (convert_to_divmod): Set the nothrow flag.
13471 * tree-if-conv.c (predicate_mem_writes): Likewise.
13472 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13473 (vectorizable_call): Likewise.
13474 (vectorizable_store): Likewise.
13475 (vectorizable_load): Likewise.
13476 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13477 (vect_recog_mask_conversion_pattern): Likewise.
13479 2017-08-29 Martin Liska <mliska@suse.cz>
13482 * gcc.c (driver_handle_option): Add new argument.
13483 * opts-common.c (handle_option): Pass
13484 target_option_override_hook.
13485 * opts-global.c (lang_handle_option): Add new option.
13486 (set_default_handlers): Add new argument.
13487 (decode_options): Likewise.
13488 * opts.c (target_handle_option): Likewise.
13489 (common_handle_option): Call target_option_override_hook.
13490 * opts.h (struct cl_option_handler_func): Add hook for
13491 target option override.
13492 (struct cl_option_handlers): Likewise.
13493 (set_default_handlers): Add new argument.
13494 (decode_options): Likewise.
13495 (common_handle_option): Likewise.
13496 (target_handle_option): Likewise.
13497 * toplev.c (toplev::main): Pass targetm.target_option.override
13500 2017-08-29 Richard Biener <rguenther@suse.de>
13501 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13503 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13504 life to the active subtree.
13506 2017-08-28 Jeff Law <law@redhat.com>
13508 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13509 derive_equivalences.
13510 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13511 Code moved into....
13512 (edge_info::derive_equivalences): New private member function
13514 * tree-ssa-dom.c (class edge_info): Changed from a struct
13515 to a class. Add ctor/dtor, methods and data members.
13516 (edge_info::edge_info): Renamed from allocate_edge_info.
13517 Initialize additional members.
13518 (edge_info::~edge_info): New.
13519 (free_dom_edge_info): Delete the edge info.
13520 (record_edge_info): Use new class & associated member functions.
13521 Tighten forms for testing for edge equivalences.
13522 (record_temporary_equivalences): Iterate over the simple
13523 equivalences rather than assuming there's only one per edge.
13524 (cprop_into_successor_phis): Iterate over the simple
13525 equivalences rather than assuming there's only one per edge.
13526 (optimize_stmt): Use operand_equal_p rather than pointer
13527 equality for mini-DSE code.
13529 2017-08-28 Nathan Sidwell <nathan@acm.org>
13531 * gcc.c (execute): Fold SIGPIPE handling into switch
13532 statement. Adjust internal error message.
13534 2017-08-28 Richard Biener <rguenther@suse.de>
13537 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13538 Do nothing for removed DIEs.
13540 2017-08-28 Richard Biener <rguenther@suse.de>
13542 PR tree-optimization/81977
13543 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13546 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13549 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13550 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13551 using targetm.gen_mem_thread_fence.
13553 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13556 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13557 predicate to register_operand. Reorder operands.
13558 (*btr<mode>): Ditto.
13559 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13560 (*btr<mode>_mask): Ditto.
13562 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13564 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13565 * config/rs6000/xmmintrin.h: New file.
13566 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13568 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13571 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13572 parameter and_insn and return it.
13573 (recombine_lvx_pattern): Insert a copy to ensure availability of
13574 the base register of the copied masking operation at the point of
13575 the instruction replacement.
13576 (recombine_stvx_pattern): Likewise.
13578 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13580 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13581 undocumented switches.
13582 (-mpower9-dform-vector): Likewise.
13583 (-mpower9-dform): Likewise.
13584 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13585 comments to delete references to -mpower9-dform* switches.
13586 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13587 Delete reference to -mpower9-dform* switches, test for
13588 -mpower9-vector instead.
13589 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13590 (OTHER_P9_VECTOR_MASKS): Likewise.
13591 (POWERPC_MASKS): Likewise.
13592 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13593 tests against -mpower9-dform* to -mpower9-vector. Delete code
13594 that checked for -mpower9-dform* consistancy with other options.
13595 Add test for -mpower9-misc to enable other power9 switches.
13596 (rs6000_init_hard_regno_mode_ok): Likewise.
13597 (rs6000_option_override_internal): Likewise.
13598 (rs6000_emit_prologue): Likewise.
13599 (rs6000_emit_epilogue): Likewise.
13600 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13601 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13602 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13604 (emit_fusion_p9_store): Likewise.
13605 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13606 resetting these macros if the assembler does not support ISA 3.0
13608 (TARGET_P9_DFORM_VECTOR): Likewise.
13609 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13610 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13612 2017-08-25 Alan Modra <amodra@gmail.com>
13614 PR rtl-optimization/81747
13615 * cse.c (cse_extended_basic_block): Don't attempt to record
13616 equivalences for degenerate conditional branches that branch
13617 to their fall-through.
13619 2017-08-24 Martin Sebor <msebor@redhat.com>
13621 PR middle-end/81908
13622 * gimple-fold.c (size_must_be_zero_p): New function.
13623 (gimple_fold_builtin_memory_op): Call it.
13625 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13627 * config/rs6000/mm_malloc.h: New file.
13629 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13631 PR tree-optimization/81913
13632 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13633 analysis when either IVs in condition can wrap.
13635 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13637 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13638 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13640 2017-08-24 Richard Biener <rguenther@suse.de>
13643 * targhooks.c (default_target_can_inline_p): Properly
13644 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13645 is present and always compare.
13646 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13647 infer -mfpmath=sse from TARGET_SSE_P.
13648 (ix86_can_inline_p): Properly use target_option_default_node when
13649 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13651 2017-08-24 Richard Biener <rguenther@suse.de>
13654 * dwarf2out.c (output_die): Handle flag_generate_offload like
13656 (output_comp_unit): Likewise.
13657 (gen_array_type_die): Likewise.
13658 (dwarf2out_early_finish): Likewise.
13659 (note_variable_value_in_expr): Likewise.
13660 (dwarf2out_finish): Likewise. Adjust assert.
13661 * cgraphunit.c (symbol_table::compile): Move setting of
13662 flag_generate_offload earlier ...
13663 (symbol_table::finalize_compilation_unit): ... here, before
13664 early debug finalization.
13666 2017-08-24 Richard Biener <rguenther@suse.de>
13668 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13669 and ipa-fnsummary.h.
13670 (ix86_can_inline_p): When ix86_fpmath flags do not match
13671 check whether the callee uses FP math at all.
13673 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13675 PR middle-end/81931
13676 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13677 instead of TYPE_PRECISION.
13679 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13680 Alan Hayward <alan.hayward@arm.com>
13681 David Sherwood <david.sherwood@arm.com>
13683 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13685 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13687 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13689 (ix86_frame::stack_realign_allocate): New field.
13690 (struct machine_frame_state): Modify comments.
13691 (machine_frame_state::sp_realigned_fp_end): New field.
13692 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13693 layout calculation.
13694 (sp_valid_at): Add assertion to assure no attempt to access invalid
13695 offset of a realigned stack.
13696 (fp_valid_at): Likewise.
13697 (choose_baseaddr): Modify comments.
13698 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13699 ix86_expand_prologue.
13700 (ix86_expand_prologue): Modify stack realignment and allocation.
13701 (ix86_expand_epilogue): Modify comments.
13702 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13703 avx2_runtime, avx512f, and avx512f_runtime.
13705 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13707 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13708 (mstackrealign): Do not init to -1.
13709 * config/i386/i386.c (ix86_option_override_internal):
13710 Check opts_set, not opts when setting default value of
13711 opts->x_ix86_force_align_arg_pointer.
13713 2017-08-23 Richard Biener <rguenther@suse.de>
13715 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13716 lang_hooks.decl_printable_name.
13717 * print-rtl-function.c (print_rtx_function): Likewise.
13718 * tree-pretty-print.c (dump_function_header): Likewise.
13720 2017-08-23 Richard Biener <rguenther@suse.de>
13723 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13724 -g0 at compile-time.
13726 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13728 PR middle-end/19706
13729 * doc/sourcebuild.texi (Other hardware attributes):
13732 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13734 PR middle-end/19706
13735 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13736 Fix single-use check.
13738 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13740 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13742 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13744 * doc/install.texi: Modify to add more details on running selected
13747 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13749 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13750 is combined with -mabi=ms.
13751 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13754 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13756 PR tree-optimization/81488
13757 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13758 and cached_basis fields.
13759 (MAX_SPREAD): New constant.
13760 (alloc_cand_and_find_basis): Initialize new fields.
13761 (clear_visited): New function.
13762 (create_phi_basis_1): Rename from create_phi_basis, set visited
13763 and cached_basis fields.
13764 (create_phi_basis): New wrapper function.
13765 (phi_add_costs_1): Rename from phi_add_costs, add spread
13766 parameter, set visited field, short-circuit when limits reached.
13767 (phi_add_costs): New wrapper function.
13768 (record_phi_increments_1): Rename from record_phi_increments, set
13770 (record_phi_increments): New wrapper function.
13771 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13772 (phi_incr_cost): New wrapper function.
13773 (all_phi_incrs_profitable_1): Rename from
13774 all_phi_incrs_profitable, set visited field.
13775 (all_phi_incrs_profitable): New wrapper function.
13777 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13778 Alan Hayward <alan.hayward@arm.com>
13779 David Sherwood <david.sherwood@arm.com>
13781 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13782 that takes the outer and inner modes.
13783 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13784 comparison as the canonical test for a paradoxical subreg.
13785 * combine.c (simplify_set): Use paradoxical_subreg_p.
13786 (make_extraction): Likewise.
13787 (force_to_mode): Likewise.
13788 (rtx_equal_for_field_assignment_p): Likewise.
13789 (gen_lowpart_for_combine): Likewise.
13790 (simplify_comparison): Likewise.
13791 * cse.c (equiv_constant): Likewise.
13792 * expmed.c (store_bit_field_1): Likewise.
13793 * final.c (alter_subreg): Likewise.
13794 * fwprop.c (propagate_rtx): Likewise.
13795 (forward_propagate_subreg): Likewise.
13796 * ira-conflicts.c (ira_build_conflicts): Likewise.
13797 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13798 * lra-constraints.c (curr_insn_transform): Likewise.
13799 (split_reg): Likewise.
13800 * lra-eliminations.c (move_plus_up): Likewise.
13801 (lra_eliminate_regs_1): Likewise.
13802 * recog.c (general_operand): Likewise.
13803 * ree.c (combine_reaching_defs): Likewise.
13804 * reload.c (push_reload): Likewise.
13805 (find_reloads): Likewise.
13806 * reload1.c (elimination_effects): Likewise.
13807 (compute_reload_subreg_offset): Likewise.
13808 (choose_reload_regs): Likewise.
13809 * rtlanal.c (subreg_lsb_1): Likewise.
13810 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13811 (simplify_subreg): Likewise.
13812 * var-tracking.c (track_loc_p): Likewise.
13813 * emit-rtl.c (byte_lowpart_offset): Likewise.
13814 (paradoxical_subreg_p): Delete out-of-line definition.
13816 2017-08-22 Jeff Law <law@redhat.com>
13818 PR tree-optimization/81741
13819 PR tree-optimization/71947
13820 * tree-ssa-dom.c: Include tree-inline.h.
13821 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13822 equivalences if one is more expensive to compute than the other.
13823 * tree-ssa-scopedtables.h (class const_or_copies): Make
13824 record_const_or_copy_raw method private.
13825 (class avail_exprs_stack): New method simplify_binary_operation.
13826 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13827 avail_exprs_stack::simplify_binary_operation as needed.
13828 (avail_exprs_stack::simplify_binary_operation): New function.
13830 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13832 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13833 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13834 (DOT_SYMBOLS): Likewise.
13835 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13836 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13837 (RS6000_ABI_NAME): Likewise.
13838 (TARGET_CMODEL): Likewise.
13839 (TOC_SECTION_ASM_OP): Likewise.
13840 (SET_CMODEL): New macro.
13841 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13843 2017-08-22 Richard Biener <rguenther@suse.de>
13845 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13846 to allow for free-lang-data replacements similar to verify_type_variant.
13848 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13849 Alan Hayward <alan.hayward@arm.com>
13850 David Sherwood <david.sherwood@arm.com>
13852 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13853 VOIDmode for the LABEL_REF.
13855 2017-08-22 Richard Biener <rguenther@suse.de>
13857 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13859 2017-08-22 Richard Biener <rguenther@suse.de>
13861 * common.opt (feliminate-dwarf2-dups): Ignore.
13862 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13863 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13864 same_die_p_wrap, compute_section_prefix,
13865 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13866 (comdat_symbol_id, comdat_symbol_number): Likewise.
13867 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13869 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13870 (output_die): Mark unreachable path unreachable.
13871 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13872 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13873 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13874 (dwarf2out_early_finish): Likewise.
13876 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13878 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13880 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13883 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13884 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13885 (avr_attribute_table) <io, io_low, address>: Initialize
13886 .decl_required with true.
13888 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13890 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13891 undocumented debugging options.
13892 (-mvsx-scalar-double): Likewise.
13893 (-mallow-df-permute): Likewise.
13894 (-mvectorize-builtins): Likewise.
13895 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13896 (rs6000_builtin_vectorized_function): Likewise.
13897 (rs6000_builtin_md_vectorized_function): Likewise.
13898 (rs6000_opt_vars): Likewise.
13900 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13903 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13904 (*btrq_imm): Rename from *btrq.
13905 (*btcq_imm): Rename from *btcq.
13906 (btsc): New code attribute.
13907 (*<btsc><mode>): New insn pattern.
13908 (*btr<mode>): Ditto.
13909 (*<btsc><mode>_mask): New insn_and_split pattern.
13910 (*btr<mode>_mask): Ditto.
13912 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13913 Alan Hayward <alan.hayward@arm.com>
13914 David Sherwood <david.sherwood@arm.com>
13916 * function.c (pad_below): Simplify padding calculation.
13918 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13919 Alan Hayward <alan.hayward@arm.com>
13920 David Sherwood <david.sherwood@arm.com>
13922 * target.def (function_prologue): Remove frame size argument.
13923 (function_epilogue): Likewise.
13924 * doc/tm.texi: Regenerate.
13925 * final.c (final_start_function): Update call to function_prologue.
13926 (final_end_function): Update call to function_epilogue.
13927 (default_function_pro_epilogue): Remove frame size argument.
13928 * output.h (default_function_pro_epilogue): Likewise.
13929 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13930 (arm_output_function_prologue): Likewise.
13931 * config/frv/frv.c (frv_function_prologue): Likewise.
13932 (frv_function_epilogue): Likewise.
13933 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13934 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13935 (ia64_output_function_epilogue): Likewise.
13936 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13937 (m32r_output_function_epilogue): Likewise.
13938 * config/microblaze/microblaze.c (microblaze_function_prologue)
13939 (microblaze_function_epilogue): Likewise.
13940 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13941 (mips_output_function_epilogue): Likewise.
13942 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13943 (mmix_target_asm_function_epilogue): Likewise.
13944 * config/msp430/msp430.c (msp430_start_function): Likewise.
13945 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13946 (nds32_asm_function_epilogue): Likewise.
13947 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13948 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13949 (pa_output_function_epilogue): Likewise.
13950 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13951 (rs6000_output_function_epilogue): Likewise.
13952 * config/rl78/rl78.c (rl78_start_function): Likewise.
13953 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13954 (rs6000_output_function_epilogue): Likewise.
13955 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13956 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13957 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13958 (sparc_asm_function_epilogue): Likewise.
13960 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13962 * tree.h (type_has_mode_precision_p): New function.
13963 * convert.c (convert_to_integer_1): Use it.
13964 * expr.c (expand_expr_real_2): Likewise.
13965 (expand_expr_real_1): Likewise.
13966 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13967 * match.pd: Likewise.
13968 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13969 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13970 * tree-tailcall.c (process_assignment): Likewise.
13971 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13972 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13973 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13974 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13975 (vectorizable_assignment): Likewise.
13976 (vectorizable_shift): Likewise.
13977 (vectorizable_operation): Likewise.
13978 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13980 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13982 * match.pd: Add pow (C, x) simplification.
13984 2017-08-21 Richard Biener <rguenther@suse.de>
13986 PR tree-optimization/81900
13987 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13988 for blocks with abnormal predecessors.
13989 (compute_antic): Do not set visited flag prematurely.
13991 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13994 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13996 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13998 * stor-layout.h (vector_type_mode): Move to...
13999 * tree.h (vector_type_mode): ...here.
14000 * stor-layout.c (vector_type_mode): Move to...
14001 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14003 2017-08-21 Richard Biener <rguenther@suse.de>
14005 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14006 register_external_die hooks.
14007 (debug_false_tree_charstarstar_uhwistar): Declare.
14008 (debug_nothing_tree_charstar_uhwi): Likewise.
14009 * debug.c (do_nothing_debug_hooks): Adjust.
14010 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14011 (debug_nothing_tree_charstar_uhwi): Likewise.
14012 * dbxout.c (dbx_debug_hooks): Adjust.
14013 (xcoff_debug_hooks): Likewise.
14014 * sdbout.c (sdb_debug_hooks): Likewise.
14015 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14016 * dwarf2out.c (macinfo_label_base): New global.
14017 (dwarf2out_register_external_die): New function for the
14018 register_external_die hook.
14019 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14020 (dwarf2_debug_hooks): Use them.
14021 (dwarf2_lineno_debug_hooks): Adjust.
14022 (struct die_struct): Add with_offset flag.
14023 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14024 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14025 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14026 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14027 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14028 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14029 defining section names for the early LTO debug variants.
14030 (reset_indirect_string): New helper.
14031 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14032 (print_dw_val): Add support for offsetted symbol references.
14033 (get_ultimate_context): Split out from is_cxx.
14034 (is_cxx): Use get_ultimate_context.
14035 (is_fortran): Add decl overload.
14036 (compute_comp_unit_symbol): Split out worker from
14037 compute_section_prefix.
14038 (compute_section_prefix): Call compute_comp_unit_symbol and
14039 set comdat_type_p here.
14040 (output_die): Skip DIE symbol output for the LTO added one.
14041 Handle DIE symbol references with offset.
14042 (output_comp_unit): Guard section name mangling properly.
14043 For LTO debug sections emit a symbol at the section beginning
14044 which we use to refer to its DIEs.
14045 (add_abstract_origin_attribute): For DIEs registered via
14046 dwarf2out_register_external_die directly refer to the early
14047 DIE rather than indirectly through the shadow one we created.
14048 Remove obsolete call to dwarf2out_abstract_function for
14049 non-function/block origins.
14050 (gen_array_type_die): When generating early LTO debug do
14051 not emit DW_AT_string_length.
14052 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14053 late when in LTO. As suggested place a gcc_unreachable for
14054 the DECL_ABSTRACT_P case.
14055 (gen_subprogram_die): Avoid another specification DIE
14056 for early built declarations/definitions for the late LTO case.
14057 (gen_variable_die): Add type references for late duplicated VLA dies
14059 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14060 we have the abstract instance already.
14061 (process_scope_var): Adjust decl DIE contexts in LTO which
14062 first puts them in limbo.
14063 (gen_decl_die): Do not generate type DIEs late apart from
14064 types for VLAs or for decls we do not yet have a DIE. Do not
14065 call dwarf2out_abstract_function late.
14066 (dwarf2out_early_global_decl): Make sure to create DIEs
14067 for abstract instances of a decl first.
14068 (dwarf2out_late_global_decl): Adjust comment.
14069 (output_macinfo_op): With multiple macro sections use
14070 macinfo_label_base to distinguish labels.
14071 (output_macinfo): Likewise. Update macinfo_label_base.
14072 Pass in the line info label.
14073 (note_variable_value_in_expr): When generating LTO resolve
14074 all variable values here by generating DIEs as needed.
14075 (init_sections_and_labels): Add early LTO debug flag parameter
14076 and generate different sections and names if set. Add generation
14077 counter for the labels so we can have multiple of them.
14078 (reset_dies): Helper to allow DIEs to be output multiple times.
14079 (dwarf2out_finish): When outputting DIEs to the fat part of an
14080 LTO object first reset DIEs.
14081 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14082 (modified_type_die): Check for decl_ultimate_origin being self
14084 (gen_type_die_with_usage): Likewise.
14085 (gen_typedef_die): Allow decl_ultimate_origin being self.
14086 (set_decl_abstract_flags): Remove.
14087 (set_block_abstract_flags): Likewise.
14088 (dwarf2out_abstract_function): Treat the early generated DIEs
14089 as the abstract copy and only add DW_AT_inline and
14090 DW_AT_artificial here and call set_decl_origin_self.
14091 If the DIE has an abstract origin don't do anything.
14092 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14093 if we have none yet (Go fails to build one, PR78628).
14094 (variably_modified_type_p): Prevent endless recursion for Ada
14095 cyclic pointer types.
14096 * lto-streamer-in.c: Include debug.h.
14097 (dref_queue): New global.
14098 (lto_read_tree_1): Stream in DIE references.
14099 (lto_input_tree): Register DIE references.
14100 (input_function): Stream DECL_DEBUG_ARGS.
14101 * lto-streamer-out.c: Include debug.h.
14102 (lto_write_tree_1): Output DIE references.
14103 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14104 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14105 (output_function): Stream DECL_DEBUG_ARGS.
14106 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14107 Stream DECL_ABSTRACT_ORIGIN.
14108 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14109 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14110 DECL_CONTEXT for file-scope decls.
14111 * lto-streamer.h (struct dref_entry): Declare.
14112 (dref_queue): Likewise.
14113 * cfgexpand.c (pass_expand::execute): Do not call the
14114 outlining_inline_function hook here.
14115 * lto-wrapper.c (debug_obj): New global.
14116 (tool_cleanup): Unlink it if required.
14117 (debug_objcopy): New function.
14118 (run_gcc): Handle early debug sections in the IL files by
14119 extracting them to separate files, partially linkin them and
14120 feeding the result back as result to the linker.
14121 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14122 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14123 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14124 sections into a separate segment.
14125 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14127 (darwin_asm_dwarf_section): Likewise.
14128 (darwin_asm_output_dwarf_offset): Likewise.
14129 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14131 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14132 Alan Hayward <alan.hayward@arm.com>
14133 David Sherwood <david.sherwood@arm.com>
14135 * read-md.h (md_reader::record_potential_iterator_use): Replace
14136 pointer argument with an rtx and an index.
14137 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14138 (apply_mode_iterator): Likewise.
14139 (apply_code_iterator): Likewise.
14140 (apply_int_iterator): Likewise.
14141 (apply_subst_iterator): Likewise.
14142 (record_iterator_use): Likewise.
14143 (record_attribute_use): Likewise.
14144 (md_reader::record_potential_iterator_use): Likewise. Update calls
14145 to record_iterator_use and apply_iterator.
14146 (iterator_use): Replace ptr with x and index.
14147 (attribute_use): Likewise.
14148 (apply_attribute_uses): Update calls to apply_iterator.
14149 (apply_iterators): Likewise. Update initialization of iterator_use.
14150 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14151 and record_potential_iterator_use.
14152 (rtx_reader::read_rtx_operand): Likewise.
14154 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14155 Alan Hayward <alan.hayward@arm.com>
14156 David Sherwood <david.sherwood@arm.com>
14158 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14161 2017-08-21 Richard Biener <rguenther@suse.de>
14163 PR middle-end/81884
14164 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14165 at struct end conservatively when comparing common bases.
14167 2017-08-21 Richard Biener <rguenther@suse.de>
14169 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14170 (mem_ref_in_stmt): Remove.
14171 (determine_max_movement): Use ref index to get at the reference.
14172 (invariantness_dom_walker::before_dom_children): Deal with
14173 lim data already initialized.
14174 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14176 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14178 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14179 (TARGET_ISA_ROUND): Ditto.
14180 (TARGET_ROUND): Ditto.
14181 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14182 * config/i386/i386.md: Ditto.
14183 * config/i386/sse.md: Ditto.
14184 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14185 with OPTION_MASK_ISA_SSE4_1.
14187 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14190 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14191 __builtin_ia32_lzcnt_u16.
14193 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14196 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14197 (rs6000_set_current_function): Rewrite function to use it.
14199 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14202 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14203 and TYPE_WARN_IF_NOT_ALIGN.
14204 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14205 (handle_warn_if_not_align): New.
14206 (place_union_field): Call handle_warn_if_not_align.
14207 (place_field): Call handle_warn_if_not_align.
14208 Copy TYPE_WARN_IF_NOT_ALIGN.
14209 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14210 (layout_type): Likewise.
14211 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14213 (tree_decl_common): Add warn_if_not_align.
14214 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14215 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14216 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14217 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14218 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14219 * doc/extend.texi: Document warn_if_not_aligned attribute.
14220 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14222 2017-08-17 Martin Liska <mliska@suse.cz>
14225 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14226 (get_data_dependence): Use it as pointer type.
14227 (distribute_loop): Likewise.
14229 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14231 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14232 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14233 (p8_vmrgow_<mode>_direct): New define_insn.
14234 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14235 handle endianness for vmrgew and vmrgow permute patterns.
14237 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14239 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14240 * config/rs6000/rs6000-cpus.def: Remove comment.
14241 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14242 (POWERPC_MASKS): Likewise.
14243 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14244 use of TARGET_VSX_TIMODE.
14245 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14246 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14247 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14248 (rs6000_option_override_internal): Remove dead code.
14249 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14250 (rs6000_legitimize_reload_address): Likewise.
14251 (rs6000_legitimate_address_p): Likewise.
14252 (rs6000_opt_masks): Delete "vsx-timode".
14253 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14254 from function comment.
14255 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14256 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14257 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14259 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14260 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14261 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14262 (VSX_TI): Likewise.
14264 (define_peephole2): Likewise.
14266 2017-08-17 Martin Sebor <msebor@redhat.com>
14269 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14270 past the end of an array.
14271 (test_pp_format): Add test cases.
14273 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14275 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14276 missing ECF_NOTHROW flags.
14278 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14281 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14282 operands residing in integer registers.
14283 (*vsx_le_perm_load_<mode>): Likewise.
14284 (*vsx_le_perm_store_<mode>): Likewise.
14285 (define_peephole2): Add peepholes to optimize the above.
14287 2017-08-17 Marek Polacek <polacek@redhat.com>
14289 PR middle-end/81814
14290 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14291 to mimic what shorten_compare did. Change the return type to bool.
14292 (fold_cond_expr_with_comparison): Update call to
14293 operand_equal_for_comparison_p.
14294 (fold_ternary_loc): Likewise.
14296 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14298 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14300 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14302 2017-08-17 Richard Biener <rguenther@suse.de>
14304 * tree-ssa-structalias.c (solve_graph): When propagating
14305 to successors update the graphs succ edges and avoid duplicate work.
14307 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14310 * config/i386/i386.c (ix86_option_override_internal): Save target
14311 specific options after ix86_stack_protector_guard_reg was changed.
14313 2017-08-17 Richard Biener <rguenther@suse.de>
14315 PR tree-optimization/81827
14316 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14317 (new_var_info): Initialize it conservatively.
14318 (get_call_vi): Mark register vars.
14319 (new_scalar_tmp_constraint_exp): Likewise.
14320 (handle_rhs_call): Likewise.
14321 (handle_const_call): Likewise.
14322 (create_function_info_for): Likewise.
14323 (solve_constraints): Sort varinfos to separate register from
14324 non-register vars to pack points-to solution bitmaps during
14327 2017-08-17 Marek Polacek <polacek@redhat.com>
14329 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14331 2017-08-17 Richard Biener <rguenther@suse.de>
14333 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14334 to true when overflow is undefined and we saturated the result.
14336 2017-08-17 Alan Modra <amodra@gmail.com>
14339 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14340 Don't use store multiple if only one reg needs saving.
14341 (interesting_frame_related_regno): New function.
14342 (rs6000_frame_related): Don't emit frame info for regs that
14344 (rs6000_emit_epilogue): Likewise.
14346 2017-08-16 Nathan Sidwell <nathan@acm.org>
14348 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14349 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14350 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14351 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14352 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14353 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14354 (verify_type): Adjust for TYPE_BINFO move.
14355 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14356 process TYPE_BINFO directly.
14357 (hash_tree): Likewise.
14358 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14360 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14363 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14365 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14367 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14370 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14371 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14372 Add ix86_binary_operator_ok to insn constraint.
14373 (*iordi_1_bts): Ditto.
14374 (*xordi_1_btc): Ditto.
14375 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14376 Update corresponding peephole2 pattern.
14380 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14382 PR tree-optimization/81832
14383 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14384 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14386 2017-08-16 Marek Polacek <polacek@redhat.com>
14389 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14390 regarding size_zero_node.
14392 2017-08-16 Martin Liska <mliska@suse.cz>
14395 * config.gcc: Respect previously set extra_objs in case
14398 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14400 PR tree-optimization/81835
14401 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14402 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14403 not depend on the phi.
14405 2017-08-16 Alan Modra <amodra@gmail.com>
14407 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14410 2017-08-16 Alan Modra <amodra@gmail.com>
14412 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14413 (save_reg_p): ..into this. Update all callers.
14414 (first_reg_to_save): Simplify.
14416 2017-08-16 Alan Modra <amodra@gmail.com>
14418 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14421 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14425 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14426 constant count if that count is less than 32.
14428 2017-08-15 Nathan Sidwell <nathan@acm.org>
14430 * gcc.c (execute): Emit friendlier message if inferior is killed
14431 by an external cause.
14433 2017-08-15 Richard Biener <rguenther@suse.de>
14435 PR tree-optimization/81790
14436 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14437 CONSTRUCTORs from simplifying and VN.
14439 2017-08-14 Martin Sebor <msebor@redhat.com>
14441 * builtin-attrs.def: Add comments.
14443 2017-08-14 Martin Sebor <msebor@redhat.com>
14446 * doc/extend.texi (attribute nonstring): Document new attribute.
14448 2017-08-14 Martin Sebor <msebor@redhat.com>
14451 * tree-diagnostic.c (default_tree_printer): Handle %G.
14452 * gimple-pretty-print.h (percent_G_format): Declare new function.
14453 * gimple-pretty-print.c (percent_G_format): Define.
14454 * tree-pretty-print.c (percent_K_format): Add argument.
14456 2017-08-14 Martin Sebor <msebor@redhat.com>
14458 PR translation/79998
14459 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14460 Remove a stray space.
14462 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14465 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14466 (*iordi_1_bts): Ditto.
14467 (*xordi_1_btc): Ditto.
14469 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14472 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14473 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14475 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14476 quoted strings, and make more translator-friendly.
14477 (darwin_rs6000_override_options): Likewise.
14478 (rs6000_option_override_internal): Likewise.
14479 (rs6000_return_in_memory): Fix overlong line.
14480 (init_cmulative_args): Use quoted strings, and make more
14481 translator-friendly.
14482 (rs6000_pass_by_reference): Fix overlong line.
14483 (def_builtin): Use quoted strings.
14484 (altivec_expand_predicate_builtin): Use quoted strings, and make
14485 more translator-friendly.
14486 (htm_expand_builtin): Use quoted strings.
14487 (cpu_expand_builtin): Use quoted strings, and make more
14488 translator-friendly.
14489 (altivec_expand_builtin): Likewise.
14490 (paired_expand_predicate_builtin): Likewise.
14491 (rs6000_invalid_builtin): Likewise.
14492 (builtin_function_type): Use quoted strings.
14493 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14494 more translator-friendly.
14495 (rs6000_trampoline_init): Likewise.
14496 (rs6000_handle_altivec_attribute): Likewise.
14497 (rs6000_inner_target_options): Use quoted strings.
14498 (rs6000_disable_incompatible_switches): Likewise.
14499 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14500 strings, and make more translator-friendly.
14501 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14503 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14505 PR tree-optimization/81799
14506 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14507 cond_expr to simple gimple operand.
14509 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14511 PR middle-end/46932
14512 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14514 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14518 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14519 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14520 TARGET_GASISR_PROLOGUES.
14521 * config/avr/avr.c (avr_option_override): Same.
14522 (avr_pass_pre_proep::execute): Same.
14524 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14527 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14528 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14530 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14532 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14533 define_insn_and_split. Split to a memory load from 0 in
14534 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14535 using PTR mode iterator.
14536 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14537 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14538 (*add_tp_<mode>): Redefine as define_insn_and_split.
14539 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14540 address space. Merge with *add_tp_x32 using PTR mode iterator.
14541 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14542 Split to an add with a memory load from 0 in
14543 DEFAULT_TLS_SEG_REG address space.
14545 2017-08-12 Andrew Pinski <apinski@cavium.com>
14547 * config/aarch64/aarch64-option-extensions.def (rdma):
14548 Fix feature string to what Linux prints out in /proc/cpuinfo.
14550 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14553 * dwarf2out.c (modified_type_die): For C typedef types that have
14554 an ultimate origin, process the ultimate origin instead of the
14556 (gen_typedef_die): Assert that input DECLs have no ultimate
14558 (gen_type_die_with_usage): For typedef variants that have an
14559 ultimate origin, just call gen_decl_die on the original DECL.
14560 (process_scope_var): Avoid creating DIEs for local typedefs and
14561 concrete static variables.
14563 2017-08-12 Alan Modra <amodra@gmail.com>
14567 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14568 match gnu-user.h startfile.
14569 (ENDFILE_LINUX_SPEC): Similarly.
14571 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14574 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14576 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14578 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14579 * config/aarch64/aarch64.md (mov<mode>): Change.
14580 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14581 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14582 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14584 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14586 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14587 for storage order barriers.
14589 2017-08-11 Martin Liska <mliska@suse.cz>
14592 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14593 variables of void type.
14595 2017-08-11 Martin Liska <mliska@suse.cz>
14597 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14598 TARGET_SUPPORTS_ALIASES.
14599 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14600 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14601 (optimize_weakref): Likewise.
14602 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14603 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14604 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14606 2017-08-11 Martin Liska <mliska@suse.cz>
14609 * config/i386/i386.c (make_resolver_func): Do complete
14610 refactoring of the function.
14612 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14615 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14616 * config/i386/i386.c (ix86_stack_protect_guard): Use
14617 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14618 * doc/invoke.texi (x86 Options): Document
14619 -mstack-protector-guard-symbol= option.
14621 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14623 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14624 * config/i386/i386.c (ix86_split_stack_guard): New function.
14625 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14626 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14627 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14628 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14629 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14630 (split_stack_space_check): Call ix86_split_stack_guard.
14632 2017-08-10 Martin Sebor <msebor@redhat.com>
14634 * print-tree.c (print_node): Print location using the established
14636 Replace spaces with colons.
14637 (debug_raw, debug): Ditto.
14639 2017-08-10 Martin Sebor <msebor@redhat.com>
14642 * pretty-print.c (pp_format): Correct the handling of %s precision.
14644 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14647 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14649 (ix86_finalize_stack_frame_flags): This. Also clear
14650 frame_pointer_needed if -fno-omit-frame-pointer is used without
14652 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14653 with ix86_finalize_stack_frame_flags.
14654 (ix86_expand_epilogue): Likewise.
14655 (ix86_expand_split_stack_prologue): Likewise.
14656 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14658 2017-08-10 Martin Liska <mliska@suse.cz>
14661 * c-attribs.c (handle_target_attribute):
14662 Report warning for an empty string argument of target attribute.
14664 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14667 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14669 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14670 or DECL_NONLOCAL labels.
14671 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14672 or DECL_NONLOCAL labels here.
14674 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14676 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14677 to indicate when early gimple folding has been disabled.
14678 (rs6000_gimple_fold_builtin): Add debug content.
14679 (rs6000_invalid_builtin): Fix whitespace.
14680 (rs6000_expand_builtin): Fix whitespace.
14681 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14683 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14686 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14687 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14689 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14691 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14692 (qdf24xx): Likewise.
14693 * config/aarch64/aarch64.md: Include falkor.md.
14694 * config/aarch64/falkor.md: New.
14696 2017-08-09 Marek Polacek <polacek@redhat.com>
14699 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14700 * diagnostic.c (emit_diagnostic): Add a comment.
14701 (emit_diagnostic_valist): New function.
14703 2017-08-09 Marek Polacek <polacek@redhat.com>
14706 * input.c (make_location): New overload.
14707 * input.h (make_location): Declare.
14709 2017-08-08 Alan Modra <amodra@gmail.com>
14710 H.J. Lu <hongjiu.lu@intel.com>
14713 * gcc.c (NO_PIE_SPEC): Delete.
14714 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14716 (LINK_PIE_SPEC): ..to here.
14717 (LINK_COMMAND_SPEC): Support -no-pie.
14718 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14719 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14721 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14722 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14723 (ENDFILE_CRTEND_SPEC): Similarly.
14725 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14728 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14729 (mstack-protector-guard-offset=): Ditto.
14730 * config/i386/i386.c (ix86_option_override): Handle
14731 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14733 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14734 ix86_stack_protect_guard_offset variables.
14735 (TARGET_STACK_PROTECT_GUARD): Always define.
14736 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14737 and -mstack-protector-guard-offset= options.
14739 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14741 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14742 boundary case for the last candidate.
14744 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14746 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14747 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14749 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14751 PR middle-end/19706
14752 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14753 * config/aarch64/aarch64-builtins.c
14754 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14755 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14756 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14758 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14759 Andrew Pinski <pinskia@gmail.com>
14761 PR middle-end/19706
14762 * internal-fn.def (XORSIGN): New.
14763 * optabs.def (xorsign_optab): New.
14764 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14765 (convert_expand_mult_copysign): New.
14766 (pass_optimize_widening_mul::execute): Call
14767 convert_expand_mult_copysign.
14769 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14771 PR tree-optimization/81354
14772 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14773 Insert on edges rather than explicitly creating landing pads.
14774 (analyze_candidates_and_replace): Commit edge inserts.
14776 2017-08-08 Richard Biener <rguenther@suse.de>
14778 PR middle-end/81719
14779 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14780 (expand_simple_operations): Also look through ADDR_EXPRs with
14781 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14783 2017-08-08 Richard Biener <rguenther@suse.de>
14785 PR tree-optimization/81723
14786 * tree-vect-slp.c (struct bst_traits): New hash traits.
14787 (bst_fail): New global.
14788 (vect_build_slp_tree_2): New worker, split out from ...
14789 (vect_build_slp_tree): ... this now wrapping it with using
14790 bst_fail set to cache SLP tree build fails. Properly handle
14792 (vect_analyze_slp_instance): Allocate and free bst_fail.
14794 2017-08-08 Martin Liska <mliska@suse.cz>
14797 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14798 LABEL_DECLs that can be from a different function.
14800 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14802 PR tree-optimization/81744
14803 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14804 loop's number of iterations.
14806 2017-08-08 Martin Liska <mliska@suse.cz>
14808 * asan.c: Include header files.
14809 * attribs.c (build_decl_attribute_variant): New function moved
14811 (build_type_attribute_qual_variant): Likewise.
14812 (cmp_attrib_identifiers): Likewise.
14813 (simple_cst_list_equal): Likewise.
14814 (omp_declare_simd_clauses_equal): Likewise.
14815 (attribute_value_equal): Likewise.
14816 (comp_type_attributes): Likewise.
14817 (build_type_attribute_variant): Likewise.
14818 (lookup_ident_attribute): Likewise.
14819 (remove_attribute): Likewise.
14820 (merge_attributes): Likewise.
14821 (merge_type_attributes): Likewise.
14822 (merge_decl_attributes): Likewise.
14823 (merge_dllimport_decl_attributes): Likewise.
14824 (handle_dll_attribute): Likewise.
14825 (attribute_list_equal): Likewise.
14826 (attribute_list_contained): Likewise.
14827 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14828 (lookup_attribute_by_prefix): Likewise.
14829 * bb-reorder.c: Include header files.
14830 * builtins.c: Likewise.
14831 * calls.c: Likewise.
14832 * cfgexpand.c: Likewise.
14833 * cgraph.c: Likewise.
14834 * cgraphunit.c: Likewise.
14835 * convert.c: Likewise.
14836 * dwarf2out.c: Likewise.
14837 * final.c: Likewise.
14838 * fold-const.c: Likewise.
14839 * function.c: Likewise.
14840 * gimple-expr.c: Likewise.
14841 * gimple-fold.c: Likewise.
14842 * gimple-pretty-print.c: Likewise.
14843 * gimple.c: Likewise.
14844 * gimplify.c: Likewise.
14845 * hsa-common.c: Likewise.
14846 * hsa-gen.c: Likewise.
14847 * internal-fn.c: Likewise.
14848 * ipa-chkp.c: Likewise.
14849 * ipa-cp.c: Likewise.
14850 * ipa-devirt.c: Likewise.
14851 * ipa-fnsummary.c: Likewise.
14852 * ipa-inline.c: Likewise.
14853 * ipa-visibility.c: Likewise.
14855 * lto-cgraph.c: Likewise.
14856 * omp-expand.c: Likewise.
14857 * omp-general.c: Likewise.
14858 * omp-low.c: Likewise.
14859 * omp-offload.c: Likewise.
14860 * omp-simd-clone.c: Likewise.
14861 * opts-global.c: Likewise.
14862 * passes.c: Likewise.
14863 * predict.c: Likewise.
14864 * sancov.c: Likewise.
14865 * sanopt.c: Likewise.
14866 * symtab.c: Likewise.
14867 * toplev.c: Likewise.
14868 * trans-mem.c: Likewise.
14869 * tree-chkp.c: Likewise.
14870 * tree-eh.c: Likewise.
14871 * tree-into-ssa.c: Likewise.
14872 * tree-object-size.c: Likewise.
14873 * tree-parloops.c: Likewise.
14874 * tree-profile.c: Likewise.
14875 * tree-ssa-ccp.c: Likewise.
14876 * tree-ssa-live.c: Likewise.
14877 * tree-ssa-loop.c: Likewise.
14878 * tree-ssa-sccvn.c: Likewise.
14879 * tree-ssa-structalias.c: Likewise.
14880 * tree-ssa.c: Likewise.
14881 * tree-streamer-in.c: Likewise.
14882 * tree-vectorizer.c: Likewise.
14883 * tree-vrp.c: Likewise.
14884 * tsan.c: Likewise.
14885 * ubsan.c: Likewise.
14886 * varasm.c: Likewise.
14887 * varpool.c: Likewise.
14888 * tree.c: Remove functions moved to attribs.[ch].
14889 * tree.h: Likewise.
14890 * config/aarch64/aarch64.c: Add attrs.h header file.
14891 * config/alpha/alpha.c: Likewise.
14892 * config/arc/arc.c: Likewise.
14893 * config/arm/arm.c: Likewise.
14894 * config/avr/avr.c: Likewise.
14895 * config/bfin/bfin.c: Likewise.
14896 * config/c6x/c6x.c: Likewise.
14897 * config/cr16/cr16.c: Likewise.
14898 * config/cris/cris.c: Likewise.
14899 * config/darwin.c: Likewise.
14900 * config/epiphany/epiphany.c: Likewise.
14901 * config/fr30/fr30.c: Likewise.
14902 * config/frv/frv.c: Likewise.
14903 * config/ft32/ft32.c: Likewise.
14904 * config/h8300/h8300.c: Likewise.
14905 * config/i386/winnt.c: Likewise.
14906 * config/ia64/ia64.c: Likewise.
14907 * config/iq2000/iq2000.c: Likewise.
14908 * config/lm32/lm32.c: Likewise.
14909 * config/m32c/m32c.c: Likewise.
14910 * config/m32r/m32r.c: Likewise.
14911 * config/m68k/m68k.c: Likewise.
14912 * config/mcore/mcore.c: Likewise.
14913 * config/microblaze/microblaze.c: Likewise.
14914 * config/mips/mips.c: Likewise.
14915 * config/mmix/mmix.c: Likewise.
14916 * config/mn10300/mn10300.c: Likewise.
14917 * config/moxie/moxie.c: Likewise.
14918 * config/msp430/msp430.c: Likewise.
14919 * config/nds32/nds32-isr.c: Likewise.
14920 * config/nds32/nds32.c: Likewise.
14921 * config/nios2/nios2.c: Likewise.
14922 * config/nvptx/nvptx.c: Likewise.
14923 * config/pa/pa.c: Likewise.
14924 * config/pdp11/pdp11.c: Likewise.
14925 * config/powerpcspe/powerpcspe.c: Likewise.
14926 * config/riscv/riscv.c: Likewise.
14927 * config/rl78/rl78.c: Likewise.
14928 * config/rx/rx.c: Likewise.
14929 * config/s390/s390.c: Likewise.
14930 * config/sh/sh.c: Likewise.
14931 * config/sol2.c: Likewise.
14932 * config/sparc/sparc.c: Likewise.
14933 * config/spu/spu.c: Likewise.
14934 * config/stormy16/stormy16.c: Likewise.
14935 * config/tilegx/tilegx.c: Likewise.
14936 * config/tilepro/tilepro.c: Likewise.
14937 * config/v850/v850.c: Likewise.
14938 * config/vax/vax.c: Likewise.
14939 * config/visium/visium.c: Likewise.
14940 * config/xtensa/xtensa.c: Likewise.
14942 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14945 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14946 constraints since the -mupper-regs-* switches have been
14948 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14949 into a vector from a double word element that was extracted from
14950 another vector, and eliminate extra XXPERMDI instructions.
14951 (vsx_concat_<mode>_2): Likewise.
14952 (vsx_concat_<mode>_3): Likewise.
14953 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14954 concat to allow optimizing inserts from previous extracts.
14956 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14958 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14959 memory reference to a SSP offset in TLS address space.
14960 (ix86_print_operand) <case '@'>: Remove.
14961 (ix86_print_operand_punct_valid_p): Remove '@' code.
14962 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14963 UNSPEC_SP_TLS_TEST.
14964 (stack_tls_protect_set_<mode>): Remove.
14965 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14966 (stack_tls_protect_test_<mode>): Remove.
14967 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14969 2017-08-07 Olivier Hainque <hainque@adacore.com>
14972 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14974 2017-08-07 Douglas Rupp <rupp@adacore.com>
14976 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14977 variable was referenced as multidir in command.
14979 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14982 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14985 2017-08-07 Martin Liska <mliska@suse.cz>
14987 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14988 * config/rl78/rl78.c: Add include of attribs.h.
14989 * config/sh/sh.c: Likewise.
14990 * config/v850/v850.c: Likewise.
14992 2017-08-07 Tom de Vries <tom@codesourcery.com>
14994 PR middle-end/78266
14995 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14997 2017-08-07 Martin Liska <mliska@suse.cz>
14999 * config/mips/mips.c: Include attribs.h.
15001 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15004 * doc/invoke.texi: Document change in behvaior for -Ofast for
15007 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15009 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15011 (aarch64_pop_regs): Likewise.
15012 (aarch64_gen_load_pair): Likewise.
15013 (aarch64_save_callee_saves): Likewise.
15014 (aarch64_restore_callee_saves): Likewise.
15016 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15018 * config/i386/i386.c: Revert the last change.
15020 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15023 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15025 (ix86_finalize_stack_frame_flags): This. Also clear
15026 frame_pointer_needed if -fno-omit-frame-pointer is used without
15028 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15029 with ix86_finalize_stack_frame_flags.
15030 (ix86_expand_epilogue): Likewise.
15031 (ix86_expand_split_stack_prologue): Likewise.
15033 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15036 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15037 to P_AES for Westmere.
15039 2017-08-07 Jonathan Yong <10walls@gmail.com>
15041 * config/i386/mingw.opt (fset-stack-executable): Removed.
15042 * config/i386/cygming.opt (fset-stack-executable): Moved
15044 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15046 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15048 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15050 2017-08-07 Marek Polacek <polacek@redhat.com>
15052 PR middle-end/81737
15053 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15055 2017-08-07 Martin Liska <mliska@suse.cz>
15057 * attribs.h (canonicalize_attr_name): New function.
15058 (cmp_attribs): Move from c-format.c and adjusted.
15059 (is_attribute_p): Moved from tree.h.
15060 * tree-inline.c: Add new includes.
15061 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15062 (private_is_attribute_p): Remove.
15063 (private_lookup_attribute): Likewise.
15064 (private_lookup_attribute_by_prefix): Simplify.
15065 (remove_attribute): Use is_attribute_p.
15066 * tree.h: Remove removed declarations.
15068 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15070 PR middle-end/81698
15071 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15072 instead of computing it in the function. Formatting fix.
15073 (expand_case): Don't rely on default_edge being the first edge,
15074 clear it if removing it, pass default_edge to
15075 emit_case_dispatch_table.
15076 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15079 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15081 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15082 insn in the function, emit NOP after the insn.
15084 2017-08-06 Tom de Vries <tom@codesourcery.com>
15086 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15089 2017-08-06 Tom de Vries <tom@codesourcery.com>
15091 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15094 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15096 PR tree-optimization/57371
15097 * match.pd: New pattern.
15099 2017-08-04 Marek Polacek <polacek@redhat.com>
15101 PR middle-end/81695
15102 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15103 perform the computation in offset_int.
15105 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15107 PR tree-optimization/81136
15108 * tree-vectorizer.h: Include tree-hash-traits.h.
15109 (vec_base_alignments): New typedef.
15110 (vec_info): Add a base_alignments field.
15111 (vect_record_base_alignments): Declare.
15112 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15114 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15115 (create_data_ref): Add an is_conditional_in_stmt argument.
15116 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15117 the is_conditional_in_stmt field.
15118 (data_ref_loc): Add an is_conditional_in_stmt field.
15119 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15120 (find_data_references_in_stmt): Update call to create_data_ref.
15121 (graphite_find_data_references_in_stmt): Likewise.
15122 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15123 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15124 (vect_record_base_alignment): New function.
15125 (vect_record_base_alignments): Likewise.
15126 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15127 for nested statements even if we fail to compute a misalignment.
15128 Use pooled base alignments for unconditional references.
15129 (vect_find_same_alignment_drs): Compare base addresses instead
15131 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15132 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15134 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15136 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15137 Add an explicit name for the enum. Use auto_vec for slp_instances
15138 and grouped_stores.
15139 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15141 (_bb_vec_info): Add a constructor and destructor.
15142 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15143 (destroy_loop_vec_info): Delete.
15144 (vect_destroy_datarefs): Likewise.
15145 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15146 (vec_info::vec_info): New function.
15147 (vec_info::~vec_info): Likewise.
15148 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15149 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15150 destroy_loop_vec_info.
15151 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15152 (_loop_vec_info::_loop_vec_info): ...this.
15153 (destroy_loop_vec_info): Replace with...
15154 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15155 the stmt_vec_infos. Leave handling of vec_info information to its
15156 destructor. Remove explicit vector releases.
15157 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15158 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15159 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15160 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15161 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15162 (destroy_bb_vec_info): Replace with...
15163 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15164 information to its destructor.
15165 (vect_slp_analyze_bb_1): Use new and delete instead of
15166 new_bb_vec_info and destroy_bb_vec_info.
15167 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15170 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15172 * tree-data-ref.h (subscript): Add access_fn field.
15173 (data_dependence_relation): Add could_be_independent_p.
15174 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15175 (same_access_functions): Move to tree-data-ref.c.
15176 * tree-data-ref.c (ref_contains_union_access_p): New function.
15177 (access_fn_component_p): Likewise.
15178 (access_fn_components_comparable_p): Likewise.
15179 (dr_analyze_indices): Add a reference to access_fn_component_p.
15180 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15182 (constant_access_functions): Likewise.
15183 (add_other_self_distances): Likewise.
15184 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15185 (initialize_data_dependence_relation): Use XCNEW and remove
15186 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15187 of access functions that have the same type. Allow the
15188 subsequence to end with different bases in some circumstances.
15189 Record the chosen access functions in SUB_ACCESS_FN.
15190 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15191 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15192 (subscript_dependence_tester_1): Likewise dra and drb.
15193 (build_classic_dist_vector): Update calls accordingly.
15194 (subscript_dependence_tester): Likewise.
15195 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15196 DDR_COULD_BE_INDEPENDENT_P.
15197 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15198 comp_alias_ddrs instead of may_alias_ddrs.
15199 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15201 (vect_analyze_data_ref_dependence): Use it if
15202 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15203 distance vectors if that fails.
15204 (dependence_distance_ge_vf): New function.
15205 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15206 LOOP_VINFO_MAY_ALIAS_DDRS.
15208 2017-08-04 Richard Biener <rguenther@suse.de>
15210 PR middle-end/81705
15211 * fold-const.c (fold_binary_loc): Properly restrict
15212 minus_var0 && minus_var1 case when associating undefined overflow
15215 2017-08-04 Tom de Vries <tom@codesourcery.com>
15217 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15219 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15221 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15222 Don't start diagnostic messages with a capital letter.
15223 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15225 (rs6000_invalid_builtin): Likewise.
15226 (rs6000_trampoline_init): Likewise.
15228 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15231 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15232 after setting changeable df flags.
15234 2017-08-03 Richard Biener <rguenther@suse.de>
15236 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15237 up if the use is in USE - X.
15239 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15241 * toplev.c (dumpfile.h): New include.
15242 (internal_error_reentered): New static function. Use it...
15243 (internal_error_function): ...here to handle reentered internal_error.
15245 2017-08-03 Richard Biener <rguenther@suse.de>
15247 PR middle-end/81148
15248 * fold-const.c (split_tree): Add minus_var and minus_con
15249 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15250 here but always use minus_*.
15251 (associate_trees): Assert we never associate with MINUS_EXPR
15252 and NULL first operand. Do not recurse for PLUS_EXPR operands
15253 when associating as MINUS_EXPR either.
15254 (fold_binary_loc): Track minus_var and minus_con.
15256 2017-08-03 Tom de Vries <tom@codesourcery.com>
15259 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15260 ACCEL_COMPILER, apply finish_options on
15261 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15263 2017-08-03 Tom de Vries <tom@codesourcery.com>
15266 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15267 function_entry_patch_area_size > 0.
15269 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15272 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15273 instead of 10??LU, perform unit multiplication in wide_int,
15274 don't change alloc_object_size_limit if the limit is larger
15277 PR tree-optimization/81655
15278 PR tree-optimization/81588
15279 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15280 the case when ranges[i].low and high are 1 for unsigned type with
15283 PR middle-end/81052
15284 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15285 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15287 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15289 * tree-vrp.h: Add include guard.
15291 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15294 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15295 (ud2): New insn pattern.
15296 * config/i386/i386.c (ix86_expand_epilogue):
15297 For naked functions, generate ud2 instead of trap insn.
15299 2017-08-02 Marek Polacek <polacek@redhat.com>
15302 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15304 2017-08-02 Tom de Vries <tom@codesourcery.com>
15305 Cesar Philippidis <cesar@codesourcery.com>
15307 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15308 Add missing edge probabilities.
15310 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15312 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15313 Correct endianness.
15315 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15317 PR middle-end/79499
15318 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15319 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15320 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15321 prologue_seq sequences - if any.
15323 2017-08-02 Richard Biener <rguenther@suse.de>
15325 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15326 via vectors if supported, integer extracts via punning if supported
15327 or otherwise vector extracts.
15329 2017-08-02 Richard Biener <rguenther@suse.de>
15331 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15333 (bitmap_insert_into_set): ... this.
15335 2017-08-02 Richard Biener <rguenther@suse.de>
15337 PR tree-optimization/81633
15339 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15341 PR tree-optimization/71752
15342 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15344 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15346 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15347 (machine_function::call_ms2sysv_pad_out): Remove field.
15348 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15349 (ix86_compute_frame_layout): Likewise.
15351 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15354 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15355 attribute with interrupt attribute.
15357 2017-08-01 Andrew Pinski <apinski@cavium.com>
15359 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15360 BIT_INSERT_EXPR's operand 1
15361 to see if the types precision matches.
15363 2017-08-01 Martin Liska <mliska@suse.cz>
15365 PR middle-end/70140
15366 * builtins.c (expand_builtin_memcpy_args): Remove.
15367 (expand_builtin_memcpy): Call newly added function
15368 expand_builtin_memory_copy_args.
15369 (expand_builtin_memcpy_with_bounds): Likewise.
15370 (expand_builtin_mempcpy): Remove last argument.
15371 (expand_builtin_mempcpy_with_bounds): Likewise.
15372 (expand_builtin_memory_copy_args): New function created from
15373 expand_builtin_mempcpy_args with small modifications.
15374 (expand_builtin_mempcpy_args): Remove.
15375 (expand_builtin_stpcpy): Remove unused argument.
15376 (expand_builtin): Likewise.
15377 (expand_builtin_with_bounds): Likewise.
15379 2017-08-01 Martin Liska <mliska@suse.cz>
15382 Make mempcpy more optimal (PR middle-end/70140).
15384 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15387 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15388 __builtin_vec_cmpne verify both arguments are compatible vectors
15389 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15390 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15391 move computation of aligned to after checking the argument types.
15395 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15396 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15399 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15400 Doug Rupp <rupp@adacore.com>
15401 Olivier Hainque <hainque@adacore.com>
15403 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15404 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15406 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15407 for TARGET_OS_CPP_BUILTIN.
15408 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15409 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15411 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15412 passing required abi options to the assembler for EABI configurations.
15413 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15414 of .text.hot and .text.unlikely sections for kernel modules when
15415 using ARM style exceptions.
15416 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15417 options. Add EXTRA_CC1_SPEC.
15418 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15420 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15422 (ARM_TARGET2_DWARF_FORMAT): Define.
15423 * config/arm/t-vxworks: Adjust multilib control to removal of the
15424 Diab command line options.
15426 2017-08-01 Martin Liska <mliska@suse.cz>
15428 PR gcov-profile/81561
15429 * gcov.c (unblock): Make unblocking safe as we need to preserve
15430 index correspondence of blocks and block_lists.
15432 2017-08-01 Richard Biener <rguenther@suse.de>
15434 PR tree-optimization/81181
15435 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15436 (compute_antic): ... end of iteration here.
15438 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15440 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15441 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15442 (ftree-slp-vectorize): Likewise.
15443 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15444 can no longer be set independent of flag_tree_loop_vectorize.
15445 * omp-general.c (emp_max_vf): Likewise.
15446 * opts.c (enable_fdo_optimizations): Remove references to
15447 flag_tree_vectorize, these are now implicit.
15448 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15449 and leave it for the options machinery.
15451 2017-08-01 Martin Liska <mliska@suse.cz>
15453 PR middle-end/70140
15454 * builtins.c (expand_builtin_memcpy_args): Remove.
15455 (expand_builtin_memcpy): Call newly added function
15456 expand_builtin_memory_copy_args.
15457 (expand_builtin_memcpy_with_bounds): Likewise.
15458 (expand_builtin_mempcpy): Remove last argument.
15459 (expand_builtin_mempcpy_with_bounds): Likewise.
15460 (expand_builtin_memory_copy_args): New function created from
15461 expand_builtin_mempcpy_args with small modifications.
15462 (expand_builtin_mempcpy_args): Remove.
15463 (expand_builtin_stpcpy): Remove unused argument.
15464 (expand_builtin): Likewise.
15465 (expand_builtin_with_bounds): Likewise.
15467 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15470 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15471 print "ds:" only for immediates in generic address space.
15473 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15476 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15477 (ix86_function_ok_for_sibcall): Return false for naked functions.
15479 2017-08-01 Richard Biener <rguenther@suse.de>
15481 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15482 (compute_antic): Seed worklist with exit block predecessors.
15483 * cfganal.c (dfs_find_deadend): For a cycle return the source
15484 of the edge closing it.
15486 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15488 * config/aarch64/aarch64.c
15489 (aarch64_can_const_movi_rtx_p): Move 0 check.
15491 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15493 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15494 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15496 * match.pd: Ditto in address comparison pattern.
15498 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15500 PR tree-optimization/81627
15501 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15502 closed ssa form for store-store chain.
15504 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15506 PR tree-optimization/81620
15507 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15508 for store-store chain.
15510 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15512 PR tree-optimization/81588
15513 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15514 ranges[i].in_p, invert comparison code ccode. For >/>=,
15515 swap rhs1 and rhs2 and comparison code unconditionally,
15516 for </<= don't do that. Don't swap rhs1/rhs2 again if
15517 ranges[i].in_p, instead invert comparison code ccode if
15518 opcode or oe->rank is BIT_IOR_EXPR.
15521 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15522 a direct optab to conversion optab.
15523 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15524 with GET_MODE_INNER as last argument instead of optab_handler.
15525 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15526 vector extraction if possible and optab is available.
15527 * expr.c (store_constructor): Use convert_optab_handler instead
15528 of optab_handler. Use vector initialization from smaller
15529 vectors if possible and optab is available.
15530 * tree-vect-stmts.c (vectorizable_load): Likewise.
15531 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15532 now have two modes.
15533 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15534 of vec_init from half-sized vectors with the same element mode.
15535 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15536 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15537 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15538 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15539 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15540 after mode in gen_vec_extract* calls.
15541 (vec_extract<mode>): Renamed to ...
15542 (vec_extract<mode><ssescalarmodelower>): ... this.
15543 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15544 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15545 element mode after mode in gen_vec_init* calls.
15546 (VEC_INIT_HALF_MODE): New mode iterator.
15547 (vec_init<mode>): Renamed to ...
15548 (vec_init<mode><ssescalarmodelower>): ... this.
15549 (vec_init<mode><ssehalfvecmodelower>): New expander.
15550 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15551 (vec_extractv2sfsf): ... this.
15552 (vec_initv2sf): Renamed to ...
15553 (vec_initv2sfsf): ... this.
15554 (vec_extractv2si): Renamed to ...
15555 (vec_extractv2sisi): ... this.
15556 (vec_initv2si): Renamed to ...
15557 (vec_initv2sisi): ... this.
15558 (vec_extractv4hi): Renamed to ...
15559 (vec_extractv4hihi): ... this.
15560 (vec_initv4hi): Renamed to ...
15561 (vec_initv4hihi): ... this.
15562 (vec_extractv8qi): Renamed to ...
15563 (vec_extractv8qiqi): ... this.
15564 (vec_initv8qi): Renamed to ...
15565 (vec_initv8qiqi): ... this.
15566 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15567 (vec_init<mode>): Renamed to ...
15568 (vec_init<mode><VEC_base_l>): ... this.
15569 (vec_extract<mode>): Renamed to ...
15570 (vec_extract<mode><VEC_base_l>): ... this.
15571 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15572 (vec_initv2sfsf): ... this.
15573 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15574 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15575 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15576 element mode after mode in gen_vec_init* calls.
15577 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15578 (vec_init<mode><Vel>): ... this.
15579 (vec_extract<mode>): Renamed to ...
15580 (vec_extract<mode><Vel>): ... this.
15581 * config/aarch64/iterators.md (Vel): New mode attribute.
15582 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15583 Add element mode after mode in gen_vec_extract* calls.
15584 * config/s390/vector.md (non_vec_l): New mode attribute.
15585 (vec_extract<mode>): Renamed to ...
15586 (vec_extract<mode><non_vec_l>): ... this.
15587 (vec_init<mode>): Renamed to ...
15588 (vec_init<mode><non_vec_l>): ... this.
15589 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15590 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15592 * config/arm/iterators.md (V_elem_l): New mode attribute.
15593 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15594 (vec_extract<mode><V_elem_l>): ... this.
15595 (vec_extractv2di): Renamed to ...
15596 (vec_extractv2didi): ... this.
15597 (vec_init<mode>): Renamed to ...
15598 (vec_init<mode><V_elem_l>): ... this.
15599 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15600 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15601 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15602 Add element mode after gen_vec_extract* calls.
15603 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15604 (vec_init<mode><unitmode>): ... this.
15605 (vec_extract<mode>): Renamed to ...
15606 (vec_extract<mode><unitmode>): ... this.
15607 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15608 (vec_init<mode><unitmode>): ... this.
15609 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15610 (vec_initv2sfsf): ... this.
15611 (vec_extractv2sf): Renamed to ...
15612 (vec_extractv2sfsf): ... this.
15613 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15614 Add element mode after gen_vec_extract* calls.
15615 * config/mips/mips.md (unitmode): New mode iterator.
15616 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15617 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15618 * config/spu/spu.md (inner_l): New mode attribute.
15619 (vec_init<mode>): Renamed to ...
15620 (vec_init<mode><inner_l>): ... this.
15621 (vec_extract<mode>): Renamed to ...
15622 (vec_extract<mode><inner_l>): ... this.
15623 * config/sparc/sparc.md (veltmode): New mode iterator.
15624 (vec_init<VMALL:mode>): Renamed to ...
15625 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15626 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15627 (vec_initv2sisi): ... this.
15628 (vec_initv2sf): Renamed to ...
15629 (vec_initv2sfsf): ... this.
15630 (vec_extractv2sf): Renamed to ...
15631 (vec_extractv2sfsf): ... this.
15632 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15633 (vec_init<mode>): Renamed to ...
15634 (vec_init<mode><VEC_base_l>): ... this.
15635 (vec_extract<mode>): Renamed to ...
15636 (vec_extract<mode><VEC_base_l>): ... this.
15637 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15638 (vec_initv2sfsf): ... this.
15639 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15640 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15641 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15642 gen_vec_init* calls.
15644 2017-08-01 Richard Biener <rguenther@suse.de>
15646 PR tree-optimization/81297
15647 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15648 TREE_OVERFLOW from INTEGER_CSTs.
15650 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15652 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15654 2017-07-31 Carl Love <cel@us.ibm.com>
15656 * config/rs6000/rs6000-c: Add support for built-in functions
15657 vector signed char vec_xl_be (signed long long, signed char *);
15658 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15659 vector signed int vec_xl_be (signed long long, signed int *);
15660 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15661 vector signed long long vec_xl_be (signed long long, signed long long *);
15662 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15663 vector signed short vec_xl_be (signed long long, signed short *);
15664 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15665 vector double vec_xl_be (signed long long, double *);
15666 vector float vec_xl_be (signed long long, float *);
15667 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15668 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15669 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15671 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15672 (altivec_expand_builtin): Add switch statement to call
15673 altivec_expand_xl_be for each builtin.
15674 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15675 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15676 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15677 __builtin_vsx_le_be_v16qi.
15678 * doc/extend.texi: Update the built-in documentation file for the
15679 new built-in functions.
15681 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15684 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15686 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15688 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15690 * config.gcc: Add z14.
15691 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15692 CPU model numbers for z13s and z14.
15693 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15695 * config/s390/s390-opts.h (enum processor_type): Rename
15696 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15697 * config/s390/s390.c (processor_table): Add field for CPU name to
15698 be passed to Binutils.
15699 (s390_asm_output_machine_for_arch): Use the new field in
15700 processor_table for Binutils.
15701 (s390_expand_builtin): Replace arch12 with z14.
15702 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15703 (s390_get_sched_attrmask): Likewise.
15704 (s390_get_unit_mask): Likewise.
15705 * config/s390/s390.opt: Add z14 to processor_type enum.
15707 2017-07-31 Martin Jambor <mjambor@suse.cz>
15710 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15711 regardless of optimization level.
15713 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15714 Martin Liska <mliska@suse.cz>
15716 * predict.def: Remove old comment and adjust probability.
15717 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15718 PREDICT statements.
15720 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15723 * config/i386/i386.c (ix86_function_naked): New function.
15724 (ix86_can_use_return_insn_p): Return false for naked functions.
15725 (ix86_expand_prologue): Skip prologue for naked functions.
15726 (ix86_expand_epilogue): Skip epilogue for naked functions
15727 and emit trap instruction.
15728 (ix86_warn_func_return): New function.
15729 (ix86_attribute_table): Add "naked" attribute specification.
15730 (TARGET_WARN_FUNC_RETURN): Define.
15731 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15733 2017-07-31 Martin Liska <mliska@suse.cz>
15735 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15736 (dump_gimple_bb_header): Always dump BB info.
15737 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15739 2017-07-31 Martin Liska <mliska@suse.cz>
15742 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15743 also with current_function_decl non-null equality.
15745 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15748 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15749 change type to the element type, instead add eltype variable and
15750 use it where we are interested in the element type.
15752 PR tree-optimization/81603
15753 * ipa-polymorphic-call.c
15754 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15755 offset arithmetic in offset_int, bail out if the resulting bit offset
15756 doesn't fit into shwi.
15758 2017-07-31 Martin Liska <mliska@suse.cz>
15760 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15761 (gimplify_save_expr): Fix comment.
15763 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15766 * config/i386/i386.c (ix86_function_arg): Update arguments for
15768 (ix86_compute_frame_layout): Set the initial stack offset to
15769 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15770 INCOMING_FRAME_SP_OFFSET.
15771 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15772 stack before exception handler returns.
15773 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15774 the 'ERROR_CODE' for exception handler.
15776 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15778 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15779 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15780 (ASM_OUTPUT_REG_POP): Ditto.
15781 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15782 instead of asm_fprintf to output pure string.
15784 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15786 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15787 to imported_module_or_decl hook.
15788 (debug_nothing_tree_tree_tree_bool): Remove.
15789 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15790 * debug.c (do_nothing_debug_hooks): Use
15791 debug_nothing_tree_tree_tree_bool_bool instead of
15792 debug_nothing_tree_tree_tree_bool.
15793 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15794 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15795 * sdbout.c (sdb_debug_hooks): Likewise.
15796 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15797 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15799 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15800 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15801 attribute, don't add anything.
15803 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15805 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15806 (fold_build2_stat_loc): Likewise.
15807 (fold_build3_stat_loc): Likewise.
15808 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15809 (fold_build1_loc): Remove macro.
15810 (fold_build2_loc): Likewise.
15811 (fold_build3_loc): Likewise.
15813 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15815 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15816 (gimple_build_debug_bind_source_stat): Likewise.
15817 * gimple.h (gimple_build_debug_bind): Remove macro.
15818 (gimple_build_debug_bind_source): Likewise.
15820 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15822 * bitmap.c (bitmap_alloc): Adjust.
15823 (bitmap_gc_alloc): Likewise.
15824 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15826 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15828 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15829 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15830 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15831 (bitmap_gc_alloc_stat): Likewise.
15832 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15834 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15836 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15837 * rtl.h (shallow_copy_rtx): Remove macro.
15839 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15841 * emit-rtl.c (gen_raw_REG): Adjust.
15842 * gengenrtl.c (gendef): Likewise.
15843 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15844 * rtl.h (rtx_alloc): Remove macro.
15846 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15848 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15849 (build_tree_list_stat): Likewise.
15850 * tree.h (build_tree_list): Remove macro.
15851 (build_tree_list_vec): Likewise.
15853 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15855 * tree.c (make_vector_stat): Remove _stat from name.
15856 (build_vector_stat): Likewise.
15857 * tree.h (make_vector_stat): Remove macro.
15858 (build_vector_stat): Likewise.
15860 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15862 * tree.h (build_var_debug_value): Remove prototype.
15864 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15866 * tree.c (tree_cons_stat): Remove _stat from name.
15867 * tree.h (tree_cons): Remove macro.
15869 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15871 * tree.c (build_vl_exp_stat): Remove _stat from name.
15872 * tree.h (build_vl_exp): Remove macro.
15874 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15876 * tree.c (build_decl_stat): Remove _stat from name.
15877 * tree.h (build_decl): Remove macro.
15879 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15881 * gimple.c (gimple_build_with_ops_stat): Adjust.
15882 (gimple_alloc_stat): Remove _stat from name.
15883 * gimple.h (gimple_alloc): Remove macro.
15885 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15887 * tree.c (make_tree_vec_stat): Remove _stat from name.
15888 (grow_tree_vec_stat): Likewise.
15889 * tree.h (make_tree_vec_stat): Adjust prototype.
15890 (grow_tree_vec_stat): Likewise.
15891 (make_tree_vec): Remove macro.
15892 (grow_tree_vec): Likewise.
15894 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15896 * fold-const.c (fold_build1_stat_loc): Adjust.
15897 (fold_build2_stat_loc): Likewise.
15898 (fold_build3_stat_loc): Likewise.
15899 * tree.c (build0_stat): Remove _stat from name.
15900 (build1_stat): Likewise.
15901 (build2_stat): Likewise.
15902 (build3_stat): Likewise.
15903 (build4_stat): Likewise.
15904 (build5_stat): Likewise.
15905 * tree.h (build1_loc): Remove macro, and rename _stat function
15907 (build2_loc): Likewise.
15908 (build3_loc): Likewise.
15909 (build4_loc): Likewise.
15910 (build5_loc): Likewise.
15912 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15914 * tree.c (make_int_cst_stat): Remove _stat from name.
15915 * tree.h (make_int_cst_stat): Adjust prototype.
15916 (make_int_cst): Remove macro.
15918 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15920 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15921 * tree.h (make_tree_binfo_stat): Adjust prototype.
15922 (make_tree_binfo): Remove.
15924 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15926 * tree.c (copy_node_stat): Rename to copy_node.
15927 (build_distinct_type_copy): Adjust.
15928 * tree.h (copy_node_stat): Adjust prototype.
15929 (copy_node): Remove macro.
15931 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15933 * tree.c (make_node_stat): rename to make_node.
15934 (build_tree_list_stat): Adjust.
15935 (build0_stat): Likewise.
15936 (build2_stat): Likewise.
15937 (build3_stat): Likewise.
15938 (build4_stat): Likewise.
15939 (build5_stat): Likewise.
15940 (build_decl_stat): Likewise.
15941 * tree.h (make_node_stat): Adjust prototype.
15942 (make_node): remove macro.
15944 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15946 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15947 (PPC_FEATURE2_SCV): Likewise.
15948 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15950 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15952 * config/aarch64/aarch64.c
15953 (aarch64_internal_mov_immediate): Add new special pattern.
15954 * config/aarch64/aarch64.md (*movdi_aarch64):
15955 Add reg/32bit const mov case.
15957 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15958 Richard Sandiford <richard.sandiford@linaro.org>
15960 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15961 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15962 Add integer and movi cases.
15963 (movi-split-hf-df-sf split, fp16): New.
15964 (enabled): Added TARGET_FP_F16INST.
15965 * config/aarch64/iterators.md (GPF_HF): New.
15966 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15968 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15970 * config/aarch64/aarch64.c
15971 (aarch64_simd_container_mode): Add prototype.
15972 (aarch64_expand_mov_immediate): Add HI support.
15973 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15974 (aarch64_can_const_movi_rtx_p): New.
15975 (aarch64_preferred_reload_class):
15976 Remove restrictions of using FP registers for certain SIMD operations.
15977 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15978 (aarch64_valid_floating_const): Add integer move validation.
15979 (aarch64_simd_imm_scalar_p): Remove.
15980 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15981 (aarch64_legitimate_constant_p): Expand list of supported cases.
15982 * config/aarch64/aarch64-protos.h
15983 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15984 (aarch64_reinterpret_float_as_int): New.
15985 (aarch64_simd_imm_scalar_p): Remove.
15986 * config/aarch64/constraints.md (Uvi): New.
15987 (Dd): Split into Ds and new Dd.
15988 * config/aarch64/aarch64.md (*movsi_aarch64):
15990 (*movdi_aarch64): Add SIMD mov case.
15992 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15994 * tree-predcom.c: (struct chain): Handle store-store chain in which
15995 stores for elimination only store loop invariant values.
15996 (execute_pred_commoning_chain): Ditto.
15997 (prepare_initializers_chain_store_elim): Ditto.
15998 (prepare_finalizers): Ditto.
15999 (is_inv_store_elimination_chain): New function.
16000 (initialize_root_vars_store_elim_1): New function.
16002 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16004 * tree-predcom.c: Revise general description of the pass.
16005 (enum chain_type): New enum type for store elimination.
16006 (struct chain): New field supporting store elimination.
16007 (struct component): Ditto.
16008 (dump_chain): Dump store-stores chain.
16009 (release_chain): Release resources.
16010 (split_data_refs_to_components): Compute and create component
16011 contains only stores for elimination.
16012 (get_chain_last_ref_at): New function.
16013 (make_invariant_chain): Initialization.
16014 (make_rooted_chain): Specify chain type in parameter and record it.
16015 (add_looparound_copies): Skip for store-stores chain.
16016 (determine_roots_comp): Compute type of chain and pass it to
16018 (initialize_root_vars_store_elim_2): New function.
16019 (finalize_eliminated_stores): New function.
16020 (remove_stmt): Handle store for elimination.
16021 (execute_pred_commoning_chain): Execute predictive commoning on
16022 store-store chains.
16023 (determine_unroll_factor): Skip unroll for store-stores chain.
16024 (prepare_initializers_chain_store_elim): New function.
16025 (prepare_initializers_chain): Hanlde store-store chain.
16026 (prepare_finalizers_chain, prepare_finalizers): New function.
16027 (tree_predictive_commoning_loop): Return integer value indicating
16028 if loop is unrolled or lcssa form is corrupted.
16029 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16031 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16033 * tree-predcom.c (initialize_root): Delete.
16034 (execute_pred_commoning_chain): Initialize root vars and replace
16035 reference of non-combined chain directly, rather than call above
16038 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16040 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16041 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16043 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16045 * tree-predcom.c (struct chain): New field init_seq.
16046 (release_chain): Release init_seq.
16047 (prepare_initializers_chain): Record intialization stmts in above
16049 (insert_init_seqs): New function.
16050 (tree_predictive_commoning_loop): Call insert_init_seqs.
16052 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16054 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16056 2017-07-28 Richard Biener <rguenther@suse.de>
16058 * match.pd: Remove superfluous :c.
16059 * genmatch.c (simplify::id): Add member.
16060 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16062 (current_id): New global.
16063 (dt_node::parent): Move from ...
16064 (dt_operand::parent): ... here. Add for_id member.
16065 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16066 (decision_tree::find_node): Relax order requirement when
16067 merging DT_TRUE nodes to ones inbetween the current simplify
16068 and the one we try to merge with. Add diagnostic whenever
16069 we need to enforce pattern order by not merging.
16070 (decision_tree::insert): Set current_id.
16071 (decision_tree::print_node): Dump parent node and for_id.
16072 (parser::last_id): Add member.
16073 (parser::push_simplify): Assign unique id.
16074 (parser::parser): Initialize last_id.
16076 2017-07-28 Martin Liska <mliska@suse.cz>
16079 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16080 gimple_build_debug_bind.
16082 2017-07-28 Richard Biener <rguenther@suse.de>
16084 PR tree-optimization/81502
16085 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16087 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16089 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16090 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16091 for BIT_FIELD_REF args.
16092 * fold-const.c (make_bit_field_ref): Likewise.
16093 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16095 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16098 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16099 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16100 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16101 Or it into SANITIZER_UNDEFINED.
16102 * ubsan.c: Include gimple-fold.h and varasm.h.
16103 (ubsan_expand_ptr_ifn): New function.
16104 (instrument_pointer_overflow): New function.
16105 (maybe_instrument_pointer_overflow): New function.
16106 (instrument_object_size): Formatting fix.
16107 (pass_ubsan::execute): Call instrument_pointer_overflow
16108 and maybe_instrument_pointer_overflow.
16109 * internal-fn.c (expand_UBSAN_PTR): New function.
16110 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16111 * sanitizer.def (__ubsan_handle_pointer_overflow,
16112 __ubsan_handle_pointer_overflow_abort): New builtins.
16113 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16114 * internal-fn.def (UBSAN_PTR): New internal function.
16115 * opts.c (sanitizer_opts): Add pointer-overflow.
16116 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16117 * fold-const.c (build_range_check): Compute pointer range check in
16118 integral type if pointer arithmetics would be needed. Formatting
16121 2017-07-28 Martin Liska <mliska@suse.cz>
16124 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16125 parameters that are of a variable-length.
16127 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16129 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16131 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16132 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16133 (CRT_CALL_STATIC_FUNCTION): Likewise.
16134 (ASM_DEFAULT_SPEC): New define.
16135 (ASM_SPEC32): Likewise.
16136 (ASM_SPEC64): Likewise.
16137 (ASM_SPEC_COMMON): Likewise.
16138 (ASM_SPEC): Likewise.
16139 (INVALID_64BIT): Likewise.
16140 (LINK_OS_DEFAULT_SPEC): Likewise.
16141 (LINK_OS_SPEC32): Likewise.
16142 (LINK_OS_SPEC64): Likewise.
16143 (POWERPC_LINUX): Likewise.
16144 (PTRDIFF_TYPE): Likewise.
16145 (RESTORE_FP_PREFIX): Likewise.
16146 (RESTORE_FP_SUFFIX): Likewise.
16147 (SAVE_FP_PREFIX): Likewise.
16148 (SAVE_FP_SUFFIX): Likewise.
16149 (SIZE_TYPE): Likewise.
16150 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16151 (TARGET_64BIT): Likewise.
16152 (TARGET_64BIT): Likewise.
16153 (TARGET_AIX): Likewise.
16154 (WCHAR_TYPE_SIZE): Likewise.
16155 (WCHAR_TYPE): Undefine.
16156 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16157 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16158 (CPP_OS_RTEMS_SPEC): Delete.
16159 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16160 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16162 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16164 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16166 PR tree-optimization/81578
16167 * tree-parloops.c (build_new_reduction): Bail out if
16168 reduction_code isn't one of the standard OpenMP reductions.
16169 Move the details printing after that decision.
16171 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16173 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16174 related to reload_in_progress.
16175 (splat_input_operand): Likewise.
16176 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16178 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16180 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16181 (TARGET_INSTANTIATE_DECLS): Likewise.
16182 (legitimate_indexed_address_p): Delete reload_in_progress code.
16183 (rs6000_debug_legitimate_address_p): Likewise.
16184 (rs6000_eliminate_indexed_memrefs): Likewise.
16185 (rs6000_emit_le_vsx_store): Likewise.
16186 (rs6000_emit_move_si_sf_subreg): Likewise.
16187 (rs6000_emit_move): Likewise.
16188 (register_to_reg_type): Likewise.
16189 (rs6000_pre_atomic_barrier): Likewise.
16190 (rs6000_machopic_legitimize_pic_address): Likewise.
16191 (rs6000_allocate_stack_temp): Likewise.
16192 (rs6000_address_for_fpconvert): Likewise.
16193 (rs6000_address_for_altivec): Likewise.
16194 (rs6000_secondary_memory_needed_rtx): Delete function.
16195 (rs6000_check_sdmode): Likewise.
16196 (rs6000_alloc_sdmode_stack_slot): Likewise.
16197 (rs6000_instantiate_decls): Likewise.
16198 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16199 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16200 Delete reload_in_progress.
16201 (*vec_reload_and_plus_<mptrsize>): Likewise.
16202 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16203 (vsx_div_v2di): Likewise.
16204 (vsx_udiv_v2di): Likewise.
16206 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16208 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16209 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16210 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16211 (rs6000_debug_reg_global): Delete print of LRA status.
16212 (rs6000_option_override_internal): Delete dead LRA related code.
16213 (rs6000_lra_p): Delete function.
16214 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16216 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16218 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16219 * config/riscv/rtems.h: New file.
16221 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16222 Sudakshina Das <sudi.das@arm.com>
16224 * config/aarch64/aarch64.md
16225 (define_split for and<mode>3nr_compare): Move
16226 non aarch64_logical_operand to a register.
16227 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16228 register immediate operand to a register.
16229 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16231 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16233 PR middle-end/81564
16234 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16236 2017-07-27 Richard Biener <rguenther@suse.de>
16238 PR tree-optimization/81573
16239 PR tree-optimization/81494
16240 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16241 multi defuse cycle case.
16243 2017-07-27 Richard Biener <rguenther@suse.de>
16245 PR tree-optimization/81571
16246 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16249 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16251 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16252 earlier and only if MASK_FPU is set. Adjust formatting.
16254 2017-07-27 Martin Liska <mliska@suse.cz>
16256 * opt-functions.awk: Add validation of value of Init.
16257 * optc-gen.awk: Pass new argument.
16259 2017-07-27 Martin Liska <mliska@suse.cz>
16261 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16262 Fix wrong condition.
16264 2017-07-27 Martin Liska <mliska@suse.cz>
16266 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16267 BBs and edges seen by autoFDO.
16269 2017-07-27 Richard Biener <rguenther@suse.de>
16271 PR tree-optimization/81502
16272 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16273 with incompatible but same sized type.
16274 (execute_update_addresses_taken): Likewise.
16276 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16278 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16279 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16281 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16284 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16285 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16286 Change s_operand to memory_operand.
16288 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16290 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16291 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16292 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16293 Emit instructions rather than returning an expression. Handle TFmode
16294 and KFmode by casting to TImode.
16295 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16296 (rs6000_emit_le_vsx_store): Likewise.
16297 * config/rs6000/vsx.md (VSX_TI): New iterator.
16298 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16299 (*vsx_le_undo_permute_<mode>): Likewise.
16300 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16301 emit the split sequence.
16302 (*vsx_le_perm_store_<mode>): Likewise.
16304 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16306 PR tree-optimization/81555
16307 PR tree-optimization/81556
16308 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16309 if true, force CHANGED for the recursive invocation.
16310 (reassociate_bb): Remember original length of ops array, pass
16311 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16313 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16314 attributes for noipa attribute. For naked attribute use
16315 lookup_attribute first before lookup_attribute_spec.
16316 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16318 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16319 for functions with noipa attribute.
16320 (cgraph_externally_visible_p): Return true for functions with noipa
16322 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16323 for functions with noipa attribute.
16324 * doc/extend.texi: Document noipa function attribute.
16325 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16326 also for functions with noipa attribute.
16327 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16329 2017-07-26 Andrew Pinski <apinski@cavium.com>
16331 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16332 vec_unalign_load_cost and vec_unalign_store_cost.
16334 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16336 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16337 -mvsx-small-integer option.
16338 (ISA_3_0_MASKS_IEEE): Likewise.
16339 (OTHER_VSX_VECTOR_MASKS): Likewise.
16340 (POWERPC_MASKS): Likewise.
16341 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16342 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16343 code, only testing for DImode being allowed in non-VSX floating
16345 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16346 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16348 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16349 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16350 TARGET_P8_VECTOR test.
16351 (rs6000_secondary_reload_simple_move): Likewise.
16352 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16353 since TARGET_P9_VECTOR was already tested.
16354 (rs6000_opt_masks): Remove -mvsx-small-integer.
16355 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16356 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16358 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16359 test for TARGET_VEXTRACTUB was used, and that uses
16361 (p9 extract splitter): Likewise.
16362 (vsx_extract_<mode>_di_p9): Likewise.
16363 (vsx_extract_<mode>_store_p9): Likewise.
16364 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16365 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16366 the elimination of TARGET_VSX_SMALL_INTEGER.
16367 (vsx_extract_<mode>_p8): Likewise.
16368 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16369 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16370 (vsx_set_<mode>_p9): Likewise.
16371 (vsx_set_v4sf_p9): Likewise.
16372 (vsx_set_v4sf_p9_zero): Likewise.
16373 (vsx_insert_extract_v4sf_p9): Likewise.
16374 (vsx_insert_extract_v4sf_p9_2): Likewise.
16375 * config/rs6000/rs6000.md (sign extend splitter): Change
16376 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16377 (floatsi<mode>2_lfiwax_mem): Likewise.
16378 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16379 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16380 since a test for TARGET_P9_VECTOR was used.
16381 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16382 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16383 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16384 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16385 TARGET_P8_VECTOR test.
16386 (fix_trunc<mode>si2_stfiwx): Likewise.
16387 (fix_trunc<mode>si2_internal): Likewise.
16388 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16389 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16391 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16392 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16393 TARGET_P8_VECTOR test.
16394 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16395 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16396 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16398 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16399 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16400 since a test for TARGET_P9_VECTOR was used.
16401 (splitter for loading small constants): Likewise.
16403 2017-07-26 Andrew Pinski <apinski@cavium.com>
16405 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16408 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16411 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16412 (fp_valid_at): Likewise.
16414 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16416 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16417 (qdf24xx_addrcost_table): Likewise.
16418 (cortexa57_tunings): Update to use generic_branch_cost.
16419 (cortexa72_tunings): Likewise.
16420 (cortexa73_tunings): Likewise.
16421 (qdf24xx_tunings): Likewise.
16423 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16425 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16426 (thunderx2t99_branch_cost): Likewise.
16427 (cortexa35_tunings): Update to use generic_branch_cost.
16428 (cortexa53_tunings): Likewise.
16429 (cortexa57_tunings): Likewise.
16430 (cortexa72_tunings): Likewise.
16431 (cortexa73_tunings): Likewise.
16432 (thunderx2t99_tunings): Likewise.
16434 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16436 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16437 (sparc_option_override): Honour MASK_FSMULD.
16438 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16439 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16440 * config/sparc/sparc.opt (mfsmuld): New option.
16441 * doc/invoke.texi (mfsmuld): Document option.
16443 2017-07-26 Marek Polacek <polacek@redhat.com>
16445 PR middle-end/70992
16446 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16448 2017-07-26 Richard Biener <rguenther@suse.de>
16450 * gimple-match-head.c (do_valueize): Return OP if valueize
16452 (get_def): New helper to get at the def stmt of a SSA name
16453 if valueize allows.
16454 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16455 do_valueize to get at the def stmt.
16456 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16458 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16460 PR middle-end/46932
16461 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16463 2017-07-26 Martin Liska <mliska@suse.cz>
16466 * function.c (expand_function_start): Make expansion of
16467 nonlocal_goto_save_area after parm_birth_insn.
16469 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16471 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16472 from all CPU target flags enable members.
16474 2017-07-26 Richard Biener <rguenther@suse.de>
16476 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16477 (decision_tree::gen): Make 'type' const.
16478 (write_predicate): Likewise.
16480 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16482 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16483 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16484 (rs6000_option_override_internal): Likewise.
16485 (rs6000_expand_vector_set): Likewise.
16486 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16487 (TARGET_UPPER_REGS_SF): Likewise.
16488 (TARGET_UPPER_REGS_DI): Likewise.
16489 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16490 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16491 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16492 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16493 (Splitters for DI constants in Altivec registers): Likewise.
16494 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16495 (vsx_set_v4sf_p9): Likewise.
16496 (vsx_set_v4sf_p9_zero): Likewise.
16497 (vsx_insert_extract_v4sf_p9): Likewise.
16498 (vsx_insert_extract_v4sf_p9_2): Likewise.
16500 2017-07-25 Carl Love <cel@us.ibm.com>
16502 * doc/extend.texi: Update the built-in documentation file for the
16503 existing built-in functions
16504 vector signed char vec_cnttz (vector signed char);
16505 vector unsigned char vec_cnttz (vector unsigned char);
16506 vector signed short vec_cnttz (vector signed short);
16507 vector unsigned short vec_cnttz (vector unsigned short);
16508 vector signed int vec_cnttz (vector signed int);
16509 vector unsigned int vec_cnttz (vector unsigned int);
16510 vector signed long long vec_cnttz (vector signed long long);
16511 vector unsigned long long vec_cnttz (vector unsigned long long);
16513 2017-07-25 Andrew Pinski <apinski@cavium.com>
16515 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16516 accesses where the use is for the first operand of a BIT_INSERT.
16518 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16521 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16522 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16524 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16526 * config/i386/gstabs.h: Delete.
16527 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16529 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16531 * config/i386/i386.c (ix86_decompose_address): Do not check for
16532 register RTX when looking at index_reg or base_reg.
16533 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16535 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16537 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16538 to update EH info here.
16540 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16542 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16544 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16546 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16548 2017-07-25 Torsten Duwe <duwe@suse.de>
16550 * common.opt: Introduce -fpatchable-function-entry
16551 command line option, and its variables function_entry_patch_area_size
16552 and function_entry_patch_area_start.
16553 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16554 including a two-value parser.
16555 * target.def (print_patchable_function_entry): New target hook.
16556 * targhooks.h (default_print_patchable_function_entry): New function.
16557 * targhooks.c (default_print_patchable_function_entry): Likewise.
16558 * toplev.c (process_options): Switch off IPA-RA if
16559 patchable function entries are being generated.
16560 * varasm.c (assemble_start_function): Look at the
16561 patchable-function-entry command line switch and current
16562 function attributes and maybe generate NOP instructions by
16563 calling the print_patchable_function_entry hook.
16564 * doc/extend.texi: Document patchable_function_entry attribute.
16565 * doc/invoke.texi: Document -fpatchable_function_entry
16566 command line option.
16567 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16569 * doc/tm.texi: Re-generate.
16571 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16574 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16575 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16577 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16579 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16581 2017-07-25 Richard Biener <rguenther@suse.de>
16583 PR tree-optimization/81455
16584 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16585 not walk in cycles when looking for guards.
16587 2017-07-25 Richard Biener <rguenther@suse.de>
16589 PR tree-optimization/81529
16590 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16591 when optimizing backedge uses.
16593 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16595 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16597 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16598 to dl_section_ref. On AIX, append an expression to subtract
16599 the size of the section length to dl_section_ref.
16601 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16603 * configure.ac: If any of the config.* scripts fail, exit 1.
16604 * configure: Regenerate.
16606 2017-07-25 Richard Biener <rguenther@suse.de>
16608 PR middle-end/81546
16609 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16610 of immediate uses, be more verbose on errors.
16612 2017-07-25 Richard Biener <rguenther@suse.de>
16614 PR tree-optimization/81510
16615 * tree-vect-loop.c (vect_is_simple_reduction): When the
16616 reduction stmt is not inside the loop bail out.
16618 2017-07-25 Richard Biener <rguenther@suse.de>
16620 PR tree-optimization/81303
16621 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16622 profitability check against LOOP_VINFO_NITERSM1.
16624 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16626 * domwalk.c (cmp_bb_postorder): Simplify.
16627 (sort_bbs_postorder): New function. Use it...
16628 (dom_walker::walk): ...here to optimize common cases.
16630 2017-07-25 Martin Liska <mliska@suse.cz>
16633 * ipa-visibility.c (function_and_variable_visibility): Make the
16634 redirection just on target that supports aliasing.
16635 Fix GNU coding style.
16637 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16640 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16642 * config/arm/rtems.h: Likewise.
16643 * config/bfin/rtems.h: Likewise.
16644 * config/i386/rtemself.h: Likewise.
16645 * config/lm32/rtems.h: Likewise.
16646 * config/m32c/rtems.h: Likewise.
16647 * config/m68k/rtemself.h: Likewise.
16648 * config/microblaze/rtems.h: Likewise.
16649 * config/mips/rtems.h: Likewise.
16650 * config/moxie/rtems.h: Likewise.
16651 * config/nios2/rtems.h: Likewise.
16652 * config/powerpcspe/rtems.h: Likewise.
16653 * config/rs6000/rtems.h: Likewise.
16654 * config/rtems.h: Likewise.
16655 * config/sh/rtems.h: Likewise.
16656 * config/sh/rtemself.h: Likewise.
16657 * config/sparc/rtemself.h: Likewise.
16659 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16662 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16663 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16664 * tree-ssa-structalias.c (alias_get_name): Same.
16666 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16669 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16670 instructions if no du chain is found.
16672 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16674 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16676 2017-07-25 Richard Biener <rguenther@suse.de>
16678 PR middle-end/81505
16679 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16682 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16684 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16685 upper-regs options.
16686 (ISA_2_7_MASKS_SERVER): Likewise.
16687 (ISA_3_0_MASKS_IEEE): Likewise.
16688 (OTHER_P8_VECTOR_MASKS): Likewise.
16689 (OTHER_VSX_VECTOR_MASKS): Likewise.
16690 (POWERPC_MASKS): Likewise.
16691 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16692 duplicate list of options.
16693 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16694 explicit -mupper-regs options.
16695 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16696 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16697 alias for -mupper-regs-df.
16698 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16699 (rs6000_init_hard_regno_mode_ok): Likewise.
16700 (rs6000_option_override_internal): Likewise.
16701 (rs6000_opt_masks): Likewise.
16702 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16703 options in terms of whether -mvsx or -mpower8-vector was used.
16704 (TARGET_UPPER_REGS_DI): Likewise.
16705 (TARGET_UPPER_REGS_SF): Likewise.
16706 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16707 -mupper-regs-* options.
16709 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16711 * passes.c (emergency_dump_function): Print some empty lines and a
16712 header before the RTL dump.
16714 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16716 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16718 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16721 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16722 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16724 2017-07-24 Carl Love <cel@us.ibm.com>
16726 * config/rs6000/rs6000-c.c: Add support for built-in functions
16727 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16728 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16729 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16730 vec_extract_fp_from_shortl): Add defines for the two builtins.
16731 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16732 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16734 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16735 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16736 * doc/extend.texi: Update the built-in documentation file for the
16737 new built-in function.
16739 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16742 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16744 * doc/generic.texi: Likewise.
16745 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16746 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16748 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16750 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16751 (aarch64_mls_elt_merge<mode>): Likewise.
16753 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16755 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16756 having __cxa_atexit.
16758 2017-07-23 Michael Collison <michael.collison@arm.com>
16760 * config/arm/arm.c (arm_option_override): Deprecate
16761 use of -mstructure-size-boundary.
16762 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16763 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16765 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16768 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16769 Reduce cost estimate for direct moves.
16771 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16774 * config/i386/i386.c (ix86_option_override_internal): Disable
16775 BMI, BMI2 and TBM instructions for -m16.
16777 2017-07-21 Carl Love <cel@us.ibm.com>
16779 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16780 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16781 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16782 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16783 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16784 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16785 VMULOSW): New enum "unspec" values.
16786 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16787 altivec_vmulosw): New patterns.
16788 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16789 VMULOSW): Add definitions.
16791 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16793 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16794 (qdf24xx): Likewise.
16795 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16796 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16797 (AARCH64_FL_V8_1): Renumber.
16798 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16799 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16800 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16801 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16802 rdma to feature modifiers list.
16804 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16806 PR middle-end/56727
16807 * ipa-visibility (function_and_variable_visibility): Convert
16808 recursive PLT call to direct call if appropriate.
16810 2017-07-21 Andrew Pinski <apinski@cavium.com>
16812 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16813 operand 1 to see if the types precision matches.
16814 * fold-const.c (operand_equal_p): Likewise.
16816 2017-07-21 Richard Biener <rguenther@suse.de>
16818 PR tree-optimization/81303
16819 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16820 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16821 (vect_peeling_hash_get_lowest_cost): Adjust.
16822 (vect_enhance_data_refs_alignment): Likewise. Use
16823 vect_get_peeling_costs_all_drs to compute the penalty for no
16824 peeling to match up costs.
16826 2017-07-21 Richard Biener <rguenther@suse.de>
16828 PR tree-optimization/81500
16829 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16830 we didn't identify a reduction path.
16832 2017-07-21 Tom de Vries <tom@codesourcery.com>
16833 Cesar Philippidis <cesar@codesourcery.com>
16835 PR gcov-profile/81442
16836 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16839 2017-07-21 Tom de Vries <tom@codesourcery.com>
16842 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16844 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16845 nvptx_override_options_after_change.
16847 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16849 * dwarf2out.c (output_file_names): Avoid double testing for
16850 dwarf_version >= 5.
16852 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16854 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16856 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16858 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16860 (try_crossjump_to_edge): Do not punt on partitioned functions.
16862 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16864 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16865 Put all BBs reachable only via paths crossing cold region to cold
16867 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16869 2016-07-21 Richard Biener <rguenther@suse.de>
16871 PR tree-optimization/81303
16872 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16873 into account prologue and epilogue iterations when raising
16874 min_profitable_iters to sth at least covering one vector iteration.
16876 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16878 * config/arm/arm.c (arm_test_cpu_arch_dat):
16881 2017-07-20 Nathan Sidwell <nathan@acm.org>
16883 Remove TYPE_METHODS.
16884 * tree.h (TYPE_METHODS): Delete.
16885 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16886 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16887 (dbxout_type_methods): Scan TYPE_FIELDS.
16888 (dbxout_type): Don't check TYPE_METHODS here.
16889 * function.c (use_register_for_decl): Always ignore register for
16890 class types when not optimizing.
16891 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16892 * tree.c (free_lang_data_in_type): Stitch out member functions and
16893 templates from TYPE_FIELDS.
16894 (build_distinct_type_copy, verify_type_variant,
16895 verify_type): Member fns are on TYPE_FIELDS.
16896 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16897 * tree-pretty-print.c (dump_generic_node): Likewise.
16899 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16902 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16903 V2TImode and V4TImode.
16904 (ix86_expand_vector_extract): Likewise.
16905 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16906 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16907 (ssescalarmode): Handle V4TImode and V2TImode.
16908 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16909 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16910 (VEXTRACTI128_MODE): New mode iterator.
16911 (splitter for *vec_extractv?ti first element): New.
16912 (VEC_INIT_MODE): New mode iterator.
16913 (vec_init<mode>): Consolidate 3 expanders into one using
16914 VEC_INIT_MODE mode iterator.
16916 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16918 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16919 non_spilled_static_chain_regno_p.
16921 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16923 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16925 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16927 * bb-reorder.c (connect_traces): Allow copying of blocks within
16930 2017-07-20 Richard Biener <rguenther@suse.de>
16932 * gimple.h (gimple_phi_result): Add gphi * overload.
16933 (gimple_phi_result_ptr): Likewise.
16934 (gimple_phi_arg): Likewise. Adjust index assert to only
16935 allow actual argument accesses rather than all slots available
16937 (gimple_phi_arg_def): Add gphi * overload.
16938 * tree-phinodes.c (make_phi_node): Initialize only actual
16940 (resize_phi_node): Clear memory not covered by old node,
16941 do not initialize excess argument slots.
16942 (reserve_phi_args_for_new_edge): Initialize new argument slot
16945 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16947 PR tree-optimization/81388
16949 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16951 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16952 by removing computation of may_be_zero.
16954 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16955 Tom de Vries <tom@codesourcery.com>
16957 PR middle-end/81030
16958 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16959 when gimple level profile disagrees with what RTL expander did.
16961 2017-07-20 Richard Biener <rguenther@suse.de>
16963 PR tree-optimization/61171
16964 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16965 (vect_analyze_stmt): Add slp instance parameter.
16966 (vectorizable_reduction): Likewise.
16967 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16968 (vect_is_simple_reduction): Deal with chains not detected
16969 as SLP reduction chain, specifically not properly associated
16970 chains containing a mix of plus/minus.
16971 (get_reduction_op): Remove.
16972 (get_initial_defs_for_reduction): Simplify, pass in whether
16973 this is a reduction chain, pass in the SLP node for the PHIs.
16974 (vect_create_epilog_for_reduction): Get the SLP instance as
16976 (vectorizable_reduction): Get the SLP instance as arg.
16977 During analysis remember the SLP node with the PHIs in the
16978 instance. Simplify getting at the vectorized reduction PHIs.
16979 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16980 through SLP instance.
16981 (vect_slp_analyze_operations): Likewise.
16982 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16983 (vect_transform_stmt): Likewise.
16985 2017-07-20 Tom de Vries <tom@codesourcery.com>
16987 PR tree-optimization/81489
16988 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16989 read of phi arg location to before loop that modifies phi.
16991 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16993 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16996 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16998 PR middle-end/81331
16999 * except.c (execute): Fix ordering issue.
17001 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17003 PR rtl-optimization/81423
17004 * combine.c (make_compound_operation_int): Don't try to optimize
17005 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17007 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17009 PR rtl-optimization/81423
17010 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17011 with a constant that is -1 in the truncated to mode.
17013 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17015 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17016 (determine_unlikely_bbs): ... here.
17017 * predict.h (propagate_unlikely_bbs_forward): Declare.
17018 * cfgexpand.c (pass_expand::execute): Use it.
17019 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17021 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17022 propagate_unlikely_bbs_forward.
17024 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17026 PR middle-end/81331
17027 * except.c (maybe_add_nop_after_section_switch): New function.
17030 2017-07-19 Tom de Vries <tom@codesourcery.com>
17032 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17034 2017-07-19 Tom de Vries <tom@codesourcery.com>
17036 * gimple.h (gimple_phi_arg): Make assert more strict.
17038 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17040 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17041 * config/rs6000/mmintrin.h: New file.
17042 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17044 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17046 PR tree-optimization/81346
17047 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17049 2017-07-19 Tom de Vries <tom@codesourcery.com>
17051 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17053 2017-07-19 Tom de Vries <tom@codesourcery.com>
17055 * config/nvptx/nvptx-modes.def: Add V2DImode.
17056 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17057 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17058 (nvptx_output_mov_insn): Handle lack of mov.b128.
17059 (nvptx_print_operand): Handle 'H' and 'L' codes.
17060 (nvptx_vector_mode_supported): Allow V2DImode.
17061 (nvptx_preferred_simd_mode): New function.
17062 (nvptx_data_alignment): New function.
17063 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17064 nvptx_preferred_simd_mode.
17065 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17067 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17069 2017-07-19 Tom de Vries <tom@codesourcery.com>
17071 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17072 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17073 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17074 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17075 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17076 (mov<VECIM>_insn): New define_insn.
17077 (define_expand "mov<VECIM>): New define_expand.
17079 2017-07-19 Tom de Vries <tom@codesourcery.com>
17081 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17083 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17085 PR tree-optimization/81346
17086 * fold-const.h (fold_div_compare, range_check_type): Declare.
17087 * fold-const.c (range_check_type): New function.
17088 (build_range_check): Use range_check_type.
17089 (fold_div_compare): No longer static, rewritten into
17090 a match.pd helper function.
17091 (fold_comparison): Don't call fold_div_compare here.
17092 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17093 as helper function.
17095 2017-07-19 Nathan Sidwell <nathan@acm.org>
17097 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17098 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17099 * tree.c (find_decls_types_r, verify_type): Use
17100 TYPE_{MIN,MAX}_VALUE_RAW.
17101 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17102 (hash_tree): Likewise.
17103 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17105 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17108 2017-07-18 Tom de Vries <tom@codesourcery.com>
17110 PR middle-end/81464
17111 * omp-expand.c (expand_omp_for_static_chunk): Handle
17112 equal-argument loop exit phi.
17114 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17117 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17118 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17119 operand 2 predicate.
17120 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17121 operand 2 predicate.
17122 (ror,rol -> rorx splitters): Use const_int_operand as
17123 operand 2 predicate.
17125 2017-06-18 Richard Biener <rguenther@suse.de>
17127 PR tree-optimization/81410
17128 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17129 the gap in the ! slp_perm SLP case after each group.
17131 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17133 PR middle-end/81463
17134 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17137 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17139 PR middle-end/81462
17140 * predict.c (set_even_probabilities): Cleanup; do not affect
17141 probabilities that are already known.
17142 (combine_predictions_for_bb): Call even when count is set.
17144 2017-07-18 Nathan Sidwell <nathan@acm.org>
17146 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17149 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17152 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17153 optimization for loop niter analysis.
17155 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17158 * config/avr/avr.c (avr_optimize_casesi): Don't use
17159 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17161 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17163 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17164 body_cost_vec from _vect_peel_extended_info.
17165 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17166 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17169 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17171 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17173 2017-07-18 Richard Biener <rguenther@suse.de>
17175 PR tree-optimization/80620
17176 PR tree-optimization/81403
17177 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17178 info when re-using a VN table entry.
17180 2017-07-18 Richard Biener <rguenther@suse.de>
17182 PR tree-optimization/81418
17183 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17184 vectype_in. Verify that with lane-reducing reduction operations
17185 we have a single def-use cycle.
17187 2017-07-17 Carl Love <cel@us.ibm.com>
17189 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17191 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17192 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17193 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17194 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17195 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17196 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17197 VMULOSW): New enum "unspec" values.
17198 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17199 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17200 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17201 altivec_vmulosw): New patterns.
17202 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17203 VMULOSW): Add definitions.
17204 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17206 * config/alpha/alpha.c: Include predict.h.
17208 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17210 * tree-vrp.c (compare_assert_loc): Fix comparison function
17211 to return predictable results.
17213 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17215 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17217 (subdi3): Likewise.
17218 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17219 * doc/invoke.texi (mexpand-adddi): Update text.
17221 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17223 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17224 that also clobbers the CC register. The old expand code is moved
17226 (*arc_clzsi2): ... here.
17227 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17228 the CC register. The old expand code is moved to ...
17229 (arc_ctzsi2): ... here.
17231 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17233 * config/arc/arc.opt (mindexed-loads): Use initial value
17234 TARGET_INDEXED_LOADS_DEFAULT.
17235 (mauto-modify-reg): Use initial value
17236 TARGET_AUTO_MODIFY_REG_DEFAULT.
17237 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17238 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17239 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17240 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17242 2017-07-17 Martin Liska <mliska@suse.cz>
17245 * opts.c (finish_options): Do not allow -fgnu-tm
17246 w/ -fsanitize={kernel-,}address. Say sorry.
17248 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17251 * tree-loop-distribution.c (classify_partition): Only assert on
17252 numer of iterations.
17253 (merge_dep_scc_partitions): Delete prameter. Update function call.
17254 (distribute_loop): Remove code handling loop with unknown niters.
17255 (pass_loop_distribution::execute): Skip loop with unknown niters.
17257 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17260 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17261 function sort_partitions_by_post_order.
17263 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17265 PR tree-optimization/81374
17266 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17267 the max index of basic blocks, rather than number of basic blocks.
17269 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17271 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17273 (arc_legitimate_pic_operand_p): Likewise.
17274 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17276 (arc_needs_pcl_p): Likewise.
17277 (arc_legitimate_pc_offset_p): Likewise.
17278 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17279 function is also used in constrains.md.
17280 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17281 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17282 PLUS. Only return true/false in known cases, otherwise assert.
17283 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17284 is already called in arc_legitimate_constant_p.
17285 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17287 (LEGITIMATE_PIC_OPERAND_P): Use
17288 arc_raw_symbolic_reference_mentioned_p function.
17289 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17294 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17295 Andrew Burgess <andrew.burgess@embecosm.com>
17297 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17298 (arc_return_address_register): New function.
17299 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17300 (arc_handle_fndecl_attribute): Add naked attribute.
17301 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17302 (TARGET_WARN_FUNC_RETURN): Likewise.
17303 (arc_allocate_stack_slots_for_args): New function.
17304 (arc_warn_func_return): Likewise.
17305 (machine_function): Change type fn_type.
17306 (arc_compute_function_type): Consider new naked function type,
17307 change function return type.
17308 (arc_must_save_register): Adapt to handle new
17309 arc_compute_function_type's return type.
17310 (arc_expand_prologue): Likewise.
17311 (arc_expand_epilogue): Likewise.
17312 (arc_return_address_regs): Delete.
17313 (arc_return_address_register): New function.
17314 (arc_epilogue_uses): Use above function.
17315 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17316 (arc_function_type): Change encoding, add naked type.
17317 (ARC_INTERRUPT_P): Change to handle the new encoding.
17318 (ARC_FAST_INTERRUPT_P): Likewise.
17319 (ARC_NORMAL_P): Define.
17320 (ARC_NAKED_P): Likewise.
17321 (arc_compute_function_type): Delete prototype.
17322 * config/arc/arc.md (in_ret_delay_slot): Use
17323 arc_return_address_register function.
17324 (simple_return): Likewise.
17325 (p_return_i): Likewise.
17327 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17329 PR tree-optimization/81428
17330 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17331 can't be built for those types.
17333 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17335 Remove stuff dead since r239246.
17337 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17338 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17339 (avr_inform_devices): Remove dead stuff.
17341 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17343 * config/arm/arm_neon.h: Fix softp typo.
17345 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17347 PR tree-optimization/81365
17348 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17349 aggregate moves onto bb predecessor edges, make sure there are no
17350 loads that could alias the lhs in between the start of bb and the
17353 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17356 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17357 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17358 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17360 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17362 PR tree-optimization/81396
17363 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17364 (init_symbolic_number): Initialize it to 1.
17365 (perform_symbolic_merge): Add n_ops from both operands into the new
17367 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17368 without base_addr as useless if they need more than one operation.
17369 (bswap_replace): Handle !bswap case for NULL base_addr.
17371 2017-07-17 Tom de Vries <tom@codesourcery.com>
17374 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17377 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17379 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17380 conditional builtin define __FIX_LEON3FT_B2BST.
17382 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17384 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17385 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17388 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17390 PR rtl-optimization/81424
17391 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17392 to remove potential trapping from operands if -fnon-call-exceptions.
17394 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17396 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17397 profile_proability for scalling.
17398 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17400 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17402 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17404 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17406 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17407 fixpoint arithmetics.
17409 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17411 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17412 fixpoint arithmetics.
17414 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17416 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17417 fixpoint arithmetics.
17419 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17421 * profile-count.h (profile_probability::from_reg_br_prob_note,
17422 profile_probability::to_reg_br_prob_note): New functions.
17423 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17424 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17425 * predict.c (probability_reliable_p): Update.
17426 (edge_probability_reliable_p): Update.
17427 (br_prob_note_reliable_p): Update.
17428 (invert_br_probabilities): Update.
17429 (add_reg_br_prob_note): New function.
17430 (combine_predictions_for_insn): Update.
17431 * asan.c (asan_clear_shadow): Update.
17432 * cfgbuild.c (compute_outgoing_frequencies): Update.
17433 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17434 (update_br_prob_note): Update.
17435 (rtl_verify_edges): Update.
17436 (purge_dead_edges): Update.
17437 (fixup_reorder_chain): Update.
17438 * emit-rtl.c (try_split): Update.
17439 * ifcvt.c (cond_exec_process_insns): Update.
17440 (cond_exec_process_if_block): Update.
17441 (dead_or_predicable): Update.
17442 * internal-fn.c (expand_addsub_overflow): Update.
17443 (expand_neg_overflow): Update.
17444 (expand_mul_overflow): Update.
17445 * loop-doloop.c (doloop_modify): Update.
17446 * loop-unroll.c (compare_and_jump_seq): Update.
17447 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17448 * predict.h: Update.
17449 * reorg.c (mostly_true_jump): Update.
17451 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17452 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17453 * config/arc/arc.c: (emit_unlikely_jump): Update.
17454 * config/arm/arm.c: (emit_unlikely_jump): Update.
17455 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17456 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17457 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17458 (ix86_print_operand): Update.
17459 (ix86_split_fp_branch): Update.
17460 (predict_jump): Update.
17461 * config/ia64/ia64.c (ia64_print_operand): Update.
17462 * config/mmix/mmix.c (mmix_print_operand): Update.
17463 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17464 (rs6000_expand_split_stack_prologue): Update.
17465 * config/rs6000/rs6000.c: Update.
17466 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17467 (s390_expand_vec_movstr): Update.
17468 (s390_expand_cs_tdsi): Update.
17469 (s390_expand_split_stack_prologue): Update.
17470 * config/sh/sh.c (sh_print_operand): Update.
17471 (expand_cbranchsi4): Update.
17472 (expand_cbranchdi4): Update.
17473 * config/sparc/sparc.c (output_v9branch): Update.
17474 * config/spu/spu.c (get_branch_target): Update.
17475 (ea_load_store_inline): Update.
17476 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17477 * config/tilepro/tilepro.c: Update.
17479 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17481 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17482 (gimplify_save_expr): Likewise.
17484 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17486 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17488 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17490 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17492 * timevar.def (TV_IPA_FNSUMMARY): Define.
17494 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17496 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17497 to back store errata sensitive sequence from being generated.
17498 (sqrtdf2_fix): Likewise.
17500 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17502 * tree-ssa-threadupdate.c (compute_path_counts,
17503 update_joiner_offpath_counts): Use profile_probability.
17505 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17508 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17510 * config/arm/arm-c.c (arm_cpu_builtins): Define
17511 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17513 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17515 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17516 array entries to represent __ieee128 versions of the
17517 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17518 scalar_extract_sig, and scalar_insert_exp built-in functions.
17519 (altivec_resolve_overloaded_builtin): Add special case handling
17520 for the __builtin_scalar_insert_exp function, as represented by
17521 the P9V_BUILTIN_VEC_VSIEDP constant.
17522 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17523 exponent support for __ieee128 argument.
17524 (VSESQP): Add scalar extract signature support for __ieee128
17526 (VSTDCNQP): Add scalar test negative support for __ieee128
17528 (VSIEQP): Add scalar insert exponent support for __int128 argument
17529 with __ieee128 result.
17530 (VSIEQPF): Add scalar insert exponent support for __ieee128
17531 argument with __ieee128 result.
17532 (VSTDCQP): Add scalar test data class support for __ieee128
17534 (VSTDCNQP): Add overload support for scalar test negative with
17535 __ieee128 argument.
17536 (VSTDCQP): Add overload support for scalar test data class
17537 __ieee128 argument.
17538 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17539 UNSPEC_VSX_SXSIGDP.
17540 (UNSPEC_VSX_SIEXPQP): New constant.
17541 (xsxexpqp): New insn for VSX scalar extract exponent quad
17543 (xsxsigqp): New insn for VSX scalar extract significand quad
17545 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17546 precision with floating point argument.
17547 (xststdcqp): New expand for VSX scalar test data class quad
17549 (xststdcnegqp): New expand for VSX scalar test negative quad
17551 (xststdcqp): New insn to match expansions for VSX scalar test data
17552 class quad precision and VSX scalar test negative quad precision.
17553 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17554 special case operand checking to enforce that second operand of
17555 VSX scalar test data class with quad precision argument is a 7-bit
17557 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17558 prototypes and descriptions of __ieee128 versions of
17559 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17560 scalar_test_data_class, and scalar_test_neg built-in functions.
17562 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17564 PR tree-optimization/81162
17565 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17566 replace a negate with an add.
17568 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17570 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17572 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17574 * config/arm/arm-c.c (arm_cpu_builtins): Define
17575 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17577 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17579 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17580 (armv8-r): Set ARM Cortex-R52 as default CPU.
17581 * config/arm/arm-tables.opt: Regenerate.
17582 * config/arm/arm-tune.md: Regenerate.
17583 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17585 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17586 extension for -mcpu=cortex-r52.
17588 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17590 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17591 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17592 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17593 (fp-armv8): Define it as FP_ARMv8 only.
17594 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17595 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17597 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17598 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17599 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17600 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17601 than TARGET_FPU_ARMV8.
17602 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17603 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17604 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17606 * config/arm/neon.md (neon_vrint): Likewise.
17607 (neon_vcvt): Likewise.
17608 (neon_<fmaxmin_op><mode>): Likewise.
17609 (<fmaxmin><mode>3): Likewise.
17610 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17611 * config/arm/predicates.md (arm_cond_move_operator): Check against
17612 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17614 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17616 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17617 to top of function.
17619 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17621 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17622 loop in comment with memset.
17624 2017-07-14 Martin Liska <mliska@suse.cz>
17626 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17627 * dwarf2out.c (is_java): Remove the function.
17628 (output_pubname): Remove usage of the function.
17629 (lower_bound_default): Remove usage of DW_LANG_Java.
17630 (gen_compile_unit_die): Likewise.
17631 * gcc.c: Remove compiler defaults for .java and .zip files.
17632 * gimple-expr.c (remove_suffix): Change as there's no longer
17633 extension than 4-letter one.
17634 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17635 (gimplify_save_expr): Likewise.
17636 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17637 as it's possible even for other languages than Java.
17638 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17639 * lto-opts.c (lto_write_options): Remove reference to Java.
17640 * opts.c (strip_off_ending): Update file extension handling.
17641 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17642 * tree-eh.c (lower_resx): Likewise.
17643 * tree.c (free_lang_data_in_type): Remove dead code.
17644 (find_decls_types_r): Likewise.
17645 (build_common_builtin_nodes): Remove Java from a comment.
17646 (verify_type): Remove dead code.
17647 * varasm.c (assemble_external): Remove Java from a comment.
17649 2017-07-14 Martin Liska <mliska@suse.cz>
17651 * opts.c (finish_options): Add quotes.
17652 (common_handle_option): Likewise.
17654 2017-07-14 Martin Liska <mliska@suse.cz>
17656 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17657 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17658 Remove N_SO_PASCAL.
17659 * dwarf2out.c (lower_bound_default): Do not handle
17661 (gen_compile_unit_die): Likewise.
17662 * gcc.c: Remove default extension binding for GNU Pascal.
17663 * stmt.c: Remove Pascal language from a comment.
17664 * xcoffout.c: Likewise.
17666 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17669 * diagnostic-show-locus.c (fixit_cmp): New function.
17670 (layout::layout): Sort m_fixit_hints.
17671 (column_range::column_range): Assert that the values are valid.
17672 (struct char_span): New struct.
17673 (correction::overwrite): New method.
17674 (struct source_line): New struct.
17675 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17676 calls in terms of classes source_line and char_span, and
17677 correction::overwrite.
17678 (selftest::test_overlapped_fixit_printing_2): New function.
17679 (selftest::diagnostic_show_locus_c_tests): Call it.
17681 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17683 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17684 early if there is no lhs.
17686 2017-07-13 Martin Liska <mliska@suse.cz>
17688 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17689 (gen_reference_type_die): Likewise.
17690 * stor-layout.c: Remove Pascal-related comment.
17692 2017-07-13 Martin Liska <mliska@suse.cz>
17694 * opts.c (finish_options): Add quotes to error messages.
17695 (parse_sanitizer_options): Likewise.
17697 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17699 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17701 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17703 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17705 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17707 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17709 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17711 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17714 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17715 provides the hardware capability bits, define the macro
17716 __BUILTIN_CPU_SUPPORTS__.
17717 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17718 if GLIBC does not provide the hardware capability bits. Add a
17719 gcc_unreachable call if the built-in cpu function is neither
17720 __builtin_cpu_is nor __builtin_cpu_supports.
17721 (rs6000_get_function_versions_dispatcher): Change the warning
17722 that an old GLIBC is used which does not export the capability
17723 bits to be an error.
17724 * doc/extend.texi (target_clones attribute): Document the
17725 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17726 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17727 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17728 the macros defined by GCC if the newer GLIBC is available.
17730 2017-07-12 Jeff Law <law@redhat.com>
17732 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17733 remaining includes slightly.
17734 * config/riscv/riscv-builtins.c: Include profile-count.h.
17736 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17739 * config/avr/avr.c (avr_set_current_function): In diagnostic
17740 messages: Quote keywords and (parts of) identifiers.
17741 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17744 2017-07-12 Carl Love <cel@us.ibm.com>
17746 * config/rs6000/rs6000-c.c: Add support for built-in functions
17747 vector bool char vec_revb (vector bool char);
17748 vector bool short vec_revb (vector short char);
17749 vector bool int vec_revb (vector bool int);
17750 vector bool long long vec_revb (vector bool long long);
17751 * doc/extend.texi: Update the built-in documentation file for the
17752 new built-in functions.
17754 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17756 * config/s390/s390.md: Remove movcc splitter.
17758 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17760 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17761 load/store on condition.
17763 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17766 * config/avr/avr.c (avr_encode_section_info)
17767 [progmem && !TREE_READONLY]: Error if progmem object needs
17770 2017-07-11 Michael Collison <michael.collison@arm.com>
17772 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17775 2017-07-11 Carl Love <cel@us.ibm.com>
17777 * config/rs6000/rs6000-c.c: Add support for builtins
17778 vector unsigned int vec_parity_lsbb (vector signed int);
17779 vector unsigned int vec_parity_lsbb (vector unsigned int);
17780 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17781 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17782 vector unsigned long long vec_parity_lsbb (vector signed long long);
17783 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17784 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17785 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17786 * doc/extend.texi: Update the built-in documentation file for the
17787 new built-in functions.
17789 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17791 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17792 (layout::m_primary_loc): New field.
17793 (layout::layout): Initialize new field. Move location filtering
17794 logic from here to...
17795 (layout::maybe_add_location_range): ...this new method. Add
17796 support for filtering to just the lines already specified by other
17798 (layout::will_show_line_p): New method.
17799 (gcc_rich_location::add_location_if_nearby): New method.
17800 (selftest::test_add_location_if_nearby): New test function.
17801 (selftest::diagnostic_show_locus_c_tests): Call it.
17802 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17805 2017-07-11 Tom de Vries <tom@codesourcery.com>
17807 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17808 (bb_first_real_insn): New function.
17809 (nvptx_single): Add extra initialization of broadcasted condition
17812 2017-07-11 Nathan Sidwell <nathan@acm.org>
17814 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17816 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17818 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17819 Binutils doc as TEXI will mess them up.
17820 * doc/invoke.texi (AVR Options): Same here.
17822 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17824 * config/sparc/sparc.opt (mfix-ut700): New option.
17825 (mfix-gr712rc): Likewise.
17826 (sparc_fix_b2bst): New variable.
17827 * doc/invoke.texi (SPARC options): Document them.
17828 (ARM options): Fix warnings.
17829 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17830 instructions to prevent sequences that can trigger the store-store
17831 errata for certain LEON3FT processors.
17832 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17833 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17834 * config/sparc/sparc.md (fix_b2bst): New attribute.
17835 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17837 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17840 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17842 (*rsqrtsf2_sse): Ditto.
17844 (div<mode>3): Macroize insn from divdf3 and divsf3
17845 using MODEF mode iterator.
17847 2017-07-10 Martin Sebor <msebor@redhat.com>
17849 PR tree-optimization/80397
17850 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17851 instead of testing for equality to INTEGER_TYPE.
17853 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17855 * config.gcc: Remove uclibc from arc target spec.
17857 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17859 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17861 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17864 * lto-wrapper.c (remove_option): New function.
17865 (merge_and_complain): Merge PIC/PIE options more realistically.
17867 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17869 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17873 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17874 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17875 * config.in: Regenerate.
17876 * configure: Regenerate.
17877 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17878 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17879 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17880 (TARGET_GASISR_PROLOGUES): ...target mask.
17881 * common/config/avr/avr-common.c
17882 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17883 Set -mgas-isr-prologues.
17884 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17885 INSERT_PASS_BEFORE for it.
17886 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17887 * config/avr/avr.c (avr_option_override)
17888 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17889 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17890 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17891 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17892 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17894 (make_avr_pass_pre_proep): New function.
17895 (emit_push_sfr) <treg>: Add argument to function and use it
17896 instead of TMP_REG.
17897 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17898 and set machine->gasisr.yes.
17899 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17900 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17901 __gcc_isr.n_pushed to .L__stack_usage.
17902 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17903 (avr_asm_final_postscan_insn): ...this new static function.
17904 * config/avr/avr.h (machine_function)
17905 <is_no_gccisr, use_L__stack_usage>: New fields.
17906 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17907 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17908 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17909 (gasisr, *gasisr): New expander and insn.
17910 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17911 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17912 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17914 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17916 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17919 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17921 Move jump-tables out of .text again.
17924 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17925 (ASM_OUTPUT_ADDR_VEC): New function.
17926 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17927 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17928 INSN_ADDRESSes as asm comment.
17929 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17930 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17931 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17932 * config/avr/avr.md (*tablejump): Adjust comment.
17933 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17934 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17936 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17937 (avr_output_addr_vec): New proto.
17938 (avr_log_t) <insn_addresses>: New field.
17940 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17943 * config/i386/i386.c (ix86_function_arg_advance): Set
17944 outgoing_args_on_stack to true if there are outgoing arguments
17946 (ix86_function_arg): Likewise.
17947 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17948 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17949 * config/i386/i386.h (machine_function): Add
17950 outgoing_args_on_stack.
17952 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17954 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17955 supporting pthreds.
17956 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17958 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17960 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17961 (REAL_H): Remove $(MACHMODE_H).
17962 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17964 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17965 $(MACHMODE_H) and double-int.h.
17966 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17968 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17971 2017-07-07 Andrew Pinski <apinski@cavium.com>
17973 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17974 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17976 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17978 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17979 Add warning if GCC was not configured to link against a GLIBC that
17980 exports the hardware capability bits.
17981 (make_resolver_func): Make resolver function private and not a
17982 COMDAT function. Create the name with clone_function_name instead
17983 of make_unique_name.
17986 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17987 correct operand in doing the split.
17989 2017-07-07 Carl Love <cel@us.ibm.com>
17991 * config/rs6000/rs6000-c: Add support for built-in function
17992 vector unsigned short vec_pack_to_short_fp32 (vector float,
17994 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17995 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17996 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17997 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17998 (convert_4f32_8i16): Add define_expand.
17999 * doc/extend.texi: Update the built-in documentation file for the
18000 new built-in function.
18002 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18004 * config/sparc/m8.md: New file.
18005 * config/sparc/sparc.md: Include m8.md.
18007 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18009 * config/sparc/sparc.opt: New option -mvis4b.
18010 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18011 (sparc_option_override): Handle VIS4B.
18012 (enum sparc_builtins): Define
18013 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18014 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18015 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18016 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18017 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18018 (check_constant_argument): New function.
18019 (sparc_vis_init_builtins): Define builtins
18020 __builtin_vis_dictunpack{8,16,32},
18021 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18022 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18023 __builtin_vis_fpcmpde{8,16,32}shl and
18024 __builtin_vis_fpcmpur{8,16,32}shl.
18025 (sparc_expand_builtin): Check that the constant operands to
18026 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18027 constant and in range.
18028 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18030 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18031 (SPARC_IMM5_P): Likewise.
18032 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18033 (enabled): Handle vis4b.
18034 (UNSPEC_DICTUNPACK): New unspec.
18035 (UNSPEC_FPCMPSHL): Likewise.
18036 (UNSPEC_FPUCMPSHL): Likewise.
18037 (UNSPEC_FPCMPDESHL): Likewise.
18038 (UNSPEC_FPCMPURSHL): Likewise.
18039 (cpu_feature): New CPU feature `vis4b'.
18040 (dictunpack{8,16,32}): New insns.
18041 (FPCSMODE): New mode iterator.
18042 (fpcscond): New code iterator.
18043 (fpcsucond): Likewise.
18044 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18045 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18046 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18047 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18048 * config/sparc/constraints.md: Define constraints `q' for unsigned
18049 2-bit integer constants and `t' for unsigned 5-bit integer
18051 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18053 (imm5_operand_dictunpack16): Likewise.
18054 (imm5_operand_dictunpack32): Likewise.
18055 (imm2_operand): Likewise.
18056 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18057 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18058 ditunpack* and fpcmp*shl builtins.
18060 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18062 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18063 * config.in: Add HAVE_AS_SPARC6 define.
18064 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18066 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18068 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18069 (CPP_CPU_SPEC): Handle m8.
18070 (ASM_CPU_SPEC): Likewise.
18071 * config/sparc/sparc-opts.h (enum processor_type): Add
18073 * config/sparc/sparc.c (m8_costs): New struct.
18074 (sparc_option_override): Handle TARGET_CPU_m8.
18075 (sparc32_initialize_trampoline): Likewise.
18076 (sparc64_initialize_trampoline): Likewise.
18077 (sparc_issue_rate): Likewise.
18078 (sparc_register_move_cost): Likewise.
18079 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18080 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18081 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18082 (CPP_CPU_SPEC): Handle M8.
18083 (ASM_CPU_SPEC): Likewise.
18084 (AS_M8_FLAG): Define.
18085 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18086 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18087 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18089 * configure: Regenerate.
18090 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18093 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18095 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18097 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18098 ("*movdi_insn_sp32"): Do not set v3pipe.
18099 ("*movsi_insn"): Likewise.
18100 ("*movdi_insn_sp64"): Likewise.
18101 ("*movsf_insn"): Likewise.
18102 ("*movdf_insn_sp32"): Likewise.
18103 ("*movdf_insn_sp64"): Likewise.
18104 ("*zero_extendsidi2_insn_sp64"): Likewise.
18105 ("*sign_extendsidi2_insn"): Likewise.
18106 ("*mov<VM32:mode>_insn"): Likewise.
18107 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18108 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18109 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18110 ("<vlop:code><VL:mode>3"): Likewise.
18111 ("*not_<vlop:code><VL:mode>3"): Likewise.
18112 ("*nand<VL:mode>_vis"): Likewise.
18113 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18114 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18115 ("one_cmpl<VL:mode>2"): Likewise.
18116 ("faligndata<VM64:mode>_vis"): Likewise.
18117 ("alignaddrsi_vis"): Likewise.
18118 ("alignaddrdi_vis"): Likweise.
18119 ("alignaddrlsi_vis"): Likewise.
18120 ("alignaddrldi_vis"): Likewise.
18121 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18122 ("bmaskdi_vis"): Likewise.
18123 ("bmasksi_vis"): Likewise.
18124 ("bshuffle<VM64:mode>_vis"): Likewise.
18125 ("cmask8<P:mode>_vis"): Likewise.
18126 ("cmask16<P:mode>_vis"): Likewise.
18127 ("cmask32<P:mode>_vis"): Likewise.
18128 ("pdistn<P:mode>_vis"): Likewise.
18129 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18131 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18133 * config/sparc/sparc.md ("subtype"): New insn attribute.
18134 ("*wrgsr_sp64"): Set insn subtype.
18135 ("*rdgsr_sp64"): Likewise.
18136 ("alignaddrsi_vis"): Likewise.
18137 ("alignaddrdi_vis"): Likewise.
18138 ("alignaddrlsi_vis"): Likewise.
18139 ("alignaddrldi_vis"): Likewise.
18140 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18141 ("fexpand_vis"): Likewise.
18142 ("fpmerge_vis"): Likewise.
18143 ("faligndata<VM64:mode>_vis"): Likewise.
18144 ("bshuffle<VM64:mode>_vis"): Likewise.
18145 ("cmask8<P:mode>_vis"): Likewise.
18146 ("cmask16<P:mode>_vis"): Likewise.
18147 ("cmask32<P:mode>_vis"): Likewise.
18148 ("fchksm16_vis"): Likewise.
18149 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18150 ("fmean16_vis"): Likewise.
18151 ("fp<plusminus_insn>64_vis"): Likewise.
18152 ("<plusminus_insn>v8qi3"): Likewise.
18153 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18154 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18155 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18156 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18157 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18158 ("*movqi_insn"): Likewise.
18159 ("*movhi_insn"): Likewise.
18160 ("*movsi_insn"): Likewise.
18161 ("movsi_pic_gotdata_op"): Likewise.
18162 ("*movdi_insn_sp32"): Likewise.
18163 ("*movdi_insn_sp64"): Likewise.
18164 ("movdi_pic_gotdata_op"): Likewise.
18165 ("*movsf_insn"): Likewise.
18166 ("*movdf_insn_sp32"): Likewise.
18167 ("*movdf_insn_sp64"): Likewise.
18168 ("*zero_extendhisi2_insn"): Likewise.
18169 ("*zero_extendqihi2_insn"): Likewise.
18170 ("*zero_extendqisi2_insn"): Likewise.
18171 ("*zero_extendqidi2_insn"): Likewise.
18172 ("*zero_extendhidi2_insn"): Likewise.
18173 ("*zero_extendsidi2_insn_sp64"): Likewise.
18174 ("ldfsr"): Likewise.
18175 ("prefetch_64"): Likewise.
18176 ("prefetch_32"): Likewise.
18177 ("tie_ld32"): Likewise.
18178 ("tie_ld64"): Likewise.
18179 ("*tldo_ldub_sp32"): Likewise.
18180 ("*tldo_ldub1_sp32"): Likewise.
18181 ("*tldo_ldub2_sp32"): Likewise.
18182 ("*tldo_ldub_sp64"): Likewise.
18183 ("*tldo_ldub1_sp64"): Likewise.
18184 ("*tldo_ldub2_sp64"): Likewise.
18185 ("*tldo_ldub3_sp64"): Likewise.
18186 ("*tldo_lduh_sp32"): Likewise.
18187 ("*tldo_lduh1_sp32"): Likewise.
18188 ("*tldo_lduh_sp64"): Likewise.
18189 ("*tldo_lduh1_sp64"): Likewise.
18190 ("*tldo_lduh2_sp64"): Likewise.
18191 ("*tldo_lduw_sp32"): Likewise.
18192 ("*tldo_lduw_sp64"): Likewise.
18193 ("*tldo_lduw1_sp64"): Likewise.
18194 ("*tldo_ldx_sp64"): Likewise.
18195 ("*mov<VM32:mode>_insn"): Likewise.
18196 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18197 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18199 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18201 * config/sparc/sparc.md ("type"): New insn type viscmp.
18202 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18204 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18205 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18206 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18207 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18209 ("n7_vis_logical_11cycle"): Likewise.
18210 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18211 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18212 * config/sparc/niagara.md ("niag_vis"): Likewise.
18213 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18214 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18216 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18218 * config/sparc/sparc.md: New instruction type `bmask'.
18219 (bmaskdi_vis): Use the `bmask' type.
18220 (bmasksi_vis): Likewise.
18221 * config/sparc/ultra3.md (us3_array): Likewise.
18222 * config/sparc/niagara7.md (n7_array): Likewise.
18223 * config/sparc/niagara4.md (n4_array): Likewise.
18224 * config/sparc/niagara2.md (niag2_vis): Likewise.
18225 (niag3_vis): Likewise.
18226 * config/sparc/niagara.md (niag_vis): Likewise.
18228 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18230 * ipa-comdats.c: Remove optimize check from gate.
18231 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18232 for functions not optimized.
18233 (ipa_fn_summary_read): Skip optimize check.
18234 (ipa_fn_summary_write): Likewise.
18235 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18237 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18239 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18240 (check_callers): Give up on caller which is not optimized.
18241 (inline_small_functions): Likewise.
18242 (ipa_inline): Do not give up when not optimizing.
18243 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18244 away unoptimizes cdtors.
18245 (whole_program_function_and_variable_visibility): Do
18246 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18247 * ipa.c (process_references): Do not check optimize.
18248 (symbol_table::remove_unreachable_nodes): Update optimize check.
18249 (set_writeonly_bit): Update optimize check.
18250 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18251 (pass_ipa_single_use::gate): Remove.
18253 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18255 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18256 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18257 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18258 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18259 permute_load, permute_store, adjust_extract, adjust_splat,
18260 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18261 replace_swap_with_copy, dump_swap_insn_table,
18262 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18263 recombine_lvx_pattern, recombine_stvx_pattern,
18264 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18265 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18266 to file rs6000-p8swap.c.
18267 * config/rs6000/rs6000-p8swap.c: New file.
18268 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18269 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18270 and rs6000*-*-* targets.
18272 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18274 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18276 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18278 * lto-wrapper.c (merge_and_complain): Do not merge
18279 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18280 fsigned_zeros, ftrapping_math, fwrapv.
18281 (append_compiler_options): Do not track these options.
18282 (append_linker_options): Likewie
18284 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18286 * cgraphunit.c (cgraph_node::finalize_function): When
18287 !flag_toplevel_reorde set no_reorder flag.
18288 (varpool_node::finalize_decl): Likewise.
18289 (symbol_table::compile): Drop no toplevel reorder path.
18291 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18293 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18294 edges; zero probability is not better than uninitialized.
18296 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18298 * asan.h (asan_sanitize_allocas_p): Declare.
18299 * asan.c (asan_sanitize_allocas_p): New function.
18300 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18301 (handle_builtin_alloca): Likewise.
18302 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18303 if !asan_sanitize_allocas_p.
18304 * params.def (asan-instrument-allocas): Add new option.
18305 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18306 * opts.c (common_handle_option): Disable allocas sanitization for
18309 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18311 * asan.c: Include gimple-fold.h.
18312 (get_last_alloca_addr): New function.
18313 (handle_builtin_stackrestore): Likewise.
18314 (handle_builtin_alloca): Likewise.
18315 (asan_emit_allocas_unpoison): Likewise.
18316 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18317 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18318 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18319 (instrument_builtin_call): Pass gimple iterator to
18320 get_mem_refs_of_builtin_call.
18321 (last_alloca_addr): New global.
18322 * asan.h (asan_emit_allocas_unpoison): Declare.
18323 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18324 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18325 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18326 if function calls alloca.
18327 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18328 * gimple-fold.h (replace_call_with_value): Declare.
18329 * internal-fn.c: Include asan.h.
18330 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18331 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18333 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18335 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18336 (C_SELFTEST_FLAGS): New.
18337 (CPP_SELFTEST_FLAGS): New.
18338 (SELFTEST_DEPS): New, from deps of s-selftest.
18339 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18340 (CPP_SELFTEST_DEPS): New.
18341 (selftest): Add dependency on s-selftest-c++.
18342 (s-selftest): Rename to...
18343 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18344 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18345 than SELFTEST_FLAGS.
18346 (selftest-gdb): Rename to...
18347 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18349 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18350 (selftest-valgrind): Rename to...
18351 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18353 (selftest-valgrind): Reintroduce as an alias for
18354 selftest-c-valgrind.
18355 (s-selftest-c++): New.
18356 (selftest-c++-gdb): New.
18357 (selftest-c++-valgrind): New.
18359 2017-07-06 Olivier Hainque <hainque@adacore.com>
18361 * gcc.c (process_command): When deciding if undefined variables
18362 should be ignored when processing specs, accept "gcc -v" as well.
18364 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18366 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18367 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18369 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18371 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18372 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18373 * config/arm/arm-tables.opt: Regenerate.
18374 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18376 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18378 2017-07-06 Carl Love <cel@us.ibm.com>
18380 * ChangeLog: Clean up from mid air collision
18382 2017-07-06 Carl Love <cel@us.ibm.com>
18384 * config/rs6000/rs6000-c.c: Add support for built-in functions
18385 vector signed int vec_subc (vector signed int, vector signed int);
18386 vector signed __int128 vec_subc (vector signed __int128,
18387 vector signed __int128);
18388 vector unsigned __int128 vec_subc (vector unsigned __int128,
18389 vector unsigned __int128);
18390 vector signed int vec_sube (vector signed int, vector signed int,
18391 vector signed int);
18392 vector unsigned int vec_sube (vector unsigned int,
18393 vector unsigned int,
18394 vector unsigned int);
18395 vector signed __int128 vec_sube (vector signed __int128,
18396 vector signed __int128,
18397 vector signed__int128);
18398 vector unsigned __int128 vec_sube (vector unsigned __int128,
18399 vector unsigned __int128,
18400 vector unsigned __int128);
18401 vector signed int vec_subec (vector signed int, vector signed int,
18402 vector signed int);
18403 vector unsigned int vec_subec (vector unsigned int,
18404 vector unsigned int,
18405 vector unsigned int);
18406 vector signed __int128 vec_subec (vector signed __int128,
18407 vector signed __int128,
18408 vector signed__int128);
18409 vector unsigned __int128 vec_subec (vector unsigned __int128,
18410 vector unsigned __int128,
18411 vector unsigned __int128);
18412 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18413 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18414 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18415 BU_ALTIVEC_OVERLOAD_X definitions.
18416 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18417 * doc/extend.texi: Update the built-in documentation file for the new
18418 built-in functions.
18420 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18423 * diagnostic-show-locus.c (layout::layout): Use start and finish
18424 spelling location for the start and finish of each range.
18425 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18426 Add unused aspect param.
18427 * input.c (expand_location_1): Add "aspect" param, and use it
18428 to access the correct part of the location.
18429 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18431 (expand_location_to_spelling_point): Likewise.
18432 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18433 param, and pass it to expand_location_1.
18435 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18437 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18438 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18439 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18440 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18441 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18442 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18443 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18444 _mm_maskz_getmant_ss): New intrinsics.
18445 (__builtin_ia32_getexpss128_mask): Changed to ...
18446 __builtin_ia32_getexpss128_round ... this.
18447 (__builtin_ia32_getexpsd128_mask): Changed to ...
18448 __builtin_ia32_getexpsd128_round ... this.
18449 * config/i386/i386-builtin-types.def
18450 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18451 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18452 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18453 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18454 __builtin_ia32_getmantss_mask_round): New builtins.
18455 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18456 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18457 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18458 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18459 * config/i386/sse.md
18460 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18461 avx512f_sgetexp<mode><mask_scalar_name>
18462 <round_saeonly_scalar_name> ... this.
18463 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18464 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18465 vgetexp<ssescalarmodesuffix>
18466 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18467 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18468 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18469 avx512f_vgetmant<mode><mask_scalar_name>
18470 <round_saeonly_scalar_name> ... this.
18471 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18472 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18473 vgetmant<ssescalarmodesuffix>
18474 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18475 %0<mask_scalar_operand4>, %1, %2
18476 <round_saeonly_scalar_mask_op4>, %3} ... this.
18477 * config/i386/subst.md (mask_scalar_operand4,
18478 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18479 round_saeonly_scalar_nimm_predicate): New subst attributes.
18481 2017-07-06 Julia Koval <julia.koval@intel.com>
18483 * config/i386/i386.c (ix86_erase_embedded_rounding):
18484 Remove code for old rounding pattern.
18486 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18488 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18490 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18492 * doc/sourcebuild.texi (Test Directives, Variants of
18493 dg-require-support): Add documentation for dg-require-stack-check.
18495 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18497 * config/i386/subst.md (mask_scalar, round_scalar,
18498 round_saeonly_scalar): New meta-templates.
18499 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18500 round_scalar_mask_operand3, round_scalar_mask_op3,
18501 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18502 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18503 round_saeonly_scalar_constraint,
18504 round_saeonly_scalar_prefix): New subst attribute.
18505 * config/i386/sse.md
18506 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18507 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18508 <round_scalar_name> ... this.
18509 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18510 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18511 <round_scalar_name> ... this.
18512 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18513 <sse>_vm<code><mode>3<mask_scalar_name>
18514 <round_saeonly_scalar_name> ... this.
18515 (v<plusminus_mnemonic><ssescalarmodesuffix>
18516 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18517 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18518 v<plusminus_mnemonic><ssescalarmodesuffix>
18519 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18520 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18521 (v<multdiv_mnemonic><ssescalarmodesuffix>
18522 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18523 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18524 v<multdiv_mnemonic><ssescalarmodesuffix>
18525 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18526 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18527 (v<maxmin_float><ssescalarmodesuffix>
18528 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18529 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18530 v<maxmin_float><ssescalarmodesuffix>
18531 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18532 %0<mask_scalar_operand3>, %1, %<iptr>2
18533 <round_saeonly_scalar_mask_op3>} ... this.
18535 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18537 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18538 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18540 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18541 Alan Hayward <alan.hayward@arm.com>
18542 David Sherwood <david.sherwood@arm.com>
18544 * combine.c (simplify_if_then_else): Remove "enum" before
18546 * compare-elim.c (can_eliminate_compare): Likewise.
18547 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18549 (aarch64_lookup_simd_builtin_type): Likewise.
18550 (aarch64_simd_builtin_type): Likewise.
18551 (aarch64_init_simd_builtin_types): Likewise.
18552 (aarch64_simd_expand_args): Likewise.
18553 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18555 (aarch64_reverse_mask): Likewise.
18556 (aarch64_simd_emit_reg_reg_move): Likewise.
18557 (aarch64_gen_adjusted_ldpstp): Likewise.
18558 (aarch64_ccmp_mode_to_code): Likewise.
18559 (aarch64_operands_ok_for_ldpstp): Likewise.
18560 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18561 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18563 (aarch64_min_divisions_for_recip_mul): Likewise.
18564 (aarch64_reassociation_width): Likewise.
18565 (aarch64_get_condition_code_1): Likewise.
18566 (aarch64_simd_emit_reg_reg_move): Likewise.
18567 (aarch64_simd_attr_length_rglist): Likewise.
18568 (aarch64_reverse_mask): Likewise.
18569 (aarch64_operands_ok_for_ldpstp): Likewise.
18570 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18571 (aarch64_gen_adjusted_ldpstp): Likewise.
18572 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18574 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18575 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18576 (arm_lookup_simd_builtin_type): Likewise.
18577 (arm_simd_builtin_type): Likewise.
18578 (arm_init_simd_builtin_types): Likewise.
18579 (arm_expand_builtin_args): Likewise.
18580 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18581 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18582 (ft32_setup_incoming_varargs): Likewise.
18583 (ft32_function_arg): Likewise.
18584 (ft32_function_arg_advance): Likewise.
18585 (ft32_pass_by_reference): Likewise.
18586 (ft32_arg_partial_bytes): Likewise.
18587 (ft32_valid_pointer_mode): Likewise.
18588 (ft32_addr_space_pointer_mode): Likewise.
18589 (ft32_addr_space_legitimate_address_p): Likewise.
18590 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18592 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18593 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18594 (iamcu_alignment): Likewise.
18595 (canonicalize_vector_int_perm): Likewise.
18596 (ix86_noce_conversion_profitable_p): Likewise.
18597 (ix86_mpx_bound_mode): Likewise.
18598 (ix86_operands_ok_for_move_multiple): Likewise.
18599 * config/microblaze/microblaze-protos.h
18600 (microblaze_expand_conditional_branch_reg): Likewise.
18601 * config/microblaze/microblaze.c
18602 (microblaze_expand_conditional_branch_reg): Likewise.
18603 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18605 (rs6000_reassociation_width): Likewise.
18606 (rs6000_invalid_binary_op): Likewise.
18607 (fusion_p9_p): Likewise.
18608 (emit_fusion_p9_load): Likewise.
18609 (emit_fusion_p9_store): Likewise.
18610 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18612 (riscv_hard_regno_mode_ok_p): Likewise.
18613 (riscv_address_insns): Likewise.
18614 (riscv_split_symbol): Likewise.
18615 (riscv_legitimize_move): Likewise.
18616 (riscv_function_value): Likewise.
18617 (riscv_hard_regno_nregs): Likewise.
18618 (riscv_expand_builtin): Likewise.
18619 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18620 (riscv_build_integer): Likewise.
18621 (riscv_split_integer): Likewise.
18622 (riscv_legitimate_constant_p): Likewise.
18623 (riscv_cannot_force_const_mem): Likewise.
18624 (riscv_regno_mode_ok_for_base_p): Likewise.
18625 (riscv_valid_base_register_p): Likewise.
18626 (riscv_valid_offset_p): Likewise.
18627 (riscv_valid_lo_sum_p): Likewise.
18628 (riscv_classify_address): Likewise.
18629 (riscv_legitimate_address_p): Likewise.
18630 (riscv_address_insns): Likewise.
18631 (riscv_load_store_insns): Likewise.
18632 (riscv_force_binary): Likewise.
18633 (riscv_split_symbol): Likewise.
18634 (riscv_force_address): Likewise.
18635 (riscv_legitimize_address): Likewise.
18636 (riscv_move_integer): Likewise.
18637 (riscv_legitimize_const_move): Likewise.
18638 (riscv_legitimize_move): Likewise.
18639 (riscv_address_cost): Likewise.
18640 (riscv_subword): Likewise.
18641 (riscv_output_move): Likewise.
18642 (riscv_canonicalize_int_order_test): Likewise.
18643 (riscv_emit_int_order_test): Likewise.
18644 (riscv_function_arg_boundary): Likewise.
18645 (riscv_pass_mode_in_fpr_p): Likewise.
18646 (riscv_pass_fpr_single): Likewise.
18647 (riscv_pass_fpr_pair): Likewise.
18648 (riscv_get_arg_info): Likewise.
18649 (riscv_function_arg): Likewise.
18650 (riscv_function_arg_advance): Likewise.
18651 (riscv_arg_partial_bytes): Likewise.
18652 (riscv_function_value): Likewise.
18653 (riscv_pass_by_reference): Likewise.
18654 (riscv_setup_incoming_varargs): Likewise.
18655 (riscv_print_operand): Likewise.
18656 (riscv_elf_select_rtx_section): Likewise.
18657 (riscv_save_restore_reg): Likewise.
18658 (riscv_for_each_saved_reg): Likewise.
18659 (riscv_register_move_cost): Likewise.
18660 (riscv_hard_regno_mode_ok_p): Likewise.
18661 (riscv_hard_regno_nregs): Likewise.
18662 (riscv_class_max_nregs): Likewise.
18663 (riscv_memory_move_cost): Likewise.
18664 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18665 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18666 (rl78_addr_space_address_mode): Likewise.
18667 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18669 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18670 (rs6000_reassociation_width): Likewise.
18671 (rs6000_invalid_binary_op): Likewise.
18672 (fusion_p9_p): Likewise.
18673 (emit_fusion_p9_load): Likewise.
18674 (emit_fusion_p9_store): Likewise.
18675 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18676 (ok_for_simple_move_operands): Likewise.
18677 (ok_for_simple_move_strict_operands): Likewise.
18678 (ok_for_simple_arith_logic_operands): Likewise.
18679 (visium_legitimize_reload_address): Likewise.
18680 (visium_select_cc_mode): Likewise.
18681 (output_cbranch): Likewise.
18682 (visium_split_double_move): Likewise.
18683 (visium_expand_copysign): Likewise.
18684 (visium_expand_int_cstore): Likewise.
18685 (visium_expand_fp_cstore): Likewise.
18686 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18687 (visium_function_arg): Likewise.
18688 (visium_function_arg_advance): Likewise.
18689 (visium_libcall_value): Likewise.
18690 (visium_setup_incoming_varargs): Likewise.
18691 (visium_legitimate_constant_p): Likewise.
18692 (visium_legitimate_address_p): Likewise.
18693 (visium_legitimize_address): Likewise.
18694 (visium_secondary_reload): Likewise.
18695 (visium_register_move_cost): Likewise.
18696 (visium_memory_move_cost): Likewise.
18697 (prepare_move_operands): Likewise.
18698 (ok_for_simple_move_operands): Likewise.
18699 (ok_for_simple_move_strict_operands): Likewise.
18700 (ok_for_simple_arith_logic_operands): Likewise.
18701 (visium_function_value_1): Likewise.
18702 (rtx_ok_for_offset_p): Likewise.
18703 (visium_legitimize_reload_address): Likewise.
18704 (visium_split_double_move): Likewise.
18705 (visium_expand_copysign): Likewise.
18706 (visium_expand_int_cstore): Likewise.
18707 (visium_expand_fp_cstore): Likewise.
18708 (visium_split_cstore): Likewise.
18709 (visium_select_cc_mode): Likewise.
18710 (visium_split_cbranch): Likewise.
18711 (output_cbranch): Likewise.
18712 (visium_print_operand_address): Likewise.
18713 * expmed.c (flip_storage_order): Likewise.
18714 * expmed.h (emit_cstore): Likewise.
18715 (flip_storage_order): Likewise.
18716 * genrecog.c (validate_pattern): Likewise.
18717 * hsa-gen.c (gen_hsa_addr): Likewise.
18718 * internal-fn.c (expand_arith_overflow): Likewise.
18719 * ira-color.c (allocno_copy_cost_saving): Likewise.
18720 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18721 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18722 (process_invariant_for_inheritance): Likewise.
18723 * lra-eliminations.c (move_plus_up): Likewise.
18724 * omp-low.c (lower_oacc_reductions): Likewise.
18725 * simplify-rtx.c (simplify_subreg): Likewise.
18726 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18727 (TARGET_CHKP_BOUND_MODE): Likewise..
18728 * targhooks.c (default_chkp_bound_mode): Likewise.
18729 (default_setup_incoming_vararg_bounds): Likewise.
18730 * targhooks.h (default_chkp_bound_mode): Likewise.
18731 (default_setup_incoming_vararg_bounds): Likewise.
18732 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18733 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18734 (have_whole_vector_shift): Likewise.
18735 * tree-vect-stmts.c (vectorizable_load): Likewise.
18736 * doc/tm.texi: Regenerate.
18738 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18740 Graceful degrade if Binutils PR21472 is not available.
18743 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18744 .rodata in flash test fails.
18745 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18746 * confgure: Regenerate.
18747 * config.in: Regenerate.
18748 * config/avr/avr.c (avr_asm_named_section)
18749 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18750 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18751 (avr_asm_init_sections): Same.
18753 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18755 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18756 (fma<VH:mode>4_intrinsic): Likewise.
18757 (*fmsub<VCVTF:mode>4): Likewise.
18758 (*fmsub<VH:mode>4_intrinsic): Likewise.
18760 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18763 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18764 Don't depend on "optimize > 0".
18765 (out_movhi_r_mr, out_movqi_mr_r): Same.
18766 (out_movhi_mr_r, out_movqi_r_mr): Same.
18767 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18768 io_address_operand on "optimize > 0".
18770 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18772 * tree-loop-distribution.c: Add general explanantion on the pass.
18773 (generate_loops_for_partition): Mark distributed loop.
18774 (pg_add_dependence_edges): New parameter. Handle alias data
18775 dependence specially and record it in the parameter if asked.
18776 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18777 (init_partition_graph_vertices, add_partition_graph_edge): New.
18778 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18779 (free_partition_graph_vdata, build_partition_graph): New.
18780 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18781 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18782 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18783 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18784 (version_for_distribution_p, finalize_partitions): New.
18785 (distribute_loop): Handle alias data dependence specially. Factor
18786 out loop fusion code as functions and call these functions.
18788 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18790 * tree-loop-distribution.c (classify_partition): New parameter and
18791 better handle reduction statement.
18792 (rdg_build_partitions): Revise comment.
18793 (distribute_loop): Compute statements in all partitions and pass it
18794 to classify_partition.
18796 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18798 * tree-loop-distribution.c (enum partition_type): New.
18799 (struct partition): New field type.
18800 (partition_merge_into): Add parameter. Update partition type.
18801 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18802 (build_rdg_partition_for_vertex): Compute partition type.
18803 (rdg_build_partitions): Dump partition type.
18804 (distribute_loop): Update calls to partition_merge_into.
18806 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18808 * tree-loop-distribution.c (struct ddr_hasher): New.
18809 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18811 (classify_partition): Call get_data_dependence.
18812 (pg_add_dependence_edges): Ditto.
18813 (distribute_loop): Release data dependence hash table.
18815 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18817 * tree-loop-distribution.c (ref_base_address): Delete.
18818 (similar_memory_accesses): Rename ...
18819 (share_memory_accesses): ... to this. Check if partitions access
18820 the same memory reference.
18821 (distribute_loop): Call share_memory_accesses.
18823 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18825 * tree-loop-distribution.c (struct partition): New field recording
18826 its data reference.
18827 (partition_alloc, partition_free): Init and release data refs.
18828 (partition_merge_into): Merge data refs.
18829 (build_rdg_partition_for_vertex): Collect data refs for partition.
18830 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18832 (distribute_loop): Remve data refs from vertice data of partition
18835 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18837 * tree-loop-distribution.c (params.h): Include header file.
18838 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18839 (datarefs_vec): New global var.
18840 (create_rdg_vertices): Use datarefs_vec directly.
18841 (free_rdg): Don't free data references.
18842 (build_rdg): Update use. Don't free data references.
18843 (distribute_loop): Compute global variable for data references.
18844 Bail out if there are too many data references.
18846 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18848 * tree-loop-distribution.c (loop_nest): New global var.
18849 (build_rdg): Use loop directly, rather than loop nest.
18850 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18852 (distribute_loop): Compute global variable loop nest. Update use.
18854 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18856 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18857 (partition_merge_into): New parameter. Dump reason for fusion.
18858 (distribute_loop): Update use of partition_merge_into.
18860 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18862 * tree-loop-distribution.c (bb_top_order_index): New.
18863 (bb_top_order_index_size, bb_top_order_cmp): New.
18864 (stmts_from_loop): Use topological order.
18865 (pass_loop_distribution::execute): Compute and release topological
18866 order for basic blocks.
18868 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18870 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18873 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18875 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18876 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18877 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18878 * internal-fn.def (LOOP_DIST_ALIAS): New.
18879 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18880 (fold_loop_internal_call): ... this.
18881 (vect_loop_dist_alias_call): New function.
18882 (set_uid_loop_bbs): Call fold_loop_internal_call.
18883 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18886 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18889 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18890 Require dead FLAGS_REG at the beginning of a peephole.
18892 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18895 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18896 arguments in the call to __builtin_ia32_sbb_u32.
18897 (_subborrow_u64): Swap _X and _Y arguments in the call to
18898 __builtin_ia32_sbb_u64.
18900 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18903 * tree-vrp.c (compare_assert_loc): Turn into a function template
18904 with stable template parameter. Only test if a->e is NULL,
18905 !a->e == !b->e has been verified already. Use e == NULL or
18906 e != NULL instead of e or ! e tests. If stable is true, don't use
18907 iterative_hash_expr, on the other side allow a or b or both NULL
18908 and sort the NULLs last.
18909 (process_assert_insertions): Sort using compare_assert_loc<false>
18910 instead of compare_assert_loc, later sort using
18911 compare_assert_loc<true> before calling process_assert_insertions_for
18912 in a loop. Use break instead of continue once seen NULL pointer.
18914 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18916 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18917 Cortex-R7 and Cortex-R8 processors.
18919 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18921 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18922 uninitialized while src is not.
18924 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18926 * common/config/arm/arm-common.c: Adjust include path for
18928 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18929 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18930 (arm-cpu-data.h): Likewise.
18931 (arm-cpu-cdata.h): Likewise.
18932 * config/arm/arm-cpu.h: Delete.
18933 * config/arm/arm-cpu-cdata.h: Delete.
18934 * config/arm/arm-cpu-data.h: Delete.
18936 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18938 * config/arm/arm-cpus.in (cortex-a55): New.
18939 (cortex-a75): Likewise.
18940 (cortex-a75.cortex-a55): Likewise.
18941 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18943 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18944 * config/arm/arm-cpu-cdata.h: Regenerate.
18945 * config/arm/arm-cpu-data.h: Regenerate.
18946 * config/arm/arm-cpu.h: Regenerate.
18947 * config/arm/arm-tables.opt: Regenerate.
18948 * config/arm/arm-tune.md: Regenerate.
18950 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18952 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18954 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18956 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18959 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18961 PR tree-optimization/81292
18962 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18963 full_string_p, also call adjust_related_strinfos if the adjustment
18964 is simple, otherwise invalidate related strinfos.
18966 2017-07-04 Martin Liska <mliska@suse.cz>
18969 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18970 newly created variable as DECL_IGNORED_P.
18972 2017-07-04 Martin Liska <mliska@suse.cz>
18975 * ipa-inline.c (inline_small_functions):
18976 Use xstrdup_for_dump.
18978 2017-07-04 Tom de Vries <tom@codesourcery.com>
18980 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18982 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18985 * config/darwin.c (darwin_function_switched_text_sections):
18988 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18990 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18992 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18994 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18996 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18998 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18999 min_profitable_iters, and th as inclusive lower bounds.
19000 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19001 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19002 for min_profitable_iters and min_profitable_estimate.
19003 (vect_transform_loop): Treat th as an inclusive lower bound.
19004 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19006 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19009 * config/darwin.c (darwin_function_switched_text_sections):
19010 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19011 in two pieces, and suppress the use of buf.
19013 2017-07-03 Nathan Sidwell <nathan@acm.org>
19015 * hash-table.h (hash_table_mod1): Fix indentation.
19017 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19019 PR middle-end/81290
19020 * predict.c (force_edge_cold): Be more careful about propagation
19022 * profile-count.h (profile_probability::guessed,
19023 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19025 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19027 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19029 * doc/invoke.texi (rcpc architecture extension): Document it.
19031 2017-07-03 Richard Biener <rguenther@suse.de>
19033 PR tree-optimization/60510
19034 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19035 the scalar reduction PHI and use it.
19036 (vectorizable_reduction): Properly guard the single_defuse_cycle
19037 path for non-SLP reduction chains where we cannot use it.
19038 Rework reduc_def/index and vector type deduction. Rework
19039 vector operand gathering during reduction op code-gen.
19040 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19041 chains dissolve the chain and leave it to non-SLP reduction
19044 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19046 * tree-data-ref.h (dr_alignment): Declare.
19047 * tree-data-ref.c (dr_alignment): New function.
19048 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19049 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19051 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19053 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19055 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19056 and base_misalignment fields.
19057 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19058 * tree-data-ref.c: Include builtins.h.
19059 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19060 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19061 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19062 * tree-vect-data-refs.c: Include tree-cfg.h.
19063 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19064 fields instead of calculating an alignment here.
19065 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19066 innermost_loop_behavior fields.
19068 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19070 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19072 (DR_STEP_ALIGNMENT): New macro.
19073 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19074 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19075 (create_data_ref): Print it.
19076 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19077 to tell whether the step preserves vector (mis)alignment.
19078 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19079 Move the check for an integer step and generalise to all INTEGER_CST.
19080 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19081 Print the outer step alignment.
19083 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19085 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19086 with offset_alignment.
19087 (DR_ALIGNED_TO): Delete.
19088 (DR_OFFSET_ALIGNMENT): New macro.
19089 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19090 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19091 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19092 (create_data_ref): Likewise.
19093 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19094 (vect_analyze_data_refs): Likewise.
19095 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19096 creating dummy innermost behavior.
19098 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19100 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19101 with a "innermost_loop_behavior *" and refeence tree.
19102 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19103 (create_data_ref): Update call accordingly.
19104 * tree-predcom.c (find_looparound_phi): Likewise.
19106 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19108 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19109 fields with dr_wrt_vec_loop.
19110 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19111 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19112 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19113 (vect_dr_behavior): New function.
19114 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19115 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19116 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19117 track whether the step preserves the misalignment.
19118 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19119 Use vect_dr_behavior.
19120 (vect_setup_realignment): Update call accordingly.
19121 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19122 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19123 call to vect_create_addr_base_for_vector_ref.
19124 (vect_create_cond_for_align_checks): Likewise.
19125 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19126 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19127 (vect_recog_mask_conversion_pattern): Likewise.
19128 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19129 (new_stmt_vec_info): Remove redundant zeroing.
19131 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19133 * common/config/arm/arm-common.c (arm_be8_option): New function.
19134 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19135 (ISA_ARMv6): Add isa_bit_be8.
19136 * config/arm/arm.h (arm_be8_option): Add prototype.
19137 (BE8_SPEC_FUNCTION): New define.
19138 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19139 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19140 (mlittle-endian): Similarly.
19141 (mbe8, mbe32): New options.
19142 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19143 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19145 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19147 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19149 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19151 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19152 (cleanup_tree_cfg_bb): Use it.
19153 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19155 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19157 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19160 * loop-doloop.c (add_test): Update profile.
19162 2017-07-03 Martin Liska <mliska@suse.cz>
19165 * sanopt.c (rewrite_usage_of_param): New function.
19166 (sanitize_rewrite_addressable_params): Likewise.
19167 (pass_sanopt::execute): Call rewrite_usage_of_param.
19169 2017-07-03 Richard Biener <rguenther@suse.de>
19171 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19172 back to using VIEW_CONVERT_EXPR.
19174 2017-07-03 Martin Liska <mliska@suse.cz>
19177 * doc/extend.texi: Document when a resolver function is
19178 generated for target_clones.
19180 2017-07-03 Martin Liska <mliska@suse.cz>
19182 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19183 and shadow memory of auto variables which are subject of
19184 use-after-scope sanitization.
19185 (asan_expand_mark_ifn): Add do set only when is_poison.
19187 2016-07-03 Richard Biener <rguenther@suse.de>
19189 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19191 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19192 (vectorizable_reduction): Perform reduction PHI creation when
19193 visiting a reduction PHI and adjust and simplify code generation
19194 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19195 (vect_transform_loop): Visit reduction PHIs.
19196 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19197 defs into the SLP tree.
19198 (vect_build_slp_tree): Reduction defs terminate the recursion.
19199 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19201 (vect_get_vec_defs_for_stmt_copy): Export.
19202 (vect_get_vec_defs): Likewise.
19203 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19205 (vect_get_vec_defs_for_stmt_copy): Declare.
19206 (vect_get_vec_defs): Likewise.
19208 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19210 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19211 parameter with a "loop" parameter and use it instead of the
19212 loop containing DR_STMT. Don't check simple_iv when doing
19213 BB analysis. Describe the two analysis modes in the comment.
19215 2017-07-03 Tom de Vries <tom@codesourcery.com>
19217 PR tree-optimization/69468
19218 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19219 (find_same_succ_bb): Handle ignore_edge_flags.
19221 2017-07-03 Tom de Vries <tom@codesourcery.com>
19223 PR tree-optimization/81192
19224 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19226 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19228 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19230 2017-07-03 Tom de Vries <tom@codesourcery.com>
19232 PR tree-optimization/81192
19233 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19234 BB_SAME_SUCC (bb) == NULL.
19236 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19238 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19241 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19243 * dumpfile.c: Include profile-count.h
19244 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19246 (insert_cond_bb): Update profile.
19247 * tree-cfg.h (insert_cond_bb): Update prototype.
19248 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19249 * tree-dump.c: Do not include tree-cfg.
19251 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19253 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19255 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19257 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19260 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19262 * tree-complex.c (expand_complex_div_wide): update profile.
19264 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19265 Alan Hayward <alan.hayward@arm.com>
19266 David Sherwood <david.sherwood@arm.com>
19268 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19269 (CORETYPES_H): New define.
19270 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19271 (insn-modes-inline.h, s-modes-inline-h): New rules.
19272 (generated_files): Add insn-modes-inline.h.
19273 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19274 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19275 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19276 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19277 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19278 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19279 (build/gencodes.o, build/genconditions.o): Likewise.
19280 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19281 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19282 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19283 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19284 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19285 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19286 * coretypes.h: Include everything up to real.h for generators.
19287 Include insn-modes.h first. Include wide-int-print.h after
19288 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19289 * machmode.h: Don't include insn-modes.h here.
19290 * function-tests.c: Remove includes of signop.h, machmode.h,
19291 double-int.h and wide-int.h.
19293 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19295 * optc-save-gen.awk: Likewise.
19296 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19297 * godump.c: Remove include of wide-int-print.h.
19298 * pretty-print.h: Likewise.
19299 * wide-int-print.cc: Likewise.
19300 * wide-int.cc: Likewise.
19301 * hash-map-tests.c: Remove include of signop.h.
19302 * hash-set-tests.c: Likewise.
19303 * rtl-tests.c: Likewise.
19304 * mkconfig.sh: Remove include of machmode.h.
19305 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19307 (emit_insn_modes_inline_h): ...this new function. Emit the code
19308 into an insn-modes-inline.h header file, adding appropriate
19309 include guards and end comments.
19310 (emit_insn_modes_c_header): Remove include of machmode.h.
19311 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19313 (main): Handle -i flag and call emit_insn_modes_inline_h when
19316 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19318 * tree-ssa-strlen.c (strinfo): Rename the length field to
19319 nonzero_chars. Add a full_string_p field.
19320 (compare_nonzero_chars, zero_length_string_p): New functions.
19321 (get_addr_stridx): Add an offset_out parameter.
19322 Use compare_nonzero_chars.
19323 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19324 (new_strinfo): Update after above changes to strinfo.
19325 (set_endptr_and_length): Set full_string_p.
19326 (get_string_length): Update after above changes to strinfo.
19327 (unshare_strinfo): Update call to new_strinfo.
19328 (maybe_invalidate): Likewise.
19329 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19330 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19331 as a uhwi instead of an shwi. Update after above changes to
19332 strinfo and new_strinfo.
19333 (zero_length_string): Assert that chainsi contains full strings.
19334 Use zero_length_string_p. Update call to new_strinfo.
19335 (adjust_related_strinfos): Update after above changes to strinfo.
19336 Copy full_string_p from origsi.
19337 (adjust_last_stmt): Use zero_length_string_p.
19338 (handle_builtin_strlen): Update after above changes to strinfo and
19339 new_strinfo. Install the lhs as the string length if the previous
19340 entry didn't describe a full string.
19341 (handle_builtin_strchr): Update after above changes to strinfo
19343 (handle_builtin_strcpy): Likewise.
19344 (handle_builtin_strcat): Likewise.
19345 (handle_builtin_malloc): Likewise.
19346 (handle_pointer_plus): Likewise.
19347 (handle_builtin_memcpy): Likewise. Track nonzero characters
19348 that aren't necessarily followed by a nul terminator.
19349 (handle_char_store): Likewise.
19351 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19353 PR tree-optimization/80769
19354 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19355 for malloc and calloc. Document the new invariant that all related
19356 strinfos have delayed lengths or none do.
19357 (verify_related_strinfos): Move earlier in file.
19358 (set_endptr_and_length): New function, split out from...
19359 (get_string_length): ...here. Also set the lengths of related
19361 (zero_length_string): Assert that chainsi has known (rather than
19363 (adjust_related_strinfos): Likewise.
19365 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19367 PR tree-optimization/81136
19368 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19369 assert that two references with the same misalignment have the same
19370 compile-time misalignment if those compile-time misalignments
19373 2017-07-01 Andi Kleen <ak@linux.intel.com>
19375 * print-tree.c (print_node): Print all attributes.
19377 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19379 * cfg.c (scale_bbs_frequencies): New function.
19380 * cfg.h (scale_bbs_frequencies): Declare it.
19381 * cfgloopanal.c (single_likely_exit): Cleanup.
19382 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19384 (scale_loop_profile): Likewise.
19385 (loop_version): Likewise.
19386 (create_empty_loop_on_edge): Update.
19387 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19388 scale_loop_frequencies, scale_loop_profile, loopify,
19389 loop_version): Update prototypes.
19390 * modulo-sched.c (sms_schedule): Update.
19391 * predict.c (unlikely_executed_edge_p): Also check probability.
19392 (probably_never_executed_edge_p): Fix typo.
19393 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19394 * tree-parloops.c (gen_parallel_loop): Update.
19395 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19396 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19397 * tree-ssa-loop-split.c (split_loop): Update.
19398 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19399 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19400 (vect_loop_versioning): Update.
19401 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19403 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19405 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19407 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19409 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19410 to keep profile consistent.
19412 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19414 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19415 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19416 * profile-count.h (max_safe_multiplier): Make unsigned.
19417 (profile_count::guessed_zero): New.
19419 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19421 * bb-reorder.c (fix_up_crossing_landing_pad,
19422 fix_crossing_conditional_branches): Use make_single_succ_edge
19423 to keep profile consistent.
19425 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19427 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19430 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19433 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19434 the right scopes, make sure cond_jump isn't preserved between multiple
19435 iterations. Search for fallthru edge whenever there are 3+ edges and
19436 use find_fallthru_edge for it.
19438 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19440 Patch by Alexander Monakov <amonakov@ispras.ru>
19441 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19442 probabilities consistently.
19444 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19446 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19447 emit_cmp_and_jump_insns.
19449 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19452 * tree-inline.c (expand_call_inline): Combine profile statuses.
19454 2017-06-30 Andrew Pinski <apinski@cavium.com>
19456 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19457 fold_stmt returned true.
19459 2017-06-30 Nathan Sidwell <nathan@acm.org>
19461 * ggc.h (empty_string): Delete.
19462 * cfgexpand.c (expand_asm_stmt): Use plain "".
19463 * optabs.c (expand_asm_memory_barrier): Likewise.
19464 * stringpool.c (empty_string): Delete.
19465 (digit_vector, digit_string): Delete.
19466 (ggc_alloc_string): Use plain "", don't optimize single digit
19467 strings. Use ggc_alloc_atomic.
19469 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19471 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19472 comparison set and one other set, use the cost of the non-comparison
19475 2017-06-30 Nathan Sidwell <nathan@acm.org>
19477 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19480 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19482 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19483 loops. Remove now unneeded calls to gimple_switch_set_label() that
19484 just set removed labels to NULL_TREE.
19486 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19488 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19489 (set_range_info): ...here. Only call set_range_info_raw if domain
19491 (set_nonzero_bits): Call set_range_info_raw.
19492 * tree-ssanames.h (set_range_info_raw): New.
19494 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19497 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19498 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19499 of nonimmediate_operand and <store_mask_constraint> instead of m
19500 for the input operand. For V8FI iterator, always split if input
19501 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19502 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19503 <store_mask_predicate> instead of register_operand and
19504 <store_mask_constraint> instead of v for the input operand. Make
19505 sure both operands aren't MEMs for if not <mask_applied>.
19507 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19509 * lto-wrapper.c (copy_file) Close both file descriptors before
19512 2017-06-30 Martin Liska <mliska@suse.cz>
19515 * multiple_target.c (create_dispatcher_calls): Make ifunc
19516 also for function that don't have calls or are not referenced.
19518 2017-06-30 Richard Biener <rguenther@suse.de>
19520 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19521 analyze the first scalar stmt. Move vector type computation
19522 for the BB case here from ...
19523 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19524 live operation processing in the SLP case properly.
19526 2017-06-30 Richard Biener <rguenther@suse.de>
19528 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19530 2017-06-30 Martin Liska <mliska@suse.cz>
19533 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19534 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19536 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19538 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19541 2017-06-30 Richard Biener <rguenther@suse.de>
19543 PR tree-optimization/81249
19544 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19545 condition reduction result to original scalar type.
19547 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19549 * profile-count.h (enum profile_quality): Fix typos and whitespace
19552 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19554 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19555 type for branch probabilities.
19557 2017-06-29 Julian Brown <julian@codesourcery.com>
19558 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19560 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19561 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19562 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19563 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19565 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19567 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19568 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19569 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19570 CC usage from generic code to here.
19571 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19572 CC usage into the target macros.
19574 2017-06-29 Maya Rashish <coypu@sdf.org>
19576 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19579 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19581 * arm/arm-builtins.c: Include profile-count.h
19582 * except.c (sjlj_emit_function_enter): Use
19583 profile_probability::unlikely.
19585 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19587 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19588 and tocrel_offset be pointer args rather than implicitly using
19590 (legitimate_constant_pool_address_p, rs6000_emit_move,
19591 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19592 tocrel_offset and use in toc_relative_expr_p call.
19593 (print_operand, print_operand_address): Use static tocrel_base_oac
19594 and tocrel_offset_oac.
19595 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19598 2017-06-29 Maya Rashish <coypu@sdf.org>
19600 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19602 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19604 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19605 objects, take into account only the alignment of 'op0' and 'mode1' if
19608 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19610 * ccmp.c (ccmp_tree_comparison_p): New function.
19611 (ccmp_candidate_p): Update to use above function.
19612 (get_compare_parts): New function.
19613 (expand_ccmp_next): Update to use new functions.
19614 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19616 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19617 take mode as argument.
19618 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19619 * expr.c (expand_expr_real_1): Pass mode as argument.
19621 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19623 * combine.c (combine_instructions): Print insns to dump_file, together
19626 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19628 * asan.c (asan_emit_stack_protection): Update.
19629 (create_cond_insert_point): Update.
19630 * auto-profile.c (afdo_propagate_circuit): Update.
19631 * basic-block.h (struct edge_def): Turn probability to
19632 profile_probability.
19633 (EDGE_FREQUENCY): Update.
19634 * bb-reorder.c (find_traces_1_round): Update.
19635 (better_edge_p): Update.
19636 (sanitize_hot_paths): Update.
19637 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19638 (make_single_succ_edge): Update.
19639 (check_bb_profile): Update.
19640 (dump_edge_info): Update.
19641 (update_bb_profile_for_threading): Update.
19642 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19643 probabilitycount to 0.
19644 * cfgbuild.c (compute_outgoing_frequencies): Update.
19645 * cfgcleanup.c (try_forward_edges): Update.
19646 (outgoing_edges_match): Update.
19647 (try_crossjump_to_edge): Update.
19648 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19649 (expand_gimple_tailcall): Update.
19650 (construct_init_block): Use make_single_succ_edge.
19651 (construct_exit_block): Use make_single_succ_edge.
19652 * cfghooks.c (verify_flow_info): Update.
19653 (redirect_edge_succ_nodup): Update.
19654 (split_edge): Update.
19655 (account_profile_record): Update.
19656 * cfgloopanal.c (single_likely_exit): Update.
19657 * cfgloopmanip.c (scale_loop_profile): Update.
19658 (set_zero_probability): Remove.
19659 (duplicate_loop_to_header_edge): Update.
19660 * cfgloopmanip.h (loop_version): Update prototype.
19661 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19662 (force_nonfallthru_and_redirect): Update.
19663 (update_br_prob_note): Update.
19664 (rtl_verify_edges): Update.
19665 (purge_dead_edges): Update.
19666 (rtl_lv_add_condition_to_bb): Update.
19667 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19668 * cgraphunit.c (init_lowered_empty_function): Update.
19669 (cgraph_node::expand_thunk): Update.
19670 * cilk-common.c: Include profile-count.h
19671 * dojump.c (inv): Remove.
19672 (jumpifnot): Update.
19673 (jumpifnot_1): Update.
19674 (do_jump_1): Update.
19676 (do_jump_by_parts_greater_rtx): Update.
19677 (do_compare_rtx_and_jump): Update.
19678 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19679 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19680 * dwarf2cfi.c: Include profile-count.h
19681 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19682 (sjlj_emit_dispatch_table): Likewise.
19683 * explow.c: Include profile-count.h
19684 * expmed.c (emit_store_flag_force): Update.
19685 (do_cmp_and_jump): Update.
19686 * expr.c (compare_by_pieces_d::generate): Update.
19687 (compare_by_pieces_d::finish_mode): Update.
19688 (emit_block_move_via_loop): Update.
19689 (store_expr_with_bounds): Update.
19690 (store_constructor): Update.
19691 (expand_expr_real_2): Update.
19692 (expand_expr_real_1): Update.
19693 * expr.h (try_casesi, try_tablejump): Update prototypes.
19694 * gimple-pretty-print.c (dump_probability): Update.
19695 (dump_profile): New.
19696 (dump_gimple_label): Update.
19697 (dump_gimple_bb_header): Update.
19698 * graph.c (draw_cfg_node_succ_edges): Update.
19699 * hsa-gen.c (convert_switch_statements): Update.
19700 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19701 (find_if_case_1): Update.
19702 (find_if_case_2): Update.
19703 * internal-fn.c (expand_arith_overflow_result_store): Update.
19704 (expand_addsub_overflow): Update.
19705 (expand_neg_overflow): Update.
19706 (expand_mul_overflow): Update.
19707 (expand_vector_ubsan_overflow): Update.
19708 * ipa-cp.c (good_cloning_opportunity_p): Update.
19709 * ipa-split.c (split_function): Use make_single_succ_edge.
19710 * ipa-utils.c (ipa_merge_profiles): Update.
19711 * loop-doloop.c (add_test): Update.
19712 (doloop_modify): Update.
19713 * loop-unroll.c (compare_and_jump_seq): Update.
19714 (unroll_loop_runtime_iterations): Update.
19715 * lra-constraints.c (lra_inheritance): Update.
19716 * lto-streamer-in.c (input_cfg): Update.
19717 * lto-streamer-out.c (output_cfg): Update.
19718 * mcf.c (adjust_cfg_counts): Update.
19719 * modulo-sched.c (sms_schedule): Update.
19720 * omp-expand.c (expand_omp_for_init_counts): Update.
19721 (extract_omp_for_update_vars): Update.
19722 (expand_omp_ordered_sink): Update.
19723 (expand_omp_for_ordered_loops): Update.
19724 (expand_omp_for_generic): Update.
19725 (expand_omp_for_static_nochunk): Update.
19726 (expand_omp_for_static_chunk): Update.
19727 (expand_cilk_for): Update.
19728 (expand_omp_simd): Update.
19729 (expand_omp_taskloop_for_outer): Update.
19730 (expand_omp_taskloop_for_inner): Update.
19731 * omp-simd-clone.c (simd_clone_adjust): Update.
19732 * optabs.c (expand_doubleword_shift): Update.
19733 (expand_abs): Update.
19734 (emit_cmp_and_jump_insn_1): Update.
19735 (expand_compare_and_swap_loop): Update.
19736 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19737 * predict.c (predictable_edge_p): Update.
19738 (edge_probability_reliable_p): Update.
19739 (set_even_probabilities): Update.
19740 (combine_predictions_for_insn): Update.
19741 (combine_predictions_for_bb): Update.
19742 (propagate_freq): Update.
19743 (estimate_bb_frequencies): Update.
19744 (force_edge_cold): Update.
19745 * profile-count.c (profile_count::dump): Add missing space into dump.
19746 (profile_count::debug): Add newline.
19747 (profile_count::differs_from_p): Explicitly convert to unsigned.
19748 (profile_count::stream_in): Update.
19749 (profile_probability::dump): New member function.
19750 (profile_probability::debug): New member function.
19751 (profile_probability::differs_from_p): New member function.
19752 (profile_probability::differs_lot_from_p): New member function.
19753 (profile_probability::stream_in): New member function.
19754 (profile_probability::stream_out): New member function.
19755 * profile-count.h (profile_count_quality): Rename to ...
19756 (profile_quality): ... this one.
19757 (profile_probability): New.
19758 (profile_count): Update.
19759 * profile.c (compute_branch_probabilities): Update.
19760 * recog.c (peep2_attempt): Update.
19761 * sched-ebb.c (schedule_ebbs): Update.
19762 * sched-rgn.c (find_single_block_region): Update.
19763 (compute_dom_prob_ps): Update.
19764 (schedule_region): Update.
19765 * sel-sched-ir.c (compute_succs_info): Update.
19766 * stmt.c (struct case_node): Update.
19767 (do_jump_if_equal): Update.
19768 (get_outgoing_edge_probs): Update.
19769 (conditional_probability): Update.
19770 (emit_case_dispatch_table): Update.
19771 (expand_case): Update.
19772 (expand_sjlj_dispatch_table): Update.
19773 (emit_case_nodes): Update.
19774 * targhooks.c: Update.
19775 * tracer.c (better_p): Update.
19776 (find_best_successor): Update.
19777 * trans-mem.c (expand_transaction): Update.
19778 * tree-call-cdce.c: Update.
19779 * tree-cfg.c (gimple_split_edge): Upate.
19780 (move_sese_region_to_fn): Upate.
19781 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19782 * tree-eh.c (lower_resx): Upate.
19783 (cleanup_empty_eh_move_lp): Upate.
19784 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19785 * tree-inline.c (copy_edges_for_bb): Update.
19786 (copy_cfg_body): Update.
19787 * tree-parloops.c (gen_parallel_loop): Update.
19788 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19789 (gimple_gen_time_profiler): Update.
19790 * tree-ssa-dce.c (remove_dead_stmt): Update.
19791 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19792 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19793 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19794 (unloop_loops): Update.
19795 (try_peel_loop): Update.
19796 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19797 * tree-ssa-loop-split.c (connect_loops): Update.
19798 (split_loop): Update.
19799 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19800 (hoist_guard): Update.
19801 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19802 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19803 (value_replacement): Update.
19804 * tree-ssa-reassoc.c (branch_fixup): Update.
19805 * tree-ssa-tail-merge.c (replace_block_by): Update.
19806 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19807 (create_edge_and_update_destination_phis): Update.
19808 (compute_path_counts): Update.
19809 (recompute_probabilities): Update.
19810 (update_joiner_offpath_counts): Update.
19811 (freqs_to_counts_path): Update.
19812 (duplicate_thread_path): Update.
19813 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19814 (struct switch_conv_info): Update.
19815 (gen_inbound_check): Update.
19816 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19817 (vect_do_peeling): Update.
19818 (vect_loop_versioning): Update.
19819 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19820 (optimize_mask_stores): Update.
19821 * ubsan.c (ubsan_expand_null_ifn): Update.
19822 * value-prof.c (gimple_divmod_fixed_value): Update.
19823 (gimple_divmod_fixed_value_transform): Update.
19824 (gimple_mod_pow2): Update.
19825 (gimple_mod_pow2_value_transform): Update.
19826 (gimple_mod_subtract): Update.
19827 (gimple_mod_subtract_transform): Update.
19828 (gimple_ic): Update.
19829 (gimple_stringop_fixed_value): Update.
19830 (gimple_stringops_transform): Update.
19831 * value-prof.h: Update.
19833 2017-06-29 Carl Love <cel@us.ibm.com>
19835 * config/rs6000/rs6000-c.c: Add support for built-in functions
19836 vector signed int vec_signed (vector float);
19837 vector signed long long vec_signed (vector double);
19838 vector signed int vec_signed2 (vector double, vector double);
19839 vector signed int vec_signede (vector double);
19840 vector signed int vec_signedo (vector double);
19841 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19842 instruction generator.
19843 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19844 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19845 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19847 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19848 vunsignede_v2df): Add define_expands.
19849 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19850 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19851 VEC_UNSIGNEDO): Add definitions.
19852 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19853 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19854 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19855 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19856 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19857 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19858 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19859 vec_unsignede, vec_unsignedo): Add builtin defines.
19860 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19862 * doc/extend.texi: Update the built-in documentation file for the
19863 new built-in functions.
19865 2017-06-29 Richard Biener <rguenther@suse.de>
19867 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19868 reduction chains to LOOP_VINFO_REDUCTIONS.
19869 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19870 SLP reductions after processing reduction chains.
19872 2017-06-29 Nathan Sidwell <nathan@acm.org>
19874 * builtins.c (fold_builtin_FUNCTION): Use
19875 lang_hooks.decl_printable_name.
19877 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19879 PR middle-end/81194
19880 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19881 with only one label.
19882 * stmt.c (expand_case): Assert NCASES is greater than one.
19884 2017-06-29 Richard Biener <rguenther@suse.de>
19886 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19888 (group_case_labels): Likewise.
19889 (find_taken_edge): Push sanity checking on val to workers...
19890 (find_taken_edge_cond_expr): ... here
19891 (find_taken_edge_switch_expr): ... and here, handle cases
19892 with just a default label.
19893 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19894 (group_case_labels): Likewise.
19895 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19896 group_case_labels does anything cleanup the CFG again.
19898 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19900 PR tree-optimization/81196
19901 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19902 exit condition comparing two IVs.
19904 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19906 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19907 profile to the dummy entry at the end of the list of architectures.
19908 * config/arm/arm-cpu-cdata.h: Regenerated.
19910 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19911 Michael Collison <michael.collison@arm.com>
19914 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19916 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19917 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19918 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19919 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19920 of shift when the shift amount is masked with constant equal to
19921 the size of the mode.
19922 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19925 2017-06-29 Martin Liska <mliska@suse.cz>
19927 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19929 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19932 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19935 (edge_growth_cache_entry::edge_growth_cache_entry): New
19937 (reset_edge_growth_cache): Update to use constructor.
19939 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19941 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19942 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19943 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19945 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19947 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19948 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19950 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19952 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19953 (*-linux-uclibc*): Add t-uclibc tmake_file.
19954 * config/t-musl: New.
19955 * config/t-uclibc: New.
19957 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19959 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19961 (gen_comm_data): Emit architectural setting of arch_prof.
19962 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19964 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19965 (armv8-m.base, armv8-m.main): Likewise.
19966 * arm-protos.h (arm_build_target): Add profile field.
19967 (arch_option): Likewise.
19968 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19970 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19971 arm_active_target.profile.
19973 2017-06-28 Richard Biener <rguenther@suse.de>
19975 PR middle-end/81227
19976 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19977 TYPE_OVERFLOW_WRAPS.
19978 * match.pd (negate_expr_p): Likewise.
19979 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19980 fold_build2, not fold_binary.
19982 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19984 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19985 Convert memory address to Pmode.
19986 (aarch64_print_operand): Assert MEM operands are always Pmode.
19988 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19991 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19992 Remove redundant if.
19993 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19994 * config/arm/aarch-common-protos.h
19995 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19996 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19998 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20001 * multiple_target.c (create_dispatcher_calls): Set the default
20002 clone to be static, not public.
20004 2017-06-28 Richard Biener <rguenther@suse.de>
20006 * tree-vect-loop.c (vectorizable_reduction): Move special
20007 cond reduction IV var creation ...
20008 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20009 parameter. Use STMT_VINFO_VECTYPE.
20010 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20013 2017-06-28 Martin Liska <mliska@suse.cz>
20016 * ipa-visibility.c (non_local_p): Handle visibility.
20018 2017-06-28 Martin Liska <mliska@suse.cz>
20021 * common.opt: Add IntegerRange to various options.
20022 * opt-functions.awk (integer_range_info): New function.
20023 * optc-gen.awk: Add integer_range_info to cl_options struct.
20024 * opts-common.c (decode_cmdline_option): Handle
20025 CL_ERR_INT_RANGE_ARG.
20026 (cmdline_handle_error): Likewise.
20027 * opts.c (print_filtered_help): Show valid interval in
20028 when --help is provided.
20029 * opts.h (struct cl_option): Add range_min and range_max fields.
20030 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20032 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20034 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20035 (x * C EQ/NE y * C): New transformation.
20037 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20039 * genmultilib (combination_space): Accept '+' in option names.
20041 2017-06-28 Martin Liska <mliska@suse.cz>
20044 * asan.c (instrument_derefs): Bail out inner references
20045 that are hard register variables.
20047 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20050 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20051 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20053 2017-06-28 Richard Biener <rguenther@suse.de>
20055 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20056 (vect_get_slp_defs): Adjust.
20057 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20059 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20061 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20062 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20063 (vectorizable_reduction): Adjust.
20064 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20066 (vect_get_slp_defs): Likewise.
20067 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20068 (vectorizable_bswap): Adjust.
20069 (vectorizable_call): Likewise.
20070 (vectorizable_conversion): Likewise.
20071 (vectorizable_assignment): Likewise.
20072 (vectorizable_shift): Likewise.
20073 (vectorizable_operation): Likewise.
20074 (vectorizable_store): Likewise.
20075 (vectorizable_condition): Likewise.
20076 (vectorizable_comparison): Likewise.
20078 2017-06-28 Michael Collison <michael.collison@arm.com>
20081 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20083 (arm_gen_movmemqi): Removed unused variable 'i'.
20084 Convert 'for' loop into 'while' loop.
20085 (arm_expand_prologue): Remove last unnecessary set of insn.
20086 (thumb_pop): Remove unused variable 'pushed_words'.
20087 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20089 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20091 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20092 * config/s390/s390-protos.h: Add prototype of
20093 s390_rel_address_ok_p.
20094 * config/s390/s390.c (s390_got_symbol): New function.
20095 (s390_rel_address_ok_p): New function.
20096 (legitimize_pic_address): Use s390_rel_address_ok_p.
20097 (s390_load_got): Use s390_got_symbol.
20098 (s390_option_override): Issue error if
20099 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20100 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20102 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20104 2017-06-27 Andrew Pinski <apinski@cavium.com>
20106 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20107 (X * copysign (1.0, X)): New pattern.
20108 (X * copysign (1.0, -X)): New pattern.
20109 (copysign (-1.0, CST)): New pattern.
20111 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20113 * genmultilib (combination_space): Remove variable.
20114 Validate reuse rules against regular expression for any sequence
20115 of multilib options in any order.
20117 2017-06-27 Michael Collison <michael.collison@arm.com>
20119 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20120 call aarch64_split_simd_combine.
20121 * (aarch64_combine_internal<mode>): Delete pattern.
20122 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20123 Allow register and subreg operands.
20125 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20127 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20128 specific need, just fallback on defaults.
20129 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20131 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20132 Olivier Hainque <hainque@adacore.com>
20134 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20136 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20137 targets. Pick a default if no particular attempt applied.
20138 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20141 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20143 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20144 (x86_64-wrs-vxworks7): Likewise.
20146 2017-06-27 Marek Polacek <polacek@redhat.com>
20149 * ubsan.c (instrument_null): Check get_base_address's result for null.
20151 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20153 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20155 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20157 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20158 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20159 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20160 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20161 New function types.
20162 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20163 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20164 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20165 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20166 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20167 BUILT_IN_FEUPDATEENV): New builtins.
20168 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20169 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20170 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20171 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20173 (builtin_structptr_types): Adjust size.
20174 * tree.c (builtin_structptr_types): Add four entries.
20176 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20177 Olivier Hainque <hainque@adacore.com>
20179 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20180 (TLS_SYM): New local macro, forcing reference to __tls__ on
20181 link command lines for VxWorks 7 RTPs, triggering initialization
20183 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20184 OS features TLS support, true for RTPs on VxWorks 7.
20185 * config/vxworks.c (vxworks_override_options): Setup emutls
20188 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20190 * predict.c (test_prediction_value_range): Use -1U instead of -1
20191 to avoid narrowing conversion warning.
20192 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20193 to avoid narrowing conversion warning.
20194 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20196 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20198 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20200 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20201 64bit configurations.
20202 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20203 (SIZE_TYPE): Likewise.
20204 * config/vxworks.c (vxworks_emutls_var_fields): Use
20205 long_unsigned_type_node instead of unsigned_type_node as the offset
20206 field type, which is "pointer" mode in emutls.c.
20208 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20211 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20213 PR middle-end/81207
20214 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20215 gimple_vuse copying separately from gimple_vdef copying.
20217 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20219 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20222 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20223 Olivier Hainque <hainque@adacore.com>
20225 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20226 for all vxworks7 targets.
20227 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20228 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20229 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20230 variations for VX6/VX7 and 32/64bits later on in ...
20231 (VXWORKS_LIB_SPEC): Leverage new macros.
20232 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20233 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20235 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20237 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20238 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20240 2017-06-26 Carl Love <cel@us.ibm.com>
20242 * config/rs6000/rs6000-c.c: Add support for built-in functions
20243 vector bool char vec_reve (vector bool char);
20244 vector signed char vec_reve (vector signed char);
20245 vector unsigned char vec_reve (vector unsigned char);
20246 vector bool int vec_reve (vector bool int);
20247 vector signed int vec_reve (vector signed int);
20248 vector unsigned int vec_reve (vector unsigned int);
20249 vector bool long long vec_reve (vector bool long long);
20250 vector signed long long vec_reve (vector signed long long);
20251 vector unsigned long long vec_reve (vector unsigned long long);
20252 vector bool short vec_reve (vector bool short);
20253 vector signed short vec_reve (vector signed short);
20254 vector double vec_reve (vector double);
20255 vector float vec_reve (vector float);
20256 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20257 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20258 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20259 (altivec_vreve): New pattern.
20260 * config/rs6000/altivec.h (vec_reve): New define.
20261 * doc/extend.texi (vec_rev): Update the built-in documentation file
20262 for the new built-in functions.
20264 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20266 PR tree-optimization/71815
20267 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20269 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20271 (slsr_process_phi): Likewise.
20272 (replace_uncond_cands_and_profitable_phis): Don't replace a
20273 multiply candidate with a stride of 1 (copy or cast).
20274 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20276 (lowest_cost_path): Likewise.
20277 (total_savings): Likewise.
20279 2017-06-26 Richard Biener <rguenther@suse.de>
20282 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20283 Use def_builtin_pure for all gather builtins.
20285 2017-06-26 Richard Biener <rguenther@suse.de>
20287 PR tree-optimization/81203
20288 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20289 non-dominating BBs.
20291 2017-06-26 Marek Polacek <polacek@redhat.com>
20294 * doc/invoke.texi: Document -Wmultistatement-macros.
20296 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20298 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20299 arm_neon_ok_no_float_abi effective target.
20301 2017-06-26 Richard Biener <rguenther@suse.de>
20303 PR tree-optimization/80928
20304 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20305 (copy_bbs): Set BB_DUPLICATED flag early.
20306 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20308 (execute_on_shrinking_pred): Likewise.
20309 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20310 BB_DUPLICATED blocks.
20311 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20312 iterate over all PHIs considering removal of *gsi.
20314 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20316 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20319 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20321 * config/rs6000/rs6000-string.c: (expand_block_clear,
20322 do_load_for_compare, select_block_compare_mode,
20323 compute_current_alignment, expand_block_compare,
20324 expand_strncmp_align_check, expand_strn_compare,
20325 expand_block_move, rs6000_output_load_multiple)
20326 Move functions related to string/block move/compare
20327 to a separate file.
20328 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20329 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20330 for this function which is now used in two files.
20331 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20332 * config.gcc: Add rs6000-string.o to extra_objs for
20333 targets powerpc*-*-* and rs6000*-*-*.
20335 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20338 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20339 32-bit, since indexed is not valid for DImode.
20340 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20341 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20342 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20343 (define_peephole2 for Altivec d-form store): Likewise.
20346 * multiple_target.c (create_dispatcher_calls): Only create the
20347 dispatcher call if the function is the default clone of a
20348 versioned function.
20350 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20352 PR middle-end/80902
20353 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20354 a call, force the call to not be a tail call.
20356 2017-06-23 Jeff Law <law@redhat.com>
20358 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20361 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20363 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20364 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20365 handling for early expansion of vector shifts (sl,sr,sra,rl).
20366 (builtin_function_type): Add vector shift right instructions
20367 to the unsigned argument list.
20369 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20371 rtl-optimizatoin/79286
20372 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20373 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20374 trap. PIC register plus a const unspec without offset can never trap.
20376 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20378 * tree.h (builtin_structptr_type): New type.
20379 (builtin_structptr_types): Declare new array.
20380 * tree.c (builtin_structptr_types): New array.
20381 (free_lang_data, build_common_tree_nodes): Use it.
20383 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20386 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20389 2017-06-22 Matt Turner <mattst88@gmail.com>
20391 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20392 Lake models to skylake case. Assume skylake for unknown
20393 models with clflushopt.
20395 2017-06-22 Jeff Law <law@redhat.com>
20397 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20398 frame sizes that do not satisfy aarch64_uimm12_shift.
20400 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20402 * profile-count.h (apply_probability,
20403 apply_scale, probability_in): Fix checks for zero.
20405 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20407 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20408 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20410 2016-06-22 Richard Biener <rguenther@suse.de>
20412 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20413 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20414 REDUC_MAX_EXPR support.
20415 (vectorizable_reduction): Likewise.
20416 (vect_create_epilog_for_reduction): Likewise.
20418 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20420 * match.pd (A / (1 << B) -> A >> B): New.
20421 * generic-match-head.c: Include optabs-tree.h.
20422 * gimple-match-head.c: Likewise.
20423 * optabs-tree.h (target_supports_op_p): New.
20424 * optabs-tree.c (target_supports_op_p): New.
20426 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20428 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20429 $gcc_cv_ld --help output.
20430 (gcc_cv_ld_demangle): Likewise.
20431 (gcc_cv_ld_eh_frame_hdr): Likewise.
20432 (gcc_cv_ld_pie): Likewise.
20433 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20434 (gcc_cv_ld_buildid): Likewise.
20435 (gcc_cv_ld_sysroot): Likewise.
20436 (ld_bndplt_support): Likewise.
20437 (ld_pushpopstate_support): Likewise.
20438 * configure: Regenerate.
20439 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20441 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20444 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20445 operands indexes to avoid gap between operands and match_dups.
20447 2017-06-21 Andrew Pinski <apinski@cavium.com>
20449 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20450 Increment Arith_shift and Arith_shift_reg by 1.
20451 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20453 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20454 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20455 (aarch64_strip_extend): Add new argument and test for it.
20456 (aarch64_cheap_mult_shift_p): New function.
20457 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20458 add a cost if it is true.
20459 Update calls to aarch64_strip_extend.
20460 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20462 2017-06-21 Andrew Pinski <apinski@cavium.com>
20464 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20466 (thunderxt88): Likewise.
20467 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20468 (thunderx_prefetch_tune): New variable.
20469 (thunderx2t99_prefetch_tune): Update for the correct values.
20470 (thunderxt88_tunings): New variable.
20471 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20472 generic_prefetch_tune.
20473 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20475 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20477 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20478 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20479 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20480 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20481 (aarch64_atomic_cas<mode>, GPI): Likewise.
20483 2017-06-21 Martin Liska <mliska@suse.cz>
20485 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20486 statements on cold and hot labels.
20487 * predict.c (tree_estimate_probability_bb): Remove the
20488 prediction from this place.
20490 2017-06-21 Martin Liska <mliska@suse.cz>
20492 PR tree-optimization/79489
20493 * gimplify.c (maybe_add_early_return_predict_stmt): New
20495 (gimplify_return_expr): Call the function.
20496 * predict.c (tree_estimate_probability_bb): Remove handling
20498 * predict.def: Update comment about early return predictor.
20499 * gimple-predict.h (is_gimple_predict): New function.
20500 * predict.def: Change default value of early return to 66.
20501 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20503 * passes.def: Put pass_strip_predict_hints to the beginning of
20506 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20508 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20509 FUNCTION_DECL declarations.
20510 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20512 (dwaf2out_decl): Likewise.
20513 * godump.c (go_early_global_decl): Skip call to the real debug hook
20514 for FUNCTION_DECL declarations.
20515 * passes.c (rest_of_decl_compilation): Skip call to the
20516 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20517 -fdump-go-spec is passed.
20519 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20521 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20522 Reorder for compactness.
20523 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20524 (def_builtin_pure, def_builtin_pure2): New functions.
20525 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20527 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20529 * match.pd (nop_convert): New predicate.
20530 ((A +- CST1) +- CST2): Allow some NOP conversions.
20532 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20535 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20536 with ctors/dtors if GOVD_SHARED is set.
20538 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20540 * config/aarch64/aarch64.md (movti_aarch64):
20541 Emit mov rather than orr.
20542 (movtf_aarch64): Likewise.
20543 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20544 Emit mov rather than orr.
20546 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20548 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20549 Swap alternatives, make integer dup more expensive.
20551 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20553 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20554 Return true for non-tls symbols.
20556 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20558 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20559 (cortex-a75): Likewise.
20560 (cortex-a75.cortex-a55): Likewise.
20561 * config/aarch64/aarch64-tune.md: Regenerate.
20562 * doc/invoke.texi (-mtune): Document new values for -mtune.
20564 2017-06-21 Tom de Vries <tom@codesourcery.com>
20566 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20567 stack_size feature.
20568 (Effective-Target Keywords, Other attributes): Suggest using
20569 dg-add-options stack_size feature to get stack limit in stack_size
20570 effective target documentation.
20572 2017-06-21 Julian Brown <julian@codesourcery.com>
20573 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20575 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20576 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20577 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20579 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20580 attribute type list for neon_multiply.
20581 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20582 attribute type list for neon_multiply.
20583 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20584 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20585 attribute type list for neon_multiply.
20586 * config/arm/types.md (crypto_pmull): Add.
20587 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20588 attribute type list.
20590 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20592 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20595 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20597 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20598 to make sure not to dereference a NULL cost_classes_ptr pointer.
20600 2017-06-20 Carl Love <cel@us.ibm.com>
20602 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20603 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20604 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20605 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20606 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20607 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20608 VMULOSW): New enum "unspec" values.
20609 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20610 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20611 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20612 altivec_vmulosw): New patterns.
20613 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20614 VMULOSW): Add definitions.
20616 2017-06-20 Julia Koval <julia.koval@intel.com>
20618 * config/i386/i386.c: Fix rounding expand for new pattern.
20619 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20621 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20623 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20624 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20626 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20628 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20631 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20633 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20634 architecture, then by implementer ID.
20635 * config/aarch64/aarch64-tune.md: Regenerate.
20637 2017-06-20 Richard Biener <rguenther@suse.de>
20639 PR middle-end/81097
20640 * fold-const.c (split_tree): Fold to type before negating.
20642 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20644 * diagnostic-show-locus.c
20645 (selftest::test_fixit_deletion_affecting_newline): New function.
20646 (selftest::diagnostic_show_locus_c_tests): Call it.
20648 2017-06-20 Andreas Schwab <schwab@suse.de>
20651 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20654 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20656 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20657 __ARM_FEATURE_COPROC according to support.
20659 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20661 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20662 Rewritten to avoid overflow for > 32-bit pointers.
20665 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20666 by removing enum keyword.
20667 (ubsan_type_descriptor): Likewise. Formatting fix.
20670 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20671 splitter): Require TARGET_SSE2 in the condition.
20673 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20676 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20677 for doing vector set of SFmode on ISA 3.0.
20678 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20679 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20681 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20682 SFmode value into a V4SF variable that was extracted from another
20683 V4SF variable without converting the element to double precision
20684 and back to single precision vector format.
20685 (vsx_insert_extract_v4sf_p9_2): Likewise.
20687 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20689 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20690 in UWHI to avoid undefined overflow.
20693 * ubsan.h (enum ubsan_encode_value_phase): New.
20694 (ubsan_encode_value): Change second argument to
20695 enum ubsan_encode_value_phase with default value of
20696 UBSAN_ENCODE_VALUE_GENERIC.
20697 * ubsan.c (ubsan_encode_value): Change second argument to
20698 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20699 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20700 create_tmp_var_raw instead of create_tmp_var and use a
20702 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20703 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20704 ubsan_encode_value callers.
20707 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20708 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20709 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20711 2017-06-19 Richard Biener <rguenther@suse.de>
20713 PR middle-end/81118
20714 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20715 estimates if we changed anything.
20717 2017-06-19 Richard Biener <rguenther@suse.de>
20719 PR tree-optimization/80887
20720 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20721 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20722 simplified lookups, then reset mprts_hook.
20723 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20725 (try_to_simplify): Likewise.
20727 2017-06-19 Martin Liska <mliska@suse.cz>
20730 * gimplify.c (gimplify_switch_expr):
20731 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20733 2017-06-19 Martin Liska <mliska@suse.cz>
20735 * doc/install.texi: Document that PGO runs in 4 stages.
20737 2017-06-19 Martin Liska <mliska@suse.cz>
20740 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20741 to dispatcher function name.
20742 * multiple_target.c (replace_function_decl): New function.
20743 (create_dispatcher_calls): Redirect both edges and references.
20745 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20747 * profile-count.c (profile_count::dump): Dump quality.
20748 (profile_count::differs_from_p): Update for unsigned val.
20749 * profile-count.h (profile_count_quality): New enum.
20750 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20752 2017-06-19 Richard Biener <rguenther@suse.de>
20754 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20755 struct function as arg.
20756 (estimate_numbers_of_iterations): Export overload with loop arg.
20757 (free_numbers_of_iterations_estimates_loop): Use an overload of
20758 free_numbers_of_iterations_estimates instead.
20759 * tree-cfg.c (remove_bb): Adjust.
20760 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20761 * tree-parloops.c (gen_parallel_loop): Likewise.
20762 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20764 (tree_unroll_loops_completely): Likewise.
20765 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20766 Use an overload instead and export.
20767 (estimated_loop_iterations): Adjust.
20768 (max_loop_iterations): Likewise.
20769 (likely_max_loop_iterations): Likewise.
20770 (estimate_numbers_of_iterations): Take struct function as arg
20772 (loop_exits_before_overflow): Adjust.
20773 (free_numbers_of_iterations_estimates_loop): Use an overload.
20774 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20775 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20777 2017-06-19 Richard Biener <rguenther@suse.de>
20780 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20781 RANGE_EXPR conservatively.
20783 2017-06-16 Carl Love <cel@us.ibm.com>
20785 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20786 definitions for vec_float, vec_float2, vec_floato,
20787 vec_floate built-ins.
20788 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20789 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20791 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20792 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20793 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20794 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20795 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20796 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20797 vec_floato): Add builtin defines.
20798 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20799 Update the built-in documentation file for the new built-in
20802 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20804 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20805 (mthumb): Mark as the negative of -marm.
20807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20809 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20811 (ARM Options, -mtune): Document that this accepts the same
20812 extension options as -mcpu.
20813 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20817 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20818 permitted extensions.
20820 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20822 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20823 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20824 (armv8-m.main): Add option +nodsp.
20825 * config/arm/arm-cpu-cdata.h: Regenerated.
20827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20829 * config/arm/t-fuchsia: New file.
20830 * config.gcc (arm*-*-fuchsia*): Use it.
20832 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20834 * config/arm/t-symbian: Rewrite for new option infrastructure.
20836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20838 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20839 (MULTILIB_REQUIRED): Likewise.
20841 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20843 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20844 (MULTILIB_RESUE): Likewise.
20845 (MULTILIB_MATCHES): Likewise.
20846 (MULTLIB_REQUIRED): Likewise.
20848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20850 * config/arm/t-rtems: Rewrite for new option framework.
20852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20854 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20855 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20856 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20857 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20858 * config/arm/t-multilib: ... here.
20859 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20860 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20861 armv7-a and armv8*-a when A-profile libraries have not been built.
20862 * config/arm/t-rmprofile: Rewrite.
20864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20866 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20867 with a backslash. Remove the backslash after substituting unescaped
20869 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20871 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20873 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20874 the driver to pass through the default setting of -mfloat-abi.
20875 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20877 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20878 (all_feat_combs): New rule.
20879 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20881 * config/arm/t-aprofile: Rewrite.
20883 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20885 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20886 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20887 fpu is not specified by the user/command-line.
20888 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20889 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20890 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20891 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20892 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20893 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20895 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20897 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20898 * config/arm/t-arm-elf: Rewritten.
20900 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20902 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20903 have some floating-point instructions.
20904 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20905 (TARGET_MAYBE_HARD_FLOAT): New macro.
20906 * config/arm/arm-builtins.c (arm_init_builtins): Use
20907 TARGET_MAYBE_HARD_FLOAT.
20908 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20910 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20912 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20913 (configargs.h): Include it.
20914 (arm_print_hint_for_fpu_option): New function.
20915 (arm_parse_fpu_option): New function.
20916 (candidate_extension): New class.
20917 (arm_canon_for_multilib): New function.
20918 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20919 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20920 (ARCH_CANONICAL_SPECS): New macro.
20921 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20923 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20925 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20926 are set after handling multilib fragments. Set target_cpu_default2
20929 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20931 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20933 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20934 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20936 (check_arch): Likewise.
20937 * config/arm/arm.c (arm_configure_build_target): Handle
20938 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20939 options in the default.
20941 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20943 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20944 when an option is an alias of another.
20945 * config/arm/parsecpu.awk (optalias): New parser token.
20946 (gen_comm_data): Mark non-alias options as such. Emit entries
20947 for extension aliases.
20948 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20949 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20950 (armv6kz, armv6zk, armv6t2): Likewise.
20951 (armv7): Make vfpv3-d16 an alias.
20952 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20954 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20955 Sort in canonical order.
20956 (armv8-a): Sort in canonical order.
20957 (armv8.1-a, armv8.2-a): Likewise.
20958 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20960 (cortex-a9): Sort in canonical order.
20961 * config/arm/arm.c (selftests.h): Include it.
20962 (arm_test_cpu_arch_data): New function.
20963 (arm_run_self_tests): New function.
20964 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20965 (targetm): Move declaration to the end of the file.
20966 * arm-cpu-cdata.h: Regenerated.
20968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20970 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20971 call to target_mode_check describing the type of option passed.
20972 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20973 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20974 arm_parse_cpu_option_name to match parameters against list of
20976 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20977 arm_arch_core_flags data structure.
20978 * config/arm/arm-cpu_cdata.h: Regenerated.
20980 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20982 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20984 (arm_print_hint_for_cpu_option): Likewise.
20985 (arm_print_hint_for_arch_option): Likewise.
20986 (arm_parse_cpu_option_name): Likewise.
20987 (arm_parse_arch_option_name): Likewise.
20988 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20989 of entries in the all_fpus list.
20990 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20991 (arm_parse_cpu_option_name): Declare.
20992 (arm_parse_arch_option_name): Declare.
20993 (arm_parse_option_features): Declare.
20994 (arm_intialize_isa): Declare.
20995 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20997 (gen_comm_data): ... here. Make definitions non-static.
20998 * config/arm/arm-cpu-data.h: Regenerated.
20999 * config/arm/arm-cpu-cdata.h: Regenerated.
21001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21003 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21004 (cpu_arch_extension): New structure.
21005 (cpu_arch_option, arch_option, cpu_option): New structures.
21006 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21007 architecture types.
21008 (gen_data): Generate new format data tables.
21009 * config/arm/arm.c (cpu_tune): New structure.
21010 (cpu_option, processors): Delete.
21011 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21012 (arm_print_hint_for_cpu_option): ... this and ...
21013 (arm_print_hint_for_arch_option): ... this.
21014 (arm_parse_arch_cpu_name): Delete. Replace with ...
21015 (arm_parse_cpu_option_name): ... this and ...
21016 (arm_parse_arch_option_name): ... this.
21017 (arm_unrecognized_feature): Change type of target parameter to
21019 (arm_parse_arch_cpu_features): Delete. Replace with ...
21020 (arm_parse_option_features): ... this.
21021 (arm_configure_build_target): Rework to use new configuration data
21023 (arm_print_tune_info): Rework for new configuration data tables.
21024 * config/arm/arm-cpu-data.h: Regenerated.
21025 * config/arm/arm-cpu.h: Regenerated.
21027 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21029 * Makefile.in (OBJS): Move sbitmap.o from here ...
21030 (OBJS-libcommon): ... to here.
21032 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21034 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21035 (ISA_ALL_CRYPTO): New macro.
21036 (ISA_ALL_SIMD): New macro
21037 (ISA_ALL_FP): New macro.
21038 * config/arm/arm.c (fpu_bitlist): Update initializer.
21039 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21041 (arm9e): Add fpu. Add option for nofp
21042 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21043 (arm926ej-s, arm1026ej-s): Likewise.
21044 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21045 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21046 neon-fp16, neon-vfpv4, nofp and nosimd.
21047 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21048 (cortex-a8): Add fpu. Add option for nofp.
21049 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21050 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21051 (cortex-r4f): Add fpu.
21052 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21053 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21055 (cortex-r8): Likewise.
21056 (cortex-m4): Add fpu. Add option for nofp.
21057 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21058 (cortex-a17.cortex-a7): Likewise.
21059 (cortex-a32): Add fpu. Add options for crypto and nofp.
21060 (cortex-a35, cortex-a53): Likewise.
21061 (cortex-a57): Add fpu. Add option for crypto.
21062 (cortex-a72, cortex-a73): Likewise.
21063 (exynos-m1): Likewise.
21064 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21065 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21066 (cortex-m33): Add fpu. Add option for nofp.
21067 * config/arm/arm-cpu-cdata.h: Regenerated
21068 * config/arm/arm-cpu-data.h: Regenerated.
21070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21072 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21073 (armv5te, armv5tej): Likewise.
21074 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21075 (armv7): Add options fp and vfpv3-d16.
21076 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21077 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21079 (armv7ve): Likewise.
21080 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21081 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21082 (armv8-a): Add nocrypto option.
21083 (armv8.1-a, armv8.2-a): Likewise.
21084 (armv8-m.main): add options fp, fp.dp and nofp.
21086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21088 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21090 (armv8-a+crc): Delete.
21091 (armv8.1-a): Add options simd, crypto and nofp.
21092 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21093 (armv8.2-a+fp16): Delete.
21094 (armv8-m.main): Add option dsp.
21095 (armv8-m.main+dsp): Delete.
21096 (cortex-a8): Add fpu. Add nofp option.
21097 (cortex-a9): Add fpu. Add nofp and nosimd options.
21098 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21099 link to main cpu and architecture data structures.
21100 (gen_comm_data): Only put isa attributes from the main architecture
21102 (option): New statement for architecture and CPU entries.
21103 * arm.c (struct cpu_option): New structure.
21104 (struct processors): Add entry for options.
21105 (arm_unrecognized_feature): New function.
21106 (arm_parse_arch_cpu_name): Ignore any characters after the first
21108 (arm_parse_arch_cpu_feature): New function.
21109 (arm_configure_build_target): Separate out any CPU and architecture
21110 features and parse separately. Don't error out if -mfpu=auto is
21111 used with only an architecture string.
21112 (arm_print_asm_arch_directives): New function.
21113 (arm_file_start): Call it.
21114 * config/arm/arm-cpu-cdata.h: Regenerated.
21115 * config/arm/arm-cpu-data.h: Likewise.
21116 * config/arm/arm-tables.opt: Likewise.
21118 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21120 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21121 assembler when it is not -mfpu=auto.
21123 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21125 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21126 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21127 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21128 (ASM_CPU_SPEC): Rewrite.
21129 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21130 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21131 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21132 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21133 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21134 copied string is NUL-terminated. Also strip any characters prefixed
21136 (arm_rewrite_selected_arch): New function.
21137 (arm_rewrite_march): New function.
21139 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21141 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21142 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21143 (march, mcpu, mtune): Convert to string-based options.
21144 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21145 (arm_parse_arch_cpu_name): New function.
21146 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21147 identify selected architecture or CPU.
21148 (arm_option_save): New function.
21149 (TARGET_OPTION_SAVE): Redefine.
21150 (arm_option_restore): Restore string options.
21151 (arm_option_print): Print string options.
21153 2017-06-16 Martin Sebor <msebor@redhat.com>
21155 PR tree-optimization/80933
21156 PR tree-optimization/80934
21157 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21158 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21159 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21160 (gimple_fold_builtin): Call them.
21162 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21164 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21165 as unlikely; update profile.
21167 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21169 * predict.c (force_edge_cold): Handle declaring edges impossible
21172 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21174 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21176 (try_unroll_loop_completely): Fix reporting.
21178 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21180 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21182 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21185 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21186 if given a non-constant argument for an intrinsic which requires a
21189 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21191 * profile.c (compare_freqs): New function.
21192 (branch_prob): Sort edge list.
21193 (find_spanning_tree): Assume that the list is priority sorted.
21195 2017-06-16 Richard Biener <rguenther@suse.de>
21197 PR tree-optimization/81090
21198 * passes.def (pass_record_bounds): Remove.
21199 * tree-pass.h (make_pass_record_bounds): Likewise.
21200 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21201 make_pass_record_bounds): Likewise.
21202 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21203 not free niter estimates at the beginning but at the end.
21204 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21206 2017-06-16 Richard Biener <rguenther@suse.de>
21208 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21209 initializer to workaround ICE in host GCC 4.8.
21211 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21213 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21215 (clone_inlined_nodes): Update.
21217 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21219 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21220 prefetch settings, and enable prefetching by default at -O3.
21222 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21224 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21225 Set flag_prefetch_loop_arrays according to tuning data.
21227 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21229 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21230 New tune structure.
21231 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21232 [Unrelated to main purpose of the patch] Place the pointer field last
21233 to enable type checking errors when tune structure are wrongly merged.
21234 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21235 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21236 (thunderx2t99_prefetch_tune): New tune constants.
21237 (tune_params *_tunings): Update all tunings (no functional change).
21238 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21239 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21240 from tunings structures.
21242 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21245 * ubsan.c (instrument_null): Add T argument, use it instead
21246 of computing it based on IS_LHS.
21247 (instrument_object_size): Likewise.
21248 (pass_ubsan::execute): Adjust instrument_null and
21249 instrument_object_size callers to pass gimple_get_lhs or
21250 gimple_assign_rhs1 result to it. Use instrument_null instead of
21251 calling get_base_address and instrument_mem_ref. Handle
21252 aggregate call arguments for object-size sanitization.
21254 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21256 PR tree-optimization/81089
21257 * tree-vrp.c (is_masked_range_test): Validate operands of
21260 2017-06-15 Martin Sebor <msebor@redhat.com>
21263 * dumpfile.c (dump_register): Avoid calling memset to initialize
21264 a class with a default ctor.
21265 * gcc.c (struct compiler): Remove const qualification.
21266 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21267 * hash-table.h: Ditto.
21268 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21270 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21271 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21273 * params.h (struct param_info): Make struct members non-const.
21274 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21275 with default initialization.
21276 * vec.h (vec_copy_construct, vec_default_construct): New helper
21278 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21279 with vec_copy_construct.
21280 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21281 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21282 * doc/invoke.texi (-Wclass-memaccess): Document.
21284 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21286 * emit-rtl.h (is_leaf): Update comment about local
21287 register allocator.
21289 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21292 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21293 for a variable to have a section before checking if the section has a
21295 Set section to.persistent if persistent attribute is set.
21296 Warn if .persistent attribute is used on an automatic variable.
21298 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21300 PR rtl-optimization/80474
21301 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21303 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21305 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21308 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21310 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21311 (casesi): Emit bounds checking as RTL.
21312 (casesi_internal_mips16_<mode>): Remove bounds checking.
21314 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21316 * config/xtensa/xtensa.c (xtensa_option_override): Append
21317 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21318 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21319 xtensa_doloop_hooks): Define unconditionally.
21320 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21322 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21323 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21324 for it in xtensa_option_override.
21325 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21326 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21328 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21330 * doc/cppopts.texi: Document '-' special value to -MF.
21332 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21334 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21335 (cortex_a53_fconst): Likewise.
21336 (cortex_a53_fpmul): Likewise.
21337 (cortex_a53_f_load_64): Likewise.
21338 (cortex_a53_f_load_many): Likewise.
21339 (cortex_a53_advsimd_alu): Likewise.
21340 (cortex_a53_advsimd_alu_q): Likewise.
21341 (cortex_a53_advsimd_mul): Likewise.
21342 (cortex_a53_advsimd_mul_q): Likewise.
21343 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21344 Add missing fmul, r2f_cvt and fconst cases.
21346 2017-06-14 Richard Biener <rguenther@suse.de>
21348 PR middle-end/81088
21349 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21351 (fold_binary_loc): When associating do not treat pre-existing
21352 TREE_OVERFLOW on literal constants as a reason to allow
21353 TREE_OVERFLOW on associated literal constants.
21355 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21357 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21358 (MASK_FEATURES): New macro.
21359 * config/sparc/sparc.c (sparc_option_override): Remove the special
21360 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21362 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21364 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21365 a division of 0 if non-call exceptions are enabled.
21367 2017-06-14 Andrew Pinski <apinski@cavium.com>
21368 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21371 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21372 Improve vector initialization code gen for only variable case.
21374 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21376 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21378 2017-06-14 Richard Biener <rguenther@suse.de>
21380 PR tree-optimization/81083
21381 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21384 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21386 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21387 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21388 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21389 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21390 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21391 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21393 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21395 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21396 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21398 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21400 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21402 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21404 * config/rs6000/t-rtems: Don't handle SPE.
21406 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21408 * config/rs6000/t-linux: Don't handle SPE.
21410 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21412 * config/rs6000/eabispe.h: Delete file.
21414 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21416 * config/rs6000/t-spe: Delete file.
21418 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21420 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21421 (rs6000_legitimate_offset_address_p): Return false for anything in
21422 V2SImode or V2SFmode.
21424 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21426 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21427 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21428 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21430 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21431 (rs6000_legitimate_offset_address_p): Ditto.
21432 (rs6000_emit_move): Ditto.
21433 (rs6000_init_builtins): Remove V4HI_type_node.
21435 2017-06-13 Martin Liska <mliska@suse.cz>
21438 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21439 (gate_asan): Likewise.
21440 * asan.h (asan_no_sanitize_address_p): Remove the function.
21441 (sanitize_flags_p): New function.
21442 * builtins.def: Fix coding style.
21443 * common.opt: Use renamed enum value.
21444 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21445 * doc/extend.texi: Document no_sanitize attribute.
21446 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21447 to SANITIZE_UNDEFINED_NONDEFAULT.
21448 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21449 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21450 Use sanitize_flags_p.
21451 * gimplify.c (gimplify_function_tree): Likewise.
21452 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21453 * opts.c (parse_no_sanitize_attribute): New function.
21454 (common_handle_option): Use renamed enum value.
21455 * opts.h (parse_no_sanitize_attribute): Declare.
21456 * tree.c (sanitize_flags_p): New function.
21457 * tree.h: Declared here.
21458 * tsan.c: Use sanitize_flags_p.
21459 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21460 (instrument_mem_ref): Likewise.
21461 (instrument_bool_enum_load): Likewise.
21462 (do_ubsan_in_current_function): Remove the function.
21463 (pass_ubsan::execute): Use sanitize_flags_p.
21464 * ubsan.h: Remove do_ubsan_in_current_function
21465 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21466 (dump_function_to_file): Use it here.
21468 2017-06-13 Martin Jambor <mjambor@suse.cz>
21470 PR tree-optimization/80803
21471 PR tree-optimization/81063
21472 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21473 (propagate_subaccesses_across_link): Enqueue subtree whenever
21474 necessary instead of relying on the caller.
21476 2017-06-13 Martin Jambor <mjambor@suse.cz>
21478 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21479 that have a first_link.
21480 (sort_and_splice_var_accesses): Do not check first_link before
21482 (subtree_mark_written_and_enqueue): Likewise.
21483 (propagate_all_subaccesses): Likewise and do not stop at first
21484 parent with a first_link.
21486 2017-06-13 Martin Jambor <mjambor@suse.cz>
21488 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21491 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21493 * match.pd: New pattern.
21495 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21497 * tree-vrp.c (is_masked_range_test): New function.
21498 (register_edge_assert_for): Determine ranges for
21501 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21503 PR tree-optimization/67328
21504 * fold-const.c (maskable_range_p): New function.
21505 (build_range_check): Generate bittests if possible.
21507 2017-06-13 Martin Liska <mliska@suse.cz>
21509 * gimple-pretty-print.c (dump_probability): Add new argument.
21510 (dump_edge_probability): Dump both probability and count.
21511 (dump_gimple_label): Likewise.
21512 (dump_gimple_bb_header): Likewise.
21514 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21517 * config/avr/avr-devices.c: Fix indentation.
21518 * config/avr/gen-avr-mmcu-specs.c: Dito.
21520 2017-06-13 Richard Biener <rguenther@suse.de>
21522 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21523 instead get vector type from stmt_info.
21524 (vectorizable_reduction): Adjust. Remove dead code.
21526 2017-06-13 Richard Biener <rguenther@suse.de>
21528 PR middle-end/81065
21529 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21530 case of C * (x * C2 + C3).
21531 (fold_addr_of_array_ref_difference): Properly fold index difference.
21533 2017-06-12 David S. Miller <davem@davemloft.net>
21536 * config/sparc/sparc.md (return expander): Emit frame blockage if
21537 function uses alloca.
21539 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21541 * combine.c (make_field_assignment): Check len rather than the mode
21542 precision when calling force_to_mode.
21544 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21546 Support multilibs and devices that see flash in RAM address range.
21549 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21550 (avr_mcu_t) <flash_pm_offset>: New field.
21551 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21552 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21553 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21554 (AVR_TINY_PM_OFFSET): Remove macro.
21555 * config/avr/avr.opt (-mshort-calls): New option.
21556 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21557 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21558 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21559 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21560 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21561 instead of avr_arch->have_jmp_call.
21562 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21563 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21564 avr_arch->flash_pm_offset to define.
21565 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21566 new field flash_pm_offset. Add entry for avrxmega3.
21567 (avr_texinfo): Add entry for avrxmega3.
21568 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21569 attiny212, attiny214,
21570 attiny412, attiny414, attiny416, attiny417,
21571 attiny814, attiny816, attiny817,
21572 attiny1614, attiny1616, attiny1617,
21573 attiny3214, attiny3216, attiny3217.
21574 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21575 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21576 (avr_print_operand_address) [AVR_TINY]: Same.
21577 (avr_asm_init_sections) <readonly_data_section>: Only patch
21578 callback if avr_arch->flash_pm_offset = 0.
21579 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21580 for rodata if avr_arch->flash_pm_offset != 0.
21581 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21582 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21583 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21584 (m_options): Append opt_rcall.
21585 (m_dirnames): Append dir_rcall.
21586 * config/avr/t-multilib: Regenerate.
21588 * configure.ac [target=avr]: Check whether avrxmega3 default
21589 linker description file works as needed.
21590 * configure: Regenerate.
21591 * doc/avr-mmcu.texi: Regenerate.
21592 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21593 <__AVR_ARCH__>: Document avrxmega3 and 103.
21594 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21595 <__AVR_SHORT_CALLS__>: Document it.
21596 <__AVR_PM_BASE_ADDRESS__>: Document it.
21597 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21598 (AVR Variable Attributes) <progmem>: Document this is
21599 not needed for avrxmega3.
21600 (AVR Named Address Spaces) <__flash>: Dito.
21602 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21604 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21606 2017-06-12 Doug Rupp <rupp@adacore.com>
21608 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21609 Append vxworks-stdint.h to the tm_file list.
21610 * config/vxworks-stdint.h: New file.
21612 2017-06-12 Martin Liska <mliska@suse.cz>
21614 PR tree-optimization/81041
21615 * tree-profile.c (gimple_gen_ic_func_profiler):
21616 Create an extra BB in profile-generate
21617 (gimple_gen_time_profiler): Likewise.
21619 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21621 PR tree-optimization/81003
21622 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21623 (update_range_test): Use it instead of force_gimple_operand_gsi.
21625 2017-06-12 Richard Biener <rguenther@suse.de>
21627 PR tree-optimization/81053
21628 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21629 with backedge value not defined in loop. Simplify def stmt
21632 2017-06-11 Tom de Vries <tom@codesourcery.com>
21635 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21637 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21638 nvptx_cannot_force_const_mem.
21640 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21642 * opts.c (finish_options): Move test for flag_split_stack after
21643 it has been initialized.
21645 2017-06-11 Jason Merrill <jason@redhat.com>
21647 * tree.h (id_equal): New.
21648 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21649 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21650 instead of strcmp of IDENTIFIER_POINTER.
21652 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21654 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21655 (mark_all_inlined_calls_cdtor): Fix formating.
21656 (inline_transform): Rescale profile before inlining.
21658 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21660 * cgraph.h (cgraph_edge::clone): Update prototype.
21661 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21662 (cgraph_node::create_clone): Update.
21663 (cgraph_node::create_version_clone): Update.
21664 * tree-inline.c (copy_bb): Update.
21665 (expand_call_inline): Update.
21667 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21669 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21670 factored out from ...
21671 (rs6000_emit_prologue): ... here.
21673 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21675 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21676 factored out from ...
21677 (rs6000_emit_prologue): ... here.
21679 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21681 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21683 (handle_missing_profiles): Fix computation of tp_first_run.
21684 (counts_to_freqs): Do not touch freqs when count is 0.
21686 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21688 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21691 2017-06-10 Tom de Vries <tom@codesourcery.com>
21693 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21694 attributes): Document signal effective target.
21696 2017-06-10 Tom de Vries <tom@codesourcery.com>
21698 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21699 Document effective target stack_size.
21701 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21703 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21704 to the edit_context if they can be auto-applied.
21706 2017-06-9 Ian Lance Taylor <iant@golang.org>
21708 * opts.c (finish_options): If -fsplit-stack, disable implicit
21709 -forder-blocks-and-partition.
21710 * doc/invoke.texi (Optimize Options): Document that when using
21711 -fsplit-stack -forder-blocks-and-partition is not implicitly
21714 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21716 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21717 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21718 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21719 * builtins.def (abort, trap, unreachable): Declare cold.
21720 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21721 * tree-core.h (ECF_COLD): New.
21722 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21723 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21725 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21727 * predict.c (unlikely_executed_stmt_p): Cleanup.
21729 2017-06-09 Richard Biener <rguenther@suse.de>
21731 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21732 model if the ref is always written to.
21734 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21736 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21738 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21740 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21743 2017-06-09 Tom de Vries <tom@codesourcery.com>
21746 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21747 "target cannot support label values" when encountering LABEL_REF.
21749 2017-06-09 Martin Liska <mliska@suse.cz>
21751 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21752 (gimple_gen_ic_func_profiler): Emit direct comparison
21753 of __gcov_indirect_call_callee with NULL.
21754 (gimple_gen_time_profiler): Change probability from
21755 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21757 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21759 * profile.c (edge_gcov_counts): Turn to pointer.
21760 (compute_branch_probabilities, compute_branch_probabilities): Update.
21761 (branch_prob): Do not clear edge_gcov_count.
21762 * profile.h (edge_gcov_counts): Turn to pointer.
21763 (edge_gcov_count): Update.
21765 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21767 * gimple.h (gimple_check_failed): Mark cold.
21769 2017-06-09 Richard Biener <rguenther@suse.de>
21771 PR tree-optimization/66623
21772 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21773 refactor check_reduction into two parts, properly computing
21774 whether we have to check reduction validity for outer loop
21777 2017-06-09 Richard Biener <rguenther@suse.de>
21779 PR tree-optimization/79483
21780 * graphite-scop-detection.c (order): New global.
21781 (get_order): Compute bb to order mapping that satisfies code
21782 generation constraints.
21783 (cmp_pbbs): New helper.
21784 (build_scops): Start domwalk at entry block, sort generated
21787 2017-06-09 Richard Biener <rguenther@suse.de>
21789 PR middle-end/81007
21790 * ipa-polymorphic-call.c
21791 (ipa_polymorphic_call_context::restrict_to_inner_class):
21792 Skip FIELD_DECLs with error_mark_node type.
21793 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21796 2017-06-09 Martin Liska <mliska@suse.cz>
21798 * predict.c (struct branch_predictor): New struct.
21799 (test_prediction_value_range): New test.
21800 (predict_c_tests): New function.
21801 * selftest-run-tests.c (selftest::run_tests): Run the function.
21802 * selftest.h: Declare new tests.
21804 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21807 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21808 gen_add3_insn did not fail.
21809 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21810 r0, construct that number in a temporary reg and add that reg to r0.
21811 If asked to put the result in r0 as well, fail.
21813 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21815 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21816 for early expansion of vec_eqv.
21818 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21820 PR middle-end/81005
21821 * ubsan.c (instrument_null): Avoid pointless code temporary.
21822 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21824 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21828 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21831 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21834 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21836 * predict.c (unlikely_executed_edge_p): Move ahead.
21837 (probably_never_executed_edge_p): Use it.
21839 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21841 PR middle-end/79988
21842 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21843 gimple_call_builtin_p call.
21845 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21847 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21848 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21849 rtl_check_failed_type2, rtl_check_failed_code1,
21850 rtl_check_failed_code2, rtl_check_failed_code_mode,
21851 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21852 rtvec_check_failed_bounds, rtl_check_failed_flag,
21853 _fatal_insn_not_found, _fatal_insn): Likewise.
21854 * tree.h (tree_contains_struct_check_failed,
21855 tree_check_failed, tree_not_check_failed,
21856 tree_class_check_failed, tree_range_check_failed,
21857 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21858 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21859 tree_operand_check_failed, omp_clause_check_failed,
21860 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21863 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21865 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21866 flag_branch_probabilities.
21867 * ipa-inline.c (edge_badness): Likewise.
21868 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21869 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21870 * predict.c (maybe_hot_frequency_p): Likewise.
21871 (probably_never_executed): Likewise.
21872 * sched-ebb.c (schedule_ebbs): Likewise.
21873 * sched-rgn.c (find_single_block_region): Likewise.
21874 * tracer.c (tail_duplicate): Likewise.
21876 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21878 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21879 longer requires x_flag_profile_use.
21881 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21883 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21884 instead of flag_reorder_blocks_and_partition.
21885 * dbxout.c (dbxout_function_end): Likewise.
21886 * dwarf2out.c (gen_subprogram_die): Likewise.
21887 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21888 * hw-doloop.c (reorg_loops): Likewise.
21889 * varasm.c (assemble_start_function,
21890 assemble_end_function): Likewise.
21891 (decide_function_section): Do not check for
21892 flag_reorder_blocks_and_partition.
21894 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21896 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21898 (chkp_get_hard_register_fake_addr_expr): Ditto.
21899 (chkp_build_addr_expr): Add check for hard reg case.
21900 (chkp_parse_array_and_component_ref): Ditto.
21901 (chkp_find_bounds_1): Ditto.
21902 (chkp_process_stmt): Don't generate bounds store for
21905 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21907 * predict.c (maybe_hot_bb_p): Do not check profile status.
21908 (maybe_hot_edge_p): Likewise.
21909 (probably_never_executed): Check for zero counts even if profile
21911 (unlikely_executed_edge_p): New function.
21912 (unlikely_executed_stmt_p): New function.
21913 (unlikely_executed_bb_p): New function.
21914 (set_even_probabilities): Use unlikely predicates.
21915 (combine_predictions_for_bb): Likewise.
21916 (predict_paths_for_bb): Likewise.
21917 (predict_paths_leading_to_edge): Likewise.
21918 (determine_unlikely_bbs): New function.
21919 (estimate_bb_frequencies): Use it.
21920 (compute_function_frequency): Use zero counts even if profile is
21922 * profile-count.h: Fix typo.
21924 2017-08-08 Julia Koval <julia.koval@intel.com>
21926 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21927 _mm512_mask_cvtsepi16_storeu_epi8,
21928 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21929 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21930 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21931 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21932 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21933 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21934 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21935 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21936 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21937 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21938 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21939 __builtin_ia32_pmovuswb256mem_mask,
21940 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21941 __builtin_ia32_pmovwb512mem_mask): New builtins.
21943 2017-08-08 Julia Koval <julia.koval@intel.com>
21945 PR target/73350,80862
21946 * config/i386/subst.md (round): Fix round pattern.
21947 * config/i386/i386.c (ix86_erase_embedded_rounding):
21948 Fix erasing rounding for the fixed pattern.
21950 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21952 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21954 2017-06-08 Martin Liska <mliska@suse.cz>
21956 PR gcov-profile/80911
21957 * gcov.c (block_info::block_info): New constructor.
21959 2017-06-07 Carl Love <cel@us.ibm.com>
21961 * config/rs6000/rs6000-c: The return type of the following
21962 built-in functions was implemented as int not long long. Fix sign
21963 of return value for the unsigned version of vec_mulo and vec_mule.
21964 vector unsigned long long vec_bperm (vector unsigned long long,
21965 vector unsigned char)
21966 vector signed long long vec_mule (vector signed int,
21968 vector unsigned long long vec_mule (vector unsigned int,
21969 vector unsigned int)
21970 vector signed long long vec_mulo (vector signed int,
21972 vector unsigned long long vec_mulo (vector unsigned int,
21973 vector unsigned int)
21974 * doc/extend.texi: Fix the documentation for the built-in
21977 2017-06-07 Carl Love <cel@us.ibm.com>
21980 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21983 2017-06-07 Carl Love <cel@us.ibm.com>
21985 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21986 support, Generate doublehv for signed int/float for BE case only.
21988 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21990 * doc/invoke.texi (mcx16): Rewrite.
21992 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21994 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21995 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21996 *mov<mode>_softfloat, and an anonymous splitter): Use
21997 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21999 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22001 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22003 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22004 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22005 (rs6000_debug_reg_global): Adjust.
22006 (rs6000_init_hard_regno_mode_ok): Adjust.
22007 (rs6000_dbx_register_number): Adjust.
22008 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22009 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22010 Remove SPE_ACC and SPEFSCR.
22011 (REG_ALLOC_ORDER): Ditto.
22012 (FRAME_POINTER_REGNUM): Change to 111.
22013 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22014 (REG_CLASS_NAMES): Ditto.
22015 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22016 (REGISTER_NAMES): Ditto.
22017 (ADDITIONAL_REG_NAMES): Ditto.
22018 (rs6000_reg_names): Ditto.
22019 * config/rs6000/rs6000.md: Renumber some register number
22022 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22024 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22026 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22027 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22029 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22030 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22031 Delete the SPE high registers.
22032 (REG_ALLOC_ORDER): Ditto.
22033 (enum reg_class): Remove SPE_HIGH_REGS.
22034 (REG_CLASS_NAMES): Ditto.
22035 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22036 (REGISTER_NAMES): Ditto.
22037 (rs6000_reg_names): Ditto.
22038 * doc/tm.texi.in: Remove SPE as example.
22039 * doc/tm.texi: Regenerate.
22041 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22043 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22044 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22045 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22046 * config/rs6000/rs6000.md (type): Remove "brinc".
22048 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22050 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22051 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22052 * config/rs6000/linuxspe.h: Delete file.
22053 * config/rs6000/rs6000.md: Don't include spe.md.
22054 * config/rs6000/spe.h: Delete file.
22055 * config/rs6000/spe.md: Delete file.
22056 * config/rs6000/t-rs6000: Remove spe.md.
22058 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22060 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22061 (reg_or_none500mem_operand): Delete.
22062 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22063 instead of reg_or_none500mem_operand.
22065 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22067 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22068 handling of SPE flags.
22069 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22071 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22073 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22075 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22076 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22077 paired_divv2sf3): Similar.
22078 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22079 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22080 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22082 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22083 Rename the paired_* instruction patterns.
22084 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22086 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22087 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22088 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22089 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22090 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22091 PAIRED_VECTOR_MODE.
22092 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22093 (spe_func_has_64bit_regs_p): Delete.
22094 (spe_expand_predicate_builtin): Delete.
22095 (spe_expand_evsel_builtin): Delete.
22096 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22097 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22098 (invalid_e500_subreg): Delete.
22099 (rs6000_legitimize_address): Always force_reg op2 as well, for
22100 paired single memory accesses.
22101 (rs6000_member_type_forces_blk): Delete.
22102 (rs6000_spe_function_arg): Delete.
22103 (rs6000_expand_unop_builtin): Delete SPE handling.
22104 (rs6000_expand_binop_builtin): Ditto.
22105 (spe_expand_stv_builtin): Delete.
22106 (bdesc_2arg_spe): Delete.
22107 (spe_expand_builtin): Delete.
22108 (spe_expand_predicate_builtin): Delete.
22109 (spe_expand_evsel_builtin): Delete.
22110 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22111 (spe_init_builtins): Delete.
22112 (spe_func_has_64bit_regs_p): Delete.
22113 (savres_routine_name): Delete "info" parameter. Adjust callers.
22114 (rs6000_emit_stack_reset): Ditto.
22115 (rs6000_dwarf_register_span): Delete.
22116 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22117 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22118 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22120 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22122 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22123 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22124 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22125 mulv2sf3, divv2sf3): Delete expanders.
22127 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22129 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22131 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22133 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22134 * config/rs6000/rs6000.c: Ditto.
22136 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22138 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22139 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22140 comparison_operator.
22142 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22144 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22145 * config/rs6000/rs6000.opt: Ditto.
22146 * config/rs6000/t-rtems: Ditto.
22148 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22150 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22151 TARGET_E500_SINGLE by 0, simplify.
22152 * config/rs6000/rs6000.c: Ditto.
22153 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22154 (spe_build_register_parallel): Delete.
22155 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22156 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22157 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22158 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22159 (E500_CONVERT): Delete.
22160 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22162 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22164 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22165 * config/rs6000/dfp.md: Ditto.
22166 (negdd2, *negdd2_fpr): Merge.
22167 (absdd2, *absdd2_fpr): Merge.
22168 (negtd2, *negtd2_fpr): Merge.
22169 (abstd2, *abstd2_fpr): Merge.
22170 * config/rs6000/e500.h: Delete file.
22171 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22172 TARGET_FPRS by 1 and simplify.
22173 * config/rs6000/rs6000-c.c: Ditto.
22174 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22175 TARGET_DF_SPE by 0.
22176 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22178 * config/rs6000/rs6000.md: Ditto.
22179 (floatdidf2, *floatdidf2_fpr): Merge.
22180 (move_from_CR_gt_bit): Delete.
22181 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22182 (E500_CR_IOR_COMPARE): Delete.
22183 (All patterns that require !TARGET_FPRS): Delete.
22184 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22186 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22188 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22190 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22192 * graphds.c (add_edge): Intitialize edge's attached data.
22193 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22194 pointer parameter. Call pointed function on each edge during
22195 graph traversing. Skip traversing the edge when the function
22197 (graphds_dfs, graphds_scc): Ditto.
22198 (for_each_edge): New parameter. Pass the new parameter to callback
22200 * graphds.h (skip_edge_callback): New function pointer type.
22201 (graphds_dfs, graphds_scc): New function pointer parameter.
22202 (graphds_edge_callback, for_each_edge): New parameter.
22204 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22206 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22207 out code checking if runtime alias check is possible to below ...
22208 Call the new function.
22209 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22210 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22212 2017-06-07 Marek Polacek <polacek@redhat.com>
22215 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22216 TYPE_OVERFLOW_WRAPS check.
22218 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22220 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22221 if versioning is required.
22222 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22223 peeling with the check for versioning.
22225 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22227 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22228 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22229 Set true to new parameter if new ssa variable is defined.
22230 (vect_gen_vector_loop_niters): Refactor. Set range information
22231 for the new vector loop bound variable.
22232 (vect_do_peeling): Ditto.
22234 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22236 * tree-affine.c (ssa.h): Include header file.
22237 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22238 has wrapping overflow behavior.
22240 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22242 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22244 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22246 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22247 (tree_to_aff_combination): ... here.
22249 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22251 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22252 reg_pressure model function.
22253 (ivopts_global_cost_for_size): Delete.
22254 (determine_set_costs, iv_ca_recount_cost): Call new model function
22255 ivopts_estimate_reg_pressure.
22257 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22259 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22260 expensive than udiv. Remove floating point cases from mod.
22262 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22264 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22265 Increase idiv cost.
22267 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22269 * config/aarch64/aarch64.md
22270 (copysignsf3): Fix mask generation.
22272 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22274 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22276 (class dump_manager): Add register_dumps method.
22277 * dumpfile.c: Include langhooks.h.
22278 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22279 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22280 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22281 (dump_manager::dump_register): Start with 512 entries instead of 32.
22282 (dump_manager::register_dumps): New method.
22283 * toplev.c (general_init): Instead of invoking register_dumps
22284 langhook, invoke register_dumps method on the dump manager.
22285 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22288 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22290 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22291 Say that # requires an associated define_split to exist, and that
22292 the define_split must be suitable for use after register allocation.
22294 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22296 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22297 (compute_outgoing_frequencies): Also initialize zero counts.
22298 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22299 around loops; preserve more of profile when nothing changes.
22301 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22303 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22305 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22306 * config/arm/arm-cpu-cdata.h: Regenerate.
22307 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22308 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22309 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22310 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22311 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22313 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22315 * config/arm/t-rmprofile: Likewise.
22316 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22318 2017-06-06 David S. Miller <davem@davemloft.net>
22321 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22322 blockage if function uses alloca.
22324 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22326 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22327 New "uid" fields to hold pretty-print IDs of group and ref.
22328 Memory references are now identified as <group_id>:<ref_id>
22329 instead of using [random] addresses.
22330 (dump_mem_details): Simplify, no functional change.
22331 (dump_mem_ref): Simplify and make output more concise.
22332 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22333 (find_or_create_group): Initialize group uid.
22334 (record_ref): Initialize ref uid. Improve debug output.
22335 (prune_group_by_reuse, should_issue_prefetch_p,)
22336 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22337 (mark_nontemporal_store, determine_loop_nest_reuse):
22338 Improve debug output.
22340 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22342 * dbgcnt.def (prefetch): New debug counter.
22343 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22344 (schedule_prefetches): Stop issueing prefetches if debug counter
22347 2017-06-06 Tom de Vries <tom@codesourcery.com>
22349 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22350 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22352 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22354 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22355 Use aarch64_reg_or_zero predicate for operand 4.
22356 (aarch64_compare_and_swap<mode> define_insn_and_split):
22357 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22358 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22360 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22362 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22363 (arm_compute_save_core_reg_mask): This.
22364 (thumb1_compute_save_reg_mask): Rename into ...
22365 (thumb1_compute_save_core_reg_mask): This.
22366 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22367 (arm_compute_frame_layout): Likewise.
22369 2017-06-06 Richard Biener <rguenther@suse.de>
22371 PR tree-optimization/80974
22372 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22373 keep or clear leaders SSA info.
22375 2017-06-06 Tom de Vries <tom@codesourcery.com>
22377 * config/nvptx/nvptx.c (split_mode_p): New function.
22378 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22380 2017-06-06 Tom de Vries <tom@codesourcery.com>
22382 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22384 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22387 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22390 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22392 * shrink-wrap.c (handle_simple_exit): Update profile.
22393 (try_shrink_wrapping): Upate profile.
22395 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22397 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22398 (tree_guess_outgoing_edge_probabilities): New.
22399 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22400 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22402 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22404 * ipa-split.c (split_function): Initialize return bb profile.
22406 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22408 * profile.c (compute_branch_probabilities): Also initialize
22409 EXIT_BLOCK profile.
22411 2017-06-06 Richard Biener <rguenther@suse.de>
22413 PR tree-optimization/80928
22414 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22415 (vect_analyze_loop_operations): Properly guard analysis for
22417 (vect_transform_loop): Likewise.
22418 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22419 (vect_model_induction_cost): Do not cost for pure SLP.
22420 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22421 of induction in inner loop vectorization.
22422 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22423 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22424 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22426 (vect_analyze_slp_cost_1): Cost induction.
22427 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22428 (vect_get_slp_vect_defs): Likewise.
22429 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22430 (vect_transform_stmt): Handle SLP reductions.
22431 * tree-vectorizer.h (vectorizable_induction): Adjust.
22433 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22435 * config/rs6000/rs6000.c (make_resolver_func): Update
22436 init_lowered_empty_function call.
22438 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22440 * doc/invoke.texi: Document the -fprofile-abs-path option.
22441 * common.opt (fprofile-abs-path): New option.
22442 * gcov-io.h (gcov_write_filename): Declare.
22443 * gcov-io.c (gcov_write_filename): New function.
22444 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22445 * profile.c (output_location): Likewise.
22447 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22449 * shring-wrap.c: Revert accidental commit.
22451 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22453 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22455 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22457 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22459 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22460 profile in callgraph edge.
22461 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22462 (apply_scale): Likewise.
22463 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22464 Also scale profile when inlining function with zero profile.
22465 (initialize_cfun): Update exit block profile even when it is zero.
22466 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22467 when profile is read.
22469 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22471 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22472 (CLONE_*): New constants to define the processors we can generate
22473 code for with the target_clone attribute.
22474 (rs6000_clone_map): New array to identify which clone processors
22475 the current program is running on.
22476 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22477 target_clone attribute.
22478 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22479 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22480 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22481 (cpu_expand_builtin): Add support for target_clone attribute.
22482 (rs6000_valid_attribute_p): Allow "default" attribute.
22483 (get_decl_name): New debug function to simplify printing the
22484 current function name in debugging statements.
22485 (rs6000_clone_priority): New functions to support the target_clone
22486 attribute, and be able to generate code to switch between ISA 2.05
22487 through ISA 3.0 (power6 through power9).
22488 (rs6000_compare_version_priority): Likewise.
22489 (rs6000_get_function_versions_dispatcher): Likewise.
22490 (make_resolver_func): Likewise.
22491 (add_condition_to_bb): Likewise.
22492 (dispatch_function_versions): Likewise.
22493 (rs6000_generate_version_dispatcher_body): Likewise.
22494 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22495 (fusion_gpr_load_p): Fix a spacing issue.
22496 * doc/extend.texi (Common Function Attributes): Document that the
22497 PowerPC supports the target_clone attribute.
22499 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22501 * config/arm/arm.h: explain F symbol found in description of ARM
22502 register allocation in its legend.
22504 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22506 * config/mips/frame-header-opt.c: Include profile-count.h.
22507 * config/riscv/riscv.c: Include profile-count.h
22509 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22511 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22513 (sm_set_flag_if_changed): Add bbs field.
22514 (execute_sm_if_changed_flag_set): Pass BBS.
22515 (execute_sm): Update.
22517 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22519 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22522 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22524 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22525 (peephole2): New peephole2 to emit the above.
22526 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22528 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22530 * config/aarch64/aarch64.c (define_peephole2 above
22531 *sub_<shift>_<mode>): New peephole.
22533 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22535 * config/i386/i386.c (make_resolver_func): Update.
22536 * Makefile.in: Add profile-count.h and profile-count.o
22537 * auto-profile.c (afdo_indirect_call): Update to new API.
22538 (afdo_set_bb_count): Update.
22539 (afdo_propagate_edge): Update.
22540 (afdo_propagate_circuit): Update.
22541 (afdo_calculate_branch_prob): Update.
22542 (afdo_annotate_cfg): Update.
22543 * basic-block.h: Include profile-count.h
22544 (struct edge_def): Turn count to profile_count.
22545 (struct basic_block_def): Likewie.
22546 (REG_BR_PROB_BASE): Move to profile-count.h
22547 (RDIV): Move to profile-count.h
22548 * bb-reorder.c (max_entry_count): Turn to profile_count.
22549 (find_traces): Update.
22550 (rotate_loop):Update.
22551 (connect_traces):Update.
22552 (sanitize_hot_paths):Update.
22553 * bt-load.c (migrate_btr_defs): Update.
22554 * cfg.c (RDIV): Remove.
22555 (init_flow): Use alloc_block.
22556 (alloc_block): Uninitialize count.
22557 (unchecked_make_edge): Uninitialize count.
22558 (check_bb_profile): Update.
22559 (dump_edge_info): Update.
22560 (dump_bb_info): Update.
22561 (update_bb_profile_for_threading): Update.
22562 (scale_bbs_frequencies_int): Update.
22563 (scale_bbs_frequencies_gcov_type): Update.
22564 (scale_bbs_frequencies_profile_count): New.
22565 * cfg.h (update_bb_profile_for_threading): Update.
22566 (scale_bbs_frequencies_profile_count): Declare.
22567 * cfgbuild.c (compute_outgoing_frequencies): Update.
22568 (find_many_sub_basic_blocks): Update.
22569 * cfgcleanup.c (try_forward_edges): Update.
22570 (try_crossjump_to_edge): Update.
22571 * cfgexpand.c (expand_gimple_tailcall): Update.
22572 (construct_exit_block): Update.
22573 * cfghooks.c (verify_flow_info): Update.
22574 (dump_bb_for_graph): Update.
22575 (split_edge): Update.
22576 (make_forwarder_block): Update.
22577 (duplicate_block): Update.
22578 (account_profile_record): Update.
22579 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22580 (get_estimated_loop_iterations): Update.
22581 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22582 (single_likely_exit): Update.
22583 * cfgloopmanip.c (scale_loop_profile): Update.
22585 (set_zero_probability): Update.
22586 (lv_adjust_loop_entry_edge): Update.
22587 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22588 (purge_dead_edges): Update.
22589 (rtl_account_profile_record): Update.
22590 * cgraph.c (cgraph_node::create): Uninitialize count.
22591 (symbol_table::create_edge): Uninitialize count.
22592 (cgraph_update_edges_for_call_stmt_node): Update.
22593 (cgraph_edge::dump_edge_flags): Update.
22594 (cgraph_node::dump): Update.
22595 (cgraph_edge::maybe_hot_p): Update.
22596 * cgraph.h: Include profile-count.h
22597 (create_clone), create_edge, create_indirect_edge): Update.
22598 (cgraph_node): Turn count to profile_count.
22599 (cgraph_edge0: Likewise.
22600 (make_speculative, clone): Update.
22601 (create_edge): Update.
22602 (init_lowered_empty_function): Update.
22603 * cgraphclones.c (cgraph_edge::clone): Update.
22604 (duplicate_thunk_for_node): Update.
22605 (cgraph_node::create_clone): Update.
22606 * cgraphunit.c (cgraph_node::analyze): Update.
22607 (cgraph_node::expand_thunk): Update.
22608 * final.c (dump_basic_block_info): Update.
22609 * gimple-streamer-in.c (input_bb): Update.
22610 * gimple-streamer-out.c (output_bb): Update.
22611 * graphite.c (print_global_statistics): Update.
22612 (print_graphite_scop_statistics): Update.
22613 * hsa-brig.c: Include basic-block.h.
22614 * hsa-dump.c: Include basic-block.h.
22615 * hsa-gen.c (T sum_slice): Update.
22616 (convert_switch_statements):Update.
22617 * hsa-regalloc.c: Include basic-block.h.
22618 * ipa-chkp.c (chkp_produce_thunks): Update.
22619 * ipa-cp.c (struct caller_statistics): Update.
22620 (init_caller_stats): Update.
22621 (gather_caller_stats): Update.
22622 (ipcp_cloning_candidate_p): Update.
22623 (good_cloning_opportunity_p): Update.
22624 (get_info_about_necessary_edges): Update.
22625 (dump_profile_updates): Update.
22626 (update_profiling_info): Update.
22627 (update_specialized_profile): Update.
22628 (perhaps_add_new_callers): Update.
22629 (decide_about_value): Update.
22630 (ipa_cp_c_finalize): Update.
22631 * ipa-devirt.c (struct odr_type_warn_count): Update.
22632 (struct decl_warn_count): Update.
22633 (struct final_warning_record): Update.
22634 (possible_polymorphic_call_targets): Update.
22635 (ipa_devirt): Update.
22636 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22637 * ipa-icf.c (sem_function::merge): Update.
22638 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22639 * ipa-inline.c (compute_uninlined_call_time): Update.
22640 (compute_inlined_call_time): Update.
22641 (want_inline_small_function_p): Update.
22642 (want_inline_self_recursive_call_p): Update.
22643 (edge_badness): Update.
22644 (lookup_recursive_calls): Update.
22645 (recursive_inlining): Update.
22646 (inline_small_functions): Update.
22647 (dump_overall_stats): Update.
22648 (dump_inline_stats): Update.
22649 * ipa-profile.c (ipa_profile_generate_summary): Update.
22650 (ipa_propagate_frequency): Update.
22651 (ipa_profile): Update.
22652 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22653 * ipa-utils.c (ipa_merge_profiles): Update.
22654 * loop-doloop.c (doloop_modify): Update.
22655 * loop-unroll.c (report_unroll): Update.
22656 (unroll_loop_runtime_iterations): Update.
22657 * lto-cgraph.c (lto_output_edge): Update.
22658 (lto_output_node): Update.
22659 (input_node): Update.
22660 (input_edge): Update.
22661 (merge_profile_summaries): Update.
22662 * lto-streamer-in.c (input_cfg): Update.
22663 * lto-streamer-out.c (output_cfg): Update.
22664 * mcf.c (create_fixup_graph): Update.
22665 (adjust_cfg_counts): Update.
22666 (sum_edge_counts): Update.
22667 * modulo-sched.c (sms_schedule): Update.
22668 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22669 * predict.c (maybe_hot_count_p): Update.
22670 (probably_never_executed): Update.
22671 (dump_prediction): Update.
22672 (combine_predictions_for_bb): Update.
22673 (propagate_freq): Update.
22674 (handle_missing_profiles): Update.
22675 (counts_to_freqs): Update.
22676 (rebuild_frequencies): Update.
22677 (force_edge_cold): Update.
22678 * predict.h: Include profile-count.h
22679 (maybe_hot_count_p, counts_to_freqs): UPdate.
22680 * print-rtl-function.c: Do not include cfg.h
22681 * print-rtl.c: Include basic-block.h
22682 * profile-count.c: New file.
22683 * profile-count.h: New file.
22684 * profile.c (is_edge_inconsistent): Update.
22685 (correct_negative_edge_counts): Update.
22686 (is_inconsistent): Update.
22687 (set_bb_counts): Update.
22688 (read_profile_edge_counts): Update.
22689 (compute_frequency_overlap): Update.
22690 (compute_branch_probabilities): Update; Initialize and deinitialize
22692 (branch_prob): Update.
22693 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22694 (edge_gcov_count): New.
22695 (bb_gcov_count): New.
22696 * shrink-wrap.c (try_shrink_wrapping): Update.
22697 * tracer.c (better_p): Update.
22698 * trans-mem.c (expand_transaction): Update.
22699 (ipa_tm_insert_irr_call): Update.
22700 (ipa_tm_insert_gettmclone_call): Update.
22701 * tree-call-cdce.c: Update.
22702 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22703 (gimple_duplicate_sese_tail): Update.
22704 (gimple_account_profile_record): Update.
22705 (execute_fixup_cfg): Update.
22706 * tree-inline.c (copy_bb): Update.
22707 (copy_edges_for_bb): Update.
22708 (initialize_cfun): Update.
22709 (freqs_to_counts): Update.
22710 (copy_cfg_body): Update.
22711 (expand_call_inline): Update.
22712 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22713 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22714 (try_unroll_loop_completely): Update.
22715 (try_peel_loop): Update.
22716 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22717 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22718 * tree-ssa-loop-split.c (connect_loops): Update.
22719 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22720 * tree-ssa-reassoc.c (branch_fixup): Update.
22721 * tree-ssa-tail-merge.c (replace_block_by): Update.
22722 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22723 (compute_path_counts): Update.
22724 (update_profile): Update.
22725 (recompute_probabilities): Update.
22726 (update_joiner_offpath_counts): Update.
22727 (estimated_freqs_path): Update.
22728 (freqs_to_counts_path): Update.
22729 (clear_counts_path): Update.
22730 (ssa_fix_duplicate_block_edges): Update.
22731 (duplicate_thread_path): Update.
22732 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22733 (struct switch_conv_info): Update.
22734 * tree-tailcall.c (decrease_profile): Update.
22735 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22736 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22737 * value-prof.c (check_counter): Update.
22738 (gimple_divmod_fixed_value): Update.
22739 (gimple_mod_pow2): Update.
22740 (gimple_mod_subtract): Update.
22741 (gimple_ic_transform): Update.
22742 (gimple_stringop_fixed_value): Update.
22743 * value-prof.h (gimple_ic): Update.
22745 2017-06-02 Carl Love <cel@us.ibm.com>
22747 * config/rs6000/rs6000-c: Add support for built-in functions
22748 vector double vec_doublee (vector signed int);
22749 vector double vec_doublee (vector unsigned int);
22750 vector double vec_doublee (vector float);
22751 vector double vec_doubleh (vector signed int);
22752 vector double vec_doubleh (vector unsigned int);
22753 vector double vec_doubleh (vector float);
22754 vector double vec_doublel (vector signed int);
22755 vector double vec_doublel (vector unsigned int);
22756 vector double vec_doublel (vector float);
22757 vector double vec_doubleo (vector signed int);
22758 vector double vec_doubleo (vector unsigned int);
22759 vector double vec_doubleo (vector float);.
22760 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22761 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22763 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22764 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22765 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22767 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22768 vec_doublel, vec_doubleh.
22769 * doc/extend.texi: Update the built-in documentation file for the
22770 new built-in functions.
22772 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22775 * ipa-inline-analysis.c (free_growth_caches): Set
22776 edge_removal_hook_holder to NULL after removing it.
22778 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22780 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22781 comparision with zero.
22783 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22784 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22785 for early expansion of vec_min and vec_max builtins.
22786 (builtin_function_type): Add min/max unsigned variants to those
22787 identified as having unsigned arguments.
22789 2017-06-02 Olivier Hainque <hainque@adacore.com>
22791 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22793 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22795 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22796 Use VALL_F16 iterator rather than VALL.
22798 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22800 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22801 Emit CBNZ inside loop when doing a strong exchange and comparing
22802 against zero. Generate the CC flags after the loop.
22804 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22806 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22807 (dl_section_ref): New.
22808 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22809 On AIX, append an expression to subtract the size of the
22810 section length to dl_section_ref.
22812 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22814 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22815 for early expansion of vector absolute builtins.
22817 2017-06-02 Richard Biener <rguenther@suse.de>
22819 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22820 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22822 2017-06-02 Richard Biener <rguenther@suse.de>
22824 PR tree-optimization/80948
22825 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22826 stmt order as well.
22828 2017-06-02 Richard Biener <rguenther@suse.de>
22830 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22832 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22833 for inductions as relevant.
22835 2017-06-02 Richard Biener <rguenther@suse.de>
22837 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22838 (vectorizable_induction): ... this. Remove dead code.
22840 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22842 * builtins. (expand_builtin_alloca): Remove second parameter and
22843 infer its value from the first parameter instead.
22844 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22846 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22848 PR rtl-optimization/80903
22849 * loop-doloop.c (add_test): Unshare sequence.
22851 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22853 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22855 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22857 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22859 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22860 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22861 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22862 (xlogue_layout::get_stub_rtx): Make static.
22863 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22864 (xlogue_layout::compute_stub_managed_regs): Rename to...
22865 (xlogue_layout::count_stub_managed_regs): ...this.
22866 (xlogue_layout::is_stub_managed_reg): New function.
22867 (xlogue_layout::m_stub_names): Rename to...
22868 (xlogue_layout::s_stub_names): ...this, make static.
22869 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22870 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22871 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22872 xlogue_layout::s_stub_names): Instantiate statics.
22873 (stub_managed_regs): Remove.
22874 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22875 (disable_call_ms2sysv_xlogues): Rename to...
22876 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22877 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22879 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22880 change after reload_completed.
22881 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22883 (ix86_expand_prologue): Likewise.
22884 (ix86_expand_epilogue): Likewise.
22885 (ix86_expand_split_stack_prologue): Likewise.
22886 (ix86_compute_frame_layout): Remove frame parameter ...
22887 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22888 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22890 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22891 (ix86_frame): Move from here ...
22892 * config/i386/i386.h (ix86_frame): ... to here.
22893 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22894 complete ix86_frame data structure instead. Remove some_ld_name.
22896 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22898 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22899 symbols that hold a DECL_VALUE_EXPR.
22901 2017-06-01 Martin Jambor <mjambor@suse.cz>
22903 PR tree-optimization/80898
22904 * tree-sra.c (process_subtree_disqualification): Removed.
22905 (disqualify_candidate): Do not acll
22906 process_subtree_disqualification.
22907 (subtree_mark_written_and_enqueue): New function.
22908 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22909 RHS has been disqualified and re-queue LHS if necessary. Apart
22910 from that, ignore disqualified RHS.
22912 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22914 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22915 address fetch for z10 or later.
22917 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22919 * config/arc/arc.md (tst_movb): Add guard when splitting.
22921 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22923 * config/arc/arc.c (arc_can_eliminate): Test against
22924 arc_frame_pointer_needed.
22926 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22928 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22929 to prevent store reordering.
22930 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22931 (type): Add block type.
22932 (stack_tie): Define special instruction to be used in
22935 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22937 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22938 constraint. It is not valid for the pattern.
22939 (noncommutative_binary_comparison): Likewise.
22941 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22943 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22946 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22948 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22949 be used by the reg-alloc.
22951 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22953 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22954 reg-alloc when having mul64 or mul32x16 instructions.
22955 (mulsidi3): Likewise.
22956 (umulsidi3): Likewise.
22957 (mulsi32x16): New pattern.
22958 (mulsi64): Likewise.
22959 (mulsidi64): Likewise.
22960 (umulsidi64): Likewise.
22961 (MUL32x16_REG): Define.
22962 (mul64_600): Use MUL32x16_REG.
22963 (mac64_600): Likewise.
22964 (umul64_600): Likewise.
22965 (umac64_600): Likewise.
22967 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22969 * config/arc/arc.md (mulsi3_700): Make it commutative.
22971 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22973 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22975 (*sign_extendsidi2_insn): Likewise for movstosw.
22977 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22979 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22980 the type of the input discriminant value. Convert the
22981 discriminant value of signedness vary.
22983 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22985 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22986 Add to -Wall section.
22988 2017-06-01 Richard Biener <rguenther@suse.de>
22990 PR middle-end/66313
22991 * fold-const.c (fold_plusminus_mult_expr): If the factored
22992 factor may be zero use a wrapping type for the inner operation.
22993 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22994 and handle moved defs.
22995 (process_assignment): Properly guard the unary op case. Return a
22996 tri-state indicating that moving the stmt before the call may allow
22997 to continue. Pass through to_move.
22998 (find_tail_calls): Handle moving unrelated defs before
23001 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23004 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23005 splitter result in the canonical way.
23007 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23009 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23010 also for 32bit target. Update insn attributes.
23011 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23013 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23015 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23016 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23017 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23018 (_mm_maskz_min_ss): New intrinsics.
23020 2017-05-31 Martin Liska <mliska@suse.cz>
23022 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23023 Change comment style to one we normally use.
23024 (vectorizable_reduction): Likewise.
23025 (vectorizable_induction): Likewise.
23026 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23027 (vectorizable_call): Likewise.
23028 (vectorizable_simd_clone_call): Likewise.
23029 (vectorizable_conversion): Likewise.
23030 (vectorizable_assignment): Likewise.
23031 (vectorizable_shift): Likewise.
23032 (vectorizable_operation): Likewise.
23033 (vectorizable_store): Likewise.
23034 (vectorizable_load): Likewise.
23035 * tree-vectorizer.h: Likewise.
23037 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23039 * passes.c (emergency_dump_function): New.
23040 * tree-pass.h (emergency_dump_function): Declare.
23041 * plugin.c (plugins_internal_error_function): Remove.
23042 * plugin.h (plugins_internal_error_function): Remove declaration.
23043 * toplev.c (internal_error_function): New static function. Use it...
23044 (general_init): ...here.
23046 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23048 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23049 (arc_rtx_costs): Add costs for new patterns.
23050 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23051 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23053 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23055 * tree-ssa-strlen.c (get_next_strinfo): New function.
23056 (get_stridx_plus_constant): Use it.
23057 (zero_length_string): Likewise.
23058 (adjust_related_strinfos): Likewise.
23059 (adjust_last_stmt): Likewise.
23061 2017-05-31 Richard Biener <rguenther@suse.de>
23064 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23065 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23067 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23069 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23070 loop_vinfo argument and use of dependence distance vectors.
23071 Check instead whether the two references differ only in their
23072 initial value and assume that they have the same alignment if the
23073 difference is a multiple of the vector alignment.
23074 (vect_analyze_data_refs_alignment): Update call accordingly.
23076 2017-05-31 Martin Liska <mliska@suse.cz>
23079 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23081 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23083 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23084 (create_intersect_range_checks): Move from ...
23085 * tree-data-ref.c (create_intersect_range_checks_index)
23086 (create_intersect_range_checks): ... to here.
23087 (create_runtime_alias_checks): New function factored from ...
23088 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23089 here. Call above function.
23090 * tree-data-ref.h (create_runtime_alias_checks): New function.
23092 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23094 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23095 segment length for dr_b and compute it in wide_int.
23097 2017-05-31 Richard Biener <rguenther@suse.de>
23099 PR tree-optimization/80906
23100 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23101 and pass through iv_map.
23102 (copy_bb_and_scalar_dependences): Adjust.
23103 (translate_pending_phi_nodes): Likewise.
23104 (copy_loop_close_phi_args): Handle code-generating IVs instead
23107 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23109 * diagnostic-color.c (color_dict): Add "type-diff".
23110 (parse_gcc_colors): Update comment.
23111 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23112 -fdiagnostics-show-template-tree and -fno-elide-type.
23113 (GCC_COLORS): Add type-diff to example.
23115 (-fdiagnostics-show-template-tree): New.
23116 (-fno-elide-type): New.
23117 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23118 the pp_format_decoder callback. Call any m_format_postprocessor's
23120 (pretty_printer::pretty_printer): Initialize
23121 m_format_postprocessor.
23122 (pretty_printer::~pretty_printer): Delete any
23123 m_format_postprocessor.
23124 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23125 (class format_postprocessor): New class.
23126 (struct pretty_printer::format_decoder): Document the new parameters.
23127 (struct pretty_printer::m_format_postprocessor): New field.
23128 * tree-diagnostic.c (default_tree_printer): Update for new
23129 bool and const char ** params.
23130 * tree-diagnostic.h (default_tree_printer): Likewise.
23132 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23134 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23135 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23136 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23137 rs6000_gen_cell_microcode code.
23138 (rs6000_final_prescan_insn): Delete.
23139 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23140 "warn-cell-microcode" entries.
23141 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23142 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23143 throughout. Change cc_reg_not_micro_cr0_operand to
23144 cc_reg_not_cr0_operand throughout.
23145 (*extendhi<mode>2_noload): Delete.
23146 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23147 (mwarn-cell-microcode): Delete.
23148 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23149 -mgen-cell-microcode and -mwarn-cell-microcode.
23151 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23154 * config/i386/constraints.md (Yd): New constraint.
23156 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23157 and (?Yd, r) alternatives. Update insn attributes.
23158 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23159 and (?*Yd, r) alternatives. Update insn attributes.
23160 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23162 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23164 * gimplify.c (gimplify_modify_expr): Don't create a
23165 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23168 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23170 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23172 2017-05-30 Richard Biener <rguenther@suse.de>
23174 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23175 and reduc_def fields.
23176 (STMT_VINFO_REDUC_TYPE): New define.
23177 (STMT_VINFO_REDUC_DEF): Likewise.
23178 (vect_force_simple_reduction): Adjust prototype.
23179 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23180 (vect_is_simple_reduction): Remove check_reduction argument.
23181 (vect_force_simple_reduction): Adjust and set
23182 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23183 (vectorizable_reduction): Do not re-do reduction analysis
23184 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23185 * tree-parloops.c (gather_scalar_reductions): Adjust.
23187 2017-05-30 Richard Biener <rguenther@suse.de>
23189 PR middle-end/80901
23190 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23193 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23195 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23196 Introduce unknown_misalignment parameter and remove vf.
23197 (vect_peeling_hash_get_lowest_cost):
23198 Pass unknown_misalignment parameter.
23199 (vect_enhance_data_refs_alignment):
23200 Fix unsupportable data ref treatment.
23202 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23204 * tree-vect-data-refs.c (vect_get_data_access_cost):
23205 Workaround for SLP handling.
23206 (vect_enhance_data_refs_alignment):
23207 Compute costs for doing no peeling at all, compare to the best
23208 peeling costs so far and avoid peeling if cheaper.
23210 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23212 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23213 Return peeling info and set costs to zero for unlimited cost
23215 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23216 with unknown misalignment. Compute and costs for unknown
23217 misalignment, compare them to the costs for known misalignment
23218 and choose the cheapest for peeling.
23220 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23222 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23223 (vect_get_peeling_costs_all_drs): Create function.
23224 (vect_peeling_hash_get_lowest_cost):
23225 Use vect_get_peeling_costs_all_drs.
23226 (vect_peeling_supportable): Create function.
23227 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23229 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23231 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23232 DR_HAS_NEGATIVE_STEP.
23233 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23234 (vect_enhance_data_refs_alignment): Use.
23235 (vect_duplicate_ssa_name_ptr_info): Use.
23236 * tree-vectorizer.h (dr_misalignment): Use.
23237 (known_alignment_for_access_p): Use.
23239 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23242 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23244 (has_section_name): New function.
23246 2017-05-30 Martin Liska <mliska@suse.cz>
23249 * auto-profile.c (get_function_decl_from_block): Fix
23252 2017-05-30 Richard Biener <rguenther@suse.de>
23254 PR middle-end/80876
23255 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23257 2017-05-30 Martin Liska <mliska@suse.cz>
23259 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23260 * dumpfile.h (struct dump_file_info): Remove ctors.
23262 2017-05-30 Martin Liska <mliska@suse.cz>
23264 * predict.def: Fix GNU coding style.
23266 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23268 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23269 Mark 'to' argument with ATTRIBUTE_UNUSED.
23271 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23273 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23274 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23275 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23278 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23280 * doc/install.texi (Options specification): Restore entry of
23281 --enable-sjlj-exceptions.
23283 2017-05-27 Michael Eager <eager@eagercon.com>
23286 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23288 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23290 * config/microblaze/microblaze.h
23291 (FIXED_REGISTERS): Update in macro.
23292 (CALL_USED_REGISTERS): Update in macro.
23294 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23296 * doc/install.texi: Add links to macOS binary distributions.
23298 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23302 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23304 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23306 2017-05-26 Martin Liska <mliska@suse.cz>
23308 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23310 2017-05-26 Martin Liska <mliska@suse.cz>
23312 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23313 always leading ';; '.
23314 (dump_bb_info): Likewise.
23315 (brief_dump_cfg): Likewise.
23316 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23317 * dumpfile.c: Remove usage of TDF_VERBOSE.
23318 * dumpfile.h (enum dump_kind): Likewise.
23319 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23320 * print-tree.c (debug_verbose): Remove.
23321 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23322 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23323 * tree-diagnostic.c (default_tree_printer): Replace
23324 TDF_DIAGNOSTIC with TDF_SLIM.
23326 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23328 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23329 in parameter loop, rather than loop_vinfo.
23330 (create_intersect_range_checks): Ditto.
23331 (vect_create_cond_for_alias_checks): Update call to above functions.
23333 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23335 PR tree-optimization/80815
23336 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23337 for merging runtime alias checks. Handle negative DR_STEPs.
23339 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23341 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23343 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23344 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23345 out code pruning runtime alias checks.
23346 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23347 factored out from above.
23348 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23350 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23352 (prune_runtime_alias_test_list): New decalaration.
23354 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23356 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23357 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23358 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23359 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23360 (operator==, comp_dr_with_seg_len_pair): Ditto.
23361 (vect_prune_runtime_alias_test_list): Ditto.
23363 2017-05-26 Martin Liska <mliska@suse.cz>
23366 * params.def: Bound partial-inlining-entry-probability param.
23368 2017-05-26 Marek Polacek <polacek@redhat.com>
23371 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23374 2017-05-26 Richard Biener <rguenther@suse.de>
23376 PR tree-optimization/80842
23377 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23380 2017-05-26 Richard Biener <rguenther@suse.de>
23382 PR tree-optimization/80844
23383 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23385 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23387 * doc/md.texi (Machine Constraints): Update x86 family
23388 machine constraints section to match 'config/i386/constraints.md'.
23390 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23392 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23394 2017-05-25 Nathan Sidwell <nathan@acm.org>
23396 * doc/invoke.texi (--enable-languages): Update documentation.
23398 2017-05-25 Martin Liska <mliska@suse.cz>
23400 * dumpfile.c: Add TDF_FOLDING.
23401 * dumpfile.h (enum dump_kind): Likewise.
23402 * genmatch.c (dt_simplify::gen_1): Use it.
23404 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23406 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23408 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23410 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23411 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23413 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23415 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23416 * match.pd (X == C): Rewrite it here.
23417 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23418 with_certain_nonzero_bits2): New predicates.
23419 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23421 2017-05-24 Nathan Sidwell <nathan@acm.org>
23423 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23426 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23429 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23431 * config/powerpcspe: New port. Files are copied from the rs6000
23432 port, with "rs6000" in filenames replaced by "powerpcspe".
23434 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23436 PR rtl-optimization/80754
23437 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23439 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23441 * config/sparc/sparc.md (length): Return the correct value for -mflat
23442 sibcalls to match output_sibcall.
23444 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23448 * config/rs6000/rs6000.c (struct machine_function): Add new field
23450 (rs6000_get_separate_components): Init that field, use it.
23451 (rs6000_components_for_bb): Use the field.
23453 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23455 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23457 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23459 PR middle-end/80823
23460 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23462 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23465 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23466 address against address_operand predicate.
23467 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23469 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23471 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23472 parameters passed indirectly.
23474 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23476 * config/i386/i386.md (*movdi_internal): Remove SSE4
23477 alternative 18 (?r, *v). Update insn attributes.
23478 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23479 Update insn attributes.
23480 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23481 Update insn attributes.
23482 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23483 alternative 1 (r, v). Remove isa attribute.
23484 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23485 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23486 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23488 2017-05-23 Tom de Vries <tom@codesourcery.com>
23490 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23493 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23495 * cgraphunit.c (symbol_table::process_new_functions): Update.
23496 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23497 (inline_generate_summary): Rename to ...
23498 (ipa_fn_summary_generate): ... this one.
23499 (inline_read_summary): Rename to ...
23500 (ipa_fn_summary_read): ... this one.
23501 (inline_write_summary): Rename to ...
23502 (ipa_fn_summary_write): ... this one.
23503 (inline_free_summary): Rename to ...
23504 (ipa_free_fn_summary): ... this one.
23505 (pass_data_local_fn_summary, pass_local_fn_summary,
23506 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23507 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23508 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23509 make_pass_ipa_fn_summary): New.
23510 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23511 inline_write_summary, inline_free_summary): Remove.
23512 (ipa_free_fn_summary) : New.
23513 * ipa-inline.c (ipa_inline): Update.
23514 (pass_ipa_inline): Do not generate summaries.
23515 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23517 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23518 and add pass_ipa_fn_summary.
23519 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23521 (make_pass_inline_parameters): Remove.
23523 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23525 * omp-low.c (struct omp_context): Remove "default_kind" member.
23528 * omp-offload.c (execute_oacc_device_lower): Remove the
23529 parallelism dimensions function attributes for unparallelized
23530 OpenACC kernels constructs.
23532 2017-05-23 Martin Liska <mliska@suse.cz>
23534 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23536 (cgraph_edge::make_speculative): Likewise.
23537 (cgraph_edge::resolve_speculation): Likewise.
23538 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23539 (cgraph_node::dump): Likewise.
23540 * cgraph.h: Likewise.
23541 * cgraphunit.c (analyze_functions): Likewise.
23542 (symbol_table::compile): Likewise.
23543 * ipa-cp.c (print_all_lattices): Likewise.
23544 (determine_versionability): Likewise.
23545 (initialize_node_lattices): Likewise.
23546 (ipcp_verify_propagated_values): Likewise.
23547 (estimate_local_effects): Likewise.
23548 (update_profiling_info): Likewise.
23549 (create_specialized_node): Likewise.
23550 (perhaps_add_new_callers): Likewise.
23551 (decide_about_value): Likewise.
23552 (decide_whether_version_node): Likewise.
23553 (identify_dead_nodes): Likewise.
23554 (ipcp_store_bits_results): Likewise.
23555 * ipa-devirt.c (dump_targets): Likewise.
23556 (ipa_devirt): Likewise.
23557 * ipa-icf.c (sem_item::dump): Likewise.
23558 (sem_function::equals): Likewise.
23559 (sem_variable::equals): Likewise.
23560 (sem_item_optimizer::read_section): Likewise.
23561 (sem_item_optimizer::execute): Likewise.
23562 (congruence_class::dump): Likewise.
23563 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23564 (dump_inline_summary): Likewise.
23565 (estimate_node_size_and_time): Likewise.
23566 (inline_analyze_function): Likewise.
23567 * ipa-inline-transform.c (inline_call): Likewise.
23568 * ipa-inline.c (report_inline_failed_reason): Likewise.
23569 (want_early_inline_function_p): Likewise.
23570 (edge_badness): Likewise.
23571 (update_edge_key): Likewise.
23572 (inline_small_functions): Likewise.
23573 * ipa-profile.c (ipa_profile): Likewise.
23574 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23575 (ipa_make_edge_direct_to_target): Likewise.
23576 (remove_described_reference): Likewise.
23577 (ipa_impossible_devirt_target): Likewise.
23578 (propagate_controlled_uses): Likewise.
23579 (ipa_print_node_params): Likewise.
23580 (ipcp_transform_function): Likewise.
23581 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23582 (propagate_pure_const): Likewise.
23583 * ipa-reference.c (generate_summary): Likewise.
23584 (read_write_all_from_decl): Likewise.
23585 (propagate): Likewise.
23586 (ipa_reference_read_optimization_summary): Likewise.
23587 * ipa-utils.c (ipa_merge_profiles): Likewise.
23588 * ipa.c (walk_polymorphic_call_targets): Likewise.
23589 (symbol_table::remove_unreachable_nodes): Likewise.
23590 (ipa_single_use): Likewise.
23591 * passes.c (execute_todo): Likewise.
23592 * predict.c (drop_profile): Likewise.
23593 * symtab.c (symtab_node::get_dump_name): New function.
23594 (symtab_node::dump_name): Likewise.
23595 (symtab_node::dump_asm_name): Likewise.
23596 (symtab_node::dump_references): Likewise.
23597 (symtab_node::dump_referring): Likewise.
23598 (symtab_node::dump_base): Likewise.
23599 (symtab_node::debug_symtab): Likewise.
23600 * tree-sra.c (convert_callers_for_node): Likewise.
23601 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23602 * value-prof.c (init_node_map): Likewise.
23604 2017-05-23 Martin Liska <mliska@suse.cz>
23606 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23607 and symtab_node::debug_symtab to symbol_table::debug.
23608 * cgraphunit.c (analyze_functions): Use the renamed function.
23609 (symbol_table::compile): Likewise.
23610 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23611 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23612 * passes.c (execute_todo): Likewise.
23613 * symtab.c (symbol_table::dump): New function.
23614 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23616 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23618 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23619 that nonconst implies exec.
23621 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23623 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23624 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23625 (inline_edge_summary_vec): Turn into ...
23626 (ipa_call_summaries): ... this one.
23627 (redirect_to_unreachable, edge_set_predicate,
23628 evaluate_properties_for_edge, inline_summary_alloc,
23629 reset_ipa_call_summary, reset_inline_summary,
23630 inline_summary_t::duplicate): Update.
23631 (inline_edge_duplication_hook): Turn to ...
23632 (ipa_call_summary_t::duplicate): ... this one.
23633 (inline_edge_removal_hook): Turn to ...
23634 (ipa_call_summary_t::remove): ... this one.
23635 (dump_inline_edge_summary): Turn to ...
23636 (dump_ipa_call_summary): ... this one.
23637 (estimate_function_body_sizes): Update.
23638 (inline_update_callee_summaries): Update.
23639 (remap_edge_change_prob): Update.
23640 (remap_edge_summaries): Update.
23641 (inline_merge_summary): Update.
23642 (do_estimate_edge_time): Update.
23643 (inline_generate_summary): Update.
23644 (inline_read_section): Update.
23645 (inline_read_summary): Update.
23646 (inline_free_summary): Update.
23647 * ipa-inline.c (can_inline_edge_p): Update.
23648 (compute_inlined_call_time): Update.
23649 (want_inline_small_function_p): Update.
23650 (edge_badness): Update.
23651 (early_inliner): Update.
23652 * ipa-inline.h (inline_edge_summary): Turn to ...
23653 (ipa_call_summary): ... this one.
23654 (ipa_call_summary_t): New class.
23655 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23656 (ipa_call_summaries): New.
23657 (inline_edge_summary): Remove.
23658 (estimate_edge_growth): Update.
23659 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23660 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23661 * ipa-split.c (execute_split_functions): Update.
23662 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23664 2017-05-23 Tom de Vries <tom@codesourcery.com>
23666 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23667 attributes): Document rdrand effective target.
23669 2017-05-23 Tom de Vries <tom@codesourcery.com>
23671 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23672 attributes): Sort alphabetically.
23674 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23676 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23678 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23681 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23682 V2DF/V2DI splat into two separate patterns, one that handles
23683 registers, and the other that only handles memory. Drop support
23684 for splatting from a GPR on ISA 2.07 and then splitting the
23685 splat into direct move and splat.
23686 (vsx_splat_<mode>_reg): Likewise.
23687 (vsx_splat_<mode>_mem): Likewise.
23689 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23691 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23693 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23695 PR middle-end/80809
23696 * omp-low.c (finish_taskreg_remap): New function.
23697 (finish_taskreg_scan): If unit size of ctx->record_type
23698 is non-constant, unshare the size expression and replace
23699 decls in it with possible outer var refs.
23701 PR middle-end/80809
23702 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23703 GOVD_SHARED rather than GOVD_PRIVATE with it.
23704 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23705 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23707 PR middle-end/80853
23708 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23709 as last argument to build_outer_var_ref for pointer bases of array
23710 section reductions.
23712 2017-05-19 Martin Sebor <msebor@redhat.com>
23714 * print-tree.c (print_node): Print DECL_READ_P flag.
23716 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23718 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23719 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23720 * cgraph.c: Likewise.
23721 * cgraphunit.c: Likewise.
23722 * gengtype.c: Likewise.
23723 * ipa-cp.c: Likewise.
23724 * ipa-devirt.c: Likewise.
23725 * ipa-icf.c: Likewise.
23726 * ipa-predicate.c: Likewise.
23727 * ipa-profile.c: Likewise.
23728 * ipa-prop.c: Likewise.
23729 * ipa-split.c: Likewise.
23731 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23732 edge_predicate_pool, dump_inline_hints,
23733 inline_summary::account_size_time, redirect_to_unreachable,
23734 edge_set_predicate, set_hint_predicate,
23735 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23736 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23737 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23738 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23739 ipa_call_summary_t::remove, initialize_growth_caches,
23740 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23741 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23742 mark_modified, unmodified_parm_1, unmodified_parm,
23743 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23744 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23745 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23746 will_be_nonconstant_predicate, record_modified_bb_info,
23747 get_minimal_bb, record_modified, param_change_prob,
23748 phi_result_unknown_predicate, predicate_for_phi_result,
23749 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23750 estimate_function_body_sizes, compute_inline_parameters,
23751 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23752 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23753 inline_update_callee_summaries, remap_edge_change_prob,
23754 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23755 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23756 inline_analyze_function, inline_summary_t::insert,
23757 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23758 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23759 inline_free_summary): Move to ipa-fnsummary.h
23760 (predicate_t): Remove.
23761 * ipa-fnsummary.c: New file.
23762 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23763 (enum inline_hints_vals, inline_hints, agg_position_info,
23764 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23765 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23766 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23767 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23768 inline_read_summary, inline_write_summary, inline_free_summary,
23769 inline_analyze_function, initialize_inline_failed,
23770 inline_merge_summary, inline_update_overall_summary,
23771 compute_inline_parameters): Move to ipa-fnsummary.h
23772 * ipa-fnsummary.h: New file.
23773 * ipa-inline-transform.h: Include ipa-inline.h.
23774 * ipa-inline.c: LIkewise.
23776 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23778 * ipa-inline.c (edge_badness): Use inlined_time instead of
23779 inline_summaries->get.
23781 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23783 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23785 2017-05-22 Nathan Sidwell <nathan@acm.org>
23787 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23788 (fdump-lang): Document 'raw' option.
23789 * dumpfile.h (TDI_tu): Delete.
23790 * dumpfile.c (dump_files): Remove translation-unit.
23791 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23793 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23795 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23796 command option from $(AWK) call.
23797 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23798 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23799 [FORMAT]: Remove handling of variable.
23800 * config/avr/t-multilib: Regenerate.
23802 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23804 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23806 (dump_inline_summary): Do not print self_time.
23807 (estimate_function_body_sizes): Do not set self_time.
23808 (compute_inline_parameters): Likewise.
23809 (inline_read_section, inline_write_summary): Do not stream self_time.
23810 * ipa-inline.h (inline_summary): Drop self_time.
23812 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23814 * ipa-inline-analysis.c (account_size_time): Rename to ...
23815 (inline_summary::account_size_time): ... this one.
23816 (reset_ipa_call_summary): Turn to ...
23817 (ipa_call_summary::reset): ... this one.
23818 (reset_inline_summary): Turn to ...
23819 (inline_summary::reset): ... this one.
23820 (inline_summary_t::remove): Update.
23821 (inline_summary_t::duplicate): Update.
23822 (ipa_call_summary_t::remove): Update.
23823 (dump_inline_summary): Update.
23824 (estimate_function_body_sizes): Update.
23825 (compute_inline_parameters): Update.
23826 (estimate_node_size_and_time): Update.
23827 (inline_merge_summary): Update.
23828 (inline_update_overall_summary): Update.
23829 (inline_read_section): Update.
23830 (inline_write_summary): Update.
23831 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23832 add account_size_time and reset member functions.
23833 (ipa_call_summary): Add reset function.
23834 * ipa-predicate.h (predicate::operator &): Constify.
23836 2017-05-22 Richard Biener <rguenther@suse.de>
23838 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23840 2017-05-19 Jason Merrill <jason@redhat.com>
23842 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23844 2017-05-19 Marek Polacek <polacek@redhat.com>
23847 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23848 TYPE_OVERFLOW_WRAPS checks.
23850 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23852 * tree-core.h (enum omp_clause_default_kind): Add
23853 "OMP_CLAUSE_DEFAULT_PRESENT".
23854 * tree-pretty-print.c (dump_omp_clause): Handle it.
23855 * gimplify.c (enum gimplify_omp_var_data): Add
23856 "GOVD_MAP_FORCE_PRESENT".
23857 (gimplify_adjust_omp_clauses_1): Map it to
23858 "GOMP_MAP_FORCE_PRESENT".
23859 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23861 * gimplify.c (oacc_default_clause): Clarify.
23863 2017-05-19 Nathan Sidwell <nathan@acm.org>
23865 LANG_HOOK_REGISTER_DUMPS
23866 * toplev.c (general_init): Call register dump lang hook.
23867 * doc/invoke.texi: Document -fdump-lang option family.
23868 * dumpfile.c (dump_files): Remove class dump here.
23869 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23870 * dumpfile.h (tree_dump_index): Remove TDI_class.
23871 * langhooks-def.h (lhd_register_dumps): Declare.
23872 (LANG_HOOKS_REGISTER_DUMPS): Define.
23873 (LANG_HOOKS_INITIALIZER): Add it.
23874 * langhooks.c (lhd_register_dumps): Define.
23875 * langhooks.h (struct lang_hooks): Add register_dumps.
23877 2017-05-19 Nathan Sidwell <nathan@acm.org>
23879 * context.h (context::set_passes): New.
23880 * context.c (context::context): Do not create pass manager.
23881 * toplev.c (general_init): Create pass manager here.
23883 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23885 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23886 use this splitter if two add or or instructions would also work for
23887 the constant we want to generate.
23889 2017-05-19 Richard Biener <rguenther@suse.de>
23892 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23893 predicate evaluation.
23895 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23897 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23899 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23902 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23904 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23905 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23906 (inline_edge_summary_vec): Turn into ...
23907 (ipa_call_summaries): ... this one.
23908 (redirect_to_unreachable, edge_set_predicate,
23909 evaluate_properties_for_edge, inline_summary_alloc,
23910 reset_ipa_call_summary, reset_inline_summary,
23911 inline_summary_t::duplicate): Update.
23912 (inline_edge_duplication_hook): Turn to ...
23913 (ipa_call_summary_t::duplicate): ... this one.
23914 (inline_edge_removal_hook): Turn to ...
23915 (ipa_call_summary_t::remove): ... this one.
23916 (dump_inline_edge_summary): Turn to ...
23917 (dump_ipa_call_summary): ... this one.
23918 (estimate_function_body_sizes): Update.
23919 (inline_update_callee_summaries): Update.
23920 (remap_edge_change_prob): Update.
23921 (remap_edge_summaries): Update.
23922 (inline_merge_summary): Update.
23923 (do_estimate_edge_time): Update.
23924 (inline_generate_summary): Update.
23925 (inline_read_section): Update.
23926 (inline_read_summary): Update.
23927 (inline_free_summary): Update.
23928 * ipa-inline.c (can_inline_edge_p): Update.
23929 (compute_inlined_call_time): Update.
23930 (want_inline_small_function_p): Update.
23931 (edge_badness): Update.
23932 (early_inliner): Update.
23933 * ipa-inline.h (inline_edge_summary): Turn to ...
23934 (ipa_call_summary): ... this one.
23935 (ipa_call_summary_t): New class.
23936 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23937 (ipa_call_summaries): New.
23938 (inline_edge_summary): Remove.
23939 (estimate_edge_growth): Update.
23940 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23941 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23942 * ipa-split.c (execute_split_functions): Update.
23943 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23945 2017-05-19 Richard Biener <rguenther@suse.de>
23947 PR middle-end/80764
23948 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23950 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23952 * config/rs6000/rs6000.c (struct machine_function): Add field
23953 fpr_is_wrapped_separately.
23954 (rs6000_get_separate_components): Use 64 components. Handle the
23955 new FPR components.
23956 (rs6000_components_for_bb): Handle the FPR components.
23957 (rs6000_emit_prologue_components): Handle the FPR components.
23958 (rs6000_emit_epilogue_components): Handle the FPR components.
23959 (rs6000_set_handled_components): Handle the FPR components.
23960 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23961 that are already separately shrink-wrapped.
23962 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23963 that are already separately shrink-wrapped.
23965 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23968 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23971 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23972 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23973 cases where the register allocator uses a move and an offsettable
23974 memory operation to/from a FPR register on ISA 2.06/2.07.
23975 (define_peephole2 for Altivec d-form store): Likewise.
23977 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23980 * config/i386/mmx.md (*mov<mode>_internal): Enable
23981 alternatives 11, 12, 13 and 14 also for 32bit targets.
23982 Remove alternatives 15, 16, 17 and 18.
23983 * config/i386/sse.md (vec_concatv2di): Change
23984 alternative (!x, *y) to (x, ?!*Yn).
23986 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23988 * dumpfile.h (enum dump_kind): Remove stray comma.
23990 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23992 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23993 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23994 predicate::num_conditions
23995 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23996 (CHANGED): turn into predicate::changed.
23997 (agg_position_info): Move to ipa-predicate.h
23998 (add_condition, predicate::add_clause, predicate::operator &=,
23999 predicate::or_with, predicate::evaluate, predicate::probability,
24000 dump_condition, dump_clause, predicate::dump,
24001 predicate::remap_after_duplication, predicate::remap_after_inlining,
24002 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24003 (evaluate_conditions_for_known_args): Update.
24004 (set_cond_stmt_execution_predicate): Update.
24005 * ipa-inline.h: Include ipa-predicate.h
24006 (condition, inline_param_summary, conditions, agg_position_info,
24007 predicate): Move to ipa-predicate.h
24008 * ipa-predicate.c: New file.
24009 * ipa-predicate.h: New file.
24011 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24013 * final.c (leaf_function_p): Check we are not in a sequence.
24015 2017-05-18 Martin Liska <mliska@suse.cz>
24017 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24018 * dumpfile.c (dump_register): Use new enum dump_kind.
24019 (get_dump_file_name): Likewise.
24020 (dump_enable_all): Likewise.
24021 (dump_switch_p_1): Likewise.
24022 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24023 * dumpfile.h (enum dump_kind): New enum type.
24024 (struct dump_file_info): Create constructor and
24025 format fields and comments.
24026 * passes.c (pass_manager::register_one_dump_file):
24028 * statistics.c (statistics_early_init): Likewise.
24029 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24030 TDF_TREE with TDF_SLIM.
24031 (gather_memory_references_ref): Likewise.
24033 2017-05-18 Martin Liska <mliska@suse.cz>
24035 * vec.h (struct vnull): Use it.
24037 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24039 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24040 (true_predicate, false_predicate, true_predicate_p,
24041 false_predicate_p): Remove.
24042 (single_cond_predicate, not_inlined_predicate): Turn to member function
24044 (add_condition): Update.
24045 (add_clause): Turn to...
24046 (predicate::add_clause): ... this one; update; allow passing NULL
24048 (and_predicates): Turn to ...
24049 (predicate::operator &=): ... this one.
24050 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24051 (or_predicates): Turn to ...
24052 (predicate::or_with): ... this one.
24053 (evaluate_predicate): Turn to ...
24054 (predicate::evaluate): ... this one.
24055 (predicate_probability): Turn to ...
24056 (predicate::probability): ... this one.
24057 (dump_condition): Update.
24058 (dump_predicate): Turn to ...
24059 (predicate::dump): ... this one.
24060 (account_size_time): Update.
24061 (edge_set_predicate): Update.
24062 (set_hint_predicate): UPdate.
24063 (evaluate_conditions_for_known_args): Update.
24064 (evaluate_properties_for_edge): Update.
24065 (remap_predicate_after_duplication): Turn to...
24066 (predicate::remap_after_duplication): ... this one.
24067 (remap_hint_predicate_after_duplication): Update.
24068 (inline_summary_t::duplicate): UPdate.
24069 (dump_inline_edge_summary): Update.
24070 (dump_inline_summary): Update.
24071 (set_cond_stmt_execution_predicate): Update.
24072 (set_switch_stmt_execution_predicate): Update.
24073 (compute_bb_predicates): Update.
24074 (will_be_nonconstant_expr_predicate): Update.
24075 (will_be_nonconstant_predicate): Update.
24076 (phi_result_unknown_predicate): Update.
24077 (predicate_for_phi_result): Update.
24078 (array_index_predicate): Update.
24079 (estimate_function_body_sizes): Update.
24080 (estimate_node_size_and_time): Update.
24081 (estimate_ipcp_clone_size_and_time): Update.
24082 (remap_predicate): Rename to ...
24083 (predicate::remap_after_inlining): ... this one.
24084 (remap_hint_predicate): Update.
24085 (inline_merge_summary): Update.
24086 (inline_update_overall_summary): Update.
24087 (estimate_size_after_inlining): Update.
24088 (read_predicate): Rename to ...
24089 (predicate::stream_in): ... this one.
24090 (read_inline_edge_summary): Update.
24091 (write_predicate): Rename to ...
24092 (predicate::stream_out): ... this one.
24093 (write_inline_edge_summary): Update.
24094 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24095 (clause_t): Turn to uint32_t
24096 (predicate): Turn to class; implement constructor and operators
24098 (size_time_entry): Update.
24099 (inline_summary): Update.
24100 (inline_edge_summary): Update.
24102 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24104 * fold-const.c (fold_binary_loc): Move transformation...
24105 * match.pd (C - X CMP X): ... here.
24107 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24109 * config/sparc/sparc.c (sparc_option_override): Set function
24110 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24111 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24113 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24115 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24117 2017-05-18 Marek Polacek <polacek@redhat.com>
24120 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24121 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24122 gimple_assign_load_p.
24124 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24126 PR middle-end/80692
24127 * real.c (do_compare): Give decimal_do_compare preference over
24128 comparing just the signs.
24130 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24132 * doc/md.texi (Canonicalization of Instructions): Describe the
24133 canonical form of instructions that inherently set a condition
24136 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24138 PR middle-end/80775
24139 * tree-cfg.c: Move deletion of unreachable case statements to after
24140 the merging of consecutive case labels.
24142 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24144 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24145 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24146 restoring of callee-saved registers.
24148 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24150 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24151 * config/visium/visium.c (single_set_and_flags): Likewise.
24152 * config/visium/visium.md (Substitutions): Likewise.
24154 2017-05-17 Martin Liska <mliska@suse.cz>
24156 * cfg.c: Introduce dump_flags_t type and
24157 use it instead of int type.
24159 * cfghooks.c: Likewise.
24160 * cfghooks.h (struct cfg_hooks): Likewise.
24161 * cfgrtl.c: Likewise.
24162 * cfgrtl.h: Likewise.
24163 * cgraph.c (cgraph_node::get_body): Likewise.
24164 * coretypes.h: Likewise.
24165 * domwalk.c: Likewise.
24166 * domwalk.h: Likewise.
24167 * dumpfile.c (struct dump_option_value_info): Likewise.
24168 (dump_enable_all): Likewise.
24169 (dump_switch_p_1): Likewise.
24170 (opt_info_switch_p): Likewise.
24171 * dumpfile.h (enum tree_dump_index): Likewise.
24172 (struct dump_file_info): Likewise.
24173 * genemit.c: Likewise.
24174 * generic-match-head.c: Likewise.
24175 * gengtype.c (open_base_files): Likewise.
24176 * gimple-pretty-print.c: Likewise.
24177 * gimple-pretty-print.h: Likewise.
24178 * graph.c (print_graph_cfg): Likewise.
24179 * graphite-scop-detection.c (dot_all_sese): Likewise.
24180 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24181 * loop-unroll.c (report_unroll): Likewise.
24182 * passes.c (pass_manager::register_one_dump_file): Likewise.
24183 * print-tree.c: Likewise.
24184 * statistics.c: Likewise.
24185 * tree-cfg.c: Likewise.
24186 * tree-cfg.h: Likewise.
24187 * tree-dfa.c: Likewise.
24188 * tree-dfa.h: Likewise.
24189 * tree-dump.c (dump_function): Likewise.
24190 * tree-dump.h (struct dump_info): Likewise.
24191 * tree-pretty-print.c: Likewise.
24192 * tree-pretty-print.h: Likewise.
24193 * tree-ssa-live.c: Likewise.
24194 * tree-ssa-live.h: Likewise.
24195 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24196 * tree-vect-loop.c: Likewise.
24197 * tree-vect-slp.c: Likewise.
24199 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24200 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24202 PR tree-optimization/80457
24203 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24204 of all arguments to a statement as scalar_to_vec operations.
24205 (vectorizable_call): Adjust call to vect_model_simple_cost for
24207 (vectorizable_conversion): Likewise.
24208 (vectorizable_assignment): Likewise.
24209 (vectorizable_shift): Likewise.
24210 (vectorizable_operation): Likewise.
24211 (vectorizable_comparison): Likewise.
24212 (vect_is_simple_cond): Record the def types for operands.
24213 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24214 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24215 for statement argument count.
24217 2017-05-16 Carl Love <cel@us.ibm.com>
24219 * config/rs6000/rs6000-c: Add support for built-in functions
24220 vector unsigned long long vec_bperm (vector unsigned long long,
24221 vector unsigned char)
24222 vector signed long long vec_mule (vector signed int,
24224 vector unsigned long long vec_mule (vector unsigned int,
24225 vector unsigned int)
24226 vector signed long long vec_mulo (vector signed int,
24228 vector unsigned long long vec_mulo (vector unsigned int,
24229 vector unsigned int)
24230 vector signed char vec_sldw (vector signed char,
24231 vector signed char,
24233 vector unsigned char vec_sldw (vector unsigned char,
24234 vector unsigned char,
24236 vector signed short vec_sldw (vector signed short,
24237 vector signed short,
24239 vector unsigned short vec_sldw (vector unsigned short,
24240 vector unsigned short,
24242 vector signed int vec_sldw (vector signed int,
24245 vector unsigned int vec_sldw (vector unsigned int,
24246 vector unsigned int,
24248 vector signed long long vec_sldw (vector signed long long,
24249 vector signed long long,
24251 vector unsigned long long vec_sldw (vector unsigned long long,
24252 vector unsigned long long,
24254 * config/rs6000/rs6000-c: Add support for built-in functions
24255 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24256 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24257 * doc/extend.texi: Update the built-in documentation for the
24258 new built-in functions.
24260 2017-05-16 Marek Polacek <polacek@redhat.com>
24264 * tree.c (save_expr): Don't fold the expression.
24266 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24268 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24269 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24270 and (?*y,m). Update insn attributes.
24272 2017-05-16 Martin Liska <mliska@suse.cz>
24274 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24275 flags argument of print_gimple_stmt, print_gimple_expr,
24276 print_generic_stmt and print_generic_expr.
24277 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24278 * coretypes.h: Likewise.
24279 * except.c (dump_eh_tree): Likewise.
24280 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24281 * gimple-pretty-print.h: Likewise.
24282 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24283 (backprop::push_to_worklist): Likewise.
24284 (backprop::pop_from_worklist): Likewise.
24285 (backprop::process_use): Likewise.
24286 (backprop::intersect_uses): Likewise.
24287 (note_replacement): Likewise.
24288 * gimple-ssa-store-merging.c
24289 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24290 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24291 (pass_store_merging::execute): Likewise.
24292 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24293 (ssa_base_cand_dump_callback): Likewise.
24294 (dump_incr_vec): Likewise.
24295 (replace_refs): Likewise.
24296 (replace_mult_candidate): Likewise.
24297 (create_add_on_incoming_edge): Likewise.
24298 (create_phi_basis): Likewise.
24299 (insert_initializers): Likewise.
24300 (all_phi_incrs_profitable): Likewise.
24301 (introduce_cast_before_cand): Likewise.
24302 (replace_one_candidate): Likewise.
24303 * gimplify.c (gimplify_expr): Likewise.
24304 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24305 (set_rename): Likewise.
24306 (rename_uses): Likewise.
24307 (copy_loop_phi_nodes): Likewise.
24308 (add_close_phis_to_merge_points): Likewise.
24309 (copy_loop_close_phi_args): Likewise.
24310 (copy_cond_phi_args): Likewise.
24311 (graphite_copy_stmts_from_block): Likewise.
24312 (translate_pending_phi_nodes): Likewise.
24313 * graphite-poly.c (print_pdr): Likewise.
24314 (dump_gbb_cases): Likewise.
24315 (dump_gbb_conditions): Likewise.
24316 (print_scop_params): Likewise.
24317 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24318 (build_cross_bb_scalars_use): Likewise.
24319 (gather_bbs::before_dom_children): Likewise.
24320 * hsa-dump.c (dump_hsa_immed): Likewise.
24321 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24322 (get_replacement_map): Likewise.
24323 * ipa-inline-analysis.c (dump_condition): Likewise.
24324 (estimate_function_body_sizes): Likewise.
24325 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24326 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24327 * ipa-prop.c (ipa_dump_param): Likewise.
24328 (ipa_print_node_jump_functions_for_edge): Likewise.
24329 (ipa_modify_call_arguments): Likewise.
24330 (ipa_modify_expr): Likewise.
24331 (ipa_dump_param_adjustments): Likewise.
24332 (ipa_dump_agg_replacement_values): Likewise.
24333 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24334 * ipa-pure-const.c (check_stmt): Likewise.
24335 (pass_nothrow::execute): Likewise.
24336 * ipa-split.c (execute_split_functions): Likewise.
24337 * omp-offload.c (dump_oacc_loop_part): Likewise.
24338 (dump_oacc_loop): Likewise.
24339 * trans-mem.c (tm_log_emit): Likewise.
24340 (tm_memopt_accumulate_memops): Likewise.
24341 (dump_tm_memopt_set): Likewise.
24342 (dump_tm_memopt_transform): Likewise.
24343 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24344 (print_loop): Likewise.
24345 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24346 (chkp_gather_checks_info): Likewise.
24347 (chkp_get_check_result): Likewise.
24348 (chkp_remove_check_if_pass): Likewise.
24349 (chkp_use_outer_bounds_if_possible): Likewise.
24350 (chkp_reduce_bounds_lifetime): Likewise.
24351 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24352 (chkp_mark_completed_bounds): Likewise.
24353 (chkp_register_incomplete_bounds): Likewise.
24354 (chkp_mark_invalid_bounds): Likewise.
24355 (chkp_maybe_copy_and_register_bounds): Likewise.
24356 (chkp_build_returned_bound): Likewise.
24357 (chkp_get_bound_for_parm): Likewise.
24358 (chkp_build_bndldx): Likewise.
24359 (chkp_get_bounds_by_definition): Likewise.
24360 (chkp_generate_extern_var_bounds): Likewise.
24361 (chkp_get_bounds_for_decl_addr): Likewise.
24362 * tree-chrec.c (chrec_apply): Likewise.
24363 * tree-data-ref.c (dump_data_reference): Likewise.
24364 (dump_subscript): Likewise.
24365 (dump_data_dependence_relation): Likewise.
24366 (analyze_overlapping_iterations): Likewise.
24367 * tree-inline.c (expand_call_inline): Likewise.
24368 (tree_function_versioning): Likewise.
24369 * tree-into-ssa.c (dump_defs_stack): Likewise.
24370 (dump_currdefs): Likewise.
24371 (dump_names_replaced_by): Likewise.
24372 (dump_update_ssa): Likewise.
24373 (update_ssa): Likewise.
24374 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24375 * tree-parloops.c (build_new_reduction): Likewise.
24376 (try_create_reduction_list): Likewise.
24377 (ref_conflicts_with_region): Likewise.
24378 (oacc_entry_exit_ok_1): Likewise.
24379 (oacc_entry_exit_single_gang): Likewise.
24380 * tree-pretty-print.h: Likewise.
24381 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24382 (get_scalar_evolution): Likewise.
24383 (add_to_evolution): Likewise.
24384 (get_loop_exit_condition): Likewise.
24385 (analyze_evolution_in_loop): Likewise.
24386 (analyze_initial_condition): Likewise.
24387 (analyze_scalar_evolution): Likewise.
24388 (instantiate_scev): Likewise.
24389 (number_of_latch_executions): Likewise.
24390 (gather_chrec_stats): Likewise.
24391 (final_value_replacement_loop): Likewise.
24392 (scev_const_prop): Likewise.
24393 * tree-sra.c (dump_access): Likewise.
24394 (disqualify_candidate): Likewise.
24395 (create_access): Likewise.
24396 (reject): Likewise.
24397 (maybe_add_sra_candidate): Likewise.
24398 (create_access_replacement): Likewise.
24399 (analyze_access_subtree): Likewise.
24400 (analyze_all_variable_accesses): Likewise.
24401 (sra_modify_assign): Likewise.
24402 (initialize_constant_pool_replacements): Likewise.
24403 (find_param_candidates): Likewise.
24404 (decide_one_param_reduction): Likewise.
24405 (replace_removed_params_ssa_names): Likewise.
24406 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24407 * tree-ssa-copy.c (dump_copy_of): Likewise.
24408 (copy_prop_visit_cond_stmt): Likewise.
24409 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24410 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24411 (record_equivalences_from_stmt): Likewise.
24412 * tree-ssa-dse.c (compute_trims): Likewise.
24413 (delete_dead_call): Likewise.
24414 (delete_dead_assignment): Likewise.
24415 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24416 (forward_propagate_into_cond): Likewise.
24417 (pass_forwprop::execute): Likewise.
24418 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24419 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24421 (move_computations_worker): Likewise.
24422 (execute_sm): Likewise.
24423 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24424 (remove_exits_and_undefined_stmts): Likewise.
24425 (remove_redundant_iv_tests): Likewise.
24426 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24427 (adjust_iv_update_pos): Likewise.
24428 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24429 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24430 (value_replacement): Likewise.
24431 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24432 * tree-ssa-pre.c (print_pre_expr): Likewise.
24433 (get_representative_for): Likewise.
24434 (create_expression_by_pieces): Likewise.
24435 (insert_into_preds_of_block): Likewise.
24436 (eliminate_insert): Likewise.
24437 (eliminate_dom_walker::before_dom_children): Likewise.
24438 (eliminate): Likewise.
24439 (remove_dead_inserted_code): Likewise.
24440 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24441 * tree-ssa-reassoc.c (get_rank): Likewise.
24442 (eliminate_duplicate_pair): Likewise.
24443 (eliminate_plus_minus_pair): Likewise.
24444 (eliminate_not_pairs): Likewise.
24445 (undistribute_ops_list): Likewise.
24446 (eliminate_redundant_comparison): Likewise.
24447 (update_range_test): Likewise.
24448 (optimize_range_tests_var_bound): Likewise.
24449 (optimize_vec_cond_expr): Likewise.
24450 (rewrite_expr_tree): Likewise.
24451 (rewrite_expr_tree_parallel): Likewise.
24452 (linearize_expr): Likewise.
24453 (break_up_subtract): Likewise.
24454 (linearize_expr_tree): Likewise.
24455 (attempt_builtin_powi): Likewise.
24456 (attempt_builtin_copysign): Likewise.
24457 (transform_stmt_to_copy): Likewise.
24458 (transform_stmt_to_multiply): Likewise.
24459 (dump_ops_vector): Likewise.
24460 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24461 (print_scc): Likewise.
24462 (set_ssa_val_to): Likewise.
24463 (visit_reference_op_store): Likewise.
24464 (visit_use): Likewise.
24465 (sccvn_dom_walker::before_dom_children): Likewise.
24466 (run_scc_vn): Likewise.
24467 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24469 (expr_hash_elt::print): Likewise.
24470 (const_and_copies::pop_to_marker): Likewise.
24471 (const_and_copies::record_const_or_copy_raw): Likewise.
24472 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24473 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24474 (dump_predicates): Likewise.
24475 (find_uninit_use): Likewise.
24476 (warn_uninitialized_phi): Likewise.
24477 (pass_late_warn_uninitialized::execute): Likewise.
24478 * tree-ssa.c (verify_vssa): Likewise.
24479 (verify_ssa): Likewise.
24480 (maybe_optimize_var): Likewise.
24481 * tree-vrp.c (dump_value_range): Likewise.
24482 (dump_all_value_ranges): Likewise.
24483 (dump_asserts_for): Likewise.
24484 (register_edge_assert_for_2): Likewise.
24485 (vrp_visit_cond_stmt): Likewise.
24486 (vrp_visit_switch_stmt): Likewise.
24487 (vrp_visit_stmt): Likewise.
24488 (vrp_visit_phi_node): Likewise.
24489 (simplify_cond_using_ranges_1): Likewise.
24490 (fold_predicate_in): Likewise.
24491 (evrp_dom_walker::before_dom_children): Likewise.
24492 (evrp_dom_walker::push_value_range): Likewise.
24493 (evrp_dom_walker::pop_value_range): Likewise.
24494 (execute_early_vrp): Likewise.
24496 2017-05-16 Richard Biener <rguenther@suse.de>
24498 * dwarf2out.c (loc_list_from_tree_1): Do not create
24499 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24501 2017-05-16 Richard Biener <rguenther@suse.de>
24503 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24505 (note_variable_value_in_expr): If we resolved the decl ref
24506 do not push to the stack.
24508 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24510 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24511 operations in fast-math mode.
24512 (vaddq_f16): Likewise.
24513 (vmul_f16): Likewise.
24514 (vmulq_f16): Likewise.
24515 (vsub_f16): Likewise.
24516 (vsubq_f16): Likewise.
24517 * config/arm/neon.md (add<mode>3): New.
24519 (fma:<VH:mode>3): New. Also remove outdated comment.
24522 2017-05-16 Martin Liska <mliska@suse.cz>
24526 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24527 (odr_types_equivalent_p): Likewise.
24529 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24531 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24533 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24536 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24537 non-interunit SSE move alternatives with '?'.
24538 (zero-extendsidi peephole2): New peephole to skip intermediate
24539 general register in SSE zero-extend sequence.
24541 2017-05-15 Jeff Law <law@redhat.com>
24543 * reorg.c (relax_delay_slots): Create a new variable to hold
24544 the temporary target rather than clobbering TARGET_LABEL.
24546 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24547 missing argument to extract_bit_field call.
24548 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24550 2017-05-15 Martin Liska <mliska@suse.cz>
24553 * gcc.c (process_command): Do not allow empty argument of -o option.
24555 2017-05-15 Renlin Li <renlin.li@arm.com>
24557 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24558 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24559 * config/aarch64/constraints.md (Usf): Add long call check.
24560 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24561 (call_value): Likewise.
24562 (sibcall): Likewise.
24563 (sibcall_value): Likewise.
24565 (call_value_insn): New.
24566 (sibcall_insn): Update rtx pattern.
24567 (sibcall_value_insn): Likewise.
24568 (call_internal): Remove.
24569 (call_value_internal): Likewise.
24570 (sibcall_internal): Likewise.
24571 (sibcall_value_internal): Likewise.
24572 (call_reg): Likewise.
24573 (call_symbol): Likewise.
24574 (call_value_reg): Likewise.
24575 (call_value_symbol): Likewise.
24577 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24580 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24582 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24584 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24585 compatible with CCGOCmode and with CCZmode.
24587 2017-05-14 Martin Sebor <msebor@redhat.com>
24589 PR middle-end/77671
24590 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24591 (gimple_fold_builtin_snprintf): Same.
24592 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24593 (gimple_fold_builtin_snprintf): Same.
24594 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24595 of character types.
24596 (is_call_safe): New function.
24597 (try_substitute_return_value): Call it.
24598 (try_simplify_call): New function.
24599 (pass_sprintf_length::handle_gimple_call): Call it.
24601 2017-05-14 Martin Sebor <msebor@redhat.com>
24603 PR middle-end/80669
24604 * builtins.c (expand_builtin_stpncpy): Simplify.
24606 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24608 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24609 * config/i386/i386.h
24610 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24611 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24612 (struct machine_function): Add new members call_ms2sysv,
24613 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24614 (struct machine_frame_state): New fields sp_realigned and
24615 sp_realigned_offset.
24616 * config/i386/i386.c
24617 (enum xlogue_stub): New enum.
24618 (enum xlogue_stub_sets): New enum.
24619 (class xlogue_layout): New class.
24620 (struct ix86_frame): New fields stack_realign_allocate_offset,
24621 stack_realign_offset and outlined_save_offset. Modify comments to
24622 detail stack layout when using out-of-line stubs.
24623 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24624 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24625 -mcall-ms2sysv-xlogues.
24626 (stub_managed_regs): New static variable.
24627 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24628 registers managed by out-of-line stub.
24629 (disable_call_ms2sysv_xlogues): New function.
24630 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24631 m->call_ms2sysv when appropriate and compute frame layout for
24633 (sp_valid_at, fp_valid_at): New inline functions.
24634 (choose_basereg): New function.
24635 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24637 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24638 Use align parameter of choose_baseaddr to generated aligned SSE movs
24640 (pro_epilogue_adjust_stack): Modify to track
24641 machine_frame_state::sp_realigned.
24642 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24643 (ix86_nsaved_sseregs): Likewise.
24644 (ix86_emit_save_regs): Likewise.
24645 (ix86_emit_save_regs_using_mov): Likewise.
24646 (ix86_emit_save_sse_regs_using_mov): Likewise.
24647 (get_scratch_register_on_entry): Likewise.
24648 (gen_frame_set): New function.
24649 (gen_frame_load): Likewise.
24650 (gen_frame_store): Likewise.
24651 (emit_outlined_ms2sysv_save): Likewise.
24652 (emit_outlined_ms2sysv_restore): Likewise.
24653 (ix86_expand_prologue): Modify stack re-alignment code and call
24654 emit_outlined_ms2sysv_save when appropriate.
24655 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24656 parameter rtx_insn *insn, which allows the function to be used to only
24657 generate the notes.
24658 (ix86_expand_epilogue): Modify validity checks of frame and stack
24659 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24660 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24661 * config/i386/predicates.md
24662 (save_multiple): New predicate.
24663 (restore_multiple): Likewise.
24664 * config/i386/sse.md
24665 (save_multiple<mode>): New pattern.
24666 (save_multiple_realign<mode>): Likewise.
24667 (restore_multiple<mode>): Likewise.
24668 (restore_multiple_and_return<mode>): Likewise.
24669 (restore_multiple_leave_return<mode>): Likewise.
24670 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24672 2017-05-14 Julia Koval <julia.koval@intel.com>
24674 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24675 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24676 (__builtin_ia32_xsetbv): New builtins.
24677 * config/i386/i386.c (ix86_expand_special_args_builtin):
24679 (ix86_expand_builtin): Special expand for new intrinsics.
24680 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24681 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24682 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24684 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24686 * cfganal.c (inverted_post_order_compute): Change argument type
24688 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24689 * df-core.c (rest_of_handle_df_initialize): Adjust.
24690 (rest_of_handle_df_finish): Likewise.
24691 (df_analyze_1): Likewise.
24692 (df_analyze): Likewise.
24693 (loop_inverted_post_order_compute): Change argument to be a vec *.
24694 (df_analyze_loop): Adjust.
24695 (df_get_n_blocks): Likewise.
24696 (df_get_postorder): Likewise.
24697 * df.h (struct df_d): Change field to be a vec.
24698 * lcm.c (compute_laterin): Adjust.
24699 (compute_available): Likewise.
24700 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24701 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24702 * tree-ssa-pre.c (compute_antic): Likewise.
24704 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24706 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24707 (depth_first_search::depth_first_search): Change structure init
24708 function to this constructor.
24709 (depth_first_search::add_bb): Rename function to this member.
24710 (depth_first_search::execute): Likewise.
24711 (flow_dfs_compute_reverse_finish): Adjust.
24713 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24715 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24716 (longest_simple_path): Likewise.
24717 * shrink-wrap.c (spread_components): Likewise.
24718 (disqualify_problematic_components): Likewise.
24719 (emit_common_heads_for_components): Likewise.
24720 (emit_common_tails_for_components): Likewise.
24721 (insert_prologue_epilogue_for_components): Likewise.
24723 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24725 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24728 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24730 * df-core.c (df_set_blocks): Start using auto_bitmap.
24731 (df_compact_blocks): Likewise.
24732 * df-problems.c (df_rd_confluence_n): Likewise.
24733 * df-scan.c (df_insn_rescan_all): Likewise.
24734 (df_process_deferred_rescans): Likewise.
24735 (df_update_entry_block_defs): Likewise.
24736 (df_update_exit_block_uses): Likewise.
24737 (df_entry_block_bitmap_verify): Likewise.
24738 (df_exit_block_bitmap_verify): Likewise.
24739 (df_scan_verify): Likewise.
24740 * lra-constraints.c (lra_constraints): Likewise.
24741 (undo_optional_reloads): Likewise.
24742 (lra_undo_inheritance): Likewise.
24743 * lra-remat.c (calculate_gen_cands): Likewise.
24744 (do_remat): Likewise.
24745 * lra-spills.c (assign_spill_hard_regs): Likewise.
24746 (spill_pseudos): Likewise.
24747 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24748 (bitmap_set_subtract_values): Likewise.
24750 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24752 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24753 management with auto_bitmap.
24754 (fix_inter_tick): Likewise.
24755 (fix_recovery_deps): Likewise.
24756 * ira.c (add_store_equivs): Likewise.
24757 (find_moveable_pseudos): Likewise.
24758 (split_live_ranges_for_shrink_wrap): Likewise.
24759 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24760 (rtx_reuse_manager::seen_def_p): Likewise.
24761 (rtx_reuse_manager::set_seen_def): Likewise.
24762 * print-rtl.h (class rtx_reuse_manager): Likewise.
24764 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24766 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24768 (migrate_btr_def): Likewise.
24769 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24770 * df-core.c (loop_post_order_compute): Likewise.
24771 (loop_inverted_post_order_compute): Likewise.
24772 * hsa-common.h: Likewise.
24773 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24774 * init-regs.c (initialize_uninitialized_regs): Likewise.
24775 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24776 (inline_small_functions): Likewise.
24777 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24778 * ira.c (combine_and_move_insns): Likewise.
24779 (build_insn_chain): Likewise.
24780 * loop-invariant.c (find_invariants): Likewise.
24781 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24782 * predict.c (tree_predict_by_opcode): Likewise.
24783 (predict_paths_leading_to): Likewise.
24784 (predict_paths_leading_to_edge): Likewise.
24785 (estimate_loops_at_level): Likewise.
24786 (estimate_loops): Likewise.
24787 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24788 (spread_components): Likewise.
24789 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24790 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24791 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24792 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24793 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24794 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24795 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24796 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24797 (mark_threaded_blocks): Likewise.
24798 (thread_through_all_blocks): Likewise.
24799 * tree-ssa.c (verify_ssa): Likewise.
24800 (execute_update_addresses_taken): Likewise.
24801 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24803 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24805 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24807 (post_order_compute): Likewise.
24808 (inverted_post_order_compute): Likewise.
24809 (pre_and_rev_post_order_compute_fn): Likewise.
24811 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24813 * genrecog.c (int_set::int_set): Explicitly construct our
24814 auto_vec base class.
24815 * vec.h (auto_vec::auto_vec): New constructor.
24817 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24819 * bitmap.h (class auto_bitmap): New constructor taking
24820 bitmap_obstack * argument.
24822 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24824 * bitmap.h (class auto_bitmap): Change type of m_bits to
24825 bitmap_head, and adjust ctor / dtor and member operators.
24827 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24829 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24830 when returned register mode doesn't match original mode.
24832 2017-05-12 Jeff Law <law@redhat.com>
24833 Jakub Jelinek <jakub@redhat.com>
24835 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24836 we look for cc setter after the compare-elim changes.
24837 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24838 within the vector to match what compare-elim now expects.
24839 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24840 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24842 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24843 after the compare-elim changes.
24844 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24845 the vector to match what compare-elim now expects.
24846 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24847 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24848 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24849 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24850 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24852 * config/visium/visium.c (single_set_and_flags): Fix where
24853 we look for cc setter after the compare-elim changes.
24854 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24855 with the vector to match what compare-elim now expects.
24856 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24857 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24858 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24859 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24860 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24861 (neg<mode>2_insn_set_overflow): Likewise.
24863 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24865 PR middle-end/79794
24866 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24867 maybe_expand_insn call, set ops[0].target. If still set after call,
24868 set alt_rtl. Add extra arg to recursive calls.
24869 (extract_bit_field): Add alt_rtl argument. Pass to
24871 * expmed.h (extract_bit_field): Fix prototype.
24872 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24873 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24874 to extract_bit_field_calls.
24875 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24876 Pass alt_rtl to extract_bit_field calls.
24877 * calls.c (store_unaligned_arguments_into_psuedos)
24878 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24879 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24881 * optabs.h (struct expand_operand): Add target bitfield.
24883 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24885 * compare-elim.c (try_eliminate_compare): Canonicalize
24886 operation with embedded compare to
24887 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24888 (set (reg) (operation)].
24890 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24892 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24895 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24896 cost of adding a carry flag for ADC instruction.
24897 [case MINUS]: Ignore the cost of subtracting a carry flag
24898 for SBB instruction.
24900 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24902 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24904 * config/rs6000/bmiintrin.h: New file.
24905 * config/rs6000/bmi2intrin.h: New file.
24906 * config/rs6000/x86intrin.h: New file.
24908 2017-05-12 Jeff Law <law@redhat.com>
24910 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24913 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24915 PR middle-end/80707
24916 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24918 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24920 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24921 early expansion of vector divide builtins.
24922 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24923 builtins identified as having unsigned arguments.
24925 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24927 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24928 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24929 expansion of vector logical operations (and, andc, or, xor,
24932 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24934 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24935 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24937 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24939 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24940 early GIMPLE expansion of vector multiplies.
24942 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24944 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24945 TARGET_HAVE_MOVT conditional.
24946 (movt splitter): Likewise.
24948 2017-05-12 Richard Biener <rguenther@suse.de>
24950 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24951 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24952 Fold all stmts not inplace.
24954 2017-05-12 Richard Biener <rguenther@suse.de>
24956 PR tree-optimization/80713
24957 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24958 inserted_exprs bit for not removed stmts.
24960 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24962 PR middle-end/69921
24963 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24964 parallelized" attribute for parallelized OpenACC kernels.
24965 * omp-offload.c (execute_oacc_device_lower): Use it.
24967 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24968 Set "oacc kernels" attribute.
24969 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24970 parameter. Adjust all users.
24971 (oacc_fn_attrib_kernels_p): Remove function.
24972 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24973 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24974 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24975 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24976 assert "oacc kernels" attribute is set.
24978 2017-05-11 Carl Love <cel@us.ibm.com>
24980 * config/rs6000/rs6000-c: Add support for built-in functions
24981 vector unsigned char vec_popcnt (vector signed char)
24982 vector unsigned char vec_popcnt (vector unsigned char)
24983 vector unsigned short vec_popcnt (vector signed short)
24984 vector unsigned short vec_popcnt (vector unsigned short)
24985 vector unsigned int vec_popcnt (vector signed int)
24986 vector unsigned int vec_popcnt (vector unsigned int)
24987 vector unsigned long long vec_popcnt (vector signed long long)
24988 vector unsigned long long vec_popcnt (vector unsigned long long)
24989 vector signed long long vec_slo (vector signed long long,
24990 vector signed char)
24991 vector signed long long vec_slo (vector signed long long,
24992 vector unsigned char)
24993 vector unsigned long long vec_slo (vector unsigned long long,
24994 vector signed char)
24995 vector unsigned long long vec_slo (vector unsigned long long,
24996 vector unsigned char)
24997 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24998 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24999 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25000 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25001 * doc/extend.texi: Update the built-in documentation file for the
25002 new built-in functions.
25004 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25006 * attribs.h (sorted_attr_string): Move machine independent
25007 functions for target clone support from the i386 port to common
25008 code. Rename ix86_function_versions to common_function_versions.
25009 Rename make_name to make_unique_name.
25010 (common_function_versions): Likewise.
25011 (make_unique_name): Likewise.
25012 (make_dispatcher_decl): Likewise.
25013 (is_function_default_version): Likewise.
25014 * attribs.c (attr_strcmp): Likewise.
25015 (sorted_attr_string): Likewise.
25016 (common_function_versions): Likewise.
25017 (make_unique_name): Likewise.
25018 (make_dispatcher_decl): Likewise.
25019 (is_function_default_version): Likewise.
25020 * config/i386/i386.c (attr_strcmp): Likewise.
25021 (sorted_attr_string): Likewise.
25022 (ix86_function_versions): Likewise.
25023 (make_name): Likewise.
25024 (make_dispatcher_decl): Likewise.
25025 (is_function_default_version): Likewise.
25026 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25028 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25031 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25032 Account for direct move costs for vec_construct of integer
25035 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25038 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25039 (UNSPEC_STX_ATOMIC): Ditto.
25040 (loaddi_via_sse): New insn.
25041 (storedi_via_sse): Ditto.
25042 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25043 Update corresponding peephole2 patterns.
25044 (atomic_storedi_fpu): Ditto.
25046 2017-05-11 Julia Koval <julia.koval@intel.com>
25048 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25049 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25051 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25052 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25053 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25055 2017-05-11 Nathan Sidwell <nathan@acm.org>
25057 * graphite-poly.c: Include dumpfile.h.
25059 * dumpfle.h (dump_function): Declare here ...
25060 * tree-dump.h (dump_function): ... not here.
25061 * dumpfile.c: #include tree-cfg.h.
25062 (dump_function): Move here from ...
25063 * tree-dump.c (dump_function): ... here.
25064 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25065 * graphite-poly.c: Don't include tree-dump.h.
25066 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25067 * print-tree.c: Likewise.
25068 * stor-layout.c: Likewise.
25069 * tree-nested.c: Likewise.
25071 * dumpfile.c (dump_start): Use TDF_FLAGS.
25072 (dump_enable_all): Fix TDF_KIND check thinko.
25074 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25076 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25077 array entries to represent two legal parameterizations of the
25078 overloaded __builtin_cmpb function, as represented by the
25079 P6_OV_BUILTIN_CMPB constant.
25080 (altivec_resolve_overloaded_builtin): Add special case handling
25081 for the __builtin_cmpb function, as represented by the
25082 P6_OV_BUILTIN_CMPB constant.
25083 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25084 (BU_P6_64BIT_2): New macro.
25085 (BU_P6_OVERLOAD_2): New macro
25086 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25087 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25088 (CMPB): Add overload support to represent both 32-bit and 64-bit
25089 compare-bytes function.
25090 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25091 support for TARGET_CMPB.
25092 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25093 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25094 documentation of the __builtin_cmpb overloaded built-in function.
25096 2017-05-11 Richard Biener <rguenther@suse.de>
25098 PR tree-optimization/80705
25099 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25100 bases are not vectorizable.
25102 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25104 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25105 when counting register pressure.
25107 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25109 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25110 (iv_ca_more_deps): Renamed to ...
25111 (iv_ca_compare_deps): ... this.
25112 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25114 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25116 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25118 (determine_group_iv_costs): ... here.
25119 (find_inv_vars_cb): Record inv var if it's not recorded before.
25121 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25123 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25124 (get_shiftadd_cost): Ditto.
25126 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25128 * tree-ssa-address.c: Include header file.
25129 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25131 (add_to_parts): Refactor.
25132 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25133 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25136 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25138 PR tree-optimization/53090
25139 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25141 (extract_cond_operands): Detect condition with IV on both sides
25142 and return COMP_IV_EXPR_2.
25143 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25144 (rewrite_use_compare): Simplify by removing call to function
25145 extract_cond_operands.
25147 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25149 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25150 (extract_cond_operands): Detect condition comparing against non-
25151 invariant bound and return appropriate enum value.
25152 (find_interesting_uses_cond): Update use of extract_cond_operands.
25153 Handle its return value accordingly.
25154 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25156 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25158 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25159 nonlinear iv_use computation in loop invariant sensitive way.
25161 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25163 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25164 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25166 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25168 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25169 (dump_cand): Support iv_cand.inv_exprs.
25170 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25172 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25175 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25177 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25179 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25180 as local function. Include necessary header files.
25181 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25183 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25185 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25187 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25189 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25190 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25191 RSHIFT_EXPR and BIT_NOT_EXPR.
25193 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25195 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25196 (adjust_setup_cost): New parameter supporting round up adjustment.
25197 (struct address_cost_data): Delete.
25198 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25199 (split_address_cost, ptr_difference_cost): Delete.
25200 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25201 (struct ainc_cost_data): New struct.
25202 (get_address_cost_ainc): New function.
25203 (get_address_cost, get_computation_cost): Reimplement.
25204 (determine_group_iv_cost_address): Record inv_expr for all uses of
25206 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25207 (iv_ca_has_deps): Reimplemented to ...
25208 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25210 (iv_ca_extend): Call iv_ca_more_deps.
25212 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25214 * tree-ssa-address.c (struct mem_address): Move to header file.
25215 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25216 * tree-ssa-address.h (struct mem_address): Move from C file.
25217 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25219 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25221 * tree-affine.h (aff_combination_type): New interface.
25222 (aff_combination_zero_p): Remove static.
25223 (aff_combination_const_p): New interface.
25224 (aff_combination_singleton_var_p): New interfaces.
25226 2017-05-11 Richard Biener <rguenther@suse.de>
25228 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25229 Skip unreachable blocks and destinations.
25230 (eliminate): Move stmt removal and fixup ...
25231 (fini_eliminate): ... here. Skip inserted exprs.
25232 (pass_pre::execute): Move fini_pre after fini_eliminate.
25233 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25234 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25235 PRE to get rid of dead code that has invalid SSA form and
25236 split critical edges again.
25238 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25240 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25242 2017-05-11 Richard Biener <rguenther@suse.de>
25244 * passes.c (execute_function_todo): Verify loops if they are
25245 said to be up-to-date.
25246 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25247 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25249 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25252 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25253 handle calling assemble_external ourself.
25256 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25257 modes with zero size. Enhance comment.
25259 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25261 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25262 built-ins for vec_xl and vec_xst with short and char pointer
25265 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25267 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25268 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25269 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25270 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25271 (_mm_maskz_min_round_ss): New intrinsics.
25272 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25273 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25274 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25275 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25276 (__builtin_ia32_minss_mask_round): New builtins.
25277 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25278 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25279 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25281 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25282 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25284 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25287 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25289 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25290 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25291 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25292 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25293 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25294 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25295 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25296 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25297 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25298 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25299 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25300 (__builtin_ia32_mulss_mask_round): New builtins.
25301 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25302 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25303 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25305 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25306 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25308 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25311 2017-05-10 Julia Koval <julia.koval@intel.com>
25313 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25314 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25315 (_mm256_setr_m128i): New intrinsics.
25317 2017-05-10 Julia Koval <julia.koval@intel.com>
25319 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25320 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25321 (_mm_maskz_rcp14_ss): New intrinsics.
25322 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25323 (__builtin_ia32_rcp14ss_mask): New builtins.
25324 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25326 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25328 PR tree-optimization/51513
25329 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25330 (assert_unreachable_fallthru_edge_p): Use it.
25331 (group_case_labels_stmt): Likewise.
25332 * tree-cfg.h: Prototype it.
25333 * stmt.c: Include cfghooks.h and tree-cfg.h.
25334 (emit_case_dispatch_table) <gap_label>: New local variable.
25335 Use it to fill dispatch table gaps.
25336 Test for default_label before updating probabilities.
25337 (expand_case) <default_label>: Remove unneeded initialization.
25338 Test for unreachable default case statement and remove its edge.
25339 Set default_label accordingly.
25340 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25342 2017-05-10 Carl Love <cel@us.ibm.com>
25344 * config/rs6000/rs6000-c: Add support for built-in functions
25345 vector signed char vec_neg (vector signed char)
25346 vector signed short int vec_neg (vector short int)
25347 vector signed int vec_neg (vector signed int)
25348 vector signed long long vec_neg (vector signed long long)
25349 vector float vec_neg (vector float)
25350 vector double vec_neg (vector double)
25351 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25353 * config/rs6000/altivec.h: Add define for vec_neg
25354 * doc/extend.texi: Update the built-in documentation for the
25355 new built-in functions.
25357 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25359 PR tree-optimization/77644
25360 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25362 2017-05-10 Nathan Sidwell <nathan@acm.org>
25364 * dumpfile.h (TDI_lang_all): New.
25365 (TDF_KIND): New. Renumber others
25366 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25368 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25370 (get_dump_file_name): Adjust suffix generation.
25371 (dump_enable_all): Use TDF_KIND.
25372 * doc/invoke.texi (-fdump-lang-all): Document.
25374 * dumpfile.h: Tabify.
25376 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25379 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25380 Move member access before delete.
25382 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25384 * tree-inline.c (expand_call_inline): Split block at stmt
25387 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25390 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25391 are now unused after splitting mov{sf,sd}_hardfloat.
25392 (f32_lr2): Likewise.
25393 (f32_lm): Likewise.
25394 (f32_lm2): Likewise.
25395 (f32_li): Likewise.
25396 (f32_li2): Likewise.
25397 (f32_lv): Likewise.
25398 (f32_sr): Likewise.
25399 (f32_sr2): Likewise.
25400 (f32_sm): Likewise.
25401 (f32_sm2): Likewise.
25402 (f32_si): Likewise.
25403 (f32_si2): Likewise.
25404 (f32_sv): Likewise.
25405 (f32_dm): Likewise.
25406 (f32_vsx): Likewise.
25407 (f32_av): Likewise.
25408 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25409 For movsf, order stores so the VSX stores occur before the GPR
25410 store which encourages the register allocator to use a traditional
25411 FPR instead of a GPR. For movsd, order the stores so that the GPR
25412 store comes before the VSX stores to allow the power6 to work.
25413 This is due to the power6 not having a 32-bit integer store
25414 instruction from a FPR.
25415 (movsf_hardfloat): Likewise.
25416 (movsd_hardfloat): Likewise.
25418 2017-05-09 Martin Sebor <msebor@redhat.com>
25420 PR translation/80280
25421 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25424 PR translation/80280
25425 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25426 data member added in r247778.
25427 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25429 2017-05-09 Nathan Sidwell <nathan@acm.org>
25431 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25433 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25436 2017-05-09 Marek Polacek <polacek@redhat.com>
25438 * doc/invoke.texi: Fix typo.
25440 2017-05-09 Richard Biener <rguenther@suse.de>
25442 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25443 (vrp_val_is_min): Likewise.
25444 (set_value_range_to_value): Likewise.
25445 (set_value_range_to_nonnegative): Likewise.
25446 (gimple_assign_nonzero_p): Likewise.
25447 (gimple_stmt_nonzero_p): Likewise.
25448 (vrp_int_const_binop): Likewise. Remove unreachable case.
25449 (adjust_range_with_scev): Adjust comments.
25450 (compare_range_with_value): Likewise.
25451 (extract_range_from_phi_node): Likewise.
25452 (test_for_singularity): Likewise.
25454 2017-05-09 Richard Biener <rguenther@suse.de>
25456 * tree-vrp.c (get_single_symbol): Add assert that we don't
25457 get overflowed constants as invariant part.
25458 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25459 checks. Use wi::cmp instead of recursing for integer constants.
25460 (compare_values): Just ignore whether we assumed undefined
25461 overflow instead of failing the compare.
25462 (extract_range_for_var_from_comparison_expr): Add comment before the
25463 TREE_NO_WARNING sets.
25464 (test_for_singularity): Likewise.
25465 (extract_range_from_comparison): Do not disable optimization
25466 when we assumed undefined overflow.
25467 (extract_range_basic): Remove init of unused var.
25469 2017-05-09 Richard Biener <rguenther@suse.de>
25471 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25472 (extract_range_from_multiplicative_op_1): Adjust.
25473 (extract_range_from_binary_expr_1): Use int_const_binop.
25475 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25478 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25479 rs6000_store_data_bypass_p in seven define_bypass directives and
25480 in several comments.
25481 * config/rs6000/rs6000-protos.h: Add prototype for
25482 rs6000_store_data_bypass_p function.
25483 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25484 function implements slightly different (rs6000-specific) semantics
25485 than store_data_bypass_p, returning false rather than aborting
25486 with assertion error when arguments do not satisfy the
25487 requirements of store data bypass.
25488 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25489 rs6000_store_data_bypass_p.
25491 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25493 * config/xtensa/xtensa-protos.h
25494 (xtensa_initial_elimination_offset): New declaration.
25495 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25496 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25497 macro definition, add case for FRAME_POINTER_REGNUM when
25498 FRAME_GROWS_DOWNWARD.
25499 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25500 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25501 xtensa_initial_elimination_offset.
25503 2017-05-08 Nathan Sidwell <nathan@acm.org>
25505 * doc/invoke.texi: Alphabetize -fdump options.
25507 2017-05-08 Martin Sebor <msebor@redhat.com>
25509 PR translation/80280
25510 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25512 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25514 * target.def (compute_frame_layout): New optional target hook.
25515 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25516 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25517 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25519 * reload1.c (verify_initial_elim_offsets): Likewise.
25520 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25521 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25522 (arm_get_frame_offsets): Split up into this ...
25523 (arm_compute_frame_layout): ... and this function.
25525 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25527 * config/aarch64/constraints.md (Usa): New constraint.
25528 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25530 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25532 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25533 with_multilib_list after it has been checked.
25535 2017-05-08 Richard Biener <rguenther@suse.de>
25537 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25538 (bitmap_set_subtract_values): Likewise.
25540 2017-05-08 Richard Biener <rguenther@suse.de>
25542 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25543 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25545 (gimple_stmt_nonzero_warnv_p): Rename to ...
25546 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25548 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25549 (extract_range_basic): Adjust, do not disable propagation on
25550 strict overflow sensitive simplification.
25551 (vrp_visit_cond_stmt): Likewise.
25553 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25555 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25556 body size unconditionally.
25558 2017-05-07 Jeff Law <law@redhat.com>
25561 2017-05-06 Jeff Law <law@redhat.com>
25562 PR tree-optimization/78496
25563 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25566 PR tree-optimization/78496
25567 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25568 (simplify_stmt_using_ranges): Call it.
25569 (vrp_dom_walker::before_dom_children): Extract equivalences
25570 from an ASSERT_EXPR with an equality comparison against a
25573 2017-05-06 Jeff Law <law@redhat.com>
25575 PR tree-optimization/78496
25576 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25579 PR tree-optimization/78496
25580 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25581 (simplify_stmt_using_ranges): Call it.
25582 (vrp_dom_walker::before_dom_children): Extract equivalences
25583 from an ASSERT_EXPR with an equality comparison against a
25586 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25588 * lra-constraints.c (lra_copy_reg_equiv): New function.
25589 (split_reg): Use it to copy equivalence information from the
25590 original register to the spill register.
25592 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25594 PR rtl-optimization/75964
25595 * simplify-rtx.c (simplify_const_relational_operation): Remove
25596 invalid handling of comparisons of integer ABS.
25598 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25600 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25601 initialize to zero.
25602 (init_regs): Remove declaration.
25603 (function_arg_advance_32): Initialize error_p as boolean variable.
25605 2017-05-05 Nathan Sidwell <nathan@acm.org>
25607 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25608 lines. Use for (;;).
25610 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25612 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25613 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25614 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25615 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25616 VF=2 that require versioning.
25618 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25620 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25623 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25625 * diagnostic.h (diagnostic_override_option_index): Convert from
25626 macro to inline function.
25628 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25630 * diagnostic.c (last_module_changed_p): New function.
25631 (set_last_module): New function.
25632 (diagnostic_report_current_module): Convert macro usage to
25633 the above functions.
25634 * diagnostic.h (diagnostic_context::last_module): Strengthen
25635 from const line_map * to const line_map_ordinary *.
25636 (diagnostic_last_module_changed): Delete macro.
25637 (diagnostic_set_last_module): Delete macro.
25639 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25641 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25642 with diagnostic_report_diagnostic.
25643 (diagnostic_n_impl_richloc): Likewise.
25644 * diagnostic.h (report_diagnostic): Delete macro.
25645 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25646 with diagnostic_report_diagnostic.
25647 * substring-locations.c (format_warning_va): Likewise.
25649 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25651 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25652 save/restor of format_spec. Move option-printing code to...
25653 (print_option_information): ...this new function, and
25654 reimplement by simply printing to the pretty_printer,
25655 rather than appending to the format string.
25657 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25659 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25660 handling logic into...
25661 (update_effective_level_from_pragmas): ...this new function.
25663 2017-05-04 Andrew Waterman <andrew@sifive.com>
25665 * config/riscv/riscv.opt (mstrict-align): New option.
25666 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25667 (SLOW_UNALIGNED_ACCESS): Define.
25668 (riscv_slow_unaligned_access): Declare.
25669 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25671 (riscv_slow_unaligned_access): New variable.
25672 (rocket_tune_info): Set slow_unaligned_access to true.
25673 (optimize_size_tune_info): Set slow_unaligned_access to false.
25674 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25675 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25676 (riscv_option_override): Set riscv_slow_unaligned_access.
25677 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25679 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25681 * config/riscv/riscv.md: Unify indentation.
25683 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25688 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25690 (extendsi<mode>2): Add support for doing sign extension via
25691 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25692 don't have ISA 3.0 instructions.
25693 (extendsi<mode>2 splitter): Likewise.
25694 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25695 generate the normal insns since SImode can now go in vector
25696 registers. Disallow the special UNSPECs needed for previous
25697 machines to hide SImode being used. Add new insns
25698 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25699 (fix_trunc<mode>si2_stfiwx): Likewise.
25700 (fix_trunc<mode>si2_internal): Likewise.
25701 (fixuns_trunc<mode>si2): Likewise.
25702 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25703 (fctiw<u>z_<mode>_smallint): Likewise.
25704 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25705 of floating point to 32-bit integer from doing a direct move to
25706 the GPR registers to do a store.
25707 (fctiwz_<mode>): Break long line.
25709 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25711 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25712 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25713 (addr_list, addr_offset_valid_p): New.
25714 (split_address_groups): Check offset validity with above function.
25715 (gt-tree-ssa-loop-ivopts.h): Include header file.
25717 2017-05-05 Nathan Sidwell <nathan@acm.org>
25719 * config.gcc (arm*-*-*): Add missing 'fi'.
25721 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25723 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25724 included in -fopt-info does not matter.
25725 * doc/optinfo.texi (-fopt-info): Fix description of default
25726 behavour. Explicitly say order of options included in -fopt-info
25729 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25731 * config.gcc: Allow combinations of aprofile and rmprofile values for
25732 --with-multilib-list.
25733 * config/arm/t-multilib: New file.
25734 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25735 variables. Remove setting of ISA and floating-point ABI in
25736 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25737 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25738 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25739 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25741 * config/arm/t-rmprofile: Likewise except for the matches changes.
25742 * doc/install.texi (--with-multilib-list): Document the combination of
25743 aprofile and rmprofile values and warn about pitfalls in doing that.
25745 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25747 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25748 (movdi_aarch64): Likewise.
25750 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25752 PR tree-optimization/80632
25753 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25755 (build_arrays): Initialize it for virtual phis.
25756 (fix_phi_nodes): Use it for virtual phis.
25758 PR tree-optimization/80558
25759 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25760 [x, y] op z into [x op, y op z] for op & or | if conditions
25763 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25764 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25767 * config/arm/arm.md (use_literal_pool): Remove.
25768 (64-bit immediate split): No longer takes cost into consideration
25769 if arm_disable_literal_pool is enabled.
25770 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25771 used when arm_disable_literal_pool is enabled.
25772 (arm_max_const_double_inline_cost): Remove use of
25773 arm_disable_literal_pool.
25774 (push_minipool_fix): Add assert.
25775 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25776 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25777 (no_literal_pool_sf_immediate): New.
25779 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25781 PR tree-optimization/80613
25782 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25783 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25785 2017-05-05 Richard Biener <rguenther@suse.de>
25787 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25789 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25791 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25792 of this flag from insn conditions due to removal from r247495.
25794 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25796 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25798 (arm_early_store_addr_dep_ptr): Likewise.
25799 * config/arm/aarch-common-protos.h
25800 (arm_early_load_addr_dep_ptr): Add prototype.
25801 (arm_early_store_addr_dep_ptr): Likewise.
25802 * config/arm/cortex-a53.md: Add new bypasses.
25804 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25806 * tree.c (next_type_uid): Change type to unsigned.
25807 (type_hash_canon): Decrement back next_type_uid if
25808 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25809 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25812 2017-05-04 Martin Sebor <msebor@redhat.com>
25814 * builtins.c: Fix a trivial typo in a comment.
25816 PR middle-end/79234
25817 * builtins.c (check_sizes): Adjust to handle reading past the end.
25818 Avoid printing excessive upper bound of ranges. Use %E to print
25819 tree nodes instead of converting them to %wu.
25820 (expand_builtin_memchr): New function.
25821 (compute_dest_size): Rename...
25822 (compute_objsize): ...to this.
25823 (expand_builtin_memcpy): Adjust.
25824 (expand_builtin_mempcpy): Adjust.
25825 (expand_builtin_strcat): Adjust.
25826 (expand_builtin_strcpy): Adjust.
25827 (check_strncat_sizes): Adjust.
25828 (expand_builtin_strncat): Adjust.
25829 (expand_builtin_strncpy): Adjust and simplify.
25830 (expand_builtin_memset): Adjust.
25831 (expand_builtin_bzero): Adjust.
25832 (expand_builtin_memcmp): Adjust.
25833 (expand_builtin): Handle memcmp.
25834 (maybe_emit_chk_warning): Check strncat just once.
25836 2017-05-04 Martin Sebor <msebor@redhat.com>
25838 PR preprocessor/79214
25839 PR middle-end/79222
25840 PR middle-end/79223
25841 * builtins.c (check_sizes): Add inlining context and issue
25842 warnings even when -Wno-system-headers is set.
25843 (check_strncat_sizes): Same.
25844 (expand_builtin_strncat): Same.
25845 (expand_builtin_memmove): New function.
25846 (expand_builtin_stpncpy): Same.
25847 (expand_builtin): Handle memmove and stpncpy.
25849 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25851 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25852 which is not used any more.
25854 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25856 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25858 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25860 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25861 (cortexa53_tunings): Likewise.
25862 (cortexa57_tunings): Likewise.
25863 (cortexa72_tunings): Likewise.
25864 (cortexa73_tunings): Likewise.
25866 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25868 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25869 Set loop alignment to 8.
25871 2017-05-04 Martin Sebor <msebor@redhat.com>
25873 PR translation/80280
25874 * builtins.c (expand_builtin_object_size): Add missing quoting to
25875 %D and like directives.
25876 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25877 (hsa_type_for_tree_type): Same.
25878 (verify_function_arguments): Same.
25879 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25880 * varasm.c (get_section): Same.
25883 2017-05-04 Martin Sebor <msebor@redhat.com>
25885 PR translation/80280
25886 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25888 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25890 * config/aarch64/aarch64.c (generic_addrcost_table):
25891 Change HI/TI mode setting.
25893 2017-05-04 Martin Jambor <mjambor@suse.cz>
25895 PR tree-optimization/80622
25896 * tree-sra.c (comes_initialized_p): New function.
25897 (build_accesses_from_assign): Only set write lazily when
25898 comes_initialized_p is false.
25899 (analyze_access_subtree): Use comes_initialized_p.
25900 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25901 instead of testing for PARM_DECL.
25903 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25905 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25906 constraint on operand 0 to allow more general addressing modes.
25907 Adjust output template.
25908 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25910 * config/aarch64/aarch64-protos.h
25911 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25912 * config/aarch64/constraints.md (Dp): New address constraint.
25913 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25916 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25918 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25919 update use of estimate_ipcp_clone_size_and_time.
25920 (estimate_local_effects): Update use of
25921 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25922 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25923 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25924 Return nonspecialized time.
25926 2017-05-04 Richard Biener <rguenther@suse.de>
25928 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25929 for the last VUSE which def dominates the PHI. Directly call
25931 (get_continuation_for_phi_1): Remove.
25933 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25935 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25936 to explain the use of truncating division. Cap the number of
25937 iterations to the maximum given by nb_iterations_upper_bound,
25940 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25942 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25943 * configure: Regenerate.
25944 * config.in: Regenerate.
25945 * config/i386/driver-mingw32.c: new file.
25946 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25947 * config.host: Link driver-mingw32.o on MinGW host.
25948 * doc/install.texi: Document new --enable-mingw-wildcard configure
25951 2017-05-04 Marek Polacek <polacek@redhat.com>
25953 PR tree-optimization/80612
25954 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25956 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25957 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25959 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25960 (movt splitter): Likewise.
25961 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25962 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25963 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25964 block for Thumb-1 with MOVT.
25965 (thumb2_legitimate_address_p): Move code block ...
25966 (can_avoid_literal_pool_for_label_p): ... into this new function.
25967 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25969 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25970 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25971 "M-profile targets with the MOVT instruction".
25973 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25975 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25976 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25977 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25979 2017-05-04 Martin Liska <mliska@suse.cz>
25981 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25982 variable cond_code.
25984 2017-05-04 Richard Biener <rguenther@suse.de>
25986 * tree.c (array_at_struct_end_p): Handle arrays at struct
25987 end with flexarrays more conservatively. Refactor and treat
25988 arrays of arrays or aggregates more strict. Fix
25989 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25990 * tree.c (array_at_struct_end_p): Adjust prototype.
25991 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25992 * gimple-fold.c (get_range_strlen): Likewise.
25993 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25995 2017-05-04 Richard Biener <rguenther@suse.de>
25997 PR tree-optimization/31130
25998 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26000 (supports_overflow_infinity): Likewise.
26001 (is_negative_overflow_infinity): Likewise.
26002 (is_positive_overflow_infinity): Likewise.
26003 (is_overflow_infinity): Likewise.
26004 (stmt_overflow_infinity): Likewise.
26005 (overflow_infinity_range_p): Likewise.
26006 (usable_range_p): Remove as always returning true.
26007 (make_overflow_infinity): Remove.
26008 (negative_overflow_infinity): Likewise.
26009 (positive_overflow_infinity): Likewise.
26010 (avoid_overflow_infinity): Likewise.
26011 (set_value_range): Adjust accordingly.
26012 (set_value_range_to_nonnegative): Likewise, remove now unused
26013 overflow_infinity arg.
26014 (vrp_operand_equal_p): Adjust.
26015 (update_value_range): Likewise.
26016 (range_int_cst_singleton_p): Likewise.
26017 (operand_less_p): Likewise.
26018 (compare_values_warnv): Likewise.
26019 (extract_range_for_var_from_comparison_expr): Likewise.
26020 (vrp_int_const_binop): Likewise.
26021 (zero_nonzero_bits_from_vr): Likewise.
26022 (extract_range_from_multiplicative_op_1): Likewise.
26023 (extract_range_from_binary_expr_1): Likewise.
26024 (extract_range_from_unary_expr): Likewise.
26025 (extract_range_from_comparison): Likewise.
26026 (extract_range_basic): Likewise.
26027 (adjust_range_with_scev): Likewise.
26028 (compare_ranges): Likewise.
26029 (compare_range_with_value): Likewise.
26030 (dump_value_range): Likewise.
26031 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26033 (simplify_cond_using_ranges): Adjust.
26035 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26037 * brig-builtins.def: Added a builtin for class_f64.
26038 * builtin-types.def: Added a builtin type needed by class_f64.
26040 2017-05-03 Jason Merrill <jason@redhat.com>
26042 * timevar.def: Add TV_CONSTEXPR.
26044 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26046 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26048 2017-05-03 Martin Jambor <mjambor@suse.cz>
26050 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26051 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26052 * ipa-cp.c (ipcp_propagate_stage): Do not call
26053 ipa_update_after_lto_read.
26054 * ipa-inline.c (ipa_inline): Likewise.
26056 2017-05-03 Martin Jambor <mjambor@suse.cz>
26058 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26059 tag. Added a default constructor and a destructor.
26060 (ipa_edge_args_sum_t): New class;
26061 (ipa_edge_args_sum): Declare.
26062 (ipa_edge_args_vector): Remove declaration.
26063 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26064 (ipa_free_edge_args_substructures): Remove declaration.
26065 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26066 (ipa_edge_args_info_available_for_edge_p): Likewise.
26067 * ipa-prop.c (ipa_edge_args_vector): Removed.
26068 (edge_removal_hook_holder): Likewise.
26069 (edge_duplication_hook_holder): Likewise.
26070 (ipa_edge_args_sum): New variable.
26071 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26072 ipa_edge_args_vector.
26073 (ipa_free_edge_args_substructures): Likewise.
26074 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26075 ipa_edge_args_vector.
26076 (ipa_edge_removal_hook): Turned into method
26077 ipa_edge_args_sum_t::remove.
26078 (ipa_edge_duplication_hook): Turned into method
26079 ipa_edge_args_sum_t::duplicate.
26080 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26081 registering edge hooks.
26082 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26083 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26084 ipa_edge_args_sum instead of ipa_edge_args_vector.
26085 * ipa-profile.c (ipa_profile): Likewise.
26087 2017-05-03 Martin Jambor <mjambor@suse.cz>
26089 * symbol-summary.h (function_summary): New method exists.
26090 (function_summary::symtab_removal): Deallocate through release.
26091 (call_summary): New class.
26092 (gt_ggc_mx): New overload.
26093 (gt_pch_nx): Likewise.
26094 (gt_pch_nx): Likewise.
26096 2017-05-03 Jeff Law <law@redhat.com>
26098 PR tree-optimization/78496
26099 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26100 from simplify_cond_using_ranges. Split off code to walk
26101 backwards through casts into ...
26102 (simplify_cond_using_ranges_2): New function.
26103 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26104 (execute_vrp): After identifying jump threads, call
26105 simplify_cond_using_ranges_2.
26107 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26110 * ipa-inline.h (inline_summary): Add ctor.
26111 (create_ggc): Do not use ggc_cleared_alloc.
26113 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26114 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26116 * gcc.c (handle_braces): Support escaping in switch matching
26118 * doc/invoke.texi (Spec Files): Document it.
26119 Remove superfluous @code markup in items.
26121 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26123 * diagnostic-show-locus.c (struct column_range): New struct.
26124 (get_affected_columns): New function.
26125 (get_printed_columns): New function.
26126 (struct correction): New struct.
26127 (correction::ensure_capacity): New function.
26128 (correction::ensure_terminated): New function.
26129 (struct line_corrections): New struct.
26130 (line_corrections::~line_corrections): New dtor.
26131 (line_corrections::add_hint): New function.
26132 (layout::print_trailing_fixits): Reimplement in terms of the new
26134 (selftest::test_overlapped_fixit_printing): New function.
26135 (selftest::diagnostic_show_locus_c_tests): Call it.
26137 2017-05-03 Nathan Sidwell <nathan@acm.org>
26139 Canonicalize canonical type hashing
26140 * tree.h (type_hash_canon_hash): Declare.
26141 * tree.c (type_hash_list, attribute_hash_list): Move into
26142 type_hash_canon_hash.
26143 (build_type_attribute_qual_variant): Break out hash code calc into
26144 type_hash_canon_hash.
26145 (type_hash_canon_hash): New. Generic type hash computation.
26146 (build_range_type_1, build_array_type_1, build_function_type,
26147 build_method_type_directly, build_offset_type, build_complex_type,
26148 make_vector_type): Call it.
26150 2017-05-03 Richard Biener <rguenther@suse.de>
26152 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26153 When all DRs have unknown misaligned do not always peel
26154 when there is a store but apply the same costing model as if
26155 there were only loads.
26157 2017-05-03 Richard Biener <rguenther@suse.de>
26160 PR tree-optimization/80492
26161 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26162 compare_base_decls returning dont-know properly.
26164 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26166 * config/arm/iterators.md (CCSI): New mode iterator.
26167 (arch): New mode attribute.
26168 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26169 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26170 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26171 code iterator for success result mode.
26172 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26173 the corresponding new insn generators.
26175 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26178 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26179 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26181 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26183 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26184 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26185 (DDR_A): Wrap DDR argument in brackets.
26186 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26187 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26188 (DDR_REVERSED_P): Likewise.
26190 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26192 PR tree-optimization/79472
26193 * tree-switch-conversion.c (struct switch_conv_info): Add
26194 contiguous_range and default_case_nonstandard fields.
26195 (collect_switch_conv_info): Compute contiguous_range and
26196 default_case_nonstandard fields, don't clear final_bb if
26197 contiguous_range and only the default case doesn't have the required
26199 (check_all_empty_except_final): Set default_case_nonstandard instead
26200 of failing if contiguous_range and the default case doesn't have empty
26202 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26203 and only the default case doesn't have the required constants. Skip
26205 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26206 if default_case_nonstandard.
26207 (build_constructors): Build constant 1 just once. Assert that default
26208 values aren't inserted in between cases if contiguous_range. Skip
26210 (build_arrays): Skip virtual phis.
26211 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26212 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26213 Handle virtual phis.
26214 (gen_inbound_check): Handle default_case_nonstandard case.
26215 (process_switch): Adjust check_final_bb caller. Call
26216 gather_default_values with the first non-default case instead of
26217 default case if default_case_nonstandard.
26219 2017-05-02 Nathan Sidwell <nathan@acm.org>
26221 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26222 check. Fix formatting.
26224 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26226 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26227 errors when comparing specialized and unspecialized times.
26229 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26231 * diagnostic-show-locus.c
26232 (layout::should_print_annotation_line_p): Make private.
26233 (layout::print_annotation_line): Make private.
26234 (layout::annotation_line_showed_range_p): Make private.
26235 (layout::show_ruler): Make private.
26236 (layout::print_source_line): Make private. Pass in line and
26237 line_width, rather than calling location_get_source_line. Drop
26239 (layout::print_leading_fixits): New method.
26240 (layout::print_any_fixits): Rename to...
26241 (layout::print_trailing_fixits): ...this, and make private.
26242 Don't print newline fixits.
26243 (diagnostic_show_locus): Move logic for printing one row into...
26244 (layout::print_line): ...this new function. Move the
26245 location_get_source_line call and error-handling from
26246 print_source_line to here. Call print_leading_fixits, and rename
26247 print_any_fixits to print_trailing_fixits.
26248 (selftest::test_fixit_insert_containing_newline): Update now that
26249 newlines are partially supported.
26250 (selftest::test_fixit_insert_containing_newline_2): New test.
26251 (selftest::test_fixit_replace_containing_newline): Update comments.
26252 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26253 * edit-context.c (class added_line): New class.
26254 (class edited_line): Describe newline handling in comment.
26255 (edited_line::actually_edited_p): New method.
26256 (edited_line::print_content): Delete redundant decl.
26257 (edited_line::m_predecessors): New field.
26258 (edited_file::print_content): Call edited_line::print_content.
26259 (edited_file::print_diff): Update to support newlines.
26260 (edited_file::print_diff_hunk): Likewise.
26261 (edited_file::print_run_of_changed_lines): New function.
26262 (edited_file::print_diff_line): Convert to...
26263 (print_diff_line): ...this.
26264 (edited_file::get_effective_line_count): New function.
26265 (edited_line::edited_line): Initialize new field m_predecessors.
26266 (edited_line::~edited_line): Clean up m_predecessors.
26267 (edited_line::apply_fixit): Handle newlines.
26268 (edited_line::get_effective_line_count): New function.
26269 (edited_line::print_content): New function.
26270 (edited_line::print_diff_lines): New function.
26271 (selftest::test_applying_fixits_insert_containing_newline): New
26273 (selftest::test_applying_fixits_replace_containing_newline): New
26275 (selftest::insert_line): New function.
26276 (selftest::test_applying_fixits_multiple_lines): Add example of
26278 (selftest::edit_context_c_tests): Call the new tests.
26280 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26282 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26283 parameter cand. Update dump information.
26284 (get_computation_cost): Update uses.
26286 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26288 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26289 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26290 (get_computation_at, rewrite_use_address): Update use of
26291 get_computation_aff.
26293 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26295 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26296 (get_computation): Delete.
26297 (get_computation_cost): Implement like get_computation_cost_at.
26298 Use get_computation_at.
26299 (get_computation_cost_at): Delete.
26300 (rewrite_use_nonlinear_expr): Use get_computation_at.
26301 (rewrite_use_compare, remove_unused_ivs): Ditto.
26303 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26305 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26307 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26309 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26310 (ivopts_global_cost_for_size): Rename parameter and update uses.
26311 (iv_ca_recount_cost): Update uses.
26312 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26313 candidates seperately in n_invs and n_cands.
26314 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26316 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26318 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26319 (find_inv_vars_cb): New.
26320 (find_depends): Renamed to ...
26321 (find_inv_vars): ... this.
26322 (add_candidate_1, force_var_cost): Call find_inv_vars.
26323 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26325 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26327 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26328 inv_vars. Add inv_exprs.
26329 (struct iv_cand): Rename depends_on to inv_vars.
26330 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26331 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26332 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26333 (dump_cand): Dump inv_vars.
26334 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26335 (record_invariant, find_depends, add_candidate_1): Ditto.
26336 (set_group_iv_cost, force_var_cost): Ditto.
26337 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26338 (get_computation_cost_at, get_computation_cost): Ditto.
26339 (determine_group_iv_cost_generic): Ditto.
26340 (determine_group_iv_cost_address): Ditto.
26341 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26342 (determine_group_iv_costs): Ditto.
26343 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26344 (iv_ca_set_remove_invariants): Renamed to ...
26345 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26346 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26347 (iv_ca_set_add_invariants): Renamed to ...
26348 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26349 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26350 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26351 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26352 (create_new_ivs): Remove useless dump.
26354 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26356 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26358 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26359 (iv_ca_set_no_cp, create_new_iv): Ditto.
26361 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26363 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26365 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26367 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26368 function tree_check2.
26370 2017-05-02 Martin Liska <mliska@suse.cz>
26372 * doc/gcov.texi: Add missing preposition.
26373 * gcov.c (function_info::function_info): Properly fill up
26374 all member variables.
26376 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26378 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26380 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26382 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26384 2017-05-02 Martin Liska <mliska@suse.cz>
26387 * lto-streamer-in.c (lto_read_tree_1): Remove
26388 LTO_STREAMER_DEBUG.
26389 * lto-streamer.c (struct tree_hash_entry): Likewise.
26390 (struct tree_entry_hasher): Likewise.
26391 (tree_entry_hasher::hash): Likewise.
26392 (tree_entry_hasher::equal): Likewise.
26393 (lto_streamer_init): Likewise.
26394 (lto_orig_address_map): Likewise.
26395 (lto_orig_address_get): Likewise.
26396 (lto_orig_address_remove): Likewise.
26397 * lto-streamer.h: Likewise.
26398 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26399 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26401 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26403 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26404 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26405 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26406 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26407 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26408 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26409 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26410 (mm_maskz_sub_ss): New intrinsics.
26411 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26412 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26413 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26414 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26415 (__builtin_ia32_subss_mask_round): New builtins.
26416 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26417 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26418 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26420 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26421 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26423 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26426 2017-05-02 Martin Jambor <mjambor@suse.cz>
26428 PR tree-optimization/78687
26429 * tree-sra.c (access): New field parent.
26430 (process_subtree_disqualification): New function.
26431 (disqualify_candidate): Call it.
26432 (build_accesses_from_assign): Reset write flag if creating an
26434 (build_access_subtree): Fill in parent field and also prpagate
26435 down grp_write flag.
26436 (create_artificial_child_access): New parameter set_grp_write, set
26437 grp_write to its value.
26438 (propagate_subaccesses_across_link): Also propagate grp_write flag
26440 (propagate_all_subaccesses): Push the closest parent back to work
26441 queue if add_access_to_work_queue returned true.
26443 2017-05-02 Richard Biener <rguenther@suse.de>
26445 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26446 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26447 -fstrict-overflow documentation.
26448 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26449 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26450 flag_strict_overflow.
26451 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26452 * lto-opts.c (lto_write_options): Do not stream it.
26453 * lto-wrapper.c (merge_and_complain): Do not handle it.
26454 * opts.c (default_options_table): Do not set -fstrict-overflow.
26455 (finish_options): Likewise do not clear it when sanitizing.
26456 * simplify-rtx.c (simplify_const_relational_operation): Do not
26457 test flag_strict_overflow.
26459 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26461 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26462 using enabled attribute.
26463 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26464 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26465 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26466 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26467 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26468 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26469 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26470 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26471 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26472 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26474 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26476 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26478 2017-05-02 Richard Biener <rguenther@suse.de>
26480 PR tree-optimization/80591
26482 2017-04-10 Richard Biener <rguenther@suse.de>
26484 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26487 2017-05-02 Richard Biener <rguenther@suse.de>
26489 PR tree-optimization/80549
26490 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26491 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26492 headers if they do not have a preheader.
26494 2017-05-02 Martin Liska <mliska@suse.cz>
26497 * common.opt: Fix typo.
26498 * doc/invoke.texi: Likewise.
26500 2017-05-01 Jan Beulich <jbeulich@suse.com>
26502 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26503 swapping, add (x,x,m,x,n) alternative.
26505 2017-05-01 Nathan Sidwell <nathan@acm.org>
26507 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26508 unnecessary unadjusted_alignment check.
26510 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26513 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26515 * gimplify.c (gimplify_cilk_detach): New function.
26516 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26517 * tree-core.h: Document EXPR_CILK_SPAWN.
26518 * tree.h (EXPR_CILK_SPAWN): Define.
26520 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26522 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26523 to use new fixit_hint representation, using the "replace" logic.
26524 (get_line_span_for_fixit_hint): Likewise.
26525 (layout::print_any_fixits): Likewise.
26526 (selftest::test_one_liner_many_fixits): Rename to...
26527 (selftest::test_one_liner_many_fixits_1): ...this, and update
26528 comment and expected output to reflect that the multiple fix-it
26529 hints are now consolidated into one insertion.
26530 (selftest::test_one_liner_many_fixits_2): New test.
26531 (selftest::test_diagnostic_show_locus_one_liner): Update for
26533 (selftest::test_fixit_consolidation): Update for fix-it API
26535 * diagnostic.c (print_parseable_fixits): Likewise.
26536 * edit-context.c (edited_line::m_line_events): Convert from
26537 auto_vec <line_event *> to auto_vec <line_event>.
26538 (class line_event): Convert from abstract base class to a concrete
26539 class, taking over the role of replace_event.
26540 (class insert_event): Delete.
26541 (class replace_event): Rename to class line_event. Convert to
26543 (edit_context::add_fixits): Reimplement.
26544 (edit_context::apply_insert): Delete.
26545 (edit_context::apply_replace): Rename to...
26546 (edit_context::apply_fixit): ...this. Convert to half-open range.
26547 (edited_file::apply_insert): Delete.
26548 (edited_file::apply_replace): Rename to...
26549 (edited_file::apply_fixit): ...this.
26550 (edited_line::~edited_line): Drop deletion of events.
26551 (edited_line::apply_insert): Delete.
26552 (edited_line::apply_replace): Rename to...
26553 (edited_line::apply_fixit): ...this. Convert to half-open range.
26554 Update for change to type of m_line_events.
26555 * edit-context.h (edit_context::apply_insert): Delete.
26556 (edit_context::apply_replace): Rename to...
26557 (edit_context::apply_fixit): ...this.
26559 2017-05-01 Martin Sebor <msebor@redhat.com>
26561 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26564 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26567 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26568 __get_cpuid_max returns 0.
26569 (__get_cpuid_count): Ditto.
26571 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26573 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26574 replacement expression is another instance of one of its arguments.
26576 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26579 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26580 check for stack push/pop autoinc.
26581 * config/i386/i386.c (ix86_agi_dependent): Return false
26582 if the only reason why modified_in_p returned true is that
26583 addr is SP based and set_insn is a push or pop.
26585 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26587 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26590 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26593 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26594 (account_size_time): Use two predicates - exec_pred and
26596 (evaluate_conditions_for_known_args): Compute both clause and
26598 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26599 (inline_summary_t::duplicate): Update.
26600 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26602 (compute_inline_parameters): Likewise.
26603 (estimate_edge_size_and_time): Update caluclation of time.
26604 (estimate_node_size_and_time): Compute both time and nonspecialized
26606 (estimate_ipcp_clone_size_and_time): Update.
26607 (inline_merge_summary): Update.
26608 (do_estimate_edge_time): Update.
26609 (do_estimate_edge_size): Update.
26610 (do_estimate_edge_hints): Update.
26611 (inline_read_section, inline_write_summary): Stream both new predicates.
26612 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26614 (compute_inlined_call_time): Cleanup.
26615 (big_speedup_p): Update.
26616 (edge_badness): Update.
26617 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26618 (size_time_entry): Replace predicate by exec_predicate and
26619 nonconst_predicate.
26620 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26621 (estimate_edge_time): Return also nonspec_time.
26622 (reset_edge_growth_cache): Update.
26624 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26626 PR rtl-optimization/80491
26627 * ifcvt.c (noce_process_if_block): When looking for x setter
26628 with missing else_bb, don't check only the insn right before
26629 cond_earliest, but look for the last insn that x is modified in
26630 within the same bb.
26632 PR rtl-optimization/80491
26633 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26635 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26637 PR tree-optimization/80487
26638 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26640 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26642 PR tree-optimization/79697
26643 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26644 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26645 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26647 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26648 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26650 2017-04-28 Martin Sebor <msebor@redhat.com>
26652 PR tree-optimization/80523
26653 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26654 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26656 (maybe_warn, format_directive, parse_directive): Use new functions.
26657 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26659 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26661 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26663 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26665 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26666 target_header_dir): Set correctly.
26667 * configure: Regenerated.
26668 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26669 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26670 instead of SYSTEM_HEADER_DIR.
26672 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26674 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26675 (estimate_local_effects): Likewise.
26676 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26677 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26678 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26679 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26680 do_estimate_edge_time, estimate_edge_time): Likewise.
26681 * ipa-inline-analysis.c (estimate_node_size_and_time,
26682 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26683 (estimate_time_after_inlining): Remove.
26685 2017-04-28 Martin Liska <mliska@suse.cz>
26687 * doc/gcov.texi: Enhance documentation of gcov.
26689 2017-04-28 Martin Liska <mliska@suse.cz>
26691 * doc/gcov.texi: Sort options in alphabetic order.
26692 * doc/gcov-dump.texi: Likewise.
26693 * doc/gcov-tool.texi: Likewise.
26694 * gcov.c (print_usage): Likewise.
26695 * gcov-dump.c (print_usage): Likewise.
26696 * gcov-tool.c (print_merge_usage_message): Likewise.
26697 (print_rewrite_usage_message): Likewise.
26698 (print_overlap_usage_message): Likewise.
26700 2017-04-28 Martin Liska <mliska@suse.cz>
26702 PR gcov-profile/53915
26703 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26705 2017-04-28 Martin Liska <mliska@suse.cz>
26707 PR gcov-profile/79891
26708 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26709 is marked by compiler as living on a line.
26710 (get_cycles_count): Remove usage of the union.
26711 (output_intermediate_file): Likewise.
26712 (find_source): Fix GNU coding style.
26713 (accumulate_line_counts): Remove old non-all block mode.
26714 (output_lines): Remove usage of the union.
26715 * profile.c (output_location): Include all BBs, even if
26716 belonging to a same line (and file) as a previous BB.
26718 2017-04-28 Martin Liska <mliska@suse.cz>
26720 * gcov.c (process_args): Handle new argument 'w'.
26721 (read_graph_file): Assign ID to BBs.
26722 (output_branch_count): Display BB # if verbose flag is set.
26723 (output_lines): Likewise for arcs.
26724 (print_usage): Add '--verbose' option help.
26725 * doc/gcov.texi: Document --verbose (-w) option.
26727 2017-04-28 Martin Liska <mliska@suse.cz>
26729 * gcov.c (struct block_location_info): New struct.
26730 (process_file): Fill up the new structure.
26731 (read_graph_file): Replace usage of encoding by the newly added
26733 (add_line_counts): Likewise.
26734 (accumulate_line_counts): Remove usage of the union.
26735 (function_info::function_info): New function.
26736 (function_info::~function_info): Likewise.
26737 (process_file): Call delete instead of release_function.
26738 (release_function): Release the function.
26739 (release_structures): Call delete instead of release_function.
26740 (solve_flow_graph): Replace usage of num_blocks.
26741 (find_exception_blocks): Likewise.
26742 (output_lines): Fix GNU coding style.
26744 2017-04-28 Martin Liska <mliska@suse.cz>
26747 * coverage.c (coverage_remove_note_file): New function.
26748 * coverage.h: Declare the function.
26749 * toplev.c (finalize): Clean if an error has been seen.
26751 2017-04-28 Martin Liska <mliska@suse.cz>
26753 PR gcov-profile/80031
26754 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26755 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26756 * gcov.c (read_graph_file): Read just number of blocks.
26757 * profile.c (branch_prob): Do not stream 0 flags per a basic
26760 2017-04-28 Martin Liska <mliska@suse.cz>
26762 * gcov-dump.c (tag_*): Add new argument to declarations.
26763 (dump_gcov_file): Likewise.
26764 (tag_blocks): Add and use new argument depth.
26765 (tag_arcs): Likewise.
26766 (tag_lines): Likewise.
26767 (tag_counters): Likewise.
26768 (tag_summary): Likewise.
26769 (dump_working_sets): Use depth to do a proper indentation.
26771 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26774 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26775 * symtab.c (symtab_node::debug_symtab): Move definition here.
26777 2017-04-28 Richard Biener <rguenther@suse.de>
26779 * lto-streamer.h (LTO_major_version): Bump to 7.
26781 2017-04-28 Richard Biener <rguenther@suse.de>
26783 * tree-vrp.c (assert_info): New struct.
26784 (add_assert_info): New helper.
26785 (register_edge_assert_for_2): Refactor to add asserts to a vector
26787 (register_edge_assert_for_1): Likewise.
26788 (register_edge_assert_for): Likewise.
26789 (finish_register_edge_assert_for): New helper actually registering
26790 asserts where live on edge.
26791 (find_conditional_asserts): Adjust.
26792 (find_switch_asserts): Likewise.
26793 (evrp_dom_walker::try_find_new_range): Generalize.
26794 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26796 2017-04-27 Marek Polacek <polacek@redhat.com>
26799 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26800 arg10 and arg11 to itype.
26802 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26804 * doc/extend.texi (Object Size Checking): Improve grammar.
26806 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26809 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26810 that the logic for permitting reciprocal estimates matches that
26813 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26816 * tree.c (type_cache_hasher::equal): Only compare
26817 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26818 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26819 non-aggregate element types.
26820 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26821 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26823 2017-04-27 Richard Biener <rguenther@suse.de>
26825 PR middle-end/80533
26826 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26827 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26828 keeping a reference to a trailing array.
26830 2017-04-27 Richard Biener <rguenther@suse.de>
26832 PR middle-end/80539
26833 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26834 being in loop-closed SSA form conservatively.
26835 (chrec_fold_multiply_poly_poly): Likewise.
26837 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26839 PR middle-end/79665
26840 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26841 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26843 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26846 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26847 (aarch64_function_arg_alignment): Return unsigned int again, but still
26848 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26849 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26850 Don't emit -Wpsabi note.
26851 (aarch64_function_arg_boundary): Likewise.
26852 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26855 2017-04-26 Nathan Sidwell <nathan@acm.org>
26857 * tree.h (crc32_unsigned_n): Declare.
26858 (crc32_unsigned, crc32_unsigned): Make inline.
26859 * tree.c (crc32_unsigned_bits): Replace with ...
26860 (crc32_unsigned_n): ... this.
26861 (crc32_unsigned, crc32_byte): Remove.
26862 (crc32_string): Remove unnecessary braces.
26864 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26866 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26867 * ipa-inline-analysis.c (MAX_TIME): Remove.
26868 (account_size_time): Use sreal for time.
26869 (dump_inline_summary): Update.
26870 (estimate_function_body_sizes): Update.
26871 (estimate_edge_size_and_time): Update.
26872 (estimate_calls_size_and_time): Update.
26873 (estimate_node_size_and_time): Update.
26874 (inline_merge_summary): Update.
26875 (inline_update_overall_summary): Update.
26876 (estimate_time_after_inlining): Update.
26877 (inline_read_section): Update.
26878 (inline_write_summary): Update.
26879 * ipa-inline.c (compute_uninlined_call_time): Update.
26880 (compute_inlined_call_time): Update.
26881 (recursive_inlining): Update.
26882 (inline_small_functions): Update.
26883 (dump_overall_stats): Update.
26884 * ipa-inline.h: Include sreal.h.
26885 (size_time_entry): Turn time to sreal.
26886 (inline_summary): Turn self_time nad time to sreal.
26888 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26890 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26892 (sreal::stream_out, sreal::stream_in): New.
26893 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26895 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26897 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26900 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26903 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26905 (dimode_scalar_chain::compute_convert_gain): Ditto.
26906 (dimode_scalar_chain::make_vector_copies): Ditto.
26907 (dimode_scalar_chain::convert_reg): Ditto.
26908 (dimode_scalar_chain::convert_insn): Ditto.
26909 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26910 (VI248_AVX512BW_1): New mode iterator.
26911 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26912 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26915 2017-04-25 Martin Sebor <msebor@redhat.com>
26917 PR tree-optimization/80497
26918 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26919 constants are representable in HOST_WIDE_INT.
26920 (parse_directive): Ditto.
26922 2017-04-25 Martin Sebor <msebor@redhat.com>
26925 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26926 (new_zero_array): Adjust signature.
26927 (dom_info::dom_init): Used unsigned rather that size_t.
26928 (dom_info::dom_info): Same.
26930 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26931 Jakub Jelinek <jakub@redhat.com>
26934 * config/arm/arm.c: Include gimple.h.
26935 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26936 returns negative, increment ncrn only if it returned positive.
26937 (arm_needs_doubleword_align): Return int instead of bool,
26938 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26939 members, but if there is any such non-FIELD_DECL
26940 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26941 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26942 returns negative, increment nregs only if it returned positive.
26943 (arm_setup_incoming_varargs): Likewise.
26944 (arm_function_arg_boundary): Emit -Wpsabi note if
26945 arm_needs_doubleword_align returns negative, return
26946 DOUBLEWORD_ALIGNMENT only if it returned positive.
26948 2017-04-25 Marek Polacek <polacek@redhat.com>
26951 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26952 first argument to type.
26954 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26957 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26958 type checks to test for compatibility instead of equality.
26960 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26961 Jakub Jelinek <jakub@redhat.com>
26964 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26966 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26967 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26968 the alignment computation, but return their maximum in warn_alignment.
26969 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26970 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26972 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26973 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26976 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26978 * config/arc/simdext.md (dmpyh): Fix typo.
26980 2017-04-25 Richard Biener <rguenther@suse.de>
26982 PR tree-optimization/80492
26983 * alias.c (compare_base_decls): Handle registers with asm
26984 specification conservatively.
26985 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26986 compare_base_decls returning dont-know properly.
26988 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26990 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26991 (legitimate_offset_address_p): New function.
26992 (arc_legitimate_address_p): Use above function.
26994 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26996 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26998 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27000 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27001 ACCH registers whenever they are available.
27003 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27005 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27006 double regs fix when not used.
27008 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27010 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27012 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27013 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27015 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27017 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27018 when emitting short ADD instructions.
27020 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27022 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27024 (cmpsi_cc_c_insn): Likewise.
27025 (cbranchsi4_scratch): Compute proper instruction length using
27026 compact_hreg_operand.
27027 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27029 2017-04-25 Richard Biener <rguenther@suse.de>
27031 PR middle-end/80509
27032 * passes.c (pass_manager::pass_manager): Initialize
27033 m_name_to_pass_map.
27035 2017-04-25 Richard Biener <rguenther@suse.de>
27037 PR tree-optimization/79201
27038 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27040 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27043 * config/s390/vector.md: Split MEM->GPR vector moves for
27044 non-s_operand addresses.
27046 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27049 * config/s390/predicates.md (reload_const_wide_int_operand): New
27051 * config/s390/s390.md ("movti"): Remove d/P alternative.
27052 ("movti_bigconst"): New pattern definition.
27054 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27057 * s390-protos.h (s390_expand_cs_hqi): Removed.
27058 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27059 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27060 modes as well as CCZ1mode and CCZmode.
27061 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27062 signature of s390_emit_compare_and_swap.
27063 (s390_expand_cs_hqi): Likewise, make static.
27064 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27065 compare-and-swap, in some cases.
27066 (s390_expand_cs): Wrapper function.
27067 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27069 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27070 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27071 patterns for small and large integers. Forbid symref memory operands.
27072 Move expander to s390.c. Require cc register.
27073 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27074 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27075 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27076 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27077 symref memory operands. Remove CC mode and call s390_match_ccmode
27079 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27081 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27083 * config/s390/s390.md (define_peephole2): New peephole to help
27084 combining the load-and-test pattern with volatile memory.
27086 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27088 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27089 with CCZmode for TARGET_Z196.
27091 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27093 PR rtl-optimization/80501
27094 * combine.c (make_compound_operation_int): Set subreg_code to SET
27095 even for AND with mask of the sign bit of mode.
27097 PR rtl-optimization/80500
27098 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27099 sum's initial value.
27101 2017-04-25 Julian Brown <julian@codesourcery.com>
27102 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27104 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27106 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27108 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27110 2017-04-25 Julian Brown <julian@codesourcery.com>
27111 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27113 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27114 (thunderx2t99_sha): New Reservation.
27116 2017-04-25 Julian Brown <julian@codesourcery.com>
27117 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27119 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27120 type for 1-element load.
27122 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27124 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27126 2017-04-24 Martin Jambor <mjambor@suse.cz>
27128 PR tree-optimization/80293
27129 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27130 char arrays not totally scalarizable if it is false.
27131 (analyze_all_variable_accesses): Pass correct value in the new
27132 parameter. Add a statistics counter.
27134 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27136 PR middle-end/79931
27137 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27139 2017-04-24 Richard Biener <rguenther@suse.de>
27141 PR tree-optimization/80494
27142 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27143 out for complex types.
27145 2017-04-24 Richard Biener <rguenther@suse.de>
27147 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27148 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27149 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27150 (DFS): Adjust and never fail.
27151 (sccvn_dom_walker::fail): Remove.
27152 (sccvn_dom_walker::before_dom_children): Adjust.
27153 (run_scc_vn): Likewise and never fail.
27154 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27155 (pass_fre::execute): Likewise.
27157 2017-04-24 Richard Biener <rguenther@suse.de>
27159 PR tree-optimization/79725
27160 * tree-ssa-sink.c (statement_sink_location): Return whether
27161 failure reason was zero uses. Move that check later.
27162 (sink_code_in_bb): Deal with zero uses by removing the stmt
27165 2017-04-24 Richard Biener <rguenther@suse.de>
27168 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27169 pointer-based references.
27171 2017-04-24 Richard Biener <rguenther@suse.de>
27174 * pass_manager.h (pass_manager::operator new): Remove.
27175 (pass_manager::operator delete): Likewise.
27176 * passes.c (pass_manager::operator new): Remove.
27177 (pass_manager::operator delete): Likewise.
27178 (pass_manager::pass_manager): Zero individual pass members.
27180 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27183 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27184 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27185 Check "XEXP (src, 1)" operand here.
27186 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27187 Check "XEXP (src, 1)" operand here.
27188 (dimode_scalar_chain::make_vector_copies): Detect count register
27189 of a shift instruction. Zero extend count register from QImode
27190 to DImode to satisfy vector shift pattern count operand predicate.
27191 Substitute vector shift count operand with a DImode copy.
27192 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27195 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27197 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27198 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27199 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27200 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27201 (UNSPEC_NOREX_MEM): Remove definition.
27203 2017-04-21 Richard Biener <rguenther@suse.de>
27205 PR tree-optimization/79547
27206 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27207 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27208 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27209 without any constraints.
27211 2017-04-21 Richard Biener <rguenther@suse.de>
27213 PR tree-optimization/78847
27214 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27216 2017-04-21 Richard Biener <rguenther@suse.de>
27218 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27219 (build_distinct_type_copy): Likewise.
27220 (build_variant_type_copy): Likewise.
27221 * tree.c (build_qualified_type): Pass down mem-stat info.
27222 (build_distinct_type_copy): Likewise.
27223 (build_variant_type_copy): Likewise.
27225 2017-04-21 Richard Biener <rguenther@suse.de>
27227 PR tree-optimization/80237
27228 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27230 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27231 for a simplified result.
27233 2016-04-21 Richard Biener <rguenther@suse.de>
27235 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27236 sth as strict as a simple_iv but a chrec without symbols and an
27237 operand defined in the loop we are peeling (and not some subloop).
27238 (propagate_constants_for_unrolling): Propagate all constants.
27240 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27243 * config/i386/i386.c (print_reg): Remove assert for disalowed
27244 regno values, call output_operand_lossage instead.
27246 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27249 * config/i386/constraints.md (Yc): New register constraint.
27250 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27251 Use Yc constraint for alternative 2 of operand 0. Remove
27252 preferred_for_speed attribute.
27254 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27256 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27257 lastprivate clauses in SIMT case.
27259 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27261 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27263 2017-04-20 Richard Biener <rguenther@suse.de>
27265 PR tree-optimization/57796
27266 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27267 as N scalar stores.
27268 (vect_model_load_cost): Cost gathers as N scalar loads.
27270 2017-04-20 Richard Biener <rguenther@suse.de>
27272 * ggc-page.c (ggc_allocated_p): Rename to ...
27273 (safe_lookup_page_table_entry): ... this and return the lookup
27275 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27277 2017-04-20 Richard Biener <rguenther@suse.de>
27279 PR tree-optimization/80453
27280 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27281 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27282 from the conditions.
27283 (vn_phi_eq): Pass them down.
27284 (vn_phi_lookup): Record them.
27285 (vn_phi_insert): Likewise.
27287 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27289 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27290 uninitialized variable warning to avoid buffer overrun.
27292 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27295 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27296 is suppressed for '{ 0 }' in C.
27298 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27300 * BASE-VER: Set to 8.0.0.
27302 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27304 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27305 priority .init_array and .fini_array section with SECTION_NOTYPE
27308 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27310 PR middle-end/80423
27311 * tree.h (build_array_type): Add typeless_storage default argument.
27312 * tree.c (type_cache_hasher::equal): Also compare
27313 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27314 (build_array_type): Add typeless_storage argument, set
27315 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27317 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27318 (build_array_type): Likewise. Add typeless_storage argument.
27320 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27321 Jakub Jelinek <jakub@redhat.com>
27323 PR tree-optimization/80426
27324 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27325 operation on symbolic operands, also compute the overflow for the
27326 invariant part when the operation degenerates into a negation.
27328 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27331 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27332 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27335 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27337 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27340 * config/avr/avr.c (tree.h): Include it.
27341 (cgraph.h): Include it.
27342 (avr_encode_section_info): Don't warn for uninitialized progmem
27343 variable if it's just an alias.
27345 2017-04-19 Richard Biener <rguenther@suse.de>
27348 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27349 when needed by AutoPGO.
27351 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27354 * doc/lto.texi: Remove an extra 'that'.
27356 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27358 PR rtl-optimization/80429
27359 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27360 are only used in debug insns.
27362 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27363 Vladimir Makarov <vmakarov@redhat.com>
27365 * config/sparc/predicates.md (input_operand): Add comment. Return
27366 true for any memory operand when LRA is in progress.
27367 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27369 2017-04-18 Jeff Law <law@redhat.com>
27372 * mips.md ({return,simple_return}_internal): Do not overwrite
27375 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27377 PR tree-optimization/80443
27378 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27379 instead of adding 1, subtract -1 and similarly instead of subtracting
27382 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27384 PR rtl-optimization/80357
27385 * haifa-sched.c (tmp_bitmap): New variable.
27386 (model_recompute): Handle duplicate use records.
27387 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27388 (free_global_sched_pressure_data): Free it.
27390 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27393 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27394 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27395 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27396 instead of SYSTEM_HEADER_DIR.
27398 2017-04-18 Jeff Law <law@redhat.com>
27400 PR middle-end/80422
27401 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27402 predecessors after walking up the insn chain.
27404 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27407 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27408 sizetype type into debug info.
27410 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27413 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27414 unneeded test for TARGET_UPPER_REGS_SF.
27415 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27417 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27420 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27421 instead of gsi_after_labels.
27423 2017-04-18 Jeff Law <law@redhat.com>
27425 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27429 2017-04-13 Jeff Law <law@redhat.com>
27430 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27431 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27433 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27436 * config/avr/avr.c (intl.h): Include it.
27437 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27439 2017-04-18 Martin Liska <mliska@suse.cz>
27441 PR gcov-profile/78783
27442 * gcov-tool.c (gcov_output_files): Validate that destination
27443 file is either removed by the tool or by a user.
27445 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27446 Guy Benyei <guybe@mellanox.com>
27448 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27449 block, and do not negate it, the stored id is already negative.
27451 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27453 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27455 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27458 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27459 masks of options that should be turned off if the VSX vector
27460 options are turned off.
27461 (OTHER_P8_VECTOR_MASKS): Likewise.
27462 (OTHER_VSX_VECTOR_MASKS): Likewise.
27463 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27464 rs6000_disable_incompatible_switches to validate no type switches
27466 (rs6000_incompatible_switch): New function to disallow turning on
27467 other vector options if -mno-vsx, -mno-power8-vector, or
27468 -mno-power9-vector are specified.
27470 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27472 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27474 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27476 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27477 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27478 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27479 (ARG_POINTER_CFA_OFFSET): Likewise.
27481 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27483 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27484 conditions to take advantage of various optimizations.
27486 2017-04-13 Jeff Law <law@redhat.com>
27488 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27489 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27490 (zero_extendsidi2_dext): Likewise.
27492 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27495 * fold-const.c (fold_ternary_loc): Revert
27496 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27499 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27501 PR rtl-optimization/80343
27502 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27503 new scratch pseudo.
27505 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27508 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27509 to ubsan_encode_value.
27511 2017-04-13 Jeff Law <law@redhat.com>
27513 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27514 appearing in DEBUG_INSNs.
27516 2017-04-13 Martin Liska <mliska@suse.cz>
27518 PR gcov-profile/80413
27519 * gcov-io.c (gcov_write_string): Copy to buffer just when
27520 allocated size is greater than zero.
27522 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27525 * dwarf2out.c (decls_for_scope): Ignore declarations of
27526 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27528 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27531 * ipa-visibility.c (non_local_p): Fix typos.
27532 (localize_node): When localizing symbol in same comdat group,
27533 dissolve the group only when we know external symbols are going
27535 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27537 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27539 PR tree-optimization/79390
27540 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27541 order does not result in usable sequence, retry with reversed operand
27547 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27548 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27549 op0 instead of fold_convert_loc (loc, type, arg0).
27551 2017-04-12 Jeff Law <law@redhat.com>
27553 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27554 has a delay slot in the generated code.
27556 * config/cris/cris.md (cris_preferred_reload_class): Return
27557 GENNONACR_REGS rather than GENERAL_REGS.
27559 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27562 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27563 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27564 signedness of the result type.
27566 2017-04-12 Richard Biener <rguenther@suse.de>
27567 Jeff Law <law@redhat.com>
27569 PR tree-optimization/80359
27570 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27571 trim stores to TARGET_MEM_REFs.
27573 2017-04-12 Richard Biener <rguenther@suse.de>
27575 PR tree-optimization/79390
27576 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27577 threading case even more.
27579 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27582 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27583 for quad_address_p for TImode, instead of just not indexed_address.
27585 2017-04-12 Richard Biener <rguenther@suse.de>
27586 Bernd Edlinger <bernd.edlinger@hotmail.de>
27588 PR middle-end/79671
27589 * alias.c (component_uses_parent_alias_set_from): Handle
27590 TYPE_TYPELESS_STORAGE.
27591 (get_alias_set): Likewise.
27592 * tree-core.h (tree_type_common): Add typeless_storage flag.
27593 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27594 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27595 for types containing members with TYPE_TYPELESS_STORAGE.
27596 (place_field): Likewise.
27597 (layout_type): Likewise for ARRAY_TYPE.
27598 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27599 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27600 TYPE_TYPELESS_STORAGE.
27601 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27603 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27606 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27607 first argument to type.
27609 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27613 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27614 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27615 (rs6000_expand_binop_builtin): Likewise.
27616 (rs6000_expand_ternop_builtin): Likewise; also add missing
27617 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27618 vshasigma built-ins.
27619 * doc/extend.texi: Document that vec_xxpermdi's third argument
27620 must be a constant.
27622 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27624 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27625 Use shift_const cost parameter when calculating gain of STV shifts.
27627 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27629 PR rtl-optimization/70478
27630 * lra-constraints.c (process_alt_operands): Check memory for
27631 disfavoring memory insn operand.
27633 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27635 PR middle-end/80100
27636 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27637 left shift in unsigned HOST_WIDE_INT type.
27639 PR rtl-optimization/80385
27640 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27641 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27644 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27645 if they have any depend clauses.
27647 2017-04-11 Martin Liska <mliska@suse.cz>
27650 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27651 * ipa-split.c (split_function): Create a local comdat symbol
27652 if caller is in a comdat group.
27654 2017-04-11 Martin Liska <mliska@suse.cz>
27657 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27660 2017-04-11 Martin Sebor <msebor@redhat.com>
27662 PR middle-end/80364
27663 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27664 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27666 (directive::set_width, directive::set_precision, format_character):
27668 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27671 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27674 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27675 conflict, set target->arch_name instead of target->cpu_name.
27677 2017-04-11 Richard Biener <rguenther@suse.de>
27679 PR tree-optimization/80374
27680 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27681 build_zero_cst, remove fold_convertible_p check again.
27683 2017-04-11 Martin Liska <mliska@suse.cz>
27686 * ubsan.c (instrument_object_size): Do not instrument register
27689 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27692 * config/i386/i386-builtin-types.def
27693 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27694 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27695 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27696 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27697 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27698 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27699 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27700 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27701 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27702 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27703 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27704 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27705 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27706 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27707 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27708 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27709 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27710 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27711 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27712 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27713 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27714 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27715 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27716 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27717 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27718 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27719 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27720 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27721 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27722 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27723 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27724 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27725 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27726 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27727 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27728 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27729 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27730 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27731 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27732 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27733 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27734 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27735 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27736 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27737 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27738 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27740 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27741 flag to second_arg_count, handle 4 argument function type _COUNT
27742 aliases, handle second_arg_count on second argument rather than last.
27744 2017-04-10 Jeff Law <law@redhat.com>
27746 PR tree-optimization/80374
27747 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27748 record anything if we can not convert integer_zero_node to the
27751 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27754 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27755 Enhance special handling given to the TARGET_P9_MINMAX option in
27756 relation to certain other options.
27758 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27760 PR tree-optimization/80153
27761 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27762 remove POINTER_PLUS_EXPR's base part directly, rather than through
27765 2017-04-10 Richard Biener <rguenther@suse.de>
27766 Bin Cheng <bin.cheng@arm.com>
27768 PR tree-optimization/80153
27769 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27770 the first element of pointer type aff_tree. Build result expr in
27772 (add_elt_to_tree): Convert to type unconditionally. Remove other
27773 fold_convert calls.
27774 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27775 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27777 2017-04-10 Richard Biener <rguenther@suse.de>
27779 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27782 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27784 PR rtl-optimization/70478
27785 * lra-constraints.c (curr_small_class_check): New.
27786 (update_and_check_small_class_inputs): New.
27787 (process_alt_operands): Update curr_small_class_check. Disfavor
27788 alternative insn memory operands. Check available regs for small
27791 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27794 * config/mips/mips.opt (-mvirt): Update description.
27795 * doc/invoke.texi (-mvirt): Likewise.
27797 2017-04-10 Richard Biener <rguenther@suse.de>
27799 PR middle-end/80362
27800 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27801 looking for NEGATE_EXPR in -A / -B to A / B folding.
27803 2017-04-10 Martin Liska <mliska@suse.cz>
27805 PR gcov-profile/80224
27806 * gcov.c (print_usage): Fix usage string.
27807 (get_gcov_intermediate_filename): Remove.
27808 (output_gcov_file): Use both for normal and intermediate format.
27809 (generate_results): Do not initialize special file for
27810 intermediate format.
27812 2017-04-10 Richard Biener <rguenther@suse.de>
27814 PR tree-optimization/80304
27815 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27818 2017-04-10 Nathan Sidwell <nathan@acm.org>
27821 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27822 (rs6000_init_builtins): Use it.
27824 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27826 * config/arm/arm.md (<mrc>): Add mode to SET source.
27827 (<mrrc>): Likewise.
27829 2017-04-10 Richard Biener <rguenther@suse.de>
27831 PR middle-end/80344
27832 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27834 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27837 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27838 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27839 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27840 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27841 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27842 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27843 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27844 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27845 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27846 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27847 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27848 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27849 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27850 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27851 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27852 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27853 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27854 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27855 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27856 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27857 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27858 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27859 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27861 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27863 PR rtl-optimization/70478
27864 * lra-constraints.c: Reverse the last patch.
27866 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27868 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27869 Add comment for WCHAR_T.
27871 2017-04-08 Martin Liska <mliska@suse.cz>
27874 2017-04-07 Martin Liska <mliska@suse.cz>
27877 * ipa-split.c (split_function): Add function part to a same comdat
27880 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27883 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27885 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27887 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27888 (vec_pairing): ...to this.
27889 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27890 (rs6000_sched_init): Adjust for name change.
27891 (struct rs6000_sched_context): Likewise.
27892 (rs6000_init_sched_context): Likewise.
27893 (rs6000_set_sched_context): Likewise.
27895 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27901 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27903 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27904 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27905 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27907 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27909 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27911 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27913 PR rtl-optimization/70703
27914 * ira-color.c (update_conflict_hard_regno_costs): Use
27915 int64_t instead of HOST_WIDE_INT.
27917 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27919 PR rtl-optimization/70478
27920 * lra-constraints.c (process_alt_operands): Disfavor alternative
27921 insn memory operands.
27923 2017-04-07 Jeff Law <law@redhat.com>
27925 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27926 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27928 2017-04-07 Martin Liska <mliska@suse.cz>
27931 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27932 Show error message instead of an ICE.
27934 2017-04-07 Martin Liska <mliska@suse.cz>
27937 * ipa-split.c (split_function): Add function part to a same comdat
27940 2017-04-07 Richard Biener <rguenther@suse.de>
27942 PR middle-end/80341
27943 * tree.c (get_unwidened): Also handle ! for_type case for
27945 * convert.c (do_narrow): Split out from ...
27946 (convert_to_integer_1): ... here. Do not pass final truncation
27947 type to get_unwidened for TRUNC_DIV_EXPR.
27949 2017-04-07 Richard Biener <rguenther@suse.de>
27951 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27953 (aff_combination_const): Adjust.
27954 (aff_combination_scale): Likewise.
27955 (aff_combination_add_elt): Likewise.
27956 (aff_combination_add_cst): Likewise.
27957 (aff_combination_convert): Likewise.
27958 (add_elt_to_tree): Likewise. Remove unused argument.
27959 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27961 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27963 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27965 * config/arm/arm.c (arm_default_short_enums): Use
27966 ARM_DEFAULT_SHORT_ENUMS.
27967 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27969 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27972 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27973 members with redundant out-of-class redeclaration.
27975 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27978 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27979 * config/i386/i386.md (*zero_extendsidi2):
27980 Add (?*x,*x) and (?*v,*v) alternatives.
27982 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27985 * config/i386/i386.c (ix86_expand_builtin)
27986 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27987 mode from insn data. Convert operands to insn operand mode.
27988 Copy operands that don't satisfy insn predicate to a register.
27990 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27992 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27995 2017-04-06 Richard Biener <rguenther@suse.de>
27997 PR tree-optimization/80334
27998 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27999 preserve alignment of accesses.
28001 2017-04-06 Richard Biener <rguenther@suse.de>
28003 PR tree-optimization/80262
28004 * tree-sra.c (build_ref_for_offset): Preserve address-space
28006 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28007 Drop useless address-space information on MEM_REF offsets.
28009 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28011 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28013 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28015 PR rtl-optimization/70703
28016 * ira-color.c (update_conflict_hard_regno_costs): Use
28017 HOST_WIDE_INT instead of long.
28019 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28022 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28023 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28025 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28026 for x86_64 target. Handle -m3dnowa option.
28028 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28030 PR rtl-optimization/70703
28031 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28032 (update_conflict_hard_regno_costs): Use long instead of unsigned
28033 arithmetic for cost calculation.
28035 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28036 Bernd Edlinger <bernd.edlinger@hotmail.de>
28039 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28042 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28045 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28046 ptr_mode with Pmode throughout.
28047 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28048 into probe_stack_range and use DImode.
28050 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28053 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28054 call_eh_return is true.
28056 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28058 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28059 Initialize last_match_fntype_index.
28061 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28064 * tree-nvr.c: Include internal-fn.h.
28065 (pass_return_slot::execute): Ignore internal calls without
28068 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28069 Richard Biener <rguenther@suse.de>
28072 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28073 captures used multiple times, except for the last use.
28074 * generic-match-head.c: Include gimplify.h.
28076 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28078 PR tree-optimization/79390
28079 * target.h (struct noce_if_info): Declare.
28080 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28081 * target.def (noce_conversion_profitable_p): New target hook.
28082 * ifcvt.h (struct noce_if_info): New type, moved from ...
28083 * ifcvt.c (struct noce_if_info): ... here.
28084 (noce_conversion_profitable_p): Renamed to ...
28085 (default_noce_conversion_profitable_p): ... this. No longer
28087 (noce_try_store_flag_constants, noce_try_addcc,
28088 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28089 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28090 instead of noce_conversion_profitable_p.
28091 * config/i386/i386.c: Include ifcvt.h.
28092 (ix86_option_override_internal): Don't override
28093 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28094 (ix86_noce_conversion_profitable_p): New function.
28095 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28096 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28097 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28098 * doc/tm.texi: Regenerated.
28100 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28102 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28105 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28108 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28109 instructions for small multiply cores.
28111 2017-04-04 Jeff Law <law@redhat.com>
28113 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28115 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28116 that are not set by the loop over the elements.
28118 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28121 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28122 int mode, convert_modes it to mode as unsigned, otherwise use
28123 lowpart_subreg to mode rather than SImode.
28124 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28125 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28126 Use DImode instead of SImode for the shift count operand.
28127 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28130 2017-04-04 Richard Biener <rguenther@suse.de>
28132 PR middle-end/80281
28133 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28134 arithmetic done for the negate or the plus. Simplify.
28135 (A - (-B) -> A + B): Likewise.
28136 * fold-const.c (split_tree): Make sure to not negate pointers.
28138 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28140 PR rtl-optimization/60818
28141 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28142 a compare of comparisons with the thing compared if this results
28143 in a different machine mode.
28145 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28147 * alias.c (base_alias_check): Fix typo in comment.
28148 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28149 * cgraphunit.c (symbol_table::compile): Likewise.
28150 * collect2.c (maybe_run_lto_and_relink): Likewise.
28151 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28152 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28153 * config/avr/avr.c (avr_map_op_t): Likewise.
28154 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28155 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28156 * config/epiphany/epiphany.md (movcc): Likewise.
28157 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28158 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28160 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28161 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28162 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28163 * config/sh/sh.c (sh_rtx_costs): Likewise.
28164 * fold-const.c (fold_truth_andor): Likewise.
28165 * genautomata.c (collapse_flag): Likewise.
28166 * gengtype.h (struct type::u::s): Likewise.
28167 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28168 * input.c (FORMAT_AMOUNT): Likewise.
28169 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28170 (known_aggs_to_agg_replacement_list): Likewise.
28171 * ipa-inline-analysis.c: Likewise.
28172 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28173 * ipa-polymorphic-call.c
28174 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28175 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28176 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28178 * modulo-sched.c (apply_reg_moves): Likewise.
28179 * omp-expand.c (build_omp_regions_1): Likewise.
28180 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28181 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28182 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28183 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28184 * value-prof.c: Likewise.
28185 * var-tracking.c (val_reset): Likewise.
28187 2017-04-03 Richard Biener <rguenther@suse.de>
28189 PR tree-optimization/80275
28190 * fold-const.c (split_address_to_core_and_offset): Handle
28193 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28195 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28196 descriptors is at least equal to that of functions.
28198 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28200 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28202 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28205 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28206 (mov<IMOD4:mode>): New expander.
28207 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28209 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28211 PR rtl-optimization/79405
28212 * fwprop.c (propagations_left): New variable.
28213 (forward_propagate_into): Decrement it.
28214 (fwprop_init): Initialize it.
28215 (fw_prop): If the variable has reached zero, stop propagating.
28216 (fwprop_addr): Ditto.
28218 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28221 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28222 a FUNCTION_DECL, pass it as decl instead of origin to
28225 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28227 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28230 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28233 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28234 TARGET_VSX_SMALL_INTEGER.
28236 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28238 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28239 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28241 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28243 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28244 extraction from odd-numbered MSA register.
28246 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28248 PR middle-end/80173
28249 * expmed.c (store_bit_field_1): Don't attempt to create
28250 a word subreg out of hard registers wider than word if they
28251 have HARD_REGNO_NREGS of 1 for their mode.
28253 PR middle-end/80163
28254 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28255 conversions to integer types wider than word and pointer.
28258 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28259 (rtx_equal_for_cselib_p): Pass 0 to it.
28260 * cselib.c (cselib_hasher::equal): Likewise.
28261 (rtx_equal_for_cselib_1): Add depth argument. If depth
28262 is 128, don't look up VALUE locs and punt. Increment
28263 depth in recursive calls when walking VALUE locs.
28265 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28267 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28268 (make_gcov_file_name): Use the canonical path name for generating
28270 (read_line): Fix handling of files with ascii null bytes.
28272 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28274 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28275 to initialise a vector register instead
28276 of using a const_int.
28278 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28280 PR translation/80189
28281 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28282 diagnostic messages.
28284 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28287 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28288 (dfp_diex_<mode>): Update mode of operand 1.
28289 * doc/extend.texi (dxex, dxexq): Document change to return type.
28290 (diex, diexq): Document change to argument type.
28292 2017-03-30 Martin Jambor <mjambor@suse.cz>
28295 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28296 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28297 it reflects the signature changes performed at the callee side.
28298 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28299 to cgraph_build_function_type_skip_args.
28300 (build_function_decl_skip_args): Adjust call to the above function.
28302 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28305 * config/i386/sse.md
28306 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28307 register as dest whenever it is a MEM not rtx_equal_p to the
28308 corresponding dup operand, and when forcing into reg move the
28309 reg into the memory afterwards.
28310 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28311 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28312 for the force_reg mode.
28313 (avx512vl_vextractf128<mode>): Use register as dest either
28314 always when a MEM, or when it is a MEM not rtx_equal_p to the
28315 corresponding dup operand, or even not when it is a CONST_VECTOR
28316 depending on the mode and lo vs. hi.
28317 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28319 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28320 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28321 Likewise. Require that operands[2] is even.
28322 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28323 Remove extraneous parens. Require that operands[2] is a multiple
28325 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28326 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28327 disallow memory then.
28329 2017-03-30 Richard Biener <rguenther@suse.de>
28331 PR tree-optimization/77498
28332 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28333 to non-constants over backedges.
28335 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28337 PR rtl-optimization/80233
28338 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28339 as last_combined_insn. Do not test for BARRIER_P separately.
28341 2017-03-29 Andreas Schwab <schwab@suse.de>
28344 * calls.c (prepare_call_address): Convert funexp to Pmode before
28345 copying to temp reg.
28347 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28349 PR tree-optimization/80158
28350 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28351 Handle possible future case of more than one alternate
28353 (replace_rhs_if_not_dup): Likewise.
28354 (replace_one_candidate): Likewise.
28356 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28358 PR rtl-optimization/80193
28359 * ira.c (ira): Do not check allocation for LRA.
28361 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28363 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28364 (nvptx_output_simt_exit): Declare.
28365 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28366 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28367 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28368 (nvptx_declare_function_name): ...here. Emit declaration of local
28369 memory space buffer for omp_simt_enter insn.
28370 (nvptx_output_unisimt_switch): New.
28371 (nvptx_output_softstack_switch): New.
28372 (nvptx_output_simt_enter): New.
28373 (nvptx_output_simt_exit): New.
28374 * config/nvptx/nvptx.h (struct machine_function): New fields
28375 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28376 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28377 (UNSPECV_SIMT_EXIT): Ditto.
28378 (omp_simt_enter_insn): New insn.
28379 (omp_simt_enter): New expansion.
28380 (omp_simt_exit): New insn.
28381 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28383 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28384 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28385 (expand_GOMP_SIMT_EXIT): New.
28386 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28387 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28388 (GOMP_SIMT_EXIT): Ditto.
28389 * target-insns.def (omp_simt_enter): New insn.
28390 (omp_simt_exit): Ditto.
28391 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28393 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28394 (lower_rec_input_clauses): Likewise.
28395 (lower_lastprivate_clauses): Handle SIMT privatization.
28397 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28398 (ompdevlow_adjust_simt_enter): New.
28399 (find_simtpriv_var_op): New.
28400 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28401 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28403 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28404 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28405 (copy_decl_for_dup_finish): Ditto.
28407 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28409 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28412 * config/i386/i386.c (ix86_option_override_internal): Always
28413 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28415 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28417 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28419 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28421 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28422 mark new edge's irreducible flag accordign to it.
28423 (vect_do_peeling): Check loop preheader edge's irreducible flag
28424 and pass it to function slpeel_add_loop_guard.
28426 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28428 PR tree-optimization/80218
28429 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28430 Update block frequencies and counts.
28432 2017-03-28 Richard Biener <rguenther@suse.de>
28434 PR tree-optimization/78644
28435 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28436 of a simplification result we may not use it at all.
28438 2017-03-28 Richard Biener <rguenther@suse.de>
28441 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28442 without arguments, generate default definition of a SSA name.
28444 2017-03-28 Richard Biener <rguenther@suse.de>
28446 PR middle-end/80222
28447 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28448 TYPE_REF_CAN_ALIAS_ALL references.
28449 * fold-const.c (fold_indirect_ref_1): Likewise.
28451 2017-03-28 Martin Liska <mliska@suse.cz>
28454 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28455 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28457 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28458 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28460 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28461 (EXTRA_SPECS): Define.
28462 (SUBTARGET_EXTRA_SPECS): Likewise.
28463 (SUBTARGET_CPP_SPEC): Likewise.
28464 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28465 SUBTARGET_EXTRA_SPECS.
28466 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28468 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28470 * config/arc/simdext.md (vst64_insn): Update pattern.
28471 (vld32wh_insn): Likewise.
28472 (vld32wl_insn): Likewise.
28473 (vld64_insn): Likewise.
28474 (vld32_insn): Likewise.
28476 2017-03-28 Marek Polacek <polacek@redhat.com>
28479 * fold-const.c (fold_comparison): Use protected_set_expr_location
28480 instead of SET_EXPR_LOCATION.
28482 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28484 * tree.c (add_expr): Avoid name lookup warning.
28486 2017-03-27 Jeff Law <law@redhat.com>
28488 PR tree-optimization/80216
28489 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28490 function name. Limit recursion depth.
28491 (record_temporary_equivalences): Corresponding changes.
28493 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28495 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28498 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28501 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28503 * cfgcleanup.c (reg_note_cfa_p): New array.
28504 (insns_have_identical_cfa_notes): New function.
28505 (old_insns_match_p): Don't cross-jump in between /f
28506 and non-/f instructions. If both i1 and i2 are frame related,
28507 verify all CFA notes, their order and content.
28509 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28512 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28513 HImode and SImode with zero extend to DImode to one insn.
28514 (bswap<mode>2_extenddi): Likewise.
28515 (bswapsi2_extenddi): Likewise.
28516 (bswaphi2_extendsi): Likewise.
28517 (bswaphi2): Combine bswap HImode and SImode into one insn.
28518 Separate memory insns from swapping register.
28519 (bswapsi2): Likewise.
28520 (bswap<mode>2): Likewise.
28521 (bswaphi2_internal): Delete, no longer used.
28522 (bswapsi2_internal): Likewise.
28523 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28524 store, and gpr<-gpr swap insns.
28525 (bswap<mode>2_store): Likewise.
28526 (bswaphi2_reg): Register only splitter, combine with the splitter.
28527 (bswaphi2 splitter): Likewise.
28528 (bswapsi2_reg): Likewise.
28529 (bswapsi2 splitter): Likewise.
28530 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28531 the insns into load, store, and register/register insns.
28532 (bswapdi2_ldbrx): Likewise.
28533 (bswapdi2_load): Likewise.
28534 (bswapdi2_store): Likewise.
28535 (bswapdi2_reg): Likewise.
28537 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28539 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28540 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28542 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28545 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28547 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28548 special handling for target option conflicts between dform
28549 options (-mpower9-dform, -mpower9-dform-vector,
28550 -mpower9-dform-scalar) and -mno-direct-move.
28552 2017-03-27 Richard Biener <rguenther@suse.de>
28554 PR tree-optimization/80181
28555 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28557 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28559 * config/arc/predicates.md (move_double_src_operand): Replace the
28560 call to move_double_src_operand with a call to address_operand.
28562 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28564 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28565 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28566 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28568 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28570 * config/arc/predicates.md (long_immediate_loadstore_operand):
28571 Consider scaled addresses cases.
28573 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28575 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28576 restored when in interrupt.
28577 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28578 doesn't have delay slot.
28580 2017-03-27 Richard Biener <rguenther@suse.de>
28583 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28584 inlined thunk clones.
28586 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28589 * asan.c (instrument_derefs): Copy over last operand from
28590 original COMPONENT_REF to the new COMPONENT_REF with
28591 DECL_BIT_FIELD_REPRESENTATIVE.
28592 * ubsan.c (instrument_object_size): Likewise.
28594 2017-03-27 Richard Biener <rguenther@suse.de>
28596 PR tree-optimization/80170
28597 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28598 sure DR/SCEV didnt fold in constants we do not see when looking
28599 at the reference base alignment.
28601 2017-03-27 Richard Biener <rguenther@suse.de>
28603 PR middle-end/80171
28604 * gimple-fold.c (fold_ctor_reference): Properly guard against
28605 NULL return value from canonicalize_constructor_val.
28607 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28610 * config/i386/i386.c (ix86_expand_builtin)
28611 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28612 flags reg setting and flags reg using instructions.
28613 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28614 clobbering instructions to zero extend op2.
28616 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28618 * doc/install.texi (Configuration) <--with-aix-soname>:
28619 Update link to AIX ld.
28621 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28623 PR rtl-optimization/80160
28624 PR rtl-optimization/80159
28625 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28626 reg_alternate_class into account.
28628 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28631 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28632 to consider in curr_insn_transform.
28634 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28636 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28637 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28638 and emit_mode_inner.
28640 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28642 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28643 argument to the overloaded builtin variants. Use the new flag to
28644 deprecate certain builtin variants.
28645 * config/s390/s390-builtin-types.def: Add new builtin types.
28646 * config/s390/s390-builtins.h: Support new flags field for
28647 overloaded builtins.
28648 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28649 (s390_macro_to_expand): Enable vector float data type.
28650 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28651 builtins by incrementing the __VEC__ version number.
28652 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28654 (s390_resolve_overloaded_builtin): Emit error messages depending
28655 on the builtin flags.
28656 * config/s390/s390.c (s390_expand_builtin): Support additional
28657 flags argument. Change error message to match the messages
28658 emitted in s390-c.c.
28659 * config/s390/s390.md: New UNSPEC_* constants.
28660 (op_type): Add new instruction types.
28661 * config/s390/vecintrin.h: Add new builtins and test data class
28663 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28664 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28665 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28666 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28667 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28668 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28670 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28671 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28672 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28673 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28675 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28676 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28677 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28678 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28680 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28681 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28682 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28683 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28684 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28685 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28686 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28688 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28689 ("vec_scatter_element<V_HW_4:mode>_DI")
28690 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28691 ("vec_fpint<mode>", "vflls")
28692 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28693 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28694 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28695 ("*vec_cmphe<mode>_cc"): ... these.
28697 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28698 mode constant instead of magic value.
28700 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28702 * config/s390/s390.c (s390_expand_vec_compare): Support other
28703 vector floating point modes than just V2DF.
28704 (s390_expand_vcond): Likewise.
28705 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28706 (s390_cannot_change_mode_class): Prevent mode changes between TF
28707 and V1TF in vector registers.
28708 * config/s390/s390.md (DF, SF): New mode attributes.
28709 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28710 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28711 SFmode support for VRs.
28712 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28714 (VFT, VF_HW): New mode iterators.
28715 (vw, sdx): New mode attributes.
28716 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28717 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28718 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28719 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28720 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28721 also the new vector floating point modes. Renaming to ...
28723 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28724 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28725 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28726 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28727 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28728 ("vec_unordered<mode>"): ... these.
28730 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28731 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28732 ("*vec_extendv2df"): New insn definitions.
28734 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28736 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28737 ("mulditi3_2", "*muldi3_sign"): New patterns.
28738 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28739 rename the pattern definition.
28741 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28743 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28745 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28747 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28749 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28750 instruction if possible.
28751 * config/s390/vector.md (vec_halfnumelts): New mode
28753 ("*vec_vllezlf<mode>"): New pattern.
28755 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28757 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28758 ("popcountv4si2", "popcountv2di2"): Rename to ...
28759 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28760 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28762 ("popcount<mode>2_vxe"): New pattern.
28764 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28766 * common/config/s390/s390-common.c (processor_flags_table): Add
28768 * config.gcc: Add arch12.
28769 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28770 Default to arch12 for unknown CPU model numbers.
28771 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28772 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28773 PROCESSOR_max sanity check.
28774 * config/s390/s390-opts.h (enum processor_type): Add
28776 * config/s390/s390.c (processor_table): Add arch12.
28777 (s390_expand_builtin): Add check for B_VXE flag.
28778 (s390_issue_rate): Add PROCESSOR_ARCH12.
28779 (s390_get_sched_attrmask): Likewise.
28780 (s390_get_unit_mask): Likewise.
28781 (s390_sched_score): Enable z13 scheduling for arch12.
28782 (s390_sched_reorder): Likewise.
28783 (s390_sched_variable_issue): Likewise.
28784 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28786 (s390_tune_attr): Use z13 scheduling also for arch12.
28787 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28788 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28789 (TARGET_VXE_P): New macros.
28790 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28791 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28792 * config/s390/s390.opt: Add arch12 as processor_type.
28794 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28796 * config/s390/s390.md
28797 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28798 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28799 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28801 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28802 Rename expanders to ...
28804 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28805 ("fixuns_truncdddi2_emu"): ... these.
28807 ("fixuns_trunc<mode>si2_emu"): New expander.
28809 ("*fixuns_truncdfdi2_z13"): Rename to ...
28810 ("*fixuns_truncdfdi2_vx"): ... this.
28812 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28814 * config/s390/2964.md: Remove the single element vector compare
28815 instructions which are no longer used.
28816 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28818 (s390_canonicalize_comparison): Remove handling of DFmode
28820 (s390_expand_vec_compare_scalar): Remove function.
28821 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28822 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28824 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28826 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28828 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28830 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28831 will anyway by matched by mov<mode>_64dfp.
28833 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28835 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28836 vlef/vstef. Add missing operand to vleif.
28838 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28840 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28841 pair for all vector types with 64 bit elements.
28842 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28843 * config/s390/vector.md (V_HW_64): ... here.
28844 (V_128_NOSINGLE): New mode iterator.
28845 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28846 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28847 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28848 ("*vec_load_pairv2di"): Change to ...
28849 ("*vec_load_pair<mode>"): ... this one.
28851 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28853 * config/s390/constraints.md: Add comments.
28854 (jKK): Reject element sizes > 8 bytes.
28855 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28857 * config/s390/s390.md: Add the s_operand checks formerly in
28858 s390_split_ok_p to various splitters where they are still
28860 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28861 for 128 bit vectors. Plus two splitters.
28863 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28865 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28868 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28871 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28872 error if the boundary argument is not constant.
28874 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28876 PR rtl-optimization/80112
28877 * loop-doloop.c (doloop_condition_get): Don't check condition
28878 if cmp isn't SET with IF_THEN_ELSE src.
28880 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28882 PR tree-optimization/80158
28883 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28884 replacing a candidate statement, also replace it for the
28885 candidate's alternate interpretation.
28886 (replace_rhs_if_not_dup): Likewise.
28887 (replace_one_candidate): Likewise.
28889 2017-03-24 Richard Biener <rguenther@suse.de>
28891 PR tree-optimization/80167
28892 * graphite-isl-ast-to-gimple.c
28893 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28895 (translate_isl_ast_to_gimple::get_rename): Likewise.
28897 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28899 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28900 handling of certain combinations of target options, including the
28901 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28902 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28904 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28907 * config/arm/arm.md (*load_multiple): Add reload_completed to
28908 matching condition.
28910 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28911 Richard Biener <rguenth@suse.de>
28913 PR tree-optimization/79908
28914 PR tree-optimization/80136
28915 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28916 been cast away, gimplify_and_add suffices.
28918 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28920 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28922 2017-03-23 Richard Biener <rguenther@suse.de>
28924 PR tree-optimization/80032
28925 * gimplify.c (gimple_push_cleanup): Forced unconditional
28926 cleanups still have to go to the conditional_cleanups
28929 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28931 PR tree-optimization/80072
28932 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28934 (next_operand_entry_id): Change type to unsigned int.
28935 (sort_by_operand_rank): Make sure to return the right return value
28936 even if unsigned fields are bigger than INT_MAX.
28937 (struct oecount): Change cnt and id type to unsigned int.
28938 (oecount_hasher::equal): Formatting fix.
28939 (oecount_cmp): Make sure to return the right return value
28940 even if unsigned fields are bigger than INT_MAX.
28941 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28944 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28945 TREE_READONLY on result if writing it more than once.
28948 * doc/invoke.texi (-fsanitize=thread): Document that with
28949 -fnon-call-exceptions atomics are not able to throw
28953 * tsan.c: Include tree-eh.h.
28954 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28955 maybe_clean_or_replace_eh_stmt where needed.
28956 (instrument_memory_accesses): Add cfg_changed argument.
28957 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28958 if it returned true.
28959 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28961 PR rtl-optimization/63191
28962 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28963 wrapper function, moved the whole old content into ...
28964 (ix86_delegitimize_address_1): ... this. New inline function.
28965 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28966 true as last argument instead of ix86_delegitimize_address.
28968 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28970 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28971 cortexa57_branch_cost.
28973 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28975 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28977 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28980 * doc/md.texi (Constraints): Document wA constraint.
28981 * config/rs6000/constraints.md (wA): New.
28982 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28983 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28984 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28985 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28987 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28990 * gimplify.c (is_oacc_declared): New function.
28991 (oacc_default_clause): Use it to set default flags for acc declared
28992 variables inside parallel regions.
28993 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28994 declared variables.
28995 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28996 declare attribute to any decl as necessary.
28998 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29001 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29002 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29003 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29004 (arm_arch_lpae): This.
29005 * config/arm/arm.c (arm_arch7ve): Rename into ...
29006 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29007 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29010 2017-03-22 Martin Liska <mliska@suse.cz>
29013 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29014 error message instead of an ICE.
29016 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29018 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29020 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29022 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29024 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29027 2017-03-21 Martin Sebor <msebor@redhat.com>
29029 * doc/extend.texi: Use "cannot" instead of "can't."
29030 * doc/hostconfig.texi: Same.
29031 * doc/install.texi: Same.
29032 * doc/invoke.texi: Same.
29033 * doc/loop.texi: Same.
29034 * doc/md.texi: Same.
29035 * doc/objc.texi: Same.
29036 * doc/rtl.texi: Same.
29037 * doc/tm.texi: Same.
29038 * doc/tm.texi.in: Same.
29039 * doc/trouble.texi: Same.
29041 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29044 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29045 (collect_checksum_attributes): Set it.
29046 (die_checksum_ordered): Use it.
29048 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29050 PR tree-optimization/79908
29051 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29052 change: For a VA_ARG whose LHS has been cast away, use
29053 force_gimple_operand to construct the side effects.
29055 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29057 PR translation/80001
29058 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29059 more amenable to translation.
29060 (oacc_loop_auto_partitions): Likewise.
29062 2017-03-21 Marek Polacek <polacek@redhat.com>
29063 Martin Sebor <msebor@redhat.com>
29065 PR tree-optimization/80109
29066 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29067 on INTEGRAL_TYPE_P.
29069 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29070 Segher Boessenkool <segher@kernel.crashing.org>
29073 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29074 check reg_used_between_p between insn and one of succ or succ2
29075 depending on if succ is artificial insn not inserted into insn
29078 2017-03-21 Martin Liska <mliska@suse.cz>
29080 PR gcov-profile/80081
29081 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29082 * doc/gcc.texi: Include gcov-dump stuff.
29083 * doc/gcov-dump.texi: New file.
29085 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29087 PR rtl-optimization/79150
29088 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29089 conditional jump, if the jump is the last insn of the loop.
29091 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29092 Richard Biener <rguenth@suse.de>
29094 PR tree-optimization/79908
29095 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29096 been cast away, use force_gimple_operand to construct the side
29099 2017-03-21 Martin Liska <mliska@suse.cz>
29101 PR libfortran/79956
29102 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29105 2017-03-21 Brad Spengler <spender@grsecurity.net>
29108 * plugin.c (htab_hash_plugin): New function.
29109 (add_new_plugin): Use it and adjust.
29110 (parse_plugin_arg_opt): Adjust.
29111 (init_one_plugin): Likewise.
29113 2017-03-21 Richard Biener <rguenther@suse.de>
29115 PR tree-optimization/80032
29116 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29117 if set force the cleanup to happen unconditionally.
29118 (gimplify_target_expr): Push inserted clobbers with force_uncond
29119 to avoid them being removed by control-dependent DCE.
29121 2017-03-21 Richard Biener <rguenther@suse.de>
29123 PR tree-optimization/80122
29124 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29125 va_arg_pack_len when the inlined call stmt requires pack
29127 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29129 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29132 * tsan.c (instrument_builtin_call): If the memory model argument
29133 is not a constant, assume it is valid.
29136 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29139 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29141 PR rtl-optimization/79910
29142 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29143 if its dest is used by an insn before I2 (other than the combined
29144 insns themselves, which are properly handled already).
29146 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29149 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29151 * combine.c (record_used_regs): New static function.
29152 (try_combine): Handle situations where there is an additional
29153 instruction between I2 and I3 which needs to have a LOG_LINK
29157 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29159 * combine.c (try_combine): Delete redundant i1 test. Call
29160 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29162 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29165 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29166 alternatives 13/14.
29168 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29170 PR tree-optimization/80054
29171 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29172 the optimization if a PHI or any of its arguments is not dominated
29173 by the candidate's basis. Use gphi* rather than gimple* as
29175 (replace_profitable_candidates): Clean up a gimple* variable that
29176 should be a gphi* variable.
29178 2017-03-20 Martin Sebor <msebor@redhat.com>
29181 * doc/extend.texi (attribute constructor): Document present limitation.
29183 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29186 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29187 __POWER9_VECTOR__ #ifdef control, change template definition to
29188 use Power9-specific built-in function.
29189 (vec_any_eq): Likewise.
29190 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29191 to control outcomes from this test.
29192 (vector_ae_<mode>p): For VEC_F modes, likewise.
29194 2017-03-20 Ian Lance Taylor <iant@google.com>
29196 * config/i386/i386.c (ix86_function_regparm): Save an extra
29197 register for -fsplit-stack with DECL_STATIC_CHAIN.
29199 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29202 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29203 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29205 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29207 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29209 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29212 2017-03-20 Marek Polacek <polacek@redhat.com>
29215 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29217 2017-03-20 Richard Biener <rguenther@suse.de>
29219 PR tree-optimization/80113
29220 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29221 allocate extra SSA name for PHI def.
29222 (add_close_phis_to_outer_loops): Likewise.
29223 (add_close_phis_to_merge_points): Likewise.
29224 (copy_loop_close_phi_args): Likewise.
29225 (copy_cond_phi_nodes): Likewise.
29227 2017-03-20 Martin Liska <mliska@suse.cz>
29229 PR middle-end/79753
29230 * tree-chkp.c (chkp_build_returned_bound): Do not build
29231 returned bounds for a LHS that's not a BOUNDED_P type.
29233 2017-03-20 Martin Liska <mliska@suse.cz>
29237 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29238 COMPLEX_CST and VECTOR_CST.
29240 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29243 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29244 target operand. A new splitter adds the clobber statement in case
29245 the target operand is dead anyway.
29247 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29249 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29250 to age-old versions of binutils and glibc.
29252 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29254 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29256 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29258 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29260 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29262 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29263 requirement for binutils 2.13.
29265 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29267 * combine.c (try_combine): Delete redundant i1 test. Call
29268 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29270 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29272 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29273 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29275 <riscv64-*-elf>: Re-arrange section
29276 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29277 <riscv32-*-linux>: Likewise.
29278 <riscv64-*-elf>: Likewise
29279 <riscv64-*-linux>: Likewise.
29281 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29284 * aarch64.opt(verbose-cost-dump): Fix typo.
29286 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29289 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29290 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29292 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29294 * reload.c (find_reloads): When reloading a nonoffsettable address,
29295 use RELOAD_OTHER for it and its address reloads.
29297 PR rtl-optimization/79910
29298 * combine.c (record_used_regs): New static function.
29299 (try_combine): Handle situations where there is an additional
29300 instruction between I2 and I3 which needs to have a LOG_LINK
29303 2017-03-17 Jeff Law <law@redhat.com>
29305 PR tree-optimization/71437
29306 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29307 conditional in the hash table first.
29308 (vrp_dom_walker::before_dom_children): Extract condition from
29309 ASSERT_EXPR. Record condition, its inverion and any implied
29310 conditions as well.
29312 2017-03-17 Marek Polacek <polacek@redhat.com>
29313 Markus Trippelsdorf <markus@trippelsdorf.de>
29315 PR tree-optimization/80079
29316 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29319 2017-03-17 Richard Biener <rguenther@suse.de>
29321 PR middle-end/80075
29322 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29323 Properly verify the LHS before the RHS possibly claims to be
29325 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29328 2017-03-17 Martin Jambor <mjambor@suse.cz>
29330 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29331 (List of -O2 options): Likewise.
29332 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29335 2017-03-17 Tom de Vries <tom@codesourcery.com>
29337 * gcov-dump.c (print_usage): Print bug_report_url.
29339 2017-03-17 Richard Biener <rguenther@suse.de>
29341 PR middle-end/80050
29342 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29343 (parser::peek): Likewise.
29345 2017-03-17 Richard Biener <rguenther@suse.de>
29347 PR tree-optimization/80048
29348 * sese.c (free_sese_info): Properly release rename_map and
29349 copied_bb_map elements.
29351 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29353 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29354 Add linked-list forward and backlinks. Insert on
29355 construction, remove on destruction.
29356 (class pass_store_merging): Add m_stores_head field.
29357 (pass_store_merging::terminate_and_process_all_chains):
29358 Iterate over m_stores_head list.
29359 (pass_store_merging::terminate_all_aliasing_chains):
29361 (pass_store_merging::execute): Check for debug stmts first.
29362 Push new chains onto the m_stores_head stack.
29364 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29367 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29368 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29369 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29371 2017-03-16 Jeff Law <law@redhat.com>
29373 PR tree-optimization/71437
29374 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29375 member function. Implementation moved into after_dom_children
29376 member function and into the threader's thread_outgoing_edges
29378 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29379 some code into new thread_outgoing_edges.
29380 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29381 definition. Simplify marker handling (do it here). Assume we always
29382 have the available expression and the const/copies tables.
29383 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29385 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29386 * tree-vrp.c (equiv_stack): No longer file scoped.
29387 (vrp_dom_walker): New class.
29388 (vrp_dom_walker::before_dom_children): New member function.
29389 (vrp_dom_walker::after_dom_children): Likewise.
29390 (identify_jump_threads): Setup domwalker. Use it rather than
29391 walking edges in a random order by hand. Simplify setup/finalization.
29392 (finalize_jump_threads): Remove.
29393 (vrp_finalize): Do not call identify_jump_threads here.
29394 (execute_vrp): Do it here instead and call thread_through_all_blocks
29397 PR tree-optimization/71437
29398 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29400 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29402 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29403 (dom_opt_dom_walker::thread_across_edge): Remove
29404 handle_dominating_asserts argument. All callers changed.
29405 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29406 changes. Remove calls to lhs_of_dominating_assert. Other
29407 uses of handle_dominating_asserts turn into unconditional code
29408 (simplify_control_stmt_condition_1): Likewise.
29409 (simplify_control_stmt_condition): Likewise.
29410 (thread_through_normal_block, thread_across_edge): Likewise.
29411 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29412 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29413 object if it is not an SSA_NAME.
29414 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29415 before calling into the VRP specific simplifiers.
29416 (identify_jump_threads): Remove handle_dominating_asserts
29419 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29422 * tree-diagnostic.c (default_tree_printer): No longer static.
29423 * tree-diagnostic.h (default_tree_printer): New prototype.
29425 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29427 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29428 Change ins into fmov.
29430 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29432 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29433 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29434 Use h_con constraint for operand 1.
29435 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29436 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29438 2017-03-15 Jeff Law <law@redhat.com>
29440 PR tree-optimization/71437
29441 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29442 (record_temporary_equivalences): Use it.
29444 PR tree-optimization/71437
29445 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29446 tree-ssa-scopedtables.
29447 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29448 (record_conditions, record_cond, vuse_eq): Likewise.
29449 (record_edge_info): Adjust to API tweak of record_conditions.
29450 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29451 (record_temporary_equivalences, optimize_stmt): Likewise.
29452 (eliminate_redundant_computations): Likewise.
29453 (record_equivalences_from_stmt): Likewise.
29454 * tree-ssa-scopedtables.c: Include options.h and params.h.
29455 (vuse_eq): New function, moved from tree-ssa-dom.c
29456 (build_and_record_new_cond): Likewise.
29457 (record_conditions): Likewise. Accept vector of conditions rather
29458 than edge_equivalence structure for first argument.
29459 for the first argument.
29460 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29461 from tree-ssa-dom.c.
29462 (avail_exprs_stack::record_cond): Likewise.
29463 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29464 from tree-ssa-dom.c.
29465 (avail_exprs_stack): Add new member functions lookup_avail_expr
29467 (record_conditions): Declare.
29469 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29472 * lra-constraints.c (process_alt_operands): Increase reject for
29473 reloading an input/output operand.
29475 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29478 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29479 insns to convert from signed/unsigned char/short to IEEE 128-bit
29481 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29483 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29486 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29487 subreg of inner mode for values already in registers.
29489 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29491 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29492 iteration reg is used after the loop.
29494 2017-03-14 Martin Sebor <msebor@redhat.com>
29496 PR tree-optimization/79800
29497 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29498 precision in negative-positive range.
29499 (format_floating): Call non-const overload with adjusted precision.
29501 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29504 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29507 2017-03-14 Martin Sebor <msebor@redhat.com>
29509 PR middle-end/80020
29510 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29511 * builtins.def (aligned_alloc): Use it.
29514 * Makefile.in (GTFILES): Add calls.c.
29515 * calls.c: Include "gt-calls.h".
29517 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29519 PR rtl-optimization/79728
29520 * regs.h (struct target_regs): New field
29521 x_contains_allocatable_regs_of_mode.
29522 (contains_allocatable_regs_of_mode): New macro.
29523 * reginfo.c (init_reg_sets_1): Initialize it, and change
29524 contains_reg_of_mode so it includes global regs as well.
29525 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29526 rather than contains_regs_of_mode.
29528 2017-03-14 Martin Liska <mliska@suse.cz>
29530 * doc/invoke.texi: Document options that can't be combined with
29531 -fcheck-pointer-bounds.
29533 2017-03-14 Martin Liska <mliska@suse.cz>
29535 PR middle-end/79831
29536 * doc/invoke.texi (-Wchkp): Document the option.
29538 2017-03-14 Martin Liska <mliska@suse.cz>
29540 * Makefile.in: Install gcov-dump.
29542 2017-03-14 Martin Liska <mliska@suse.cz>
29544 * multiple_target.c (expand_target_clones): Bail out for
29545 an invalid attribute.
29547 2017-03-14 Richard Biener <rguenther@suse.de>
29549 * alias.c (struct alias_set_entry): Pack properly.
29550 * cfgloop.h (struct loop): Likewise.
29551 * cse.c (struct set): Likewise.
29552 * ipa-utils.c (struct searchc_env): Likewise.
29553 * loop-invariant.c (struct invariant): Likewise.
29554 * lra-remat.c (struct cand): Likewise.
29555 * recog.c (struct change_t): Likewise.
29556 * rtl.h (struct address_info): Likewise.
29557 * symbol-summary.h (function_summary): Likewise.
29558 * tree-loop-distribution.c (struct partition): Likewise.
29559 * tree-object-size.c (struct object_size_info): Likewise.
29560 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29561 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29562 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29563 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29564 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29565 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29566 (struct _stmt_vec_info): Likewise.
29568 2017-03-14 Martin Liska <mliska@suse.cz>
29571 * multiple_target.c (create_dispatcher_calls): Check that
29572 a target can create a function dispatcher.
29574 2017-03-14 Martin Liska <mliska@suse.cz>
29577 * multiple_target.c (expand_target_clones): Drop local.local
29578 flag for default implementation.
29580 2017-03-14 Richard Biener <rguenther@suse.de>
29582 PR tree-optimization/80030
29583 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29585 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29587 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29588 gcc_fallthrough() instead of __attribute__((fallthrough));
29590 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29592 * doc/gcc.texi: Remove "up" link to (DIR).
29593 * doc/gccint.texi: Ditto.
29595 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29597 * doc/install.texi (Specific) <avr>: Remove reference to
29600 2017-03-13 Jeff Law <law@redhat.com>
29602 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29603 attribute rather than comments.
29605 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29606 match_scratch operand is highest.
29608 2017-03-13 Martin Liska <mliska@suse.cz>
29610 PR middle-end/78339
29611 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29612 is a CHKP clone, use original declaration.
29614 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29616 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29617 (arc_conditional_register_usage): Use a different allocation order
29618 when optimizing for size.
29619 * common/config/arc/arc-common.c (arc_option_optimization_table):
29620 Section anchors default on when optimizing for size.
29622 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29624 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29626 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29628 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29629 * config/arc/arc.md (cpu_facility): Add cd variant.
29630 (*movqi_insn): Add code density variant.
29631 (*movhi_insn): Likewise.
29632 (*movqi_insn): Likewise.
29633 (*addsi3_mixed): Likewise.
29634 (subsi3_insn): Likewise.
29636 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29638 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29640 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29642 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29643 expressions with MINUS and UNARY ops.
29645 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29648 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29650 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29651 between vec_select and vector argument.
29652 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29653 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29654 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29655 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29656 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29657 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29659 2017-03-13 Richard Biener <rguenther@suse.de>
29662 * params.def (vect-max-peeling-for-alignment): Fix typo.
29664 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29666 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29667 issue that only occurred with binutils below 2.18.
29669 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29671 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29672 refer to binutils 2.11/2.12 minimum.
29674 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29676 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29677 ftp.kernel.org and simplify binutils requirement.
29679 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29681 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29683 (Optimize Options): Ditto. Also remove redundancy.
29685 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29687 PR translation/79848
29688 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29690 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29691 to G_ to avoid double translation.
29693 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29695 PR translation/79923
29696 * auto-profile.c (get_combined_location): Convert leading
29697 character of diagnostics to lower case and remove trailing period.
29698 (read_profile): Likewise for various diagnostics.
29699 * config/arm/arm.c (arm_option_override): Remove trailing period
29700 from various diagnostics.
29701 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29702 (msp430_expand_delay_cycles): Likewise.
29704 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29707 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29708 full command-line argument, rather than just "str".
29709 (aarch64_validate_march): Likewise.
29710 (aarch64_validate_mtune): Likewise.
29712 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29714 PR rtl-optimization/78911
29715 * lra-assigns.c (must_not_spill_p): New function.
29716 (spill_for): Use it.
29718 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29720 PR tree-optimization/79981
29721 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29722 ATOMIC_COMPARE_EXCHANGE ifn result.
29723 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29724 IFN_ATOMIC_COMPARE_EXCHANGE.
29726 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29729 * opts.c (parse_sanitizer_options): Add missing question mark to
29730 "did you mean" message.
29732 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29734 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29736 (VMULEUH_UNS): Likewise.
29737 (VMULOUB_UNS): Likewise.
29738 (VMULOUH_UNS): Likewise.
29739 * config/rs6000/rs6000.c (builtin_function_type): Remove
29740 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29742 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29745 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29746 x with result of extra_parsing_for_operand_code_0.
29747 (function_reader::extra_parsing_for_operand_code_0): Convert
29748 return type from void to rtx, returning x. When reading
29749 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29750 larger size containing struct block_symbol.
29752 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29754 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29755 -mfloat128-hardware without -m64.
29757 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29760 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29761 entries to the case statement that marks unsigned arguments to
29762 overloaded functions.
29764 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29766 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29767 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29769 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29772 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29773 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29775 2017-03-10 Martin Liska <mliska@suse.cz>
29779 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29780 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29783 2017-03-10 Olivier Hainque <hainque@adacore.com>
29785 * tree-switch-conversion (array_value_type): Start by resetting
29786 candidate type to it's main variant.
29788 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29790 PR rtl-optimization/79909
29791 * combine.c (try_combine): Use simplify_replace_rtx on individual
29792 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29793 of the whole CALL_INSN_FUNCTION_USAGE.
29795 PR tree-optimization/79972
29796 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29797 get_range_info on SSA_NAMEs. Formatting fixes.
29799 2017-03-10 Richard Biener <rguenther@suse.de>
29800 Jakub Jelinek <jakub@redhat.com>
29802 PR tree-optimization/77975
29803 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29804 edge to be constant.
29805 (get_val_for): For constant x return it. Formatting fix.
29806 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29807 would use the same bases as the current one.
29809 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29811 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29812 instead of vec_select for V1TImode.
29813 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29815 (VSX_LE_128): Add V1TI to this mode iterator.
29816 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29817 (*vsx_le_perm_store_<mode>): Likewise.
29818 (pre-reload splitter for VSX stores): Likewise.
29819 (post-reload splitter for VSX stores): Likewise.
29820 (*vsx_xxpermdi2_le_<mode>): Likewise.
29821 (*vsx_lxvd2x2_le_<mode>): Likewise.
29822 (*vsx_stxvd2x2_le_<mode>): Likewise.
29824 2017-03-09 Michael Eager <eager@eagercon.com>
29826 Correct failures with --enable-checking=yes,rtl.
29828 * config/microblaze/microblaze.c (microblaze_expand_shift):
29829 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29830 test for const0_rtx.
29831 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29832 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29834 2017-03-09 Richard Biener <rguenther@suse.de>
29836 PR tree-optimization/79977
29837 * graphite-scop-detection.c (scop_detection::merge_sese):
29838 Handle the case of extra exits to blocks dominating the entry.
29840 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29842 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29845 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29847 PR rtl-optimization/79949
29848 * lra-constraints.c (process_alt_operands): Check memory when
29849 trying to predict a cycle. Print about the overall increase.
29851 2017-03-09 Richard Biener <rguenther@suse.de>
29853 PR middle-end/79971
29854 * gimple-expr.c (useless_type_conversion_p): Preserve
29855 TYPE_SATURATING for fixed-point types.
29857 2017-03-09 Richard Biener <rguenther@suse.de>
29860 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29861 alignment of BLKmode params.
29863 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29866 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29867 (VALL_NO_V2Q): Likewise.
29869 * config/aarch64/aarch64-simd.md
29870 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29872 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29873 VALL_NO_V2Q mode iterator.
29874 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29876 2017-03-09 Martin Liska <mliska@suse.cz>
29878 PR tree-optimization/79631
29879 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29880 tree_int_cst_sign_bit just for INTEGER constants.
29882 2017-03-09 Martin Liska <mliska@suse.cz>
29886 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29889 2017-03-09 Marek Polacek <polacek@redhat.com>
29892 * tree.c (inchash::add_expr): Handle TREE_VEC.
29894 2017-03-09 Martin Liska <mliska@suse.cz>
29897 (chkp_narrow_size_and_offset): New function.
29898 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29899 (void chkp_parse_bit_field_ref): New function.
29900 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29901 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29903 2017-03-09 Martin Liska <mliska@suse.cz>
29906 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29907 (chkp_find_bounds_1): Remove gcc_unreachable.
29909 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29912 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29913 BUILT_IN_SYNC*, determine the access type from the size suffix and
29914 always build a MEM_REF with that type. Handle forgotten
29915 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29918 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29919 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29920 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29921 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29922 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29923 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29924 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29925 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29926 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29927 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29928 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29929 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29930 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29931 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29932 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29933 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29934 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29935 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29936 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29937 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29938 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29939 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29940 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29941 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29942 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29943 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29944 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29945 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29946 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29947 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29948 definitions outside of __OPTIMIZE__ guarded section.
29951 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29952 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29953 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29954 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29957 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29959 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29960 ("vfenez<mode>"): Add missing constraints.
29962 2017-03-08 Martin Sebor <msebor@redhat.com>
29965 * config/nds32/nds32.c (nds32_option_override):
29966 Fix misspelled diagnostic.
29968 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29971 * gimplify.c (gimplify_omp_for): Replace index var in outer
29972 taskloop statement with an artificial variable and add
29973 OMP_CLAUSE_PRIVATE clause for it.
29975 2017-03-08 Richard Biener <rguenther@suse.de>
29977 PR tree-optimization/79955
29978 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29979 for accesses that are completely outside of the variable.
29981 2017-03-08 Andrew Haley <aph@redhat.com>
29983 PR tree-optimization/79943
29984 * tree-ssa-loop-split.c (compute_new_first_bound): When
29985 calculating the new upper bound, (END-BEG) should be added, not
29988 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29990 * config/avr/avr.md (setmemhi): Make sure match_dup
29991 operand number comes before match_scratch.
29993 2017-03-08 Richard Biener <rguenther@suse.de>
29995 PR tree-optimization/79920
29996 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29997 with ncopies == 1 to ...
29998 (vect_transform_slp_perm_load): ... here. Properly compute
29999 all element loads by iterating VF times over the group. Do
30000 not handle ncopies (computed in a broken way) in
30001 vect_create_mask_and_perm.
30003 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30006 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30007 is a uniform vector, use uniform_vector_p return value instead of
30008 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30010 2017-03-07 Marek Polacek <polacek@redhat.com>
30012 PR middle-end/79809
30013 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30014 (alloca_call_type): Likewise.
30016 2017-03-07 Martin Liska <mliska@suse.cz>
30018 * gcov.c (process_args): Put comment to correct location.
30020 2017-03-07 Martin Liska <mliska@suse.cz>
30022 PR middle-end/68270
30023 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30024 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30025 (chkp_narrow_bounds_for_field): Likewise.
30026 (chkp_parse_array_and_component_ref): Pass one more argument to
30029 2017-03-07 Richard Biener <rguenther@suse.de>
30031 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30034 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30036 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30037 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30039 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30042 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30043 to end of description.
30044 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30046 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30048 PR rtl-optimization/79901
30049 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30051 (*avx512f_<code><mode>3<mask_name>): ... this.
30052 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30053 iterator instead of VI8_AVX2_AVX512BW.
30055 PR rtl-optimization/79901
30056 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30057 min/max expander, expand it using expand_vec_cond_expr.
30060 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30063 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30066 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30067 to void * for PCH reasons.
30068 * dwarf2out.c (output_loc_operands, output_die): Cast
30069 v.val_vec.array to unsigned char *.
30071 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30074 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30077 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30079 PR rtl-optimization/79571
30080 * lra-constraints.c (process_alt_operands): Calculate static
30081 reject and subtract it from overall when only addresses will be
30084 2017-03-06 Julia Koval <julia.koval@intel.com>
30087 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30088 incoming stack boundary to 128 for 64-bit targets.
30090 2017-03-06 Richard Biener <rguenther@suse.de>
30092 PR tree-optimization/79894
30093 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30094 to NULL after folding it.
30096 2017-03-06 Richard Biener <rguenther@suse.de>
30098 PR tree-optimization/79824
30099 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30100 check disabling peeling for gaps.
30102 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30104 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30105 attributes): Document gettimeofday.
30107 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30109 * config/s390/s390.c (s390_option_override_internal): Set
30110 PARAM_MIN_VECT_LOOP_BOUND
30112 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30114 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30115 * config/s390/s390.md: Likewise.
30117 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30120 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30121 (<avx2_avx512>_perm<mode>): Rename to ...
30122 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30124 (<avx512>_perm<mode>_mask): Rename to ...
30125 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30127 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30128 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30129 instead of VI8F_256_512.
30130 (avx512f_perm<mode>): New define_expand.
30131 (avx512f_perm<mode>_mask): Likewise.
30132 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30133 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30135 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30137 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30138 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30140 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30142 2017-03-06 Martin Liska <mliska@suse.cz>
30145 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30146 when having a SSA NAME w/o VAR_DECL assigned to it.
30148 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30150 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30151 msa_dpsub_<su>_d): Fix MODE for vec_select.
30153 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30155 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30157 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30159 2017-03-06 Richard Biener <rguenther@suse.de>
30161 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30162 * plugin.c (register_plugin_info): Likewise.
30163 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30165 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30167 * config/i386/sse.md (sse_storehps, sse_storelps,
30168 avx_<castmode><avxsizesuffix>_<castmode>,
30169 avx512f_<castmode><avxsizesuffix>_<castmode>,
30170 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30171 in condition that at least one operand is not a MEM.
30173 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30175 PR middle-end/79805
30176 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30177 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30179 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30180 gimple_call_nothrow_p flag based on whether original builtin can throw.
30181 If it can, emit following stmts on the fallthrough edge.
30182 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30183 don't create new bb if inserting just debug stmts on the edge, try to
30184 insert them on the fallthru bb or just reset debug stmts.
30186 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30189 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30190 restore recog_data (including the operand rtxes inside it) around
30191 the call to get_insn_template.
30193 2017-03-03 Martin Sebor <msebor@redhat.com>
30195 PR tree-optimization/79699
30196 * context.c (context::~context): Free MPFR caches to avoid
30197 a memory leak on program exit.
30199 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30201 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30202 Use wide_int::ulow () instead of .elt (0).
30204 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30206 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30207 (*pushxf): Limit oF constraint to 32bit targets and add oC
30208 constraint for 64bit targets.
30209 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30210 (*pushdf): Change rmF constraint to rmC.
30212 2017-03-03 Martin Liska <mliska@suse.cz>
30214 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30215 Remove unused variable.
30217 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30220 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30221 is a memory operand, increase num_memory.
30222 (ix86_expand_args_builtin): Likewise.
30224 2017-03-03 Jan Hubicka <jh@suse.cz>
30227 * ipa-devirt.c (maybe_record_node): Properly handle
30228 __cxa_pure_virtual visibility.
30230 2017-03-03 Martin Liska <mliska@suse.cz>
30232 PR tree-optimization/79803
30233 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30235 (pass_loop_prefetch::execute): Disabled optimization if an
30236 assumption about L1 cache size is not met.
30238 2017-03-03 Martin Liska <mliska@suse.cz>
30240 PR rtl-optimization/79574
30241 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30242 (hash_scan_set): Likewise.
30243 (dump_hash_table): Likewise.
30244 (hoist_code): Likewise.
30246 2017-03-03 Richard Biener <rguenther@suse.de>
30248 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30249 in place of uhigh ().
30250 (fixed_convert_from_real): Likewise.
30252 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30255 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30257 2017-03-03 Richard Biener <rguenther@suse.de>
30259 PR middle-end/79818
30260 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30261 TYPE_OVERFLOW_UNDEFINED check.
30263 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30265 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30267 (vector_ae_<mode>_p): Likewise.
30268 (vector_nez_<mode>_p): Likewise.
30269 (vector_ne_v2di_p): Likewise.
30270 (vector_ae_v2di_p): Likewise.
30271 (vector_ne_<mode>_p): Likewise.
30272 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30274 (vsx_tsqrt<mode>2_fe): Likewise.
30276 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30279 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30281 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30283 PR rtl-optimization/79780
30284 * cprop.c (one_cprop_pass): When second and further conditional trap
30285 in a single basic block is turned into an unconditional trap, turn it
30286 into a deleted note to avoid RTL verification failures.
30288 2017-03-02 Richard Biener <rguenther@suse.de>
30290 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30292 2017-03-02 Richard Biener <rguenther@suse.de>
30294 PR tree-optimization/79345
30296 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30297 param and abort the walk, returning -1 if it is hit.
30298 (walk_aliased_vdefs): Take a limit param and pass it on.
30299 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30300 defaulting to 0 and return a signed int.
30301 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30302 (check_defs): New helper.
30303 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30304 about uninitialized memory.
30305 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30306 bogus uninitialized warning.
30307 (fixed_convert_from_real): Likewise.
30309 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30311 PR tree-optimization/66768
30312 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30313 iv_use if base object can't be determined.
30315 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30317 PR tree-optimization/79345
30318 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30319 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30320 (get_pattern_stats): Initialize it.
30321 * genemit.c (gen_expand): Verify match_scratch numbers come after
30322 match_operand/match_dup numbers.
30323 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30324 match_scratch numbers.
30325 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30327 * config/s390/s390.md (trunctdsd2): Likewise.
30329 2017-03-02 Richard Biener <rguenther@suse.de>
30331 * wide-int.h (wide_int_storage::operator=): Implement in terms
30334 2017-03-02 Richard Biener <rguenther@suse.de>
30336 PR tree-optimization/79777
30337 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30338 the to insert expression to sth existing.
30340 2017-03-01 Martin Sebor <msebor@redhat.com>
30342 PR middle-end/79692
30343 * gimple-ssa-sprintf.c
30344 (directive::known_width_and_precision): New function.
30345 (format_integer): Use it.
30346 (get_mpfr_format_length): Consider the full range of precision
30347 when computing %g output with the # flag. Set the likely byte
30348 count to 3 rather than 1 when precision is indeterminate.
30349 (format_floating): Correct the lower bound of precision.
30351 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30353 * doc/invoke.texi: Document default code model for 64-bit Linux.
30355 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30358 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30359 udiv rather than div since input pattern is unsigned.
30361 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30363 * config/i386/i386.c (print_reg): Warn for values of
30364 unsupported size in integer register.
30366 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30369 * config/rs6000/predicates.md (current_file_function_operand): Do
30370 not allow self calls to be local if the function is replaceable.
30372 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30375 * config/rs6000/altivec.h (vec_ctz and others): Change the
30376 preprocessor macro that controls conditional compilation from
30377 _ARCH_PWR9 to __POWER9_VECTOR__.
30378 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30379 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30380 control (instead of _ARCH_PWR9 control) so that template
30381 definition uses power9-specific function.
30382 (vec_any_eq): Likewise.
30383 (vec_all_ne): Change macro definition to use a power9-specific
30384 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30385 _ARCH_PWR9 control).
30386 (vec_any_eq) Likewise.
30387 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30388 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30389 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30390 support for xvcmpnedp instruction.
30391 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30392 macro expansion so that Power9 implementation of vec_all_ne does
30393 not use the AltiVec predicate framework.
30394 (VCMPNEH_P): Likewise.
30395 (VCMPNEW_P): Likewise.
30396 (VCMPNED_P): Likewise.
30397 (VCMPNEFP_P): Likewise.
30398 (VCMPNEDP_P): Likewise.
30399 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30400 implementation of vec_any_eq to not use AltiVec predicate
30402 (VCMPAEH_P): Likewise.
30403 (VCMPAEW_P): Likewise.
30404 (VCMPAED_P): Likewise.
30405 (VCMPAEFP_P): Likewise.
30406 (VCMPAEDP_P): Likewise.
30407 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30408 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30409 not use the AltiVec predicate framework.
30410 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30411 of vec_any_eq to not use AltiVec predicate framework.
30412 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30413 support for predefined __POWER9_VECTOR__ macro to indicate that
30414 Power9 instruction selection is enabled.
30415 (altivec_overloaded_builtins): Remove extraneous
30416 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30417 function argument types RS6000_BTI_bool_V16QI and
30418 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30419 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30420 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30421 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30422 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30423 Power9 for implementations of vec_cmpne. Change the signature for
30424 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30425 (representing vec_all_ne) to remove the previously described first
30426 argument of type RS6000_BTI_INTSI, as this was an artifact of
30427 reliance on the AltiVec predicate framework, which is no longer
30428 used in the implementation of these functions. Add
30429 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30430 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30431 since, unlike the AltiVec predicate framework implementation, we
30432 do not share function descriptors between vec_alle and vec_anyeq.
30433 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30434 set of modes that receive special treatment even when
30435 TARGET_P9_VECTOR is true. The special treatment emits code that
30436 does not depend on Power9 instructions.
30437 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30438 define_expand to not rely on AltiVec predicate framework.
30439 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30441 (vector_ne_v2di_p): Change this define_expand to not rely on
30442 AltiVec predicate framework.
30443 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30445 (vector_ne_<mode>_p): Change this define_expand to not rely on
30446 AltiVec predicate framework.
30447 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30449 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30450 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30451 define_insn pattern.
30452 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30453 define_insn pattern because the xvcmpne<VSs>. instruction is not
30455 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30456 instruction is not supported.
30458 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30460 * config/nvptx/nvptx.c: Include intl.h.
30462 2017-03-01 Martin Jambor <mjambor@suse.cz>
30465 * ipa-prop.h (ipa_bits): Removed field known.
30466 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30467 to pointers. Adjusted their comments to warn about their sharing.
30468 (ipcp_transformation_summary): Change bits to a vector of pointers.
30469 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30470 (ipa_get_ipa_bits_for_value): Declare.
30471 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30472 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30473 (ipa_bits_hash_table): Likewise.
30474 (ipa_vr_ggc_hash_traits): Likewise.
30475 (ipa_vr_hash_table): Likewise.
30476 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30477 being pointers and vr_known being removed.
30478 (ipa_set_jf_unknown): Likewise.
30479 (ipa_get_ipa_bits_for_value): New function.
30480 (ipa_set_jfunc_bits): Likewise.
30481 (ipa_get_value_range): New overloaded functions.
30482 (ipa_set_jfunc_vr): Likewise.
30483 (ipa_compute_jump_functions_for_edge): Use the above functions to
30484 construct bits and vr parts of jump functions.
30485 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30486 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30488 (ipcp_grow_transformations_if_necessary): Also allocate
30489 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30491 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30492 them. Fix too long lines.
30493 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30494 vr_known being removed.
30495 (ipa_read_jump_function): Use new setter functions to construct bits
30496 and vr parts of jump functions or set them to NULL.
30497 (write_ipcp_transformation_info): Adjust for bits being pointers.
30498 (read_ipcp_transformation_info): Likewise.
30499 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30501 Include gt-ipa-prop.h.
30502 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30504 (ipcp_store_bits_results): Likewise.
30505 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30506 Do not write to existing jump functions but use a temporary instead.
30508 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30511 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30512 attempt to use its first operand as BIT_FIELD_REF base.
30514 2017-03-01 Richard Biener <rguenther@suse.de>
30516 PR middle-end/79721
30517 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30518 interpolating formula in wrapping arithmetic.
30519 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30521 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30523 PR tree-optimization/79734
30524 * tree-vect-generic.c (expand_vector_condition): Optimize
30525 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30526 Handle VEC_COND_EXPR where comparison has different inner width from
30527 type's inner width.
30529 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30531 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30532 markup, and similar issues. Remove @opindex entries for things
30533 that aren't options. Add missing -mmpy-option entries.
30535 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30537 PR tree-optimization/79737
30538 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30539 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30540 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30541 instead of byte_size. Formatting fix.
30542 (shift_bytes_in_array_right): Formatting fix.
30544 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30547 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30548 condition on optimize for the leaf function test.
30550 2017-02-28 Martin Liska <mliska@suse.cz>
30553 * read-rtl-function.c (function_reader::handle_unknown_directive):
30554 Bail out when one uses -flto.
30556 2017-02-28 Martin Liska <mliska@suse.cz>
30558 * common.opt: Replace space with tabular for options of <number>
30560 * config/i386/i386.opt: Show <number> value for
30561 -mlarge-data-threshold.
30562 * opts.c (print_filtered_help): Do not display number in hexadecimal
30565 2017-02-28 Martin Liska <mliska@suse.cz>
30567 * common.opt: Fix --help=option -Q for options which are of
30570 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30572 * config/i386/i386.c (print_reg): Error out for values
30573 of 8-bit size in invalid integer register.
30575 2017-02-28 Martin Sebor <msebor@redhat.com>
30577 PR tree-optimization/79691
30578 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30580 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30583 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30584 gcc_unreachable with output_operand_lossage.
30586 2017-02-28 Richard Biener <rguenther@suse.de>
30588 PR tree-optimization/79740
30589 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30591 (visit_nary_op): Insert the nary into the hashtable if we
30592 pattern-matched sth.
30593 * tree-ssa-pre.c (eliminate_insert): Robustify.
30595 2017-02-28 Richard Biener <rguenther@suse.de>
30597 PR middle-end/79731
30598 * fold-const.c (decode_field_reference): Reject out-of-bound
30601 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30603 * config/i386/i386.c: Include intl.h.
30604 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30605 instead of just cond ? "..." : "...".
30606 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30607 * coverage.c (read_counts_file): Likewise.
30608 * omp-offload.c: Include intl.h.
30609 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30610 of just cond ? "..." : "...".
30611 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30612 of just cond ? "..." : "...".
30614 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30617 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30619 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30620 'tune for' CPU name.
30621 * config/arm/arm-cpu-data.h: Regenerated.
30623 2017-02-28 Richard Biener <rguenther@suse.de>
30625 PR tree-optimization/79732
30626 * tree-inline.c (expand_call_inline): Do not shadow var.
30628 2017-02-28 Richard Biener <rguenther@suse.de>
30630 PR tree-optimization/79723
30631 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30632 address-space properly.
30634 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30636 * doc/optinfo.texi (Optimization groups): Fix option used for
30638 * doc/invoke.texi (-fopt-info): Document "omp".
30639 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30640 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30641 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30642 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30643 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30645 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30647 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30648 OPTGROUP_OMP with "omp".
30650 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30653 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30654 for arithmetic shift of unsigned V2DI.
30656 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30658 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30659 arc/linux.h headers.
30660 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30661 (LINK_SPEC): Likewise.
30662 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30663 (EXTRA_SPECS): Likewise.
30664 (STARTFILE_SPEC): Likewise.
30665 (ENDFILE_SPEC): Likewise.
30666 (LIB_SPEC): Likewise.
30667 (TARGET_SDATA_DEFAULT): Likewise.
30668 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30669 (MULTILIB_DEFAULTS): Likewise.
30670 (DWARF2_UNWIND_INFO): Likewise.
30671 * config/arc/big.h: New file.
30672 * config/arc/elf.h: Likewise.
30673 * config/arc/linux.h: Likewise.
30674 * config/arc/t-uClibc: Remove.
30676 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30678 PR tree-optimization/77536
30679 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30680 (tree_transform_and_unroll_loop): Use above function to compute the
30681 estimated niter of unrolled loop and use it when scaling profile.
30682 Also use count info rather than frequency if it's non-zero.
30683 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30684 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30685 (vect_transform_loop): Call above function.
30687 2017-02-27 Richard Biener <rguenther@suse.de>
30689 PR tree-optimization/45397
30690 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30691 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30692 (visit_nary_op): Add pattern matching for CSEing sign-changed
30693 or truncated operations with wider ones.
30695 2017-02-27 Richard Biener <rguenther@suse.de>
30697 PR tree-optimization/79690
30698 * tree-vect-stmts.c (vectorizable_store): Use vector type
30699 built from the DR with address-space.
30701 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30703 * doc/invoke.texi (Optimize Options): Refine the description
30704 of asan-use-after-return.
30706 2017-02-25 Alan Modra <amodra@gmail.com>
30708 PR rtl-optimization/79584
30709 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30710 base, not ad->base_term, the reg within base. Remove assertion
30711 that ad->base == ad->base_term. Replace gen_int_mode using
30712 bogus mode with const0_rtx.
30714 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30716 PR middle-end/79396
30717 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30718 FMA_EXPR like tcc_binary or tcc_unary.
30720 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30723 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30725 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30726 (output_loc_operands): Handle DW_OP_call_ref and
30727 DW_OP_GNU_variable_value.
30728 (struct variable_value_struct): New type.
30729 (struct variable_value_hasher): Likewise.
30730 (variable_value_hash): New variable.
30731 (string_types): Remove.
30732 (copy_loc_descr): New function.
30733 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30734 (prepend_loc_descr_to_each): New function.
30735 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30736 instead of add_loc_descr_to_each if the first argument is single
30737 location list and the second has multiple.
30738 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30739 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30740 when looking for variable value which doesn't have other location info.
30741 (loc_list_from_tree): Formatting fix.
30742 (gen_array_type_die): Simplify DW_AT_string_length handling.
30743 (adjust_string_types): Remove.
30744 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30745 string_types. Call resolve_variable_values.
30746 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30747 (resolve_addr_in_expr): Likewise. Add A argument.
30748 (copy_deref_exprloc): Remove deref argument. Adjust for the
30749 original expression being DW_OP_GNU_variable_value with optionally
30750 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30751 optionally after it.
30752 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30753 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30754 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30755 (variable_value_hasher::hash, variable_value_hasher::equal): New
30757 (resolve_variable_value_in_expr, resolve_variable_value,
30758 resolve_variable_values, note_variable_value_in_expr,
30759 note_variable_value): New functions.
30760 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30763 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30766 * opts.h (handle_generated_option): Add GENERATED_P argument.
30767 * opts-common.c (handle_option): Adjust function comment.
30768 (handle_generated_option): Add GENERATED_P argument, pass it to
30770 (control_warning_option): Pass false to handle_generated_option
30772 * opts.c (maybe_default_option): Pass true to handle_generated_option
30774 * optc-gen.awk: Likewise.
30776 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30778 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30779 a REG, look at the REG it is a SUBREG of.
30780 (splitter for cmpeqsi_t): Ditto.
30782 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30784 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30785 the special USEs with the pattern of the insn, not the insn itself.
30787 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30790 * doc/invoke.texi: Document -mload-store-pairs.
30792 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30793 Sandra Loosemore <sandra@codesourcery.com>
30795 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30796 argument isn't a CONST_INT.
30797 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30798 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30799 (nios2_validate_compare): Bypass alternate compare logic if *op2
30800 is not a CONST_INT.
30801 (ldstwm_operation_p): Return false if first_base is not a REG or
30802 if first_offset is not a CONST_INT.
30804 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30806 * config/cris/cris.md: Use correct operand in a define_peephole2.
30808 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30810 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30812 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30814 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30815 this_insn if it is an INSN or JUMP_INSN.
30816 (force_offsettable): Look at base, not at addr.
30817 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30818 on things that aren't necessarily CONST_INTs.
30820 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30822 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30823 -mfpmath=sse is the default also for x86-32 targets with SSE2
30824 instruction set when @option{-ffast-math} is enabled
30826 2017-02-24 Jeff Law <law@redhat.com>
30828 PR rtl-optimizatoin/79286
30829 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30832 2017-02-24 Richard Biener <rguenther@suse.de>
30834 PR tree-optimization/79389
30835 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30838 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30840 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30841 function comment to reflect reality.
30842 (loop_exits_before_overflow): Fix typo in function description.
30844 2017-02-24 Richard Biener <rguenther@suse.de>
30846 PR tree-optimization/79389
30847 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30848 properly that a threading opportunity exists. Detect conditional
30849 copy/constant propagation opportunities.
30851 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30853 * config/visium/visium.md (type): Add trap.
30854 (b): New mode attribute.
30855 (*btst): Rename into...
30856 (*btst<mode>): ...this and adjust.
30857 (*cbranchsi4_btst_insn): Rename into...
30858 (*cbranch<mode>4_btst_insn): ...this and adjust.
30859 (trap): New define_insn.
30861 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30863 PR tree-optimization/79389
30864 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30865 (noce_reversed_cond_code): New function.
30866 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30867 reversed_comparison_code. Formatting fix.
30868 (noce_try_store_flag): Test rev_cond != NULL in addition to
30869 reversed_comparison_code.
30870 (noce_try_store_flag_constants): Likewise.
30871 (noce_try_store_flag_mask): Likewise.
30872 (noce_try_addcc): Use rev_cond if non-NULL instead of
30873 reversed_comparison_code.
30874 (noce_try_cmove_arith): Likewise. Formatting fixes.
30875 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30876 (noce_find_if_block): Initialize rev_cond.
30877 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30878 instead of false as last argument never attempt to reverse it
30881 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30883 PR tree-optimization/79663
30884 * tree-predcom.c (combine_chains): Process refs in reverse order
30885 only for ZERO length chains, and add explaining comment.
30887 2017-02-23 Jeff Law <law@redhat.com>
30889 PR tree-optimization/79578
30890 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30891 in call to operand_equal_p.
30893 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30896 * config/i386/cpuid.h: Fix another undefined behavior.
30898 2017-02-23 Richard Biener <rguenther@suse.de>
30900 PR tree-optimization/79683
30901 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30902 vector types for data-refs.
30904 2017-02-23 Martin Liska <mliska@suse.cz>
30906 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30908 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30910 PR middle-end/79665
30911 * internal-fn.c (get_range_pos_neg): Moved to ...
30912 * tree.c (get_range_pos_neg): ... here. No longer static.
30913 * tree.h (get_range_pos_neg): New prototype.
30914 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30915 are known to be in between 0 and signed maximum inclusive, try to
30916 expand both unsigned and signed divmod and use the cheaper one from
30919 2017-02-22 Jeff Law <law@redhat.com>
30921 PR tree-optimization/79578
30922 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30923 to compare base operands.
30925 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30928 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30929 gpc_reg_operand instead of fpr_reg_operand.
30931 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30933 * config/mips/mips.c (mips_return_in_memory): Force FP
30934 vector types to be returned in memory for o32 ABI.
30936 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30938 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30939 instead of DW_TAG_member for static data member declarations and don't
30940 set no_linkage_name for static inline data members.
30941 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30944 2017-02-22 Martin Liska <mliska@suse.cz>
30946 * doc/invoke.texi: Replace inequality signs with square brackets
30949 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30952 * lra-constraints.c (simplify_operand_subreg): Handle
30953 WORD_REGISTER_OPERATIONS targets.
30955 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30958 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30959 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30960 elimination by swapping fld*.
30962 2017-02-22 Richard Biener <rguenther@suse.de>
30964 PR tree-optimization/79673
30965 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30966 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30967 irrelevant address-space qualifiers and avoiding a
30968 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30970 2017-02-22 Richard Biener <rguenther@suse.de>
30972 PR tree-optimization/79666
30973 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30974 to not symbolically negate if that may introduce undefined
30977 2017-02-22 Martin Liska <mliska@suse.cz>
30980 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30981 * data-streamer-out.c (streamer_write_gcov_count_stream):
30983 * value-prof.c (stream_out_histogram_value): Make assert more
30984 precise based on type of counter.
30986 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30989 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30990 Use nonimmediate_operand instead of memory_operand for operand 1.
30991 (float-extend standard_x87sse_constant_load splitter): Ditto.
30993 2017-02-21 Jeff Law <law@redhat.com>
30995 PR tree-optimization/79621
30996 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30997 blocks with edges to themselves.
30999 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31002 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31003 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31004 Use gimple_call_builtin_p.
31007 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31008 on temporarily removed DEBUG_INSNs.
31010 PR tree-optimization/79649
31011 * tree-loop-distribution.c (classify_partition): Give up on
31012 non-generic address space loads/stores.
31014 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31016 * doc/loop.texi (Loop manipulation): Remove nonexistent
31017 tree_ssa_loop_version from the documentation.
31018 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31020 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31023 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31024 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31025 * config/rs6000/rs6000.c: Include except.h.
31026 (rs6000_expand_split_stack_prologue): Call
31027 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31029 2017-02-21 Martin Jambor <mjambor@suse.cz>
31032 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31033 have been analyzed.
31035 2017-02-21 Martin Jambor <mjambor@suse.cz>
31037 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31038 for backward compatibility only.
31039 * doc/invoke.texi (Option Summary): Remove all references to
31040 -fipa-cp-alignment.
31042 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31046 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31048 * lra-constraints.c (curr_insn_transform): Handle
31049 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31051 2017-02-21 Martin Liska <mliska@suse.cz>
31053 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31055 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31057 PR translation/79638
31058 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31060 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31063 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31064 (arm_function_ok_for_sibcall): Return false for an indirect call by
31065 descriptor if all the argument registers are used.
31066 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31067 alignment of the function.
31069 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31071 PR tree-optimization/61441
31072 * simplify-rtx.c (simplify_const_unary_operation): For
31073 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31074 the sNaN unmodified.
31076 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31078 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31079 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31080 instead of SYSTEM_HEADER_DIR.
31082 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31083 Martin Liška <mliska@suse.cz>
31085 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31086 Fix typos and grammar, use active voice, and clarify.
31088 2017-02-20 Marek Polacek <polacek@redhat.com>
31090 PR middle-end/79537
31091 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31094 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31096 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31099 * config/i386/i386.c (ix86_expand_builtin): Handle
31100 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31101 ix86_builtins_isa[fcode].isa as a requirement of those
31102 flags and any other flag in the bitmask.
31103 (ix86_init_mmx_sse_builtins): Use 0 instead of
31104 ~OPTION_MASK_ISA_64BIT as mask.
31105 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31106 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31107 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31108 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31110 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31113 * lra-constraints.c (split_reg): Check requested split mode
31114 is supported by the register.
31116 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31118 * lra-constraints.c (simplify_operand_subreg): Remove early
31121 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31124 * lra-constraints.c (curr_insn_transform): Tighten condition
31125 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31127 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31130 * lra-constraints.c (curr_insn_transform): Handle
31131 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31133 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31136 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31138 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31140 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31143 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31146 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31148 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31151 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31153 * final.c (last_columnnum, override_columnnum): New variables.
31154 (final_start_function): Set last_columnnum, pass it to begin_prologue
31155 hook and pass 0 to dwarf2out_begin_prologue.
31156 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31157 to source_line debug hook.
31158 (notice_source_line): Compute last_columnnum and for debug_column_info
31159 return true on column changes.
31160 * debug.h (struct gcc_debug_hooks): Add column argument to
31161 source_line and begin_prologue hooks.
31162 (debug_nothing_int_charstar_int_bool): Remove prototype.
31163 (debug_nothing_int_int_charstar,
31164 debug_nothing_int_int_charstar_int_bool): New prototypes.
31165 (dwarf2out_begin_prologue): Add column argument.
31166 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31167 begin_prologue hooks.
31168 (debug_nothing_int_charstar_int_bool): Remove.
31169 (debug_nothing_int_int_charstar,
31170 debug_nothing_int_int_charstar_int_bool): New functions.
31171 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31172 through to dwarf2out_source_line.
31173 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31174 (dwarf2out_source_line): Add column argument, emit it if requested.
31175 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31177 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31178 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31179 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31180 through to dwarf2out_begin_prologue.
31181 (vmsdbgout_source_line): Add column argument, pass it through to
31182 dwarf2out_source_line.
31183 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31184 dbxout_source_line caller.
31185 (dbxout_source_line): Add column argument.
31187 * common.opt (gno-column-info, gcolumn-info): New options.
31188 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31189 (check_die): Also test for multiple DW_AT_decl_column attributes.
31190 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31191 DW_AT_decl_column if requested.
31192 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31194 (gen_variable_die): Likewise.
31195 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31196 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31199 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31200 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31201 (ix86_handle_option): Handle OPT_m3dnowa.
31202 * doc/invoke.texi (-m3dnowa): Document.
31203 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31204 -m3dnowa instead of -m3dnow -march=athlon.
31207 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31208 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31210 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31213 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31214 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31215 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31216 generator for vsx_xxpermdi_<mode>_be.
31217 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31218 force big-endian semantics.
31219 (vsx_xxpermdi_<mode>_be): New define_expand with same
31220 implementation as previous version of vsx_xxpermdi_<mode>.
31222 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31224 PR tree-optimization/79327
31225 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31226 variable, its initialization and use.
31228 2017-02-17 Julia Koval <julia.koval@intel.com>
31230 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31231 (OPTION_MASK_ISA_PKU_UNSET): New.
31232 (ix86_handle_option): Handle -mrdpid.
31233 * config/i386/cpuid.h (bit_RDPID): New.
31234 * config/i386/driver-i386.c (host_detect_local_cpu):
31235 Detect RDPID feature.
31236 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31237 * config/i386/i386-c.c (ix86_target_macros_internal):
31239 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31240 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31241 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31242 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31243 * config/i386/i386.md (define_insn "rdpid"): New.
31244 * config/i386/i386.opt Add -mrdpid.
31245 * config/i386/immintrin.h (_rdpid_u32): New.
31247 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31249 PR rtl-optimization/79541
31250 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31251 instead of transforming it into USE.
31253 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31255 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31256 If HONOR_SNANS (SFmode) force the input to a register.
31257 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31258 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31259 an frsp or similar insn.
31261 2017-02-17 Martin Liska <mliska@suse.cz>
31263 PR rtl-optimization/79577
31264 * params.def (selsched-max-sched-times): Increase minimum to 1.
31266 2017-02-17 Martin Liska <mliska@suse.cz>
31268 PR rtl-optimization/79574
31269 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31271 2017-02-17 Martin Liska <mliska@suse.cz>
31273 PR tree-optimization/79529
31274 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31275 ssa_defined_default_def_p to handle cases which are implicitly
31277 * tree-ssa.c (ssa_defined_default_def_p): New function.
31278 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31279 which are implicitly defined.
31280 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31282 2017-02-17 Richard Biener <rguenther@suse.de>
31284 PR middle-end/79576
31285 * params.def (max-ssa-name-query-depth): Limit to 10.
31287 2017-02-17 Richard Biener <rguenther@suse.de>
31289 PR tree-optimization/79552
31290 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31293 2017-02-17 Richard Biener <rguenther@suse.de>
31296 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31298 2017-02-17 Marek Polacek <polacek@redhat.com>
31300 PR middle-end/79536
31301 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31302 (fold_negate_expr): New wrapper.
31304 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31306 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31307 Correct terminology and de-emphasize pre-standard behavior.
31309 2017-02-16 Alan Modra <amodra@gmail.com>
31311 PR rtl-optimization/79286
31312 * ira.c (def_dominates_uses): New function.
31313 (update_equiv_regs): Don't create an equivalence for insns that
31314 may trap where the register def does not dominate the use.
31316 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31318 PR rtl-optimization/78127
31319 * lra.c (lra): Call lra_eliminate before finish the loop after
31322 2017-02-16 Richard Biener <rguenther@suse.de>
31324 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31326 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31327 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31328 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31329 (isl_val_int_from_wi): New function.
31330 (extract_affine_gmp): Rename to ...
31331 (extract_affine_wi): ... this, take a widest_int.
31332 (extract_affine_int): Just wrap extract_affine_wi.
31333 (add_param_constraints): Use isl_val_int_from_wi.
31334 (add_loop_constraints): Likewise, and extract_affine_wi.
31336 2017-02-15 Jeff Law <law@redhat.com>
31338 PR middle-end/79521
31339 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31340 ira_init_register_move_cost_if_necessary.
31342 2017-02-15 Martin Sebor <msebor@redhat.com>
31344 PR middle-end/32003
31345 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31346 removed in a prior commit.
31348 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31350 PR tree-optimization/79347
31351 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31352 counters during peeling.
31354 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31356 * Makefile.in (site.exp): Remove "set ISLVER".
31358 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31361 * real.c (real_from_integer): Call real_convert even for decimal.
31363 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31366 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31368 2017-02-14 Andrew Pinski <apinski@cavium.com>
31370 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31371 cores and change the partno/implementer to be correct.
31372 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31373 the 'B" as the implementer.
31374 * config/aarch64/aarch64-tune.md: Regenerate.
31376 2017-02-14 Carl Love <cel@us.ibm.com>
31378 * config/rs6000/rs6000.c: Add case statement entry to make the
31379 xvcvuxdsp built-in argument unsigned.
31380 * config/rs6000/vsx.md: Fix the source and return operand types so they
31381 match the instruction definitions from the ISA document. Fix typo
31382 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31385 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31388 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31389 member early_clobber_alts.
31390 * lra-lives.c (reg_early_clobber_p): New.
31391 (process_bb_lives): Use it.
31392 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31393 (debug_operand_data): Initialize early_clobber_alts.
31394 (setup_operand_alternative): Set up early_clobber_alts.
31395 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31396 alternatives to new_insn_reg.
31397 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31399 (lra_update_insn_regno_info): Pass the new arg.
31401 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31403 PR middle-end/79505
31404 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31405 (new_oacc_loop_raw): Don't clear already cleared fields.
31408 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31409 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31410 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31412 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31415 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31417 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31420 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31421 the extra instruction to the right place to store 128-bit constant
31424 2017-02-14 Martin Sebor <msebor@redhat.com>
31426 PR middle-end/79448
31427 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31428 warning for strings of unknown length.
31430 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31432 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31434 2017-02-14 Jeff Law <law@redhat.com>
31437 * ira-costs.c (scan_one_insn): Initialize register move costs
31438 for pseudos seen in USE/CLOBBER insns.
31440 PR tree-optimization/79095
31441 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31442 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31443 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31444 if the operands are known to be not equal, then the resulting range
31446 (intersect_ranges): If the new range is ~[0,0] and the old range is
31447 wide, then prefer ~[0,0].
31448 * tree-vrp.c (overflow_comparison_p_1): New function.
31449 (overflow_comparison_p): New function.
31450 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31451 if NAME is used in an overflow test.
31452 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31453 overflow check that can be expressed as an equality test, then adjust
31454 ops to be that equality test.
31456 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31458 * config/s390/s390-builtin-types.def: Remove flags argument.
31459 * config/s390/s390.c (s390_init_builtins): Likewise.
31461 2017-02-14 Martin Liska <mliska@suse.cz>
31463 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31464 vector. Fix trailing white spaces.
31466 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31468 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31471 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31473 PR rtl-optimization/68664
31474 * config/arm/arm.c (arm_sched_can_speculate_insn):
31475 New function. Declare prototype.
31476 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31478 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31480 PR rtl-optimization/68664
31481 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31483 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31485 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31487 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31488 max skip bytes for function, loop and jump.
31490 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31492 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31493 ABS_EXPR for gimple dump.
31495 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31498 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31500 PR tree-optimization/79408
31501 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31502 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31503 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31504 also if rhs1 is INTEGER_CST.
31506 2017-02-14 Richard Biener <rguenther@suse.de>
31508 PR middle-end/79432
31509 * tree-into-ssa.c (insert_phi_nodes): When the function can
31510 have abnormal edges rewrite SSA names with broken use-def
31511 dominance out of SSA and register them for PHI insertion.
31513 2017-02-13 Martin Sebor <msebor@redhat.com>
31515 PR middle-end/79496
31516 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31517 clearing info.nowrite flag when snprintf size argument is a range.
31519 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31521 * cprop.c (cprop_jump): Add missing space in string literal.
31522 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31523 (get_constraint_for_component_ref): Likewise.
31524 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31525 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31526 * lra-constraints.c (process_alt_operands): Likewise.
31527 * ipa-inline.c (inline_small_functions): Likewise.
31528 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31529 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31530 * trans-mem.c (diagnose_tm_1_op): Likewise.
31531 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31532 (grid_parallel_clauses_gridifiable): Likewise.
31534 * config/nvptx/mkoffload.c (process): Add space in between
31537 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31538 "MOD4_SSE_REGS" and "ALL_REGS".
31540 * spellcheck.c (test_data): Add , in between "foo" and "food".
31542 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31545 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31546 boundary crossing check and subsequent code generation agree.
31548 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31550 * config/aarch64/aarch64.c (has_memory_op): Delete.
31551 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31554 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31556 PR rtl-optimization/79388
31557 PR rtl-optimization/79450
31558 * combine.c (distribute_notes): When removing TEM_INSN for which
31559 corresponding dest has last value recorded, invalidate that last
31562 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31564 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31565 of explicit '@'. Add missing assembly comment marker on branch costs
31568 2017-02-13 Nathan Sidwell <nathan@acm.org>
31570 * gengtype-lex.l (<in_struct>): Add '/'.
31572 2017-02-13 Martin Liska <mliska@suse.cz>
31575 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31577 2017-02-13 Richard Biener <rguenther@suse.de>
31579 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31581 * configure: Re-generate.
31582 * config.in: Likewise.
31583 * graphite-dependences.c: Simplify as if
31584 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31585 * graphite-isl-ast-to-gimple.c: Likewise.
31586 * graphite-optimize-isl.c: Likewise.
31587 * graphite-poly.c: Likewise.
31588 * graphite-sese-to-poly.c: Likewise.
31589 * graphite.h: Likewise.
31590 * toplev.c: Include isl/version.h and use isl_version () for
31591 printing the ISL version.
31592 * doc/install.texi: Update ISL requirement.
31594 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31596 * doc/standards.texi (Standards): Update reference to
31599 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31601 * doc/extend.texi (Named Address Spaces): sourceware.org now
31603 * doc/install.texi (Binaries): Ditto.
31606 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31608 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31609 standard terminology "stringize"/"stringizing" throughout.
31610 * doc/cppinternals.texi: Likewise.
31612 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31614 * doc/extend.texi: Fix some spelling mistakes and typos.
31615 * doc/invoke.texi: Likewise.
31617 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31620 * params.def (inline-min-speedup) Change from 10 to 8.
31622 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31624 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31627 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31630 * ipa-inline-analysis.c (get_minimal_bb): New function.
31631 (record_modified): Use it.
31632 (remap_edge_change_prob): Handle also ancestor functions.
31634 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31636 * doc/contrib.texi (Contributors): Remove broken link into
31637 the Mauve CVS repository.
31639 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31641 PR middle-end/79454
31642 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31643 result computation whenever lhs doesn't have vector mode, not
31644 just when it has BLKmode.
31646 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31648 * doc/makefile.texi (profiledbootstrap): Refer to the
31649 installation instructions only in textual form.
31651 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31654 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31656 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31658 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31659 (Specific): Update mingw-w64 reference.
31661 (Specific): Remove broken link to Renesas RX processor.
31663 2017-02-10 Richard Biener <rguenther@suse.de>
31665 * toplev.c (process_options): Do not mention obsolete graphite
31666 options when printing sorry message about missing graphite support.
31667 Mention -floop-nest-optimize.
31669 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31671 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31672 (vtst_p16): Likewise.
31673 (vtstq_p8): Likewise.
31674 (vtstq_p16): Likewise.
31676 (vtstq_p64): Likewise.
31677 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31678 (vset_lane_p64): New.
31679 (vsetq_lane_p64): New.
31681 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31683 PR tree-optimization/79411
31684 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31685 stmt operands are SSA_NAMEs used in abnormal phis.
31686 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31689 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31692 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31695 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31697 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31699 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31701 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31705 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31706 not arbitrary TREE_CONSTANT.
31709 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31710 "omp declare target link" attribute unless is_global_var.
31711 * omp-offload.c (find_link_var_op): Likewise.
31713 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31714 Chung-Lin Tang <cltang@codesourcery.com>
31716 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31718 (gimplify_adjust_omp_clauses): Don't delete TILE.
31719 (gimplify_omp_for): Deal with TILE.
31720 * internal-fn.c (expand_GOACC_TILE): New function.
31721 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31723 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31724 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31726 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31727 avoid DIV for outermost collapse var.
31728 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31729 Remove out of date comments, fix whitespace.
31730 * omp-general.c (omp_extract_for_data): Deal with tiling.
31731 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31732 adjust OLF_DIM_BASE value.
31733 (struct omp_for_data): Add tiling field.
31734 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31735 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31736 for auto loops. Remove default auto determining, moved to
31737 oacc_loop_fixed_partitions.
31738 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31739 stmts, add e_mask field.
31740 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31741 (oacc_thread_numbers): Use oacc_dim_call.
31742 (oacc_xform_tile): New.
31743 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31744 (finish_oacc_loop): Adjust for ifns vector.
31745 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31746 add case for GOACC_TILE fns.
31747 (oacc_loop_xform_loop): Delete.
31748 (oacc_loop_process): Iterate over call list directly, and add
31749 handling for GOACC_TILE fns.
31750 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31752 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31753 vector partitioning to outer loops. Assign 2 partitions to loops
31754 when available. Add TILE handling.
31755 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31756 (execite_oacc_device_lower): Process GOACC_TILE fns,
31757 ignore unknown specs.
31758 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31759 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31760 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31762 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31764 * configure.ac (ACX_BUGURL): Update.
31765 * configure: Regenerate.
31767 2017-02-09 Richard Biener <rguenther@suse.de>
31769 PR tree-optimization/69823
31770 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31771 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31773 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31775 * config/arc/arc-c.def: Add __NPS400__ definition.
31776 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31777 (TARGET_NPS400): Define.
31779 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31781 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31783 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31784 pointer, arch_info.
31785 (arc_cpu_types): Fill the arch_info field with a pointer into the
31786 arc_arch_types table.
31787 (arc_selected_cpu): Declare.
31788 * config/arc/arc.c (arc_selected_cpu): Make global.
31789 (arc_selected_arch): Delete.
31790 (arc_base_cpu): Delete.
31791 (arc_override_options): Remove references to deleted variables,
31792 update access to arch information.
31793 (ARC_OPT): Update access to arch information.
31794 (ARC_OPTX): Likewise.
31795 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31796 (TARGET_ARC600): Update access to arch information.
31797 (TARGET_ARC601): Likewise.
31798 (TARGET_ARC700): Likewise.
31799 (TARGET_EM): Likewise.
31800 (TARGET_HS): Likewise.
31801 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31804 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31807 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31808 condition/operands for integer GE/LE/GEU/LEU operations.
31810 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31812 PR translation/79397
31813 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31816 2017-02-08 Martin Jambor <mjambor@suse.cz>
31819 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31820 whether allocation happened.
31821 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31822 nothing was allocated.
31824 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31826 PR tree-optimization/79408
31827 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31828 constant, but SSA_NAME with a known integer range, use the minimum
31829 of that range instead of op1 to determine if modulo can be replaced
31830 with its first operand.
31832 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31834 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31836 2017-02-08 Richard Biener <rguenther@suse.de>
31838 PR tree-optimization/71824
31839 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31840 Check all loops contained in the merged region.
31842 2017-02-07 Andrew Pinski <apinski@cavium.com>
31844 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31846 2017-02-07 Andrew Pinski <apinski@cavium.com>
31848 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31849 (thunderxt88): Likewise.
31850 (thunderxt81): Disable LSE and change v8.1 to v8.
31851 (thunderxt83): Likewise.
31853 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31854 Richard Biener <rguenther@suse.de>
31856 PR middle-end/79399
31857 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31858 type from int to size_t.
31859 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31861 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31863 PR rtl-optimization/79386
31864 * cprop.c (bypass_conditional_jumps): Initialize
31865 bypass_last_basic_block already before splitting bbs after
31866 unconditional traps...
31867 (bypass_conditional_jumps): ... rather than here.
31870 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31871 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31872 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31873 fix -masm=intel patterns.
31875 2017-02-07 Richard Biener <rguenther@suse.de>
31877 PR tree-optimization/79256
31878 PR middle-end/79278
31879 * builtins.c (get_object_alignment_2): Use min_align_of_type
31880 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31881 and ADJUST_FIELD_ALIGN.
31883 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31885 * doc/tm.texi: Regenerate.
31886 * stor-layout.c (layout_decl): Adjust.
31887 (update_alignment_for_field): Likewise.
31888 (place_field): Likewise.
31889 (min_align_of_type): Likewise.
31890 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31891 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31892 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31893 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31894 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31895 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31896 * config/i386/i386.c (x86_field_alignment): Likewise.
31897 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31898 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31899 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31900 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31901 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31902 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31906 2017-01-30 Richard Biener <rguenther@suse.de>
31908 PR tree-optimization/79256
31909 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31910 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31913 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31915 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31916 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31917 builtins to SImode and emit a zero-extend, if necessary.
31919 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31921 * docs/invoke.texi (RISC-V Options): Alphabetize.
31923 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31925 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31928 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31930 * config/riscv/riscv.c: New file.
31931 * common/config/riscv/riscv-common.c: Likewise.
31932 * config.gcc: Likewise.
31933 * config/riscv/constraints.md: Likewise.
31934 * config/riscv/elf.h: Likewise.
31935 * config/riscv/generic.md: Likewise.
31936 * config/riscv/linux.h: Likewise.
31937 * config/riscv/multilib-generator: Likewise.
31938 * config/riscv/peephole.md: Likewise.
31939 * config/riscv/pic.md: Likewise.
31940 * config/riscv/predicates.md: Likewise.
31941 * config/riscv/riscv-builtins.c: Likewise.
31942 * config/riscv/riscv-c.c: Likewise.
31943 * config/riscv/riscv-ftypes.def: Likewise.
31944 * config/riscv/riscv-modes.def: Likewise.
31945 * config/riscv/riscv-opts.h: Likewise.
31946 * config/riscv/riscv-protos.h: Likewise.
31947 * config/riscv/riscv.h: Likewise.
31948 * config/riscv/riscv.md: Likewise.
31949 * config/riscv/riscv.opt: Likewise.
31950 * config/riscv/sync.md: Likewise.
31951 * config/riscv/t-elf-multilib: Likewise.
31952 * config/riscv/t-linux: Likewise.
31953 * config/riscv/t-linux-multilib: Likewise.
31954 * config/riscv/t-riscv: Likewise.
31955 * configure.ac: Likewise.
31956 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31957 Waterman as RISC-V maintainers.
31958 * doc/install.texi: Add RISC-V entries.
31959 * doc/invoke.texi: Add RISC-V options section.
31960 * doc/md.texi: Add RISC-V constraints section.
31961 * configure: Regenerated.
31963 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31966 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31967 false values to be constant vectors with all 0 or all 1 bits set.
31968 (vcondu<mode><mode>): Likewise.
31969 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31971 (fpmask_comparison_operator): Update comment.
31972 (vecint_comparison_operator): New predicate.
31973 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31974 vector conditionals when the true and false values are constant
31975 vectors with all 0 bits or all 1 bits set.
31977 2017-02-06 Martin Sebor <msebor@redhat.com>
31979 PR tree-optimization/79376
31980 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31982 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31984 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31985 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31986 to simplify split condition.
31988 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31990 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31991 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31994 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31996 PR rtl-optimization/68664
31997 * target.def (can_speculate_insn): New hook.
31998 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31999 * doc/tm.texi: Regenerate.
32000 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32001 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32002 (rs6000_sched_can_speculate_insn): New function.
32004 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32006 PR tree-optimization/79284
32007 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32008 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32009 vectorizable_mask_load_store, vectorizable_operation,
32010 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32011 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32012 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32013 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32014 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32015 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32016 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32017 is_gimple_assign (stmt). Replace another such test with
32018 is_gimple_assign (stmt).
32020 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32023 * config/avr/avr.c (rtl-iter.h): Include it.
32024 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32025 (avr_legitimate_combined_insn): ...and implementation.
32027 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32029 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32030 * config/s390/s390.c (s390_const_operand_ok)
32031 (s390_canonicalize_comparison, s390_extract_part)
32032 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32033 (s390_contiguous_bitmask_p, s390_rtx_costs)
32034 (legitimize_pic_address): Likewise.
32035 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32036 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32037 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32038 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32039 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32041 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32043 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32044 REGNO($0) == REGNO($1).
32046 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32048 * config/s390/linux.h(SIZE_TYPE): Add comment.
32050 2017-02-06 Julian Brown <julian@codesourcery.com>
32051 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32052 Virendra Pathak <virendra.pathak@broadcom.com>
32054 * config/aarch64/aarch64-cores.def: Change the scheduler
32056 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32057 * config/aarch64/thunderx2t99.md: New file.
32059 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32061 * doc/standards.texi (Go Language): Update link to language
32064 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32066 * tree-eh.c (lower_resx): Sanitize profile.
32067 (cleanup_empty_eh_move_lp): Likewise.
32069 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32072 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32074 * cfgloopmanip.h (loop_version): Update prototype.
32075 * modulo-sched.c (sms_schedule): Update call of loop_version.
32076 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32077 * tree-parloops.c (gen_parallel_loop): Likewise.
32078 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32079 * tree-ssa-loop-split.c (split_loop): Likewise.
32080 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32081 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32083 2017-02-05 Martin Liska <mliska@suse.cz>
32086 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32088 (print_operand_address): Initialize a struct to zero.
32090 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32092 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32093 garbage collector only in textual form.
32095 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32097 * doc/extend.texi (x86 specific memory model extensions for
32098 transactional memory): Simplify a phrase.
32100 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32103 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32104 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32105 (atomic_storedi_1): Likewise.
32107 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32109 PR tree-optimization/79338
32110 * tree-parloops.c (gather_scalar_reductions): Don't call
32111 vect_analyze_loop_form for loop->inner before destroying loop's
32114 2017-02-03 Martin Sebor <msebor@redhat.com>
32116 PR tree-optimization/79327
32117 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32118 when precision has resulted in leading zeros.
32119 (format_integer): Adjust the likely counter to assume an unknown
32120 argument that may be zero is non-zero.
32122 2017-02-03 Jason Merrill <jason@redhat.com>
32125 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32126 avoid copying non-taken branch.
32128 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32130 PR tree-optimization/79340
32131 * tree-vect-loop.c (vectorizable_reduction): Release
32132 vec_defs elements after safe_splicing them into other vectors.
32135 PR tree-optimization/79327
32136 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32137 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32139 (format_integer): Use wide_int_to_tree instead of build_int_cst
32140 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32141 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32142 of shortest and longest sequence.
32144 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32146 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32147 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32149 2017-02-03 Walter Lee <walt@tilera.com>
32152 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32153 after initial stackframe link reg save.
32154 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32156 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32159 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32160 wu for stxssp alternative.
32162 2017-02-03 Martin Sebor <msebor@redhat.com>
32164 PR tree-optimization/79352
32165 * gimple-fold.c (get_range_strlen): Add argument.
32166 (get_range_strlen): Change return type to bool.
32167 (get_maxval_strlen): Pass in a dummy argument.
32168 * gimple-fold.h (get_range_strlen): Change return type to bool.
32169 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32170 * tree.h (array_at_struct_end_p): Add argument.
32171 * tree.c (array_at_struct_end_p): Handle it.
32173 2017-02-03 Martin Liska <mliska@suse.cz>
32176 * multiple_target.c (create_dispatcher_calls): Redirect edge
32177 from a caller of a dispatcher.
32178 (expand_target_clones): Make the clones local.
32179 (ipa_target_clone): Do both target clones and resolvers.
32180 (ipa_dispatcher_calls): Remove the pass.
32181 (pass_dispatcher_calls::gate): Likewise.
32182 (make_pass_dispatcher_calls): Likewise.
32183 * passes.def (pass_target_clone): Put as very first IPA early
32186 2017-02-03 Martin Liska <mliska@suse.cz>
32188 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32189 in case of a function with ifunc attribute.
32191 2017-02-03 Martin Liska <mliska@suse.cz>
32193 * cgraph.c (cgraph_node::dump): Dump function version info.
32194 * symtab.c (symtab_node::dump_base): Add missing new line.
32196 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32198 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32199 (ifcombine_ifandif): Use it.
32201 2017-02-03 Martin Liska <mliska@suse.cz>
32203 * doc/invoke.texi: Document default value for
32204 use-after-scope-direct-emission-threshold.
32206 2017-02-03 Martin Liska <mliska@suse.cz>
32208 PR tree-optimization/79339
32209 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32210 (format_floating): Likewise.
32212 2017-02-03 Martin Liska <mliska@suse.cz>
32215 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32217 (ipa_node_params_t::remove): Likewise.
32218 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32219 initialization from removed ipa_node_params_t::insert.
32220 (ipa_node_params::~ipa_node_params): Move from removed
32221 ipa_node_params_t::release.
32222 * symbol-summary.h (symbol_summary::m_released): New member.
32223 Do not release a summary twice. Do not allow to call finalizer
32224 for types of a summary that live in GGC memory.
32226 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32228 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32231 2017-02-02 Martin Sebor <msebor@redhat.com>
32233 PR middle-end/79275
32234 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32235 (format_string): Tighten up the range of output for non-constant
32236 strings and correct the expected range for wide non-constant strings.
32238 2017-02-02 Martin Sebor <msebor@redhat.com>
32240 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32242 PR middle-end/32003
32243 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32245 (-fdump-tree-@var): Add to index and document how to come up
32246 with pass-specific option and dump file names.
32247 (-fdump-passes): Clarify where to look for output.
32249 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32251 PR middle-end/77445
32252 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32253 statistics of the analyzed path; allow threading for speed when
32254 any of BBs along the path are optimized for speed.
32256 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32258 PR middle-end/78468
32259 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32260 settings of the virtual registers.
32263 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32265 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32266 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32269 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32271 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32272 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32274 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32276 * config/s390/s390.md: Add missing comments with the expanded
32278 * config/s390/vector.md: Likewise.
32279 * config/s390/vx-builtins.md: Likewise.
32281 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32284 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32285 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32286 conditions on a single line.
32288 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32290 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32291 __S390_VX__ to __VX__.
32293 2017-02-01 Andrew Pinski <apinski@cavium.com>
32295 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32296 stmt_info to record_stmt_cost.
32297 (vect_get_known_peeling_cost): Pass stmt_info if known to
32299 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32300 cpu_vector_cost field into
32301 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32302 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32303 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32304 splitting of scalar_stmt_cost and vec_stmt_cost.
32305 (thunderx_vector_cost): Likewise.
32306 (cortexa57_vector_cost): LIkewise.
32307 (exynosm1_vector_cost): Likewise.
32308 (xgene1_vector_cost): Likewise.
32309 (thunderx2t99_vector_cost): Improve after the splitting of the two
32311 (aarch64_builtin_vectorization_cost): Update for the splitting of
32312 scalar_stmt_cost and vec_stmt_cost.
32314 2017-02-01 Torvald Riegel <triegel@redhat.com>
32315 Richard Henderson <rth@redhat.com>
32317 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32318 conditional on existance of a fast atomic load.
32319 * optabs-query.c (can_atomic_load_p): New function.
32320 * optabs-query.h (can_atomic_load_p): Declare it.
32321 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32322 no fast atomic load is available for the particular size of access.
32323 (expand_atomic_compare_and_swap): Likewise.
32324 (expand_atomic_load): Likewise.
32325 (expand_atomic_store): Likewise.
32326 (expand_atomic_fetch_op): Likewise.
32327 * testsuite/lib/target-supports.exp
32328 (check_effective_target_sync_int_128): Remove x86 because it provides
32329 no fast atomic load.
32330 (check_effective_target_sync_int_128_runtime): Likewise.
32332 2017-02-01 Richard Biener <rguenther@suse.de>
32334 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32335 (graphite_transform_loops): Provide opt-info for optimized nests.
32336 * tree-parloop.c (parallelize_loops): Provide opt-info for
32337 parallelized loops.
32339 2017-02-01 Richard Biener <rguenther@suse.de>
32341 PR middle-end/79315
32342 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32343 was not set before.
32345 2017-02-01 Richard Biener <rguenther@suse.de>
32347 PR tree-optimization/71824
32348 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32349 Verify the loops are valid in the merged SESE region.
32350 (scop_detection::can_represent_loop_1): Check analyzing the
32351 evolution of the number of iterations in the region succeeds.
32353 2017-01-31 Ian Lance Taylor <iant@golang.org>
32355 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32356 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32358 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32360 PR preprocessor/79210
32361 * input.c (get_substring_ranges_for_loc): Replace line_width
32362 assertion with error-handling.
32364 2017-01-31 Richard Biener <rguenther@suse.de>
32366 PR tree-optimization/77318
32367 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32368 (create_pw_aff_from_tree): Take loop parameter.
32369 (add_condition_to_pbb): Pass loop of the condition to
32370 create_pw_aff_from_tree.
32372 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32374 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32375 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32377 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32381 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32383 (convert_int_to_float128): Likewise.
32384 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32385 (convert_int_to_float128): Likewise.
32386 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32387 (UNSPEC_IEEE128_CONVERT): Likewise.
32388 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32389 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32390 Use local variables for IBM extended format.
32391 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32392 (fix_trunc<mode>si2_fprs): Likewise.
32393 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32394 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32395 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32396 to know that we can now have integers of all sizes in vector
32398 (fix<uns>_<mode>di2_hw): Likewise.
32399 (float<uns>_<mode>si2_hw): Likewise.
32400 (fix_<mode>si2_hw): Likewise.
32401 (fixuns_<mode>si2_hw): Likewise.
32402 (float<uns>_<mode>di2_hw): Likewise.
32403 (float_<mode>di2_hw): Likewise.
32404 (float_<mode>si2_hw): Likewise.
32405 (floatuns_<mode>di2_hw): Likewise.
32406 (floatuns_<mode>si2_hw): Likewise.
32407 (xscvqp<su>wz_<mode>): Delete, no longer used.
32408 (xscvqp<su>dz_<mode>): Likewise.
32409 (xscv<su>dqp_<mode>): Likewise.
32410 (ieee128_mfvsrd_64bit): Likewise.
32411 (ieee128_mfvsrd_32bit): Likewise.
32412 (ieee128_mfvsrwz): Likewise.
32413 (ieee128_mtvsrw): Likewise.
32414 (ieee128_mtvsrd_64bit): Likewise.
32415 (ieee128_mtvsrd_32bit): Likewise.
32417 2017-01-31 Martin Liska <mliska@suse.cz>
32420 * ipa-prop.c (ipa_free_all_node_params): Call release method
32421 instead of ~sumbol_summary to not to trigger double times
32424 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32426 PR tree-optimization/71691
32427 * bitmap.h (class auto_bitmap): New.
32428 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32429 is_maybe_undefined instead of ssa_undefined_value_p.
32431 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32433 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32434 __S390_ARCH_LEVEL__ to __ARCH__.
32436 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32438 PR tree-optimization/79267
32439 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32440 if should_remove_lhs_p is true.
32442 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32445 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32446 (add_alignment_attribute): New.
32447 (base_type_die): Add alignment attribute.
32448 (subrange_type_die): Likewise.
32449 (modified_type_die): Likewise.
32450 (gen_array_type_die): Likewise.
32451 (gen_descr_array_type_die: Likewise.
32452 (gen_enumeration_type_die): Likewise.
32453 (gen_subprogram_die): Likewise.
32454 (gen_variable_die): Likewise.
32455 (gen_field_die): Likewise.
32456 (gen_ptr_to_mbr_type_die): Likewise.
32457 (gen_struct_or_union_type_die): Likewise.
32458 (gen_subroutine_type_die): Likewise.
32459 (gen_typedef_die): Likewise.
32460 (base_type_cmp): Compare alignment attribute.
32462 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32465 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32466 (setb_unsigned) New pattern for setb with CCUNS.
32467 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32468 subfc./subfe sequence to avoid overflow problems. Generate a
32469 shorter sequence with cmpld/setb for power9.
32470 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32471 for generating subfc. instruction.
32472 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32473 now uses this instruction.
32475 2017-01-30 Ian Lance Taylor <iant@google.com>
32478 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32479 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32481 2017-01-30 Martin Sebor <msebor@redhat.com>
32483 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32484 Move constant to the right of a relational operator.
32485 (get_mpfr_format_length, format_character, format_string): Ditto.
32486 (should_warn_p, maybe_warn): Same.
32488 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32490 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32493 * asan.c (get_translation_unit_decl): Remove function.
32494 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32496 2017-01-30 Martin Liska <mliska@suse.cz>
32498 PR gcov-profile/79259
32499 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32500 -fprofile-generate.
32502 2017-01-30 Martin Liska <mliska@suse.cz>
32505 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32506 Initialize variables with NULL value.
32508 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32511 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32513 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32515 2017-01-30 Richard Biener <rguenther@suse.de>
32517 PR tree-optimization/79276
32518 * tree-vrp.c (process_assert_insertions): Properly adjust common
32519 when removing a duplicate.
32521 2017-01-30 Richard Biener <rguenther@suse.de>
32523 PR tree-optimization/79256
32524 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32525 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32527 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32529 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32532 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32533 ("*r<noxa>sbg_<mode>_sll_bitmask")
32534 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32535 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32536 Use contiguous_bitmask_nowrap_operand.
32538 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32541 * config/rs6000/altivec.h (vec_xl): Revise #define.
32542 (vec_xst): Likewise.
32544 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32546 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32548 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32550 PR rtl-optimization/79194
32551 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32552 traps before call to bypass_conditional_jumps.
32554 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32556 PR tree-optimization/71374
32557 * lra-constraints.c (check_conflict_input_operands): New.
32558 (match_reload): Use it.
32560 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32563 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32564 account to calculate conflict_set.
32566 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32568 PR rtl-optimization/78559
32569 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32570 other_insn in combine.
32572 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32574 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32575 uint16_type_node for BT_UINT16.
32577 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32579 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32580 "RTL Tests" to menu.
32581 (GIMPLE Tests): New node.
32582 (RTL Tests): New node.
32584 2017-01-27 Richard Biener <rguenther@suse.de>
32586 PR tree-optimization/79245
32587 * tree-loop-distribution.c (distribute_loop): Apply cost
32588 modeling also to detected patterns.
32590 2017-01-27 Richard Biener <rguenther@suse.de>
32592 PR tree-optimization/71433
32593 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32594 (compare_assert_loc): New function.
32595 (process_assert_insertions): Sort and optimize assert locations
32596 to remove duplicates and push down identical assertions on
32597 edges to their destination block.
32599 2017-01-27 Richard Biener <rguenther@suse.de>
32601 PR tree-optimization/79244
32602 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32603 out SSA names even if abnormal.
32605 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32607 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32608 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32609 instead of MPFR_RNDN.
32611 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32614 * arm.c (arm_option_override): Don't call build_target_option_node
32615 until after doing all option overrides.
32616 (arm_valid_target_attribute_tree): Likewise.
32618 2017-01-27 Martin Liska <mliska@suse.cz>
32620 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32621 for {cd}tors and C++ {cd}tors.
32623 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32625 * config/s390/s390.md ("*setmem_long_and")
32626 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32628 2017-01-26 Martin Sebor <msebor@redhat.com>
32630 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32633 2017-01-26 Martin Sebor <msebor@redhat.com>
32635 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32636 HAVE_DFmode before using XFmode or DFmode.
32637 (parse_directive): Avoid using the z length modifier to avoid
32638 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32640 PR middle-end/78703
32641 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32642 to accept adjustment as an array.
32643 (get_int_range): New function.
32644 (struct directive): Make width and prec arrays.
32645 (directive::set_width, directive::set_precision): Call get_int_range.
32646 (format_integer, format_floating): Handle width and precision ranges.
32647 (format_string, parse_directive): Same.
32649 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32652 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32653 comdat_type_p set, just clone them, but keep the children in the
32657 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32658 which have direct callers with -fvar-tracking-assignments enabled
32660 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32661 inside of type units.
32663 2017-01-26 Martin Sebor <msebor@redhat.com>
32665 PR middle-end/78703
32666 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32668 (struct format_result): Replace number_chars, number_chars_min,
32669 and number_chars_max with a single member of struct result_range.
32671 (format_result::operator+=): Adjust.
32672 (struct fmtresult): Remove bounded. Handle likely and unlikely
32674 (fmtresult::adjust_for_width_or_precision): New function.
32675 (fmtresult:type_max_digits): New function.
32676 (bytes_remaining): Handle likely and unlikely counters.
32677 (min_bytes_remaining): Remove.
32678 (format_percent): Simplify.
32679 (format_integer, format_floating): Set likely and unlikely counters.
32680 (get_string_length, format_character, format_string): Same.
32681 (format_plain, should_warn_p): New function.
32682 (maybe_warn): Call should_warn_p. Update diagnostic messages
32683 and handle those for all directives, including plain strings.
32684 (format_directive): Handle likely and unlikely counters.
32685 Remove unnecessary quoting from diagnostics. Add an informational
32687 (add_bytes): Remove.
32688 (pass_sprintf_length::compute_format_length): Simplify.
32689 (try_substitute_return_value): Handle likely and unlikely counters.
32691 2017-01-26 Carl Love <cel@us.ibm.com>
32693 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32694 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32696 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32699 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32700 endianess for subregs into account.
32701 * lra-constraints.c (lra_constraints): Do risky transformations
32702 always on the first iteration.
32703 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32704 last_call_used_reg_set.
32705 (process_bb_lives): Define and use last_call_used_reg_set.
32706 * lra.c (lra): Always continue after lra_constraints on the first
32709 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32711 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32713 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32715 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32717 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32718 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32719 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32720 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32721 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32722 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32723 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32724 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32725 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32727 2017-01-26 Marek Polacek <polacek@redhat.com>
32730 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32731 for the third operand.
32733 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32735 PR middle-end/79236
32736 * omp-low.c (struct omp_context): Add simt_stmt field.
32737 (scan_omp_for): Return omp_context *.
32738 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32739 context to the _simt_ SIMD stmt.
32740 (lower_omp_for): For combined SIMD with sibling _simt_
32741 SIMD, make sure to use the same decls in _looptemp_
32742 clauses as in the sibling.
32744 2017-01-26 David Sherwood <david.sherwood@arm.com>
32746 PR middle-end/79212
32747 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32750 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32753 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32754 emit fld b; fld a; if possible.
32756 * brig-builtins.def: Update copyright years.
32757 * config/arm/arm_acle_builtins.def: Update copyright years.
32759 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32762 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32763 constraint instead of o for the stxsd instruction.
32765 2017-01-25 Carl Love <cel@us.ibm.com>
32767 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32768 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32770 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32772 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32774 2017-01-25 Richard Biener <rguenther@suse.de>
32776 PR tree-optimization/69264
32777 * target.def (vector_alignment_reachable): Improve documentation.
32778 * doc/tm.texi: Regenerate.
32779 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32781 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32782 earlier changes with respect to TYPE_USER_ALIGN.
32783 (vector_alignment_reachable_p): Likewise. Improve dumping.
32785 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32788 * config/arm/arm.md (xordi3): Force constant operand into a register
32791 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32793 * doc/invoke.texi (-fstore-merging): Correct default optimization
32794 levels at which it is enabled.
32795 (-O): Move -fstore-merging from list to...
32798 2017-01-25 Richard Biener <rguenther@suse.de>
32801 * omp-expand.c: Include debug.h.
32802 (expand_omp_taskreg): Make sure to generate early debug before
32803 outlining anything from a function.
32804 (expand_omp_target): Likewise.
32805 (grid_expand_target_grid_body): Likewise.
32807 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32810 * asan.c (get_translation_unit_decl): New function.
32811 (asan_add_global): Extract modules file name from globals
32812 TRANSLATION_UNIT_DECL name.
32814 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32817 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32818 for long calls with APCS frame and VFP.
32820 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32822 * cfg.c (original_copy_tables_initialized_p): New function.
32823 * cfg.h (original_copy_tables_initialized_p): New decl.
32824 * cfgrtl.c (relink_block_chain): Guard the call to
32825 free_original_copy_tables with a call to
32826 original_copy_tables_initialized_p.
32827 * cgraph.h (symtab_node::native_rtl_p): New decl.
32828 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32829 (symtab_node::needed_p): Don't assert for early assembly output
32830 for __RTL functions.
32831 (cgraph_node::finalize_function): Set "force_output" for __RTL
32833 (cgraph_node::analyze): Bail out early for __RTL functions.
32834 (analyze_functions): Update assertion to support __RTL functions.
32835 (cgraph_node::expand): Bail out early for __RTL functions.
32836 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32838 * function.h (struct function): Update comment for field
32840 * gimple-expr.c: Include "tree-pass.h".
32841 (gimple_has_body_p): Return false for __RTL functions.
32842 * Makefile.in (OBJS): Add run-rtl-passes.o.
32843 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32845 (gcc::pass_manager::get_clean_slate): New accessor.
32846 * passes.c: Include "insn-addr.h".
32847 (should_skip_pass_p): Add logging. Update logic for running
32848 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32849 property-provider override so it is only done for gimple passes.
32851 (skip_pass): New function.
32852 (execute_one_pass): Call skip_pass when skipping passes.
32853 * read-md.c (md_reader::read_char): Support filtering
32854 the input to a subset of line numbers.
32855 (md_reader::md_reader): Initialize fields
32856 m_first_line and m_last_line.
32857 (md_reader::read_file_fragment): New function.
32858 * read-md.h (md_reader::read_file_fragment): New decl.
32859 (md_reader::m_first_line): New field.
32860 (md_reader::m_last_line): New field.
32861 * read-rtl-function.c (function_reader::create_function): Only
32862 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32863 curr_properties. Set DECL_INITIAL to a dummy block.
32864 (read_rtl_function_body_from_file_range): New function.
32865 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32867 * run-rtl-passes.c: New file.
32868 * run-rtl-passes.h: New file.
32870 2017-01-24 Jeff Law <law@redhat.com>
32872 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32875 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32877 PR tree-optimization/79159
32878 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32879 (record_nonwrapping_iv): Improve boundary using above function if no
32880 value range information.
32882 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32883 Martin Jambor <mjambor@suse.cz>
32885 * brig-builtins.def: New file.
32886 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32887 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32888 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32889 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32890 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32891 * builtin-types.def (BT_INT8): New.
32892 (BT_INT16): Likewise.
32893 (BT_UINT8): Likewise.
32894 (BT_UINT16): Likewise.
32895 (BT_FN_ULONG): Likewise.
32896 (BT_FN_UINT_INT): Likewise.
32897 (BT_FN_UINT_ULONG): Likewise.
32898 (BT_FN_UINT_LONG): Likewise.
32899 (BT_FN_UINT_PTR): Likewise.
32900 (BT_FN_ULONG_PTR): Likewise.
32901 (BT_FN_INT8_FLOAT): Likewise.
32902 (BT_FN_INT16_FLOAT): Likewise.
32903 (BT_FN_UINT32_FLOAT): Likewise.
32904 (BT_FN_UINT16_FLOAT): Likewise.
32905 (BT_FN_UINT8_FLOAT): Likewise.
32906 (BT_FN_UINT64_FLOAT): Likewise.
32907 (BT_FN_UINT16_UINT32): Likewise.
32908 (BT_FN_UINT32_UINT16): Likewise.
32909 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32910 (BT_FN_INT_PTR_INT): Likewise.
32911 (BT_FN_UINT_PTR_UINT): Likewise.
32912 (BT_FN_LONG_PTR_LONG): Likewise.
32913 (BT_FN_ULONG_PTR_ULONG): Likewise.
32914 (BT_FN_VOID_UINT64_UINT64): Likewise.
32915 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32916 (BT_FN_INT8_INT8_INT8): Likewise.
32917 (BT_FN_INT16_INT16_INT16): Likewise.
32918 (BT_FN_INT_INT_INT): Likewise.
32919 (BT_FN_UINT_FLOAT_UINT): Likewise.
32920 (BT_FN_FLOAT_UINT_UINT): Likewise.
32921 (BT_FN_ULONG_UINT_UINT): Likewise.
32922 (BT_FN_ULONG_UINT_PTR): Likewise.
32923 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32924 (BT_FN_UINT_UINT_UINT): Likewise.
32925 (BT_FN_VOID_UINT_PTR): Likewise.
32926 (BT_FN_UINT_UINT_PTR: Likewise.
32927 (BT_FN_UINT32_UINT64_PTR): Likewise.
32928 (BT_FN_INT_INT_UINT_UINT): Likewise.
32929 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32930 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32931 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32932 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32933 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32934 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32935 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32936 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32937 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32938 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32939 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32940 * doc/frontends.texi: List BRIG FE.
32941 * doc/install.texi (Testing): Add BRIG tesring requirements.
32942 * doc/invoke.texi (Overall Options): Mention BRIG.
32943 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32945 2017-01-24 Richard Biener <rguenther@suse.de>
32947 PR translation/79208
32948 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32950 2017-01-24 Martin Jambor <mjambor@suse.cz>
32953 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32954 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32955 and known_contexts.
32957 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32959 PR middle-end/79123
32960 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32961 casts from signed to unsigned really don't have a range.
32963 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32965 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32966 GMP_RNDx for compatiblity.
32968 2017-01-24 Martin Liska <mliska@suse.cz>
32971 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32972 that would prevent us to call alloca with -1 as argument.
32974 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32976 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32977 Avoid -Wformat-security warning.
32979 2017-01-23 Andrew Pinski <apinski@cavium.com>
32981 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32984 2017-01-23 Martin Sebor <msebor@redhat.com>
32986 PR middle-end/78703
32987 * gimple-ssa-sprintf.c (warn_level): New global.
32988 (format_integer): Use it here and throughout the rest of the file.
32989 Use the same switch to compute sign as base.
32990 (maybe_warn): New function.
32991 (format_directive): Factor out warnings into maybe_warn.
32992 Add debugging output. Use warn_level.
32993 (add_bytes): Use warn_level.
32994 (pass_sprintf_length::compute_format_length): Add debugging output.
32995 (try_substitute_return_value): Same.
32996 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32998 PR middle-end/78703
32999 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33000 (struct fmtresult, format_integer, format_floating): Adjust.
33001 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33002 (get_string_length, format_string,format_directive): Same.
33003 (pass_sprintf_length::compute_format_length): Same.
33004 (try_substitute_return_value): Simplify slightly.
33006 PR middle-end/78703
33007 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33008 (fmtresult::operator+=): Outlined.
33009 (struct fmtresult): Add ctors.
33010 (struct conversion_spec): Rename...
33011 (struct directive): ...to this. Add and remove data members.
33012 (directive::set_width, directive::set_precision): New functions.
33013 (format_percent): Use fmtresult ctor.
33014 (get_width_and_precision): Remove.
33015 (format_integer): Make naming changes. Avoid computing width and
33017 (format_floating): Same. Adjust indentation.
33018 (format_character, format_none): New functions.
33019 (format_string): Moved character handling to format_character.
33020 (format_directive): Remove arguments, change return type.
33021 (parse_directive): New function.
33022 (pass_sprintf_length::compute_format_length): Move directive
33023 parsing to parse_directive.
33025 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33027 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33028 (assign_assembler_name_if_needed): ... this.
33029 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33030 (assign_assembler_name_if_needed): ... this.
33031 (free_lang_data_in_cgraph): Adjust callers.
33032 * cgraphunit.c (cgraph_node::analyze): Likewise.
33033 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33036 2017-01-23 Richard Biener <rguenther@suse.de>
33038 PR tree-optimization/79088
33039 PR tree-optimization/79188
33040 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33041 resetting loop bounds after last path deletion. Reset loop
33042 bounds of the target loop, make code match the comments.
33043 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33044 Make sure loops need no fixups.
33046 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33048 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33049 exponent support with double type for first argument.
33050 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33051 type returned by __builtin_vec_extract_sig,
33052 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33053 functions from "vector int" to "vector unsigned int" or from
33054 "vector long long int" to "vector unsigned long long int".
33055 Changed type returned by __builtin_vec_extract_exp,
33056 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33057 functions from "vector int" to "vector unsigned int" or from
33058 "vector long long int" to "vector unsigned long long int".
33059 Changed return type of __builtin_vec_test_data_class,
33060 __builtin_vec_test_data_class_sp, and
33061 __builtin_vec_test_data_class_dp from "vector int" to
33062 "vector bool int" or from "vector long long int" to "vector bool
33063 long long int" and changed second argument type from "unsigned
33064 int" to "int". Added new overloaded function forms "vector float
33065 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33066 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33067 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33068 double, vector unsigned long long int)" and "vector double
33069 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33070 long int)". Changed return type of
33071 __builtin_scalar_test_data_class and
33072 __builtin_scalar_test_data_class_sp and
33073 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33074 int" and changed second argument from "unsigned int" to "int".
33075 Changed type returned by __builtin_scalar_test_neg,
33076 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33077 from "int" to "bool int". Added new overloaded function form
33078 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33079 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33080 exponent double-precision with floating point first argument.
33081 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33082 documentation of scalar_test_data_class, scalar_test_neg,
33083 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33084 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33085 vec_test_data_class built-in functions to reflect refinements in
33086 their type signatures.
33088 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33090 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33092 (aarch64_elf_asm_destructor): Likewise.
33094 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33096 PR rtl-optimization/78634
33097 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33098 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33099 * ifcvt.c (noce_try_cmove): Add missing cost check.
33101 PR rtl-optimization/71724
33102 * combine.c (if_then_else_cond): Look for situations where it is
33103 beneficial to undo the work of one of the recursive calls.
33105 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33107 PR tree-optimization/70754
33108 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33109 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33110 combined stmt before it if not NULL.
33111 (combine_chains): Process refs reversely and compute dominance point
33114 2017-01-23 Martin Liska <mliska@suse.cz>
33116 PR tree-optimization/79196
33117 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33118 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33120 (strlen_optimize_stmt): Call the renamed function.
33122 2017-01-23 Michael Matz <matz@suse.de>
33124 PR tree-optimization/78384
33125 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33127 2017-01-23 Richard Biener <rguenther@suse.de>
33129 PR tree-optimization/79186
33130 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33131 both incoming edges before moving an assert.
33133 2017-01-23 Martin Jambor <mjambor@suse.cz>
33135 * ipa-prop.c (load_from_param_1): Removed.
33136 (load_from_unmodified_param): Bits from load_from_param_1 put back
33138 (load_from_param): Removed.
33139 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33140 with stmt. Reverted back to use of load_from_unmodified_param.
33142 2017-01-23 Martin Jambor <mjambor@suse.cz>
33145 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33146 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33147 field a pointer to garbage collected vector, mark lattices and
33148 ipcp_orig_node with GTY((skip)).
33149 (ipa_get_param_count): Adjust to descriptors being a pointer.
33150 (ipa_get_param): Likewise.
33151 (ipa_get_type): Likewise.
33152 (ipa_get_param_move_cost): Likewise.
33153 (ipa_set_param_used): Likewise.
33154 (ipa_get_controlled_uses): Likewise.
33155 (ipa_set_controlled_uses): Likewise.
33156 (ipa_is_param_used): Likewise.
33157 (ipa_node_params_t): Move into garbage collector. New methods insert
33159 (ipa_node_params_sum): Annotate wth GTY(()).
33160 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33162 (ipa_load_from_parm_agg): Adjust declaration.
33163 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33164 * ipa-profile.c (ipa_profile): Likewise.
33165 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33166 (ipa_populate_param_decls): Make descriptors parameter garbage
33168 (ipa_dump_param): Adjust to descriptors being a pointer.
33169 (ipa_alloc_node_params): Likewise.
33170 (ipa_initialize_node_params): Likewise.
33171 (load_from_param_1): Make descriptors parameter garbage collected.
33172 (load_from_unmodified_param): Likewise.
33173 (load_from_param): Likewise.
33174 (ipa_load_from_parm_agg): Likewise.
33175 (ipa_node_params::~ipa_node_params): Removed.
33176 (ipa_free_all_node_params): Remove call to delete operator.
33177 (ipa_node_params_t::insert): New.
33178 (ipa_node_params_t::remove): Likewise.
33179 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33180 copy known_csts and known_contexts vectors.
33181 (ipa_read_node_info): Adjust to descriptors being a pointer.
33182 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33184 (ipcp_transform_function): Make descriptors variable garbage
33187 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33189 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33190 * config/i386/avx512dqintrin.h: Ditto.
33191 * config/i386/avx512fintrin.h: Ditto.
33192 * config/i386/i386.c: Handle new builtins.
33193 * config/i386/i386-builtin.def: Add new builtins.
33194 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33195 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33197 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33198 Martin Liska <mliska@suse.cz>
33200 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33201 * asan.c (asan_expand_poison_ifn): Support stores and use
33202 appropriate ASAN report function.
33203 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33204 * internal-fn.def (ASAN_POISON_USE): Declare.
33205 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33206 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33207 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33208 ASAN_POISON calls w/o LHS.
33209 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33210 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33211 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33212 * gimplify.c (asan_poison_variables): Add attribute
33213 use_after_scope_memory to variables that really needs to live
33215 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33216 having the attribute.
33218 2017-01-23 Martin Liska <mliska@suse.cz>
33220 * asan.c (create_asan_shadow_var): New function.
33221 (asan_expand_poison_ifn): Likewise.
33222 * asan.h (asan_expand_poison_ifn): New declaration.
33223 * internal-fn.c (expand_ASAN_POISON): Likewise.
33224 * internal-fn.def (ASAN_POISON): New builtin.
33225 * sanopt.c (pass_sanopt::execute): Expand
33226 asan_expand_poison_ifn.
33227 * tree-inline.c (copy_decl_for_dup_finish): Make function
33229 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33230 * tree-ssa.c (is_asan_mark_p): New function.
33231 (execute_update_addresses_taken): Rewrite local variables
33232 (identified just by use-after-scope as addressable) into SSA.
33234 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33236 * doc/install.texi (Specific): opensource.apple.com uses https
33237 now. Remove trailing slash.
33239 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33241 * README.Portability: Remove note on an Irix compatibility issue.
33243 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33245 * gcov.c (INCLUDE_ALGORITHM): Define.
33246 (INCLUDE_VECTOR): Define.
33247 No longer include <vector> and <algorithm> directly.
33249 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33251 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33253 * doc/invoke.texi (Code Gen Options): Ditto.
33255 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33258 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33260 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33262 rtl-optimization/79125
33263 * cprop.c (local_cprop_pass): Handle cases where we make an
33264 unconditional trap.
33266 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33270 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33271 read from, for big endian.
33273 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33275 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33276 register pauth builtins for LP64 only.
33278 2017-01-20 Marek Polacek <polacek@redhat.com>
33281 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33284 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33286 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33288 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33289 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33290 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33292 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33295 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33296 in big-endian mode when they are not a single duplicated value.
33298 2017-01-20 Richard Biener <rguenther@suse.de>
33300 * BASE-VER: Bump to 7.0.1.
33302 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33304 * omp-low.c (omplow_simd_context): New struct. Use it...
33305 (lower_rec_simd_input_clauses): ...here and...
33306 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33307 references to idx, lane, max_vf, is_simt.
33309 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33311 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33314 2017-01-20 Martin Jambor <mjambor@suse.cz>
33316 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33317 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33319 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33320 (GTFILES): Rename hsa.c to hsa-common.c.
33321 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33322 * hsa-dump.c: Likewise.
33323 * hsa-gen.c: Likewise.
33324 * hsa-regalloc.c: Likewise.
33325 * ipa-hsa.c: Likewise.
33326 * omp-expand.c: Likewise.
33327 * omp-low.c: Likewise.
33328 * toplev.c: Likewise.
33330 2017-01-20 Marek Polacek <polacek@redhat.com>
33333 * doc/invoke.texi: Document -Wduplicated-branches.
33334 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33335 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33336 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33337 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33338 return 0 only when not OEP_LEXICOGRAPHIC.
33339 (fold_build_cleanup_point_expr): Use the expression
33340 location when building CLEANUP_POINT_EXPR.
33341 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33342 * tree.c (add_expr): Handle error_mark_node.
33344 2017-01-20 Martin Liska <mliska@suse.cz>
33347 * tree-profile.c (init_ic_make_global_vars): Do not call
33349 (gimple_init_gcov_profiler): Likewise.
33351 2017-01-20 Martin Liska <mliska@suse.cz>
33354 * cgraph.h (maybe_create_reference): Remove argument and
33356 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33358 * ipa-cp.c (create_specialized_node): Likewise.
33359 * symtab.c (symtab_node::maybe_create_reference): Handle
33360 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33362 2017-01-20 Martin Liska <mliska@suse.cz>
33364 * read-rtl-function.c (function_reader::create_function): Use
33365 build_decl instread of build_decl_stat.
33367 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33369 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33370 * config/i386/avx512dqintrin.h: Ditto.
33371 * config/i386/avx512fintrin.h: Ditto.
33372 * config/i386/i386-builtin-types.def: Add new types.
33373 * config/i386/i386.c: Handle new types.
33374 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33375 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33376 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33377 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33378 (__builtin_ia32_kshiftridi): New.
33379 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33381 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33385 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33386 define to rs6000_init_stack_protect_guard.
33387 (rs6000_init_stack_protect_guard): New function.
33389 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33390 Yunqiang Su <yunqiang.su@imgtec.com>
33392 * config.gcc (supported_defaults): Add madd4.
33393 (with_madd4): Add validation.
33394 (all_defaults): Add madd4.
33395 * config/mips/mips.opt (mmadd4): New option.
33396 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33398 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33400 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33401 (ISA_HAS_FUSED_MADD4): Likewise.
33402 * doc/invoke.texi (-mmadd4): Document the new option.
33403 * doc/install.texi (--with-madd4): Document the new option.
33405 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33407 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33408 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33409 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33410 (aarch64_init_pauth_hint_builtins): New.
33411 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33412 (aarch64_expand_builtin): Expand new builtins.
33414 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33416 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33417 * combine-stack-adj.c (no_unhandled_cfa): Handle
33418 REG_CFA_TOGGLE_RA_MANGLE.
33419 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33420 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33421 info for return address signing.
33422 (aarch64_expand_epilogue): Likewise.
33424 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33426 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33427 * config/aarch64/aarch64-protos.h
33428 (aarch64_return_address_signing_enabled): New declaration.
33429 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33431 (aarch64_expand_prologue): Sign return address before it's pushed onto
33433 (aarch64_expand_epilogue): Authenticate return address fetched from
33435 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33436 (aarch64_attributes): New function attributes for "sign-return-address".
33437 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33438 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33439 ("*do_return"): Generate combined instructions according to key index.
33440 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33441 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33443 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33444 * config/aarch64/aarch64.opt (msign-return-address=): New.
33445 * doc/extend.texi (AArch64 Function Attributes): Documents
33446 "sign-return-address=".
33447 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33449 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33451 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33452 overall option summary.
33454 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33456 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33457 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33458 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33459 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33461 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33463 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33464 -mpower9-minmax by default for -mcpu=power9.
33465 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33466 128-bit floating point.
33468 2017-01-20 Alan Modra <amodra@gmail.com>
33470 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33471 optimizing for size.
33473 2017-01-20 Alan Modra <amodra@gmail.com>
33476 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33477 for strcmp and strncmp from corresponding builtin decl.
33479 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33481 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33482 instead of i386/rtems-64.h.
33483 * config/i386/rtems-64.h: Remove.
33485 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33489 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33491 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33493 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33495 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33496 Change int to HOST_WIDE_INT.
33497 * config/aarch64/aarch64-protos.h
33498 (aarch64_simd_gen_const_vector_dup): Likewise.
33499 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33501 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33503 * langhooks-def.h (lhd_type_for_size): New decl.
33504 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33505 * langhooks.c (lhd_type_for_size): New function, taken from
33508 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33510 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33511 define_bypass for CR latency.
33512 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33513 (power9-alu2): Add define_bypass for CR latency.
33515 (power9-mul): Update insn latency.
33516 (power9-mul-compare): Update insn latency, bypass latency and remove
33519 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33521 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33523 * config/aarch64/aarch64.md
33524 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33525 aarch64_nopcrelative_literal_loads.
33526 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33528 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33530 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33531 TARGET_LOONGSON_3A.
33532 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33534 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33537 * config.gcc (supported_defaults): Add lxc1-sxc1.
33538 (with_lxc1_sxc1): Add validation.
33539 (all_defaults): Add lxc1-sxc1.
33540 * config/mips/mips.opt (mlxc1-sxc1): New option.
33541 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33543 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33544 __mips_no_lxc1_sxc1.
33545 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33546 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33547 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33549 2017-01-19 Richard Biener <rguenther@suse.de>
33551 PR tree-optimization/72488
33552 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33553 sure to restore SSA info.
33554 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33556 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33558 PR rtl-optimization/79121
33559 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33560 of the inner type when shifting an extended value.
33562 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33565 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33566 interposable aliases.
33568 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33571 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33572 Use the evmergelohi instruction.
33573 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33574 (mov_sitf_e500_subreg8_2_be): Likewise.
33575 (mov_sitf_e500_subreg12_2_le): Likewise.
33576 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33577 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33578 (mov_sitf_e500_subreg8_2_le): Likewise.
33579 (mov_sitf_e500_subreg12_2_be): Likewise.
33581 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33583 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33584 attribute from vecsimple to vecperm.
33585 (altivec_vbpermq2): Likewise.
33587 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33590 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33592 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33593 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33594 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33595 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33596 case where N arg is SIZE_MAX.
33597 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33598 (cmpstrsi): Add pattern.
33600 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33602 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33603 __builtin_vec_revb builtins.
33604 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33605 built-in functions to support generation of the ISA 3.0 XXBR<x>
33606 vector byte reverse instructions.
33607 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33608 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33609 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33610 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33611 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33612 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33613 (P9V_BUILTIN_VEC_REVB): Likewise.
33614 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33615 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33616 (p9_xxbrq_v16qi): Likewise.
33617 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33618 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33619 (p9_xxbrh_v8hi): Likewise.
33620 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33621 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33622 vec_revb built-in functions.
33624 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33626 PR rtl-optimization/78952
33627 * config/i386/i386.md (any_extract): New code iterator.
33628 (*insvqi_2): Use any_extract for source operand.
33629 (*insvqi_3): Use any_shiftrt for source operand.
33631 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33633 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33635 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33637 2017-01-18 Matthias Klose <doko@ubuntu.com>
33639 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33641 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33643 * config/rs6000/altivec.h (vec_bperm): Change #define.
33644 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33645 (altivec_vbpermq2): New define_insn.
33646 (altivec_vbpermd): Likewise.
33647 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33648 function interface.
33649 (VBPERMD): Likewise.
33650 (VBPERM): New polymorphic function interface.
33651 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33652 Add entries for P9V_BUILTIN_VEC_VBPERM.
33653 * doc/extend.texi: Add interfaces for vec_bperm.
33655 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33657 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33658 first letter of error messages.
33659 (s390_resolve_overloaded_builtin): Likewise.
33660 * config/s390/s390.c (s390_expand_builtin): Likewise.
33661 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33662 (s390_valid_target_attribute_inner_p): Likewise.
33663 * config/s390/s390.md ("tabort"): Likewise.
33665 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33667 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33668 (ISA_AVOID_DIV_HILO): New macro.
33669 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33670 (ISA_HAS_DDIV): Likewise.
33672 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33674 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33676 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33678 * doc/invoke.texi (fabi-version): Spelling fix.
33680 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33683 * doc/invoke.texi (fabi-version): Mention mangling fix for
33686 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33689 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33691 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33694 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33695 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33697 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33699 (stack_protect_test): Ditto.
33700 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33701 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33703 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33704 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33705 -mstack-protector-guard-offset=.
33706 (RS/6000 and PowerPC Options): Ditto.
33708 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33710 * config/i386/i386.h (MASK_CLASS_P): New define.
33711 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33712 there are no registers from different register sets also when
33713 mask registers are used. Update function comment.
33714 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33715 to (*k/*r) and (*k/*km) alternatives.
33717 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33719 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33720 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33721 (EH_RETURN_HANDLER_RTX): New define.
33722 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33723 Force frame pointer in EH return functions.
33724 (aarch64_expand_epilogue): Add barrier for eh_return.
33725 (aarch64_final_eh_return_addr): Remove.
33726 (aarch64_eh_return_handler_rtx): New function.
33727 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33729 (aarch64_eh_return_handler_rtx): New prototype.
33731 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33733 * config/rs6000/altivec.h (vec_rlmi): New #define.
33734 (vec_vrlnm): Likewise.
33735 (vec_rlnm): Likewise.
33736 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33737 (UNSPEC_VRLNM): Likewise.
33738 (VIlong): New mode iterator.
33739 (altivec_vrl<VI_char>mi): New define_insn.
33740 (altivec_vrl<VI_char>nm): Likewise.
33741 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33743 (VRLDNM): Likewise.
33744 (RLNM): New polymorphic function entry.
33745 (VRLWMI): New monomorphic function entry.
33746 (VRLDMI): Likewise.
33747 (RLMI): New polymorphic function entry.
33748 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33749 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33750 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33753 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33756 * dwarf2out.c (field_byte_offset): Restore the
33757 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33758 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33759 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33762 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33765 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33767 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33770 * dwarf2out.c (add_data_member_location_attribute): For constant
33771 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33772 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33773 DW_AT_byte_size attributes.
33775 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33777 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33778 after forcing to constant memory when the code model is medium.
33780 2017-01-17 Julia Koval <julia.koval@intel.com>
33783 * config/i386/avx512fintrin.h
33784 (_mm512_i32gather_ps): Change __addr type to void const*.
33785 (_mm512_mask_i32gather_ps): Ditto.
33786 (_mm512_i32gather_pd): Ditto.
33787 (_mm512_mask_i32gather_pd): Ditto.
33788 (_mm512_i64gather_ps): Ditto.
33789 (_mm512_mask_i64gather_ps): Ditto.
33790 (_mm512_i64gather_pd): Ditto.
33791 (_mm512_mask_i64gather_pd): Ditto.
33792 (_mm512_i32gather_epi32): Ditto.
33793 (_mm512_mask_i32gather_epi32): Ditto.
33794 (_mm512_i32gather_epi64): Ditto.
33795 (_mm512_mask_i32gather_epi64): Ditto.
33796 (_mm512_i64gather_epi32): Ditto.
33797 (_mm512_mask_i64gather_epi32): Ditto.
33798 (_mm512_i64gather_epi64): Ditto.
33799 (_mm512_mask_i64gather_epi64): Ditto.
33800 (_mm512_i32scatter_ps): Change __addr type to void*.
33801 (_mm512_mask_i32scatter_ps): Ditto.
33802 (_mm512_i32scatter_pd): Ditto.
33803 (_mm512_mask_i32scatter_pd): Ditto.
33804 (_mm512_i64scatter_ps): Ditto.
33805 (_mm512_mask_i64scatter_ps): Ditto.
33806 (_mm512_i64scatter_pd): Ditto.
33807 (_mm512_mask_i64scatter_pd): Ditto.
33808 (_mm512_i32scatter_epi32): Ditto.
33809 (_mm512_mask_i32scatter_epi32): Ditto.
33810 (_mm512_i32scatter_epi64): Ditto.
33811 (_mm512_mask_i32scatter_epi64): Ditto.
33812 (_mm512_i64scatter_epi32): Ditto.
33813 (_mm512_mask_i64scatter_epi32): Ditto.
33814 (_mm512_i64scatter_epi64): Ditto.
33815 (_mm512_mask_i64scatter_epi64): Ditto.
33816 * config/i386/avx512pfintrin.h
33817 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33818 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33819 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33820 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33821 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33822 (_mm512_prefetch_i32scatter_ps): Ditto.
33823 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33824 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33825 (_mm512_prefetch_i64scatter_pd): Ditto.
33826 (_mm512_prefetch_i64scatter_ps): Ditto.
33827 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33828 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33829 * config/i386/avx512vlintrin.h
33830 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33831 (_mm_mmask_i32gather_ps): Ditto.
33832 (_mm256_mmask_i32gather_pd): Ditto.
33833 (_mm_mmask_i32gather_pd): Ditto.
33834 (_mm256_mmask_i64gather_ps): Ditto.
33835 (_mm_mmask_i64gather_ps): Ditto.
33836 (_mm256_mmask_i64gather_pd): Ditto.
33837 (_mm_mmask_i64gather_pd): Ditto.
33838 (_mm256_mmask_i32gather_epi32): Ditto.
33839 (_mm_mmask_i32gather_epi32): Ditto.
33840 (_mm256_mmask_i32gather_epi64): Ditto.
33841 (_mm_mmask_i32gather_epi64): Ditto.
33842 (_mm256_mmask_i64gather_epi32): Ditto.
33843 (_mm_mmask_i64gather_epi32): Ditto.
33844 (_mm256_mmask_i64gather_epi64): Ditto.
33845 (_mm_mmask_i64gather_epi64): Ditto.
33846 (_mm256_i32scatter_ps): Change __addr type to void*.
33847 (_mm256_mask_i32scatter_ps): Ditto.
33848 (_mm_i32scatter_ps): Ditto.
33849 (_mm_mask_i32scatter_ps): Ditto.
33850 (_mm256_i32scatter_pd): Ditto.
33851 (_mm256_mask_i32scatter_pd): Ditto.
33852 (_mm_i32scatter_pd): Ditto.
33853 (_mm_mask_i32scatter_pd): Ditto.
33854 (_mm256_i64scatter_ps): Ditto.
33855 (_mm256_mask_i64scatter_ps): Ditto.
33856 (_mm_i64scatter_ps): Ditto.
33857 (_mm_mask_i64scatter_ps): Ditto.
33858 (_mm256_i64scatter_pd): Ditto.
33859 (_mm256_mask_i64scatter_pd): Ditto.
33860 (_mm_i64scatter_pd): Ditto.
33861 (_mm_mask_i64scatter_pd): Ditto.
33862 (_mm256_i32scatter_epi32): Ditto.
33863 (_mm256_mask_i32scatter_epi32): Ditto.
33864 (_mm_i32scatter_epi32): Ditto.
33865 (_mm_mask_i32scatter_epi32): Ditto.
33866 (_mm256_i32scatter_epi64): Ditto.
33867 (_mm256_mask_i32scatter_epi64): Ditto.
33868 (_mm_i32scatter_epi64): Ditto.
33869 (_mm_mask_i32scatter_epi64): Ditto.
33870 (_mm256_i64scatter_epi32): Ditto.
33871 (_mm256_mask_i64scatter_epi32): Ditto.
33872 (_mm_i64scatter_epi32): Ditto.
33873 (_mm_mask_i64scatter_epi32): Ditto.
33874 (_mm256_i64scatter_epi64): Ditto.
33875 (_mm256_mask_i64scatter_epi64): Ditto.
33876 (_mm_i64scatter_epi64): Ditto.
33877 (_mm_mask_i64scatter_epi64): Ditto.
33878 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33879 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33880 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33881 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33882 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33883 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33884 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33885 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33886 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33887 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33888 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33889 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33890 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33891 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33892 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33893 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33894 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33895 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33896 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33897 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33898 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33899 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33900 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33901 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33902 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33903 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33904 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33905 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33906 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33907 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33908 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33909 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33910 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33911 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33912 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33913 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33914 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33915 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33916 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33917 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33918 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33919 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33920 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33921 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33922 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33923 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33924 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33925 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33926 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33927 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33928 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33929 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33930 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33931 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33932 definitions accordingly.
33934 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33935 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33938 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33941 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33944 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33945 hard regs for inner regno.
33947 2017-01-17 Martin Liska <mliska@suse.cz>
33950 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33951 assumption and add comment.
33953 2017-01-17 Nathan Sidwell <nathan@acm.org>
33955 * ipa-visibility.c (localize_node): New function, broken out of ...
33956 (function_and_variable_visibility): ... here. Call it.
33958 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33960 PR middle-end/77445
33961 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33962 correctly set frequency of oudgoing edge.
33963 (duplicate_thread_path): Fix profile updating.
33965 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33968 * configure.ac: Add GCC_BASE_VER.
33969 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33970 version from BASE-VER file.
33971 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33972 (gcc.o): Depend on $(BASEVER).
33973 * common.opt (dumpfullversion): New option.
33974 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33975 * doc/invoke.texi: Document -dumpfullversion.
33976 * doc/install.texi: Document --with-gcc-major-version-only.
33977 * configure: Regenerated.
33979 2017-01-17 Richard Biener <rguenther@suse.de>
33981 PR tree-optimization/71433
33982 * tree-vrp.c (register_new_assert_for): Merge same asserts
33983 on all incoming edges.
33984 (process_assert_insertions_for): Handle insertions at the
33987 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33989 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33990 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33992 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33995 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33998 2017-01-17 Alan Modra <amodra@gmail.com>
34001 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34002 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34003 symbolic stack limit when pic.
34005 2017-01-16 Martin Sebor <msebor@redhat.com>
34007 PR tree-optimization/78608
34008 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34010 2017-01-16 Jeff Law <law@redhat.com>
34013 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34014 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34015 for several include directories that may be relative to sysroot.
34016 * config/i386/x-mingw32 (gplus_includedir): Define.
34017 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34018 (native_system_includedir): Likewise.
34019 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34020 override if TARGET_SYSTEM_ROOT is defined.
34021 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34023 PR tree-optimization/79090
34024 PR tree-optimization/33562
34025 PR tree-optimization/61912
34026 PR tree-optimization/77485
34027 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34028 and computed trims into the dump file.
34030 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34032 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34034 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34037 * gimplify.c (gimplify_init_constructor): If want_value and
34038 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34042 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34043 sequence. Formatting fixes.
34044 (doloop_optimize): Formatting fixes.
34047 * gcc.c (debug_level_greater_than_spec_func): New function.
34048 (static_spec_functions): Add debug-level-gt spec function.
34049 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34051 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34052 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34053 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34054 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34055 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34056 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34058 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34060 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34061 QImode fixups to general and mask registers only.
34063 2017-01-16 Carl Love <cel@us.ibm.com>
34065 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34066 for built-in functions
34067 vector signed char vec_nabs (vector signed char)
34068 vector signed short vec_nabs (vector signed short)
34069 vector signed int vec_nabs (vector signed int)
34070 vector signed long long vec_nabs (vector signed long long)
34071 vector float vec_nabs (vector float)
34072 vector double vec_nabs (vector double)
34073 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34075 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34076 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34077 * doc/extend.texi: Update the documentation file for the new built-in
34080 2017-01-16 Martin Sebor <msebor@redhat.com>
34082 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34085 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34087 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34088 UNSPEC_VSX__XXSPLTD to require special splat handling.
34090 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34093 * system.h: Poison strndup.
34095 2017-01-16 Alan Modra <amodra@gmail.com>
34098 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34101 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34103 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34105 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34107 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34108 call recog here. Assert that INSN_CODE (insn) is non-negative.
34110 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34113 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34115 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34116 in the currently scheduled RTL fragment.
34118 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34120 PR rtl-optimization/78751
34121 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34124 2017-01-14 Jeff Law <law@redhat.com>
34126 PR tree-optimization/79090
34127 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34128 variable length stores.
34129 (compute_trims): Delete dead assignment to *trim_tail.
34130 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34133 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34135 PR rtl-optimization/78626
34136 PR rtl-optimization/78727
34137 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34138 of a block, and split such blocks after everything else is finished.
34140 2017-01-14 Alan Modra <amodra@gmail.com>
34143 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34144 target legitimate_combined_insn.
34145 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34146 (rs6000_legitimate_combined_insn): New function.
34147 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34149 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34150 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34151 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34153 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34155 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34157 2017-01-13 Jeff Law <law@redhat.com>
34159 PR tree-optimization/33562
34160 PR tree-optimization/61912
34161 PR tree-optimization/77485
34162 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34164 (delete_dead_assignment): Likewise.
34165 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34166 statement to delete_dead_call and delete_dead_assignment.
34168 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34171 * substring-locations.c (format_warning_va): Strengthen case 1 so
34172 that both endpoints of the substring must be within the format
34173 range for just the substring to be printed.
34175 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34177 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34178 * config/i386/i386.c (ix86_target_string): Add missing options
34179 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34180 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34181 flags_other and ix86_target_other to flags2_other. Display unknown
34183 (ix86_valid_target_attribute_inner_p): Add missing options and
34184 reorder options by implied ISAs, as in ix86_target_string.
34186 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34188 * hash-table.h (hash_table::too_empty_p): New function.
34189 (hash_table::expand): Use it.
34190 (hash_table::traverse): Likewise.
34191 (hash_table::empty_slot): Use sizeof (value_type) instead of
34192 sizeof (PTR) to convert bytes to elements. Shrink the table
34193 if the current size is excessive for the current number of
34196 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34198 * ira-costs.c (record_reg_classes): Break from the inner loop
34199 early once alt_fail is known to be true. Update outer loop
34200 handling accordingly.
34202 2017-01-13 Jeff Law <law@redhat.com>
34204 * tree-ssa-dse.c (decrement_count): New function.
34205 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34206 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34207 when we know the partially dead statement is a mem* function.
34209 PR tree-optimization/61912
34210 PR tree-optimization/77485
34211 * tree-ssa-dse.c: Include expr.h.
34212 (maybe_trim_constructor_store): New function.
34213 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34215 PR tree-optimization/33562
34216 PR tree-optimization/61912
34217 PR tree-optimization/77485
34218 * doc/invoke.texi: Document new dse-max-object-size param.
34219 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34220 * tree-ssa-dse.c: Include params.h.
34221 (dse_store_status): New enum.
34222 (initialize_ao_ref_for_dse): New, partially extracted from
34224 (valid_ao_ref_for_dse, normalize_ref): New.
34225 (setup_live_bytes_from_ref, compute_trims): Likewise.
34226 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34227 (maybe_trim_partially_dead_store): Likewise.
34228 (maybe_trim_complex_store): Likewise.
34229 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34230 Track what bytes live from the original store. Return tri-state
34231 for dead, partially dead or live.
34232 (dse_dom_walker): Add constructor, destructor and new private members.
34233 (delete_dead_call, delete_dead_assignment): New extracted from
34235 (dse_optimize_stmt): Make a member of dse_dom_walker.
34236 Use initialize_ao_ref_for_dse.
34238 PR tree-optimization/33562
34239 PR tree-optimization/61912
34240 PR tree-optimization/77485
34241 * sbitmap.h (bitmap_count_bits): Prototype.
34242 (bitmap_clear_range, bitmap_set_range): Likewise.
34243 * sbitmap.c (bitmap_clear_range): New function.
34244 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34246 2017-01-13 Martin Liska <mliska@suse.cz>
34249 * function.c (set_cfun): Add new argument force.
34250 * function.h (set_cfun): Likewise.
34251 * ipa-inline-transform.c (inline_call): Use the function when
34252 strict alising from is dropped for function we inline to.
34254 2017-01-13 Richard Biener <rguenther@suse.de>
34256 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34257 for dumping GIMPLE INTEGER_CSTs.
34259 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34261 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34262 to 201112L since C++17.
34264 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34267 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34268 if -fsanitize=kernel-address is present.
34270 2017-01-13 Richard Biener <rguenther@suse.de>
34272 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34273 as _Literal ( type ) number in case usual suffixes do not
34274 preserve all information.
34276 2017-01-13 Richard Biener <rguenther@suse.de>
34278 PR tree-optimization/77283
34279 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34280 and ssa-iterators.h.
34281 (is_feasible_trace): Implement a cost model based on joiner
34284 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34287 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34288 char or short to __float128/_Float128 directly.
34290 2017-01-12 Martin Sebor <msebor@redhat.com>
34292 to -Wformat-overflow.
34293 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34294 (min_bytes_remaining): Same.
34295 (get_string_length): Same.
34296 (format_string): Same.
34297 (format_directive): Same.
34299 (pass_sprintf_length::handle_gimple_call): Same.
34301 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34303 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34304 info.nowrite calls with no lhs that can't throw. Return bool
34305 whether gsi_remove has been called or not.
34306 (pass_sprintf_length::handle_gimple_call): Return bool whether
34307 try_substitute_return_value called gsi_remove. Formatting fix.
34308 (pass_sprintf_length::execute): Don't use gsi_remove if
34309 handle_gimple_call returned true.
34312 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34313 be removed due to side-effects, don't remove following barrier nor
34314 turn the successor edge into fallthru edge.
34316 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34319 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34320 element-reversing loads and stores as not swappable.
34322 2017-01-12 Nathan Sidwell <nathan@acm.org>
34323 Nicolai Stange <nicstange@gmail.com>
34325 * combine.c (try_combine): Don't ignore result of overlap checking
34326 loop. Combine overlap & asm check into single loop.
34328 2017-01-12 Richard Biener <rguenther@suse.de>
34330 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34331 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34333 2017-01-12 Richard Biener <rguenther@suse.de>
34335 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34336 and TS_TARGET_OPTION directly derive from TS_BASE.
34337 * tree-core.h (tree_optimization_option): Derive from tree_base.
34338 (tree_target_option): Likewise.
34340 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34342 * config/i386/i386.c (memory_address_length): Increase len
34343 only when rip_relative_addr_p returns false.
34345 2017-01-11 Julia Koval <julia.koval@intel.com>
34347 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34348 (OPTION_MASK_ISA_SGX_SET): New.
34349 (ix86_handle_option): Handle OPT_msgx.
34350 * config.gcc: Added sgxintrin.h.
34351 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34352 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34353 * config/i386/i386.c (ix86_target_string): Add -msgx.
34355 (ix86_option_override_internal): Handle new options.
34356 (ix86_valid_target_attribute_inner_p): Add sgx.
34357 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34358 * config/i386/i386.opt: Add msgx.
34359 * config/i386/sgxintrin.h: New file.
34360 * config/i386/x86intrin.h: Add sgxintrin.h.
34362 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34365 * fold-const.c (maybe_nonzero_address): Return 1 for function
34367 (tree_single_nonzero_warnv_p): Don't handle function local objects
34371 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34374 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34377 * opts.c: Include "spellcheck.h"
34378 (struct string_fragment): New struct.
34379 (struct edit_distance_traits<const string_fragment &>): New
34381 (get_closest_sanitizer_option): New function.
34382 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34384 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34386 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34388 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34389 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34390 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34391 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34392 for initial die_offset if dwarf_split_debug_info.
34393 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34394 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34396 (output_skeleton_debug_sections): Formatting fix. Use
34397 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34398 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34400 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34402 * config/arm/cortex-a53.md: Add bypasses for
34403 cortex_a53_r2f_cvt.
34404 (cortex_a53_r2f): Only use for transfers.
34405 (cortex_a53_f2r): Likewise.
34406 (cortex_a53_r2f_cvt): Add reservation for conversions.
34407 (cortex_a53_f2r_cvt): Likewise.
34409 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34411 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34412 to all inlined functions, change static to extern.
34414 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34417 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34419 (arm_assemble_integer): Likewise.
34421 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34423 * config.gcc: Use new awk script to check CPU, FPU and architecture
34424 parameters for --with-... options.
34425 * config/arm/parsecpu.awk: New file
34426 * config/arm/arm-cpus.in: New file.
34427 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34429 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34431 * config/arm/t-arm: Update dependency rules.
34432 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34433 of processing .def files.
34434 * config/arm/genopt.sh: Deleted.
34435 * config/arm/gentune.sh: Deleted.
34436 * config/arm/arm-cores.def: Deleted.
34437 * config/arm/arm-arches.def: Deleted.
34438 * config/arm/arm-fpus.def: Deleted.
34439 * config/arm/arm-tune.md: Regenerated.
34440 * config/arm/arm-tables.opt: Regenerated.
34441 * config/arm/arm-cpu.h: New generated file.
34442 * config/arm/arm-cpu-data.h: New generated file.
34443 * config/arm/arm-cpu-cdata.h: New generated file.
34445 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34448 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34450 (input_varpool_node): Unpack dynamically_initialized bit.
34452 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34454 PR rtl-optimization/79032
34455 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34456 the alignment of the adjusted memory reference against that of MODE,
34457 instead of the alignment of the original memory reference.
34459 2017-01-11 Martin Jambor <mjambor@suse.cz>
34461 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34463 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34464 decorated functions.
34466 2017-01-11 Richard Biener <rguenther@suse.de>
34468 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34469 set range/nonnull info for PHI results. Do not set it on
34470 stmts marked for removal.
34472 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34474 * expr.c (store_field): In the bitfield case, fetch the return value
34475 from the registers before applying a single big-endian adjustment.
34476 Always do a final load for a BLKmode value not larger than a word.
34478 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34481 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34482 that we correctly handle column numbers greater than
34483 LINE_MAP_MAX_COLUMN_NUMBER.
34485 2017-01-10 Martin Sebor <msebor@redhat.com>
34487 PR middle-end/78245
34488 * gimple-ssa-sprintf.c (get_destination_size): Call
34489 {init,fini}object_sizes.
34490 * tree-object-size.c (addr_object_size): Adjust.
34491 (pass_through_call): Adjust.
34492 (pass_object_sizes::execute): Adjust.
34493 * tree-object-size.h (fini_object_sizes): Declare.
34495 2017-01-10 Martin Sebor <msebor@redhat.com>
34497 PR tree-optimization/78775
34498 * builtins.c (get_size_range): Move...
34499 * calls.c: ...to here.
34500 (alloc_max_size): Accept zero argument.
34501 (operand_signed_p): Remove.
34502 (maybe_warn_alloc_args_overflow): Call get_size_range.
34503 * calls.h (get_size_range): Declare.
34505 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34507 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34508 from TI's devices.csv file as of September 2016.
34509 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34511 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34513 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34514 * doc/invoke.texi: Likewise.
34515 * doc/md.texi: Likewise.
34516 * doc/objc.texi: Likewise.
34518 2017-01-10 Joshua Conner <joshconner@google.com>
34520 * config/arm/fuchsia-elf.h: New file.
34521 * config/fuchsia.h: New file.
34522 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34523 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34525 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34527 2016-01-10 Richard Biener <rguenther@suse.de>
34529 PR tree-optimization/79034
34530 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34531 Propagate out degenerate PHIs in the joiner.
34533 2017-01-10 Martin Liska <mliska@suse.cz>
34535 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34536 (sort_congruence_classes_by_decl_uid): Likewise.
34537 (sort_congruence_class_groups_by_decl_uid): Likewise.
34538 (sem_item_optimizer::merge_classes): Sort class, groups in these
34539 classes and members in the groups by DECL_UID of declarations.
34540 This would make merge operations stable.
34542 2017-01-10 Martin Liska <mliska@suse.cz>
34544 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34545 usage of m_classes_vec.
34546 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34547 (sem_item_optimizer::get_group_by_hash): Likewise.
34548 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34549 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34550 (sem_item_optimizer::verify_classes): Likewise.
34551 (sem_item_optimizer::process_cong_reduction): Likewise.
34552 (sem_item_optimizer::dump_cong_classes): Likewise.
34553 (sem_item_optimizer::merge_classes): Likewise.
34554 * ipa-icf.h (congruence_class_hash): Rename from
34555 congruence_class_group_hash. Remove declaration of m_classes_vec.
34557 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34559 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34560 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34561 * config.gcc: Add avx512vpopcntdqintrin.h.
34562 * config/i386/avx512vpopcntdqintrin.h: New.
34563 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34564 * config/i386/i386-builtin-types.def: Add new types.
34565 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34566 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34567 __builtin_ia32_vpopcountq_v8di_mask): New.
34568 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34569 __AVX512VPOPCNTDQ__.
34570 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34571 (PTA_AVX512VPOPCNTDQ): Define.
34572 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34573 TARGET_AVX512VPOPCNTDQ_P): Define.
34574 * config/i386/i386.opt: Add mavx512vpopcntdq.
34575 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34576 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34578 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34580 PR middle-end/77484
34581 * predict.def (PRED_CALL): Set to 67.
34583 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34585 * expr.c (store_field): In the bitfield case, if the value comes from
34586 a function call and is of an aggregate type returned in registers, do
34587 not modify the field mode; extract the value in all cases if the mode
34588 is BLKmode and the size is not larger than a word.
34590 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34593 * config/i386/cpuid.h: Fix undefined behavior.
34595 2017-01-04 Jeff Law <law@redhat.com>
34597 PR tree-optimization/79007
34598 PR tree-optimization/67955
34599 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34600 conservative for pt.null when flag_non_call_exceptions is on.
34602 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34604 PR translation/79019
34605 PR translation/79020
34606 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34607 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34608 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34610 * config/avr/avr.opt (maccumulate-args): Likewise.
34611 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34612 * common.opt (freport-bug): Likewise.
34613 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34614 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34615 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34616 translatable string.
34617 * config/i386/i386.c (function_value_32): Likewise.
34618 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34619 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34621 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34622 * common/config/msp430/msp430-common.c (msp430_handle_option):
34624 * symtab.c (symtab_node::verify_base): Likewise.
34625 * opts.c (set_debug_level): Likewise.
34626 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34627 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34628 missing whitespace to translatable strings.
34629 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34630 * config/sh/superh.h: Likewise.
34631 * config/i386/xopintrin.h: Likewise.
34632 * config/i386/znver1.md: Likewise.
34633 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34634 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34635 * double-int.h (struct double_int): Likewise.
34636 * double-int.c (div_and_round_double): Likewise.
34637 * wide-int.cc: Likewise.
34638 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34639 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34640 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34641 (crossjumps_occurred): ... this.
34642 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34645 PR tree-optimization/78899
34646 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34647 returning bool return struct loop *, NULL for failure and the new
34649 (versionable_outer_loop_p): Don't version outer loop if it has
34650 dont_vectorized bit set.
34651 (tree_if_conversion): When versioning outer loop, ensure
34652 tree_if_conversion is performed also on the inner loop of the
34653 non-vectorizable outer loop copy.
34654 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34655 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34656 prevent vectorization of it.
34657 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34658 the outer loop vectorization of the non-scalar version is attempted
34659 before vectorization of the inner loop in scalar version. If
34660 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34661 vectorization of its inner loop.
34662 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34663 has 2 inner loops, rename also on edges from bb whose single pred
34664 is outer_loop->header. Fix typo in function comment.
34666 2017-01-09 Martin Sebor <msebor@redhat.com>
34669 * asan.c (asan_emit_stack_protection): Increase local buffer size
34670 to avoid snprintf truncation warning.
34672 2017-01-09 Andrew Pinski <apinski@cavium.com>
34674 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34675 to reference thunderx2t99 for the tuning structure
34676 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34678 (thunderx2t99_extra_costs): This.
34679 * config/aarch64/aarch64-tune.md: Regenerate.
34680 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34681 (vulcan_addrcost_table): This.
34682 (vulcan_regmove_cost): Rename to ...
34683 (thunderx2t99_regmove_cost): This.
34684 (vulcan_vector_cost): Rename to ...
34685 (thunderx2t99_vector_cost): this.
34686 (vulcan_branch_cost): Rename to ...
34687 (thunderx2t99_branch_cost): This.
34688 (vulcan_tunings): Rename to ...
34689 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34690 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34692 2017-01-09 Martin Jambor <mjambor@suse.cz>
34696 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34697 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34698 (propagate_vr_accross_jump_function): Use the above function for all
34699 value range computations for pass-through jump functions and type
34700 converasion from explicit value range values.
34701 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34702 parameters from TYPE_ARG_TYPES.
34703 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34704 (ipa_write_node_info): Stream type of the actual argument.
34705 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34707 2017-01-09 Martin Liska <mliska@suse.cz>
34710 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34711 (lookup_compiler): Do not show error message with have_E.
34713 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34715 PR tree-optimization/78938
34716 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34717 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34718 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34719 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34722 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34724 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34727 2017-01-09 Richard Biener <rguenther@suse.de>
34729 PR tree-optimization/78997
34730 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34731 name condition properly.
34733 2017-01-09 Richard Biener <rguenther@suse.de>
34736 * dwarf2out.c (is_cxx): New overload with context.
34737 (is_naming_typedef_decl): Use it.
34739 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34741 * invoke.texi (Option Summary): Correct spacing in option lists
34742 and add line breaks to fix over-long lines.
34744 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34746 PR middle-end/17660
34748 * extend.texi (Common Variable Attributes): Add xref to GCC
34749 Internals manual to explain mode attribute keywords.
34751 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34754 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34755 and Preprocessor Options.
34756 (Options for Linking): Document -pthread here....
34757 (RS/6000 and PowerPC Options): ...not here.
34758 (Solaris 2 Options): ...or here.
34759 * doc/cppopts.texi: Document -pthread.
34761 2017-01-08 Martin Sebor <msebor@redhat.com>
34763 PR middle-end/77708
34764 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34765 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34766 New member functions.
34767 (format_directive): Used them.
34769 (pass_sprintf_length::handle_gimple_call): Same.
34770 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34771 to avoid truncation for any argument.
34772 (extract_affine_mul): Same.
34773 * tree.c (get_file_function_name): Same.
34775 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34777 PR middle-end/77484
34778 * predict.def (PRED_INDIR_CALL): Set to 86.
34780 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34782 PR preprocessor/54124
34783 * doc/cppopts.texi: Reformat -d subtable to list the full name
34784 of the options. Add cross-reference to the docs for the general
34785 compiler -d options.
34786 * doc/invoke.texi (Developer Options): Add cross-reference to the
34787 preprocessor-specific -d option documentation.
34789 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34791 PR preprocessor/13498
34792 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34793 redudant material, and reflect new command-line options.
34794 (System Headers): Likewise.
34796 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34798 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34799 -isystem, and -idirafter. Copy-edit.
34800 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34801 default for -ftrack-macro-expansion. Delete obsolete and
34802 badly-formatted implementation details about -fdebug-cpp output.
34803 * doc/cppwarnopts.texi: Copy-edit.
34805 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34808 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34809 that the transition from a max line width >= 1<<10 to narrower
34810 lines works correctly.
34812 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34814 * doc/options.texi (PerFunction): New.
34815 * opt-functions.awk (switch_flags): Map both Optimization and
34816 PerFunction to CL_OPTIMIZATION.
34817 * opth-gen.awk: Test for PerFunction flag along with
34819 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34820 it only when the latter is present. Skip those that don't in
34821 the hash function generator.
34822 * common.opt (fvar-tracking): Mark as PerFunction instead of
34824 (fvar-tracking-assignments): Likewise.
34825 (fvar-tracking-assignments-toggle): Likewise.
34826 (fvar-tracking-uninit): Likewise.
34828 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34830 PR translation/79018
34831 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34834 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34837 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34838 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34839 TARGET_LONG_JUMP_TABLE_OFFSETS.
34840 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34841 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34842 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34843 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34844 * config/m68k/m68k.md (tablejump expander): Likewise.
34845 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34846 TARGET_LONG_JUMP_TABLE_OFFSETS.
34847 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34848 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34850 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34851 David Holsgrove <david.holsgrove@xilinx.com>
34853 * common/config/microblaze/microblaze-common.c
34854 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34855 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34857 * config/microblaze/microblaze.c (microblaze_must_save_register)
34858 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34860 (microblaze_eh_return): New function.
34861 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34862 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34863 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34864 * config/microblaze/microblaze.md (eh_return): New pattern.
34866 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34868 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34869 GCC_DIAGNOSTIC_STRINGIFY): Define.
34871 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34873 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34875 * config/arm/arm.md (<mcrr>): New.
34877 * config/arm/arm.c (arm_arch5te): New.
34878 (arm_option_override): Set arm_arch5te.
34879 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34881 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34882 (arm_mcrr_qualifiers): ... this. New.
34883 (MRRC_QUALIFIERS): Define to...
34884 (arm_mrrc_qualifiers): ... this. New.
34885 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34887 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34888 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34889 (MRRCI, mrrc, MRRC): New.
34890 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34891 VUNSPEC_MRRC2): New.
34893 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34895 * config/arm/arm.md (<mcr>): New.
34897 * config/arm/arm.c (arm_coproc_builtin_available): Add
34898 support for mcr, mrc, mcr2 and mrc2.
34899 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34900 (arm_mcr_qualifiers): ... this. New.
34901 (MRC_QUALIFIERS): Define to ...
34902 (arm_mrc_qualifiers): ... this. New.
34903 (MCR_QUALIFIERS): Define to ...
34904 (arm_mcr_qualifiers): ... this. New.
34905 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34907 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34908 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34909 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34910 VUNSPEC_MRC2): New.
34912 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34914 * config/arm/arm.md (*ldc): New.
34918 * config/arm/arm.c (arm_coproc_builtin_available): Add
34919 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34920 (arm_coproc_ldc_stc_legitimate_address): New.
34921 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34922 'qualifier_const_pointer'.
34923 (LDC_QUALIFIERS): Define to...
34924 (arm_ldc_qualifiers): ... this. New.
34925 (STC_QUALIFIERS): Define to...
34926 (arm_stc_qualifiers): ... this. New.
34927 * config/arm/arm-protos.h
34928 (arm_coproc_ldc_stc_legitimate_address): New.
34929 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34930 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34931 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34932 stc2, stcl, stc2l): New.
34933 * config/arm/constraints.md (Uz): New.
34934 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34935 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34936 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34937 VUNSPEC_STC2L): New.
34939 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34941 * config/arm/arm.md (<cdp>): New.
34942 * config/arm/arm.c (neon_const_bounds): Rename this ...
34943 (arm_const_bounds): ... this.
34944 (arm_coproc_builtin_available): New.
34945 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34946 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34947 (CDP_QUALIFIERS): Define to...
34948 (arm_cdp_qualifiers): ... this. New.
34950 (arm_expand_builtin_args): Add case for 6 arguments.
34951 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34952 (arm_const_bounds): ... this.
34953 (arm_coproc_builtin_available): New.
34954 * config/arm/arm_acle.h (__arm_cdp): New.
34956 * config/arm/arm_acle_builtins.def (cdp): New.
34958 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34959 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34960 'arm_const_bounds'.
34961 * config/arm/types.md (coproc): New.
34962 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34963 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34964 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34965 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34967 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34969 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34970 (UBINOP_QUALIFIERS): New.
34972 (acle_builtin_data): New. Change comment.
34973 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34974 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34975 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34976 arm_acle_builtins.def.
34977 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34978 (arm_init_acle_builtins): New.
34979 (CRC32_BUILTIN): Remove.
34980 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34981 crc32cb, crc32ch and crc32cw.
34982 (arm_init_crc32_builtins): Remove.
34983 (arm_init_builtins): Use arm_init_acle_builtins rather
34984 than arm_init_crc32_builtins.
34985 (arm_expand_acle_builtin): New.
34986 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34987 * config/arm/arm_acle_builtins.def: New.
34989 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34991 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34992 (arm_builtin_datum): ... this.
34993 (arm_init_neon_builtin): Rename to ...
34994 (arm_init_builtin): ... this. Add a new parameters PREFIX
34995 and USE_SIG_IN_NAME.
34996 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34997 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34998 'arm_builtin_datum'.
34999 (arm_init_vfp_builtins): Likewise.
35000 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35001 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35002 (arm_expand_neon_args): Rename to ...
35003 (arm_expand_builtin_args): ... this. Rename builtin_arg
35004 enum values and differentiate between ARG_BUILTIN_MEMORY
35005 and ARG_BUILTIN_NEON_MEMORY.
35006 (arm_expand_neon_builtin_1): Rename to ...
35007 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35008 values, arm_expand_builtin_args and add bool parameter NEON.
35009 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35010 (arm_expand_vfp_builtin): Likewise.
35011 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35013 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35015 PR middle-end/77484
35016 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35017 * predict.c (tree_estimate_probability_bb): Reverse direction of
35018 polymorphic call predictor.
35020 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35022 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35023 (determine_pass_name_match): ...this new function and...
35024 (should_skip_pass_p): ...this new function.
35026 2017-01-06 Nathan Sidwell <nathan@acm.org>
35028 * ipa-visibility.c (function_and_variable_visibility): Reformat
35029 comments and long lines. Remove extrneous if.
35030 * symtab.c (symtab_node::make_decl_local): Fix code format.
35031 (symtab_node::set_section_for_node): Fix comment typo.
35033 2017-01-06 Martin Liska <mliska@suse.cz>
35036 * lra-constraints.c: Rename invariant to lra_invariant.
35037 * predict.c (set_even_probabilities): Initialize e to NULL.
35039 2017-01-05 Martin Sebor <msebor@redhat.com>
35041 PR tree-optimization/78910
35042 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35043 (format_integer): Correct off-by-one error in the handling
35044 of precision with negative numbers in signed conversions..
35046 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35048 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35050 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35052 PR tree-optimization/71016
35053 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35054 factor_out_conditional_conversion. Formatting fix.
35055 (factor_out_conditional_conversion): Add cond_stmt argument.
35056 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35057 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35060 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35062 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35063 read-rtl-function.o, and selftest-rtl.o.
35064 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35065 (selftest::aarch64_test_loading_full_dump): New function.
35066 (selftest::aarch64_run_selftests): New function.
35067 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35068 selftest::aarch64_run_selftests.
35069 * config/i386/i386.c
35070 (selftest::ix86_test_loading_dump_fragment_1): New function.
35071 (selftest::ix86_test_loading_call_insn): New function.
35072 (selftest::ix86_test_loading_full_dump): New function.
35073 (selftest::ix86_test_loading_unspec): New function.
35074 (selftest::ix86_run_selftests): Call the new functions.
35075 * emit-rtl.c (maybe_set_max_label_num): New function.
35076 * emit-rtl.h (maybe_set_max_label_num): New decl.
35077 * function.c (instantiate_decls): Guard call to
35078 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35079 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35081 * gensupport.c (gen_reader::gen_reader): Pass "false"
35082 for new "compact" param of rtx_reader.
35083 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35084 rather than an empty string for NULL strings.
35085 * read-md.c: Potentially include config.h rather than bconfig.h.
35086 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35087 (have_error): New global, copied from errors.c.
35088 (md_reader::read_name): Rename to...
35089 (md_reader::read_name_1): ...this, adding "out_loc" param,
35090 and converting "missing name or number" to returning false, rather
35092 (md_reader::read_name): Reimplement in terms of read_name_1.
35093 (md_reader::read_name_or_nil): New function.
35094 (md_reader::read_string): Handle "(nil)" by returning NULL.
35095 (md_reader::md_reader): Add new param "compact".
35096 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35097 (md_reader::read_file): New method.
35098 * read-md.h (md_reader::md_reader): Add new param "compact".
35099 (md_reader::read_file): New method.
35100 (md_reader::is_compact): New accessor.
35101 (md_reader::read_name): Convert return type from void to file_location.
35102 (md_reader::read_name_or_nil): New decl.
35103 (md_reader::read_name_1): New decl.
35104 (md_reader::m_compact): New field.
35105 (noop_reader::noop_reader): Pass "false" for new "compact" param
35107 (rtx_reader::rtx_reader): Add new "compact" param.
35108 (rtx_reader::read_rtx_operand): Make virtual and convert return
35109 type from void to rtx.
35110 (rtx_reader::read_until): New decl.
35111 (rtx_reader::handle_any_trailing_information): New virtual function.
35112 (rtx_reader::postprocess): New virtual function.
35113 (rtx_reader::finalize_string): New virtual function.
35114 (rtx_reader::m_in_call_function_usage): New field.
35115 (rtx_reader::m_reuse_rtx_by_id): New field.
35116 * read-rtl-function.c: New file.
35117 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35118 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35119 (selftest::verify_three_block_rtl_cfg): New decl.
35120 * read-rtl-function.h: New file.
35121 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35122 For host, include function.h, memmodel.h, and emit-rtl.h.
35123 (one_time_initialization): New function.
35124 (struct compact_insn_name): New struct.
35125 (compact_insn_names): New array.
35126 (find_code): Handle insn codes in compact dumps.
35127 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35128 (bind_subst_iter_and_attr): Likewise.
35129 (add_condition_to_string): Likewise.
35130 (add_condition_to_rtx): Likewise.
35131 (apply_attribute_uses): Likewise.
35132 (add_current_iterators): Likewise.
35133 (apply_iterators): Likewise.
35134 (initialize_iterators): Guard usage of apply_subst_iterator with
35135 #ifdef GENERATOR_FILE.
35136 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35137 (md_reader::read_mapping): Likewise.
35138 (add_define_attr_for_define_subst): Likewise.
35139 (add_define_subst_attr): Likewise.
35140 (read_subst_mapping): Likewise.
35141 (check_code_iterator): Likewise.
35142 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35144 (one_time_initialization): New function.
35145 (rtx_reader::read_until): New method.
35146 (read_flags): New function.
35147 (parse_reg_note_name): New function.
35148 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35149 Handle reuse_rtx ids.
35150 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35151 Add parsing support for RTL dumps, mirroring the special-cases in
35152 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35153 values, and calling handle_any_trailing_information.
35154 (rtx_reader::read_rtx_operand): Convert return type from void
35155 to rtx, returning return_rtx. Handle case 'e'. Call
35156 finalize_string on XSTR and XTMPL fields.
35157 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35158 "(nil)" values were omitted. Call the postprocess vfunc on the
35160 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35161 class ctor. Initialize m_in_call_function_usage. Call
35162 one_time_initialization.
35163 * rtl-tests.c (selftest::test_uncond_jump): Call
35164 set_new_first_and_last_insn.
35165 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35166 * selftest-rtl.c: New file.
35167 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35168 (selftest::get_insn_by_uid): New decl.
35169 * selftest-run-tests.c (selftest::run_tests): Call
35170 read_rtl_function_c_tests.
35171 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35172 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35175 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35177 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35178 operands in a special way. Assert that pos+len <= mode precision.
35180 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35182 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35183 3 argument Alias with unlimited for the negative form.
35184 (fno-vect-cost-model): Removed.
35186 2017-01-05 Martin Liska <mliska@suse.cz>
35188 * hsa-gen.c (gen_hsa_divmod): New function.
35189 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35191 2017-01-05 Martin Liska <mliska@suse.cz>
35194 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35197 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35199 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35200 small constant length operands.
35202 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35204 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35205 between loop iterations.
35207 2017-01-05 Martin Liska <mliska@suse.cz>
35210 * gimplify.c (gimplify_decl_expr): Compare to
35211 asan_poisoned_variables instread of checking flags.
35212 (gimplify_target_expr): Likewise.
35213 (gimplify_expr): Likewise.
35214 (gimplify_function_tree): Conditionally initialize
35215 asan_poisoned_variables.
35217 2017-01-04 Jeff Law <law@redhat.com>
35219 PR tree-optimizatin/78812
35220 * rtl.h (contains_mem_rtx_p): Prototype.
35221 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35222 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35223 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35224 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35225 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35227 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35229 * input.c (assert_char_at_range): Default-initialize actual_range.
35231 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35233 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35234 to match the caller.
35236 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35238 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35239 insns after final jump in test to emit dummy move.
35241 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35243 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35244 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35246 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35248 * multiple_target.c (create_dispatcher_calls): Init e_next.
35249 * tree-ssa-loop-split.c (split_loop): Init border.
35250 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35253 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35258 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35259 (altivec_register_operand): Do not return true if the operand
35260 contains a SUBREG mixing SImode and SFmode.
35261 (vsx_register_operand): Likewise.
35262 (vsx_reg_sfsubreg_ok): New predicate.
35263 (vfloat_operand): Do not return true if the operand contains a
35264 SUBREG mixing SImode and SFmode.
35265 (vint_operand): Likewise.
35266 (vlogical_operand): Likewise.
35267 (gpc_reg_operand): Likewise.
35268 (int_reg_operand): Likewise.
35269 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35270 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35271 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35273 (rs6000_emit_move_si_sf_subreg): New helper function.
35274 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35275 fixup SUBREGs involving SImode and SFmode.
35276 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35277 numbers for the new peephole2 optimization.
35278 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35279 the GLIBC math library that do AND/IOR/XOR operations on single
35280 precision floating point.
35281 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35282 target macros to say whether we need to avoid SUBREGs mixing
35284 (TARGET_ALLOW_SF_SUBREG): Likewise.
35285 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35286 (UNSPEC_SI_FROM_SF): Likewise.
35287 (iorxor): Change spacing.
35288 (and_ior_xor): New iterator for AND, IOR, and XOR.
35289 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35290 (movdi_from_sf_zero_ext): Likewise.
35291 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35292 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35293 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35294 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35295 (fms<mode>4): Likewise.
35296 (fnma<mode>4): Likewise.
35297 (fnms<mode>4): Likewise.
35298 (nfma<mode>4): Likewise.
35299 (nfms<mode>4): Likewise.
35301 2017-01-04 Marek Polacek <polacek@redhat.com>
35304 * doc/invoke.texi: Document -Wpointer-compare.
35306 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35308 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35311 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35312 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35315 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35318 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35319 documentation of the powerpc_popcntb_ok attribute.
35320 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35321 code to issue warning messages if a requested CPU configuration is
35322 not supported by the binary (assembler and loader) toolchain.
35323 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35324 made to define a built-in function that has been disabled.
35325 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35326 made to define a built-in function that has been disabled.
35327 (altivec_init_builtins): Add comment explaining why definition
35328 of the DST built-in functions is not preceded by an assertion
35329 check. Add assertions to prevent ICE if attempts are made to
35330 define an altivec predicate or an abs* built-in function that has
35332 (htm_init_builtins): Add comment explaining why definition of the
35333 htm built-in functions is not preceded by an assertion check.
35335 2017-01-04 Jeff Law <law@redhat.com>
35337 PR tree-optimizatin/67955
35338 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35339 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35340 the points-to solution does not include pt_null. Use DECL_PT_UID
35343 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35345 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35346 Use gen_int_mode instead of gen_lopwart for const_int operands.
35348 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35350 PR tree-optimization/71563
35351 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35352 out of range value - has low bits known to be zero.
35354 2017-01-04 Alan Modra <amodra@gmail.com>
35356 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35357 * configure: Regenerate.
35358 * config.in: Regenerate.
35360 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35363 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35364 a substring of the message, but strcmp with the whole message. Ifdef
35365 ENABLE_NLS, translate the message first using dgettext.
35367 2017-01-03 Jeff Law <law@redhat.com>
35369 PR tree-optimizatin/78856
35370 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35371 (mark_threaded_blocks): Remove code to truncate thread paths that
35372 cross multiple loop headers. Instead invalidate the cached loop
35373 iteration information and handle case of a thread path walking
35374 into an irreducible region.
35376 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35379 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35380 assertions. Add support for doing the signbit if the IEEE 128-bit
35381 floating point value is in a GPR.
35382 * config/rs6000/rs6000.md (Fsignbit): Delete.
35383 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35384 Update the length attribute if the value is in a GPR.
35385 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35386 the sign or zero extension instruction, since the value is always 0/1.
35387 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35390 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35391 extracting SImode to a GPR register so that we can generate a
35392 store, limit the vector to be in a traditional Altivec register
35393 for the vextuwrx instruction.
35395 2017-01-03 Ian Lance Taylor <iant@google.com>
35397 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35399 2017-01-03 Martin Sebor <msebor@redhat.com>
35401 PR tree-optimization/78696
35402 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35403 precision. Use MPFR for %f for greater fidelity. Correct handling
35405 (pass_sprintf_length::compute_format_length): Set width and precision
35406 specified by asrerisk to void_node for vararg functions.
35407 (try_substitute_return_value): Adjust dump output.
35409 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35411 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35413 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35415 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35416 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35417 -mlra/-mno-lra was passed to the compiler.
35419 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35421 PR rtl-optimization/65618
35422 * emit-rtl.c (try_split): Move initialization of "before" and
35423 "after" to just before the call to emit_insn_after_setloc.
35425 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35427 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35429 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35431 * dwarf2out.c (gen_enumeration_type_die): When
35432 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35434 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35436 PR tree-optimization/78965
35437 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35438 Change first argument from const call_info & to call_info &. For %n
35439 set info.nowrite to false.
35441 PR middle-end/78901
35442 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35443 possibly throwing calls.
35445 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35446 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35447 and fns handling, rather than in a separate case SSA_NAME.
35449 2017-01-02 Jeff Law <law@redhat.com>
35451 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35452 fixes for first_period and second_period variables.
35454 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35457 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35458 (*insvqi_1): New insn pattern.
35459 (*insvqi_1_mem_rex64): Ditto.
35460 (*insvqi_2): Ditto.
35461 (*insvqi_3): Rename from *insvqi.
35463 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35465 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35467 * doc/cfg.texi (Edges): Remove reference to Java.
35468 (Maintaining the CFG): Ditto.
35470 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35472 PR middle-end/77674
35473 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35474 transparent aliases.
35476 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35478 PR middle-end/77484
35479 * predict.def (PRED_CALL): Update hitrate.
35480 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35481 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35482 into direct/indirect/polymorphic variants.
35484 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35486 Update copyright years.
35488 * gcc.c (process_command): Update copyright notice dates.
35489 * gcov-dump.c (print_version): Ditto.
35490 * gcov.c (print_version): Ditto.
35491 * gcov-tool.c (print_version): Ditto.
35492 * gengtype.c (create_file): Ditto.
35493 * doc/cpp.texi: Bump @copying's copyright year.
35494 * doc/cppinternals.texi: Ditto.
35495 * doc/gcc.texi: Ditto.
35496 * doc/gccint.texi: Ditto.
35497 * doc/gcov.texi: Ditto.
35498 * doc/install.texi: Ditto.
35499 * doc/invoke.texi: Ditto.
35501 Copyright (C) 2017 Free Software Foundation, Inc.
35503 Copying and distribution of this file, with or without modification,
35504 are permitted in any medium without royalty provided the copyright
35505 notice and this notice are preserved.