1 2017-11-17 Jeff Law <law@redhat.com>
3 * vr-values.h (get_output_for_vrp): Prototype.
4 * vr-values.c (get_output_for_vrp): New function extracted from
5 vrp_visit_assignment_or_call and extract_range_from_stmt.
6 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
8 2017-11-17 Luis Machado <luis.machado@linaro.org>
10 * config/aarch64/aarch64.c
11 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
12 (qdf24xx_tunings) <autoprefetcher_model>: Set to
13 tune_params::AUTOPREFETCHER_WEAK.
15 2017-11-17 Tamar Christina <tamar.christina@arm.com>
18 * config/arm/arm.c (arm_valid_target_attribute_rec):
19 Parse "arch=" and "+<ext>".
20 (arm_valid_target_attribute_tree): Re-init global options.
21 (arm_option_override): Make non-static.
22 (arm_options_perform_arch_sanity_checks): Make errors fatal.
23 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
24 (__ARM_FEATURE_CRC32): Support undef.
25 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
26 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
28 2017-11-17 David Malcolm <dmalcolm@redhat.com>
30 * gdbinit.in (break-on-diagnostic): New command.
32 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
34 * config/i386/i386.c (ix86_expand_epilogue): Change simple
35 return to indirect jump for EH return if control-flow
36 protection is enabled. Change explicit 'false' argument in
37 pro_epilogue_adjust_stack with a value of
39 * config/i386/i386.md (simple_return_indirect_internal):
40 Remove SImode restriction to support 64-bit.
42 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
44 * combine.c (added_notes_insn): New.
45 (try_combine): Handle added_notes_insn like added_links_insn.
46 Rewrite return value code.
47 (distribute_notes): Set added_notes_insn to the earliest insn we added
50 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
52 PR rtl-optimization/82621
53 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
54 dest of one of those SETs is unused.
56 2017-11-17 Richard Biener <rguenther@suse.de>
59 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
60 * tree-pretty-print.c (dump_generic_node): Handle
61 annot_expr_parallel_kind.
62 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
63 * gimplify.c (gimple_boolify): Likewise.
65 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
67 * config.gcc (extra_headers): Add cet.h for x86 targets.
68 * config/i386/cet.h: New file.
69 * doc/install.texi: Add --enable-cet/--disable-cet.
71 2017-11-17 Richard Biener <rguenther@suse.de>
73 PR tree-optimization/83017
74 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
75 (gen_parallel_loop): Properly count iterations.
76 (parallelize_loops): Handle loop->can_be_parallel independent
77 of flag_loop_parallelize_all. Make static profitability test match
79 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
80 * invoke.texi (parloops-min-per-thread): Document.
82 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
84 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
85 upstreaming review comments.
87 2017-11-17 Tamar Christina <tamar.christina@arm.com>
89 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
90 with fast unaligned access.
91 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
93 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
95 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
96 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
98 2017-11-17 Richard Biener <rguenther@suse.de>
100 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
101 folding of references.
103 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
106 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
107 of replacing call to strncmp with corresponding call to strcmp when
110 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
112 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
113 option prefer-avx256 for skylake-avx512 configuration.
114 * config/i386/i386.c (ix86_option_override_internal): Ditto.
115 (get_builtin_code_for_version): Ditto.
117 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
118 Monk Chiang <sh.chiang04@gmail.com>
120 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
121 (FIXED_REGISTERS): Reserve more register numbers.
122 (CALL_USED_REGISTERS): Likewise.
123 (REG_ALLOC_ORDER): Likewise.
124 (REG_CLASS_CONTENTS): Likewise.
125 (REGISTER_NAMES): Likewise.
127 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
128 Kito Cheng <kito.cheng@gmail.com>
130 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
132 * config/nds32/iterators.md: Add vector mode iterators and attributes.
134 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
136 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
137 parameter list for vec_splats.
139 2017-11-16 Joseph Myers <joseph@codesourcery.com>
141 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
143 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
146 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
149 * doc/install.texi: Document incompatibility of Perl >=5.6.26
150 with the required version of automake 1.11.6.
152 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
154 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
156 (DU_C2_3_power9): Correct reservation combinations.
157 (FP_DIV_power9, VEC_DIV_power9): New.
158 (power9-alu): Split out rotate/shift...
159 (power9-rot): ...to here, correct dispatch resource.
160 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
162 (power9-fp): Correct latency.
163 (power9-sdiv): Add div/sqrt resource.
164 (power9-ddiv): Correct latency, add div/sqrt resource.
165 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
166 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
168 (power9-qpdiv, power9-qpmul): Adjust resource usage.
170 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
172 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
173 switch statement mapping KF built-ins to TF built-ins if we don't
174 have the proper ISA 3.0 assembler support.
176 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
178 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
179 (lower_emutls_function_body): Do not compute it.
181 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
183 * ipa-split.c (split_bb_info): Turn time to sreal.
184 (split_point): Likewise.
185 (dump_split_point): Likewise.
186 (fine_split_points): Likewise.
187 (execute_split_functions): Only zero split_bbs; turn time to sreals.
189 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
191 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
194 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
196 * predict.c (combine_predictions_for_bb): Preserve zero predicted
198 (expensive_function_p): Remove useless assert.
199 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
201 2017-11-16 Martin Sebor <msebor@redhat.com>
203 PR tree-optimization/82588
204 PR tree-optimization/82583
205 * tree-vrp.c (check_array_ref): Handle flexible array members,
206 string literals, and inner indices.
207 (search_for_addr_array): Add detail to diagnostics.
209 2017-11-16 Nathan Sidwell <nathan@acm.org>
213 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
214 (SET_DECL_ASSEMBLER_NAME): Forward to
215 overwrite_decl_assembler_name.
216 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
217 (overwrite_decl_assembler_name): Declare.
218 * tree.c (overwrite_decl_assembler_name): New.
219 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
220 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
221 (LANG_HOOKS_INITIALIZER): Add it.
222 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
223 * langhooks.c (lhd_set_decl_assembler_name): Use
224 SET_DECL_ASSEMBLER_NAME.
225 (lhd_overwrite_decl_assembler_name): Default implementation.
227 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
228 Jackson Woodruff <jackson.woodruff@arm.com>
230 PR tree-optimization/71026
231 * match.pd: Canonicalize constant multiplies in division.
233 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
235 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
236 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
237 Remove OPT_fomit_frame_pointer entry.
238 * common/config/arc/arc-common.c: Likewise.
239 * common/config/arm/arm-common.c: Likewise.
240 * common/config/avr/avr-common.c: Likewise.
241 * common/config/c6x/c6x-common.c: Likewise.
242 * common/config/cr16/cr16-common.c: Likewise.
243 * common/config/cris/cris-common.c: Likewise.
244 * common/config/epiphany/epiphany-common.c: Likewise.
245 * common/config/fr30/fr30-common.c: Likewise.
246 * common/config/frv/frv-common.c: Likewise.
247 * common/config/ia64/ia64-common.c: Likewise.
248 * common/config/iq2000/iq2000-common.c: Likewise.
249 * common/config/lm32/lm32-common.c: Likewise.
250 * common/config/m32r/m32r-common.c: Likewise.
251 * common/config/mcore/mcore-common.c: Likewise.
252 * common/config/microblaze/microblaze-common.c: Likewise.
253 * common/config/mips/mips-common.c: Likewise.
254 * common/config/mmix/mmix-common.c: Likewise.
255 * common/config/mn10300/mn10300-common.c: Likewise.
256 * common/config/nios2/nios2-common.c: Likewise.
257 * common/config/pa/pa-common.c: Likewise.
258 * common/config/pdp11/pdp11-common.c: Likewise.
259 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
260 * common/config/riscv/riscv-common.c: Likewise.
261 * common/config/rs6000/rs6000-common.c: Likewise.
262 * common/config/rx/rx-common.c: Likewise.
263 * common/config/s390/s390-common.c: Likewise.
264 * common/config/sh/sh-common.c: Likewise.
265 * common/config/sparc/sparc-common.c: Likewise.
266 * common/config/tilegx/tilegx-common.c: Likewise.
267 * common/config/tilepro/tilepro-common.c: Likewise.
268 * common/config/v850/v850-common.c: Likewise.
269 * common/config/visium/visium-common.c: Likewise.
270 * common/config/xstormy16/xstormy16-common.c: Likewise.
271 * common/config/xtensa/xtensa-common.c: Likewise.
272 * invoke.texi (-fomit-frame-pointer): Update documentation.
274 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
276 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
278 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
280 * cfg.c (scale_bbs_frequencies_int,
281 cale_bbs_frequencies_gcov_type): Remove.
282 * cfg.h (scale_bbs_frequencies_int,
283 cale_bbs_frequencies_gcov_type): Remove.
285 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
287 * tree-ssa-loop-manip.c
288 (scale_dominated_blocks_in_loop): Update to profile counts.
289 (tree_transform_and_unroll_loop): Likewise.
291 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
293 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
294 scale_bbs_frequencies_int.
296 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
298 * final.c (compute_alignments): Use counts rather than frequencies.
300 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
302 * cfgloopanal.c: Include sreal.h
303 (average_num_loop_insns): Use counts and sreal for accounting.
305 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
307 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
310 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
311 Kito Cheng <kito.cheng@gmail.com>
313 * config/nds32/constraints.md: Provide more constraints.
314 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
315 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
316 support constraints usage.
318 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
320 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
322 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
323 Kito Cheng <kito.cheng@gmail.com>
325 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
326 * config/nds32/nds32.opt: Refine the layout.
327 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
328 TARGET_EXT_STRING): Support new options.
329 * config/nds32/nds32.h: Likewise.
330 * config/nds32/nds32.md: Likewise.
331 * config/nds32/nds32-predicates.c: Likewise.
332 * config/nds32/constraints.md: Likewise.
333 * common/config/nds32/nds32-common.c: Likewise.
335 2017-11-16 Julia Koval <julia.koval@intel.com>
338 * config/i386/gfniintrin.h: Add sse check.
339 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
341 2017-11-16 Julia Koval <julia.koval@intel.com>
343 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
344 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
345 (ix86_handle_option): Handle -mavx512vbmi2.
346 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
347 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
348 * config/i386/i386-c.c (__AVX512VBMI2__): New.
349 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
350 (ix86_valid_target_attribute_inner_p): Ditto.
351 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
352 * config/i386/i386.opt (mavx512vbmi2): New option.
353 * doc/invoke.texi: Add new option.
355 [2017-11-16 Julia Koval <julia.koval@intel.com>
357 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
358 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
359 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
360 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
361 _mm512_gf2p8mul_epi8): New intrinsics.
362 * config/i386/i386-builtin-types.def
363 (V64QI_FTYPE_V64QI_V64QI): New type.
364 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
365 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
366 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
367 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
368 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
369 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
371 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
373 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
374 explicitly as a stream of bytes.
376 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
378 * config/rs6000/altivec.h (vec_xst_be): New #define.
379 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
380 and externalize from *altivec_vperm_<mode>_internal.
381 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
383 (XL_BE_V8HI): Likewise.
384 (XL_BE_V4SI): Likewise.
385 (XL_BE_V4SI): Likewise.
386 (XL_BE_V2DI): Likewise.
387 (XL_BE_V4SF): Likewise.
388 (XL_BE_V2DF): Likewise.
389 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
390 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
391 all array entries with these keys: VSX_BUILTIN_VEC_XL,
392 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
393 VSX_BUILTIN_VEC_XST_BE.
394 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
395 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
397 (altivec_init_builtins): Replace conditional calls to def_builtin
398 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
399 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
400 with unconditional calls. Remove calls to def_builtin for
401 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
402 __builtin_vec_xst_be.
403 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
404 to define_expand, and add alternate RTL generation for P8.
405 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
407 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
408 add alternate RTL generation for P8.
409 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
410 vsx_ld_elemrev_v16qi.
411 (vsx_st_elemrev_v8hi): Convert define_insn
412 to define_expand, and add alternate RTL generation for P8.
413 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
415 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
416 add alternate RTL generation for P8.
417 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
418 vsx_st_elemrev_v16qi.
420 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
423 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
424 TARGET_AVX512ER check.
425 (ix86_option_override_internal): Set MASK_VZEROUPPER if
426 neither -mzeroupper nor -mno-zeroupper is used and
427 TARGET_EMIT_VZEROUPPER is set.
428 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
429 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
431 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
433 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
434 folding of vector compares.
435 (fold_build_vec_cmp): New helper function.
436 (fold_compare_helper): New helper function.
437 (builtin_function_type): Add compare builtins to the list of functions
438 having unsigned arguments. Cosmetic updates to comment indentation.
439 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
440 the not+eq combination.
442 2017-11-15 Bin Cheng <bin.cheng@arm.com>
444 PR tree-optimization/82726
445 PR tree-optimization/70754
446 * tree-predcom.c (order_drefs_by_pos): New function.
447 (combine_chains): Move code setting has_max_use_after to...
448 (try_combine_chains): ...here. New parameter. Sort combined chains
449 according to position information.
450 (tree_predictive_commoning_loop): Update call to above function.
451 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
453 2017-11-15 Bin Cheng <bin.cheng@arm.com>
455 PR tree-optimization/82726
457 2017-01-23 Bin Cheng <bin.cheng@arm.com>
459 PR tree-optimization/70754
460 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
461 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
462 combined stmt before it if not NULL.
463 (combine_chains): Process refs reversely and compute dominance point
467 2017-02-23 Bin Cheng <bin.cheng@arm.com>
469 PR tree-optimization/79663
470 * tree-predcom.c (combine_chains): Process refs in reverse order
471 only for ZERO length chains, and add explaining comment.
473 2017-11-15 Tamar Christina <tamar.christina@arm.com>
475 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
476 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
477 * doc/invoke.texi (ARM Options): Add armv8.3-a.
479 2017-11-15 Tamar Christina <tamar.christina@arm.com>
481 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
483 2017-11-15 Martin Liska <mliska@suse.cz>
485 * tree-cfg.c (pass_warn_function_return::execute):
486 Compare warn_return_type for greater than zero.
488 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
492 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
493 to return true on Xeon and not on Xeon Phi.
494 (ix86_check_avx256_register): Changed to ...
495 (ix86_check_avx_upper_register): ... this. Add extra check for
496 VALID_AVX512F_REG_OR_XI_MODE.
497 (ix86_avx_u128_mode_needed): Changed
498 ix86_check_avx256_register to ix86_check_avx_upper_register.
499 (ix86_check_avx256_stores): Changed to ...
500 (ix86_check_avx_upper_stores): ... this. Changed
501 ix86_check_avx256_register to ix86_check_avx_upper_register.
502 (ix86_avx_u128_mode_after): Changed
503 avx_reg256_found to avx_upper_reg_found. Changed
504 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
505 (ix86_avx_u128_mode_entry): Changed
506 ix86_check_avx256_register to ix86_check_avx_upper_register.
507 (ix86_avx_u128_mode_exit): Ditto.
508 * config/i386/i386.h: (host_detect_local_cpu): New define.
510 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
512 * config/arm/xgene1.md (xgene1): Split into automatons
513 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
514 (xgene1_f_load): Adjust reservations and/or types.
515 (xgene1_f_store): Likewise.
516 (xgene1_load_pair): Likewise.
517 (xgene1_store_pair): Likewise.
518 (xgene1_fp_load1): Likewise.
519 (xgene1_load1): Likewise.
520 (xgene1_store1): Likewise.
521 (xgene1_move): Likewise.
522 (xgene1_alu): Likewise.
523 (xgene1_simd): Likewise.
524 (xgene1_bfm): Likewise.
525 (xgene1_neon_load1): Likewise.
526 (xgene1_neon_store1): Likewise.
527 (xgene1_neon_logic): Likewise.
528 (xgene1_neon_st1): Likewise.
529 (xgene1_neon_ld1r): Likewise.
530 (xgene1_alu_cond): Added.
531 (xgene1_shift_reg): Likwise.
532 (xgene1_bfx): Likewise.
533 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
535 2017-11-15 Jakub Jelinek <jakub@redhat.com>
538 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
540 (can_widen_mult_without_libcall): New function.
541 (expand_mul_overflow): If only checking unsigned mul overflow,
542 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
543 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
544 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
545 (expand_DIVMOD): Formatting fix.
546 * expmed.h (expand_mult): Add NO_LIBCALL argument.
547 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
548 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
550 PR tree-optimization/82977
551 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
552 constructed temporary to strlen_to_stridx.put.
554 2017-11-15 Martin Liska <mliska@suse.cz>
556 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
557 * configure: Regenerate.
559 2017-11-15 Martin Liska <mliska@suse.cz>
562 * config/sh/sh-mem.cc: Use proper probability for
565 2017-11-14 Jeff Law <law@redhat.com>
567 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
568 the red zone for stack_clash_protection_final_dynamic_probe targets
569 when the total dynamic stack size is zero bytes.
571 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
572 blocks is post order.
574 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
576 * dumpfile.h (TDF_COMPARE_DEBUG): New.
577 * final.c (rest_of_clean_state): Set it for the
578 -fcompare-debug dump.
579 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
580 class when TDF_COMPARE_DEBUG is set.
582 * dwarf2out.c (gen_producer_string): Discard
585 2017-11-15 Joseph Myers <joseph@codesourcery.com>
588 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
589 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
590 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
591 (__TGMATH_REAL_2_3): Remove macros.
592 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
593 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
595 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
597 (remquo): Define using __TGMATH_REAL_3.
599 2017-11-14 Jeff Law <law@redhat.com>
601 * vr-values.c: New file with contents extracted from tree-vrp.c.
602 * Makefile.in (OBJS): Add vr-values.o
603 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
604 (set_value_range, set_and_canonicalize_value_range): Likewise.
605 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
606 (value_range_constant_singleton, symbolic_range_p): Likewise.
607 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
608 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
609 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
610 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
611 (find_case_label_range, find_case_label_index): Likewise.
612 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
613 (range_int_cst_singleton_p, value_inside_range): Likewise.
614 (get_single_symbol): Likewise.
615 (switch_update): Move structure definition here.
616 (to_remove_edges, to_update_switch_stmts): Provide externs.
617 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
618 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
619 (vrp_val_is_min, set_value_range): Likewise.
620 (set_and_canonicalize_value_range, copy_value_range): Likewise.
621 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
622 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
623 (range_is_nonnull, range_int_cst_p): Likewwise.
624 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
625 (get_single_symbol, operand_less_p): Likewise
626 (compare_values_warnv, compare_values): Likewise.
627 (value_inside_range, value_range_constant_singleton): Likewise.
628 (zero_nonzero_bitgs_from_vr): Likewise.
629 (extract_range_from_binary_expr_1): Likewise.
630 (overflow_comparison_p): Likewise.
631 (to_remove_edges, to_update_switch_stmts): Likewise.
632 (find_case_label-index, find_case_label_range): Likewise.
633 (switch_update, set_value_range_to_nonnegative): Remove.
634 (set_value_range_to_truthvalue): Likewise.
635 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
636 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
637 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
638 (find_case_label_ranges, test_for_singularity): Likewise.
639 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
640 (x_vr_values): Move to its remaining use site.
642 2017-11-10 Jeff Law <law@redhat.com>
644 * vr-values.h (VR_INITIALIZER): Move #define here.
645 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
646 * Makefile.in (OBJS): Add tree-evrp.o
647 * tree-vrp.h (assert_info): Move structure definition here.
648 (set_value_range_to_varying): Prototype.
649 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
650 (infer_value_range, register_edge_assert_for): Likewise.
651 (stmt_interesting_for_vrp): Likewise.
652 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
653 (set_value_range_to_varying): No longer static.
654 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
655 (infer_value_range, register_edge_assert_for): Likewise.
657 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
659 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
660 generate the XXBRD instruction.
662 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
663 (rs6000_builtin_type_compatible): Treat _Float128 and long double
664 as being compatible if -mabi=ieeelongdouble.
665 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
666 to setup float128 built-ins with hardware support.
667 (BU_FLOAT128_HW_2): Likewise.
668 (BU_FLOAT128_HW_3): Likewise.
669 (BU_FLOAT128_HW_VSX_1): Likewise.
670 (BU_FLOAT128_HW_VSX_2): Likewise.
671 (scalar_extract_expq): Change float128 built-in functions to
672 accommodate having both KFmode and TFmode functions. Use the
673 KFmode variant as the default.
674 (scalar_extract_sigq): Likewise.
675 (scalar_test_neg_qp): Likewise.
676 (scalar_insert_exp_q): Likewise.
677 (scalar_insert_exp_qp): Likewise.
678 (scalar_test_data_class_qp): Likewise.
679 (sqrtf128_round_to_odd): Delete processing the round to odd
680 built-in functions as special built-in functions, and define them
681 as float128 built-ins. Use the KFmode variant as the default.
682 (truncf128_round_to_odd): Likewise.
683 (addf128_round_to_odd): Likewise.
684 (subf128_round_to_odd): Likewise.
685 (mulf128_round_to_odd): Likewise.
686 (divf128_round_to_odd): Likewise.
687 (fmaf128_round_to_odd): Likewise.
688 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
689 support for KFmode and TFmode xststdcqp calls.
690 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
691 point, switch the built-in handlers for the get/set float128
692 exponent, get float128 mantissa, float128 test built-ins, and the
693 float128 round to odd built-in functions. Eliminate creating the
694 float128 round to odd built-in functions as special built-ins.
695 (rs6000_init_builtins): Eliminate special creation of the float128
696 round to odd built-in functions.
697 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
698 function insns to support both TFmode and KFmode variants.
699 (xsxsigqp_<mode>): Likewise.
700 (xsiexpqpf_<mode>): Likewise.
701 (xsiexpqp_<mode>): Likewise.
702 (xststdcqp_<mode>): Likewise.
703 (xststdcnegqp_<mode>): Likewise.
704 (xststdcqp_<mode>): Likewise.
706 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
708 * tree-ssa-threadupdate.c (compute_path_counts): Remove
709 unused path_in_freq_ptr parameter.
710 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
712 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
714 * ipa-inline.c (edge_badness): Dump sreal frequency.
715 (compute_inlined_call_time): Match natural implementaiton ...
716 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
717 forgotten division by CGRAPH_FREQ_BASE.
719 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
721 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
722 Solaris 11. Update comment.
723 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
725 * config/sol2.h (STARTFILE_SPEC): Likewise.
726 * configure: Regenerate.
728 2017-11-14 Carl Love <cel@us.ibm.com>
730 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
731 le_ and be_ prefixes to swap* variables. Remove
732 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
735 2017-11-14 Jason Merrill <jason@redhat.com>
737 Support GTY((cache)) on hash_map.
738 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
739 (ggc_cache_remove): Override it instead of ggc_mx.
740 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
741 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
742 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
743 (simple_cache_map_traits): Override maybe_mx.
744 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
745 (hash_map): Friend gt_cleare_cache.
746 (gt_cleare_cache): New.
747 * tree.h (tree_cache_traits): New hash_map traits class.
748 (tree_cache_map): New typedef.
750 2017-11-14 Richard Biener <rguenther@suse.de>
752 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
753 paramter and handling.
754 (cleanup_control_flow_bb): Likewise.
755 (cleanup_control_flow_pre): New helper performing a DFS walk
756 to call cleanup_control_flow_bb in PRE order.
757 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
758 via cleanup_control_flow_pre.
760 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
762 * config/aarch64/aarch64-simd.md
763 (aarch64_simd_bsl<mode>_internal): Remove DImode.
764 (*aarch64_simd_bsl<mode>_alt): Likewise.
765 (aarch64_simd_bsldi_internal): New.
766 (aarch64_simd_bsldi_alt): Likewise.
768 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
770 * tracer.c (better_p): Do not compare frequencies.
771 * reg-stack.c (better_edge): Likewise.
772 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
775 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
777 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
778 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
779 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
780 * lto-streamer-in.c (input_function): Use update_max_bb_count.
781 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
782 * predict.c (maybe_hot_frequency_p): Inline to ...
783 (maybe_hot_count_p): ... here; rewrite to counts.
784 (counts_to_freqs): Rename to ...
785 (update_max_bb_count): ... this one.
786 (expensive_function_p): Use counts.
787 (estimate_bb_frequencies): Update.
788 (rebuild_frequencies): Update.
789 * predict.h (counts_to_freqs): Rename to ...
790 (update_max_bb_count): ... this one.
791 * profile.c (compute_branch_probabilities): Add debug info
792 * tree-inline.c (expand_call_inline): Update debug info.
793 (optimize_inline_calls): Use update_max_bb_count..
794 (tree_function_versioning): Use update_max_bb_count..
795 * value-prof.c (gimple_value_profile_transformations):
796 Do not use update_max_bb_count.
798 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
800 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
801 always use frequencies.
803 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
805 * bb-reorder.c: Remove frequencies from comments.
806 (better_edge_p): Use profile counts.
807 (find_traces): Dump profile counts.
808 (rotate_loop): Use profile counts.
809 (find_traces_1_round): Likewise.
810 (connect_better_edge_p): Use counts instead of probabilities for
812 (copy_bb_p): Drop early check for non-0 frequency.
813 (sanitize_hot_paths): Update comments.
815 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
817 * ipa-split.c (struct split_point): Add count.
818 (consider_split): Do not compute incoming frequency; compute incoming
819 count and store it to split_point.
820 (split_function): Set count of the call to split part correctly.
822 2017-11-13 Carl Love <cel@us.ibm.com>
824 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
826 2017-11-13 Tom Tromey <tom@tromey.com>
828 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
830 2017-11-13 Carl Love <cel@us.ibm.com>
832 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
833 Add support for builtins:
834 unsigned int vec_first_{,miss}_match_{,or_eos}index,
835 vector {un,}signed {char,int,short},
836 vector {un,}signed {char,int,short}) arguments.
837 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
838 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
839 Add BU_P9V_AV_2 expansions for the builtins.
840 * config/rs6000/altivec.h (vec_first_match_index,
841 vec_first_mismatch_index, vec_first_match_or_eos_index,
842 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
843 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
844 new extern declaration.
845 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
846 * config/rs6000/vsx.md (first_match_index_<mode>,
847 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
848 first_mismatch_or_eos_index_<mode>): Add define expand.
849 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
850 * doc/extend.texi: Update the built-in documenation file for the new
853 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
855 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
856 into the min/max operations for _Float<N> and _Float<N>X types.
858 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
861 * dwarf2out.c (do_eh_frame): New static variable.
862 (dwarf2out_begin_prologue): Set it.
863 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
865 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
867 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
869 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
871 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
874 * debug.h (dwarf2out_do_eh_frame): Declare.
875 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
876 (dwarf2out_do_frame): Use it.
877 (dwarf2out_do_cfi_asm): Likewise.
878 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
879 (dwarf2out_assembly_start): Likewise.
880 (dwarf2out_begin_prologue): Fix comment.
881 * toplev.c (compile_file): Always call dwarf2out_frame_finish
882 if the target needs either debug or unwind DWARF2 info.
883 * lto-opts.c (lto_write_options): Do not save -fexceptions,
884 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
885 -ftrapping-math, -ftrapv and -fwrapv.
887 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
889 * cgraph.c (cgraph_edge::sreal_frequency): New function.
890 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
891 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
892 (estimate_edge_size_and_time): Likewise.
893 (ipa_merge_fn_summary_after_inlining): Likewise.
894 * ipa-inline.c (cgraph_freq_base_rec): Remove.
895 (compute_uninlined_call_time): Use sreal_frequency.
896 (compute_inlined_call_time): Likewise.
897 (ipa_inline): Do not initialize cgraph_freq_base_rec.
898 * profile-count.c: Include sreal.h.
899 (profile_count::to_sreal_scale): New.
900 * profile-count.h: Forward declare sreal.
901 (profile_count::to_sreal_scale): Declare.
903 2017-11-13 Nathan Sidwell <nathan@acm.org>
905 * diagnostic.c (maybe_line_and_column): New.
906 (diagnostic_get_location_text): Use it.
907 (diagnostic_report_current_module): Likewise.
908 (test_diagnostic_get_location_text): Add tests.
910 2017-11-13 Luis Machado <luis.machado@linaro.org>
912 * doc/md.texi (Specifying processor pipeline description): Fix
913 incorrect latency for the div instruction example.
915 2017-11-13 Jakub Jelinek <jakub@redhat.com>
917 PR tree-optimization/78821
918 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
919 that bit_not_p is the same.
920 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
921 (split_group): Count precisely bit_not_p bits in each statement.
922 (invert_op): New function.
923 (imm_store_chain_info::output_merged_store): Use invert_op to
924 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
925 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
927 2017-11-13 Martin Liska <mliska@suse.cz>
929 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
930 (struct source_info): Likewise.
931 (add_branch_counts): Likewise.
932 (add_line_counts): Likewise.
933 (function_summary): Likewise.
934 (output_intermediate_line): Likewise.
935 (generate_results): Likewise.
937 2017-11-13 Martin Liska <mliska@suse.cz>
939 * gcov.c (struct block_info): Remove typedef for block_t.
940 (struct line_info): Likewise.
941 (line_info::has_block): Likewise.
942 (EXIT_BLOCK): Likewise.
945 (get_cycles_count): Likewise.
946 (process_file): Likewise.
947 (read_graph_file): Likewise.
948 (solve_flow_graph): Likewise.
949 (find_exception_blocks): Likewise.
950 (add_line_counts): Likewise.
951 (accumulate_line_info): Likewise.
952 (output_line_details): Likewise.
954 2017-11-13 Martin Liska <mliska@suse.cz>
956 * gcov.c (struct arc_info): Remove typedef for arc_t.
957 (struct line_info): Likewise.
958 (add_branch_counts): Likewise.
959 (output_branch_count): Likewise.
960 (function_info::~function_info): Likewise.
962 (output_intermediate_line): Likewise.
963 (read_graph_file): Likewise.
964 (solve_flow_graph): Likewise.
965 (find_exception_blocks): Likewise.
966 (add_line_counts): Likewise.
967 (accumulate_line_info): Likewise.
968 (output_line_details): Likewise.
969 (output_function_details): Likewise.
971 2017-11-13 Martin Liska <mliska@suse.cz>
973 * gcov.c (struct function_info): Remove typedef for function_t.
974 (struct source_info): Likewise.
975 (source_info::get_functions_at_location): Likewise.
976 (solve_flow_graph): Likewise.
977 (find_exception_blocks): Likewise.
978 (add_line_counts): Likewise.
979 (output_intermediate_file): Likewise.
980 (process_file): Likewise.
981 (generate_results): Likewise.
982 (release_structures): Likewise.
983 (read_graph_file): Likewise.
984 (read_count_file): Likewise.
985 (accumulate_line_counts): Likewise.
986 (output_lines): Likewise.
988 2017-11-13 Martin Liska <mliska@suse.cz>
990 * gcov.c (function_info::function_info): Remove num_counts
991 and add vector<gcov_type>.
992 (function_info::~function_info): Use the vector.
993 (process_file): Likewise.
994 (read_graph_file): Likewise.
995 (read_count_file): Likewise.
996 (solve_flow_graph): Likewise.
998 2017-11-13 Martin Liska <mliska@suse.cz>
1000 * gcov.c (function_info::is_artificial): New function.
1001 (process_file): Erase all artificial early.
1002 (generate_results): Skip as all artificial are already
1005 2017-11-13 Martin Liska <mliska@suse.cz>
1007 * gcov.c (read_graph_file): Store to global vector of functions.
1008 (read_count_file): Iterate the vector.
1009 (process_file): Likewise.
1010 (generate_results): Likewise.
1011 (release_structures): Likewise.
1013 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1015 PR tree-optimization/82954
1016 * gimple-ssa-store-merging.c
1017 (imm_store_chain_info::coalesce_immediate_stores): If
1018 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1020 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1022 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1023 Upddate call to ENDIAN_LANE_N.
1024 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1025 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1026 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1027 and use aarch64_endian_lane_rtx.
1028 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1030 2017-11-12 Tom de Vries <tom@codesourcery.com>
1032 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1034 2017-11-12 Tom de Vries <tom@codesourcery.com>
1036 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1038 2017-11-12 Tom de Vries <tom@codesourcery.com>
1040 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1042 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1043 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1044 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1046 2017-11-11 Martin Sebor <msebor@redhat.com>
1049 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1052 * prefic.c (translate_name): Replace strncpy with memcpy to
1053 avoid -Wstringop-truncation.
1055 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1057 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1059 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1061 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1063 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1065 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1068 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1070 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1071 enable generating XXBRH if the value is in a vector register.
1072 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1073 value is in a vector register.
1074 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1075 register to register bswap64's instead of doing the GPR sequence
1076 used on previous machines.
1077 (bswapdi2_xxbrd): New insn.
1078 (bswapdi2_reg): Disallow on ISA 3.0.
1079 (register to register bswap64 splitter): Do not split the insn on
1080 ISA 3.0 systems that use XXBRD.
1082 2017-11-10 Martin Sebor <msebor@redhat.com>
1085 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1087 (find_subframework_file): Same.
1089 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1091 * auto-profile.c (afdo_indirect_call): Drop frequency.
1092 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1093 (cgraph_node::create_edge): Drop frequency argument.
1094 (cgraph_node::create_indirect_edge): Drop frequency argument.
1095 (cgraph_edge::make_speculative): Drop frequency arguments.
1096 (cgraph_edge::resolve_speculation): Do not update frequencies
1097 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1098 (cgraph_node::dump): Check consistency in IPA mode.
1099 (cgraph_edge::maybe_hot_p): Use IPA counter.
1100 (cgraph_edge::verify_count_and_frequency): Rename to ...
1101 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1102 (cgraph_node::verify_node): Update.
1103 * cgraph.h (struct cgraph_edge): Drop frequency.
1104 (cgraph_edge::frequency): New function.
1105 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1107 (cgraph_edge::rebuild_edges): Likewise.
1108 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1109 (duplicate_thunk_for_node): Do not pass frequency.
1110 (cgraph_node::create_clone): Scale only counts.
1111 (cgraph_node::create_virtual_clone): Do not pass frequency.
1112 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1113 (cgraph_node::create_version_clone): Do not pass frequency.
1114 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1115 (cgraph_node::expand_thunk): Do not pass frequency.
1116 (cgraph_node::create_wrapper): Do not pass frequency.
1117 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1119 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1120 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1121 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1122 (gather_caller_stats): Use frequency function.
1123 (ipcp_cloning_candidate_p): Use frequency function.
1124 (ipcp_propagate_stage): Use frequency function.
1125 (get_info_about_necessary_edges): Use frequency function.
1126 (update_profiling_info): Update only IPA profile.
1127 (update_specialized_profile): Use frequency functoin.
1128 (perhaps_add_new_callers): Update only IPA profile.
1129 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1130 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1131 (dump_ipa_call_summary): Use frequency function.
1132 (estimate_edge_size_and_time): Use frequency function.
1133 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1134 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1135 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1136 (update_noncloned_counts): ... ths one; scale counts only.
1137 (clone_inlined_nodes): Do not scale frequency.
1138 (inline_call): Do not pass frequency.
1139 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1140 (compute_inlined_call_time): Use IPA profile.
1141 (want_inline_small_function_p): Use IPA profile.
1142 (want_inline_self_recursive_call_p): Use IPA profile.
1143 (edge_badness): Use IPA profile.
1144 (lookup_recursive_calls): Use IPA profile.
1145 (recursive_inlining): Do not pass frequency.
1146 (resolve_noninline_speculation): Do not update frequency.
1147 (inline_small_functions): Collect max of IPA profile.
1148 (dump_overall_stats): Dump IPA porfile.
1149 (dump_inline_stats): Dump IPA porfile.
1150 (ipa_inline): Collect IPA stats.
1151 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1152 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1153 (ipa_propagate_frequency): Use frequency function.
1154 (ipa_profile): Cleanup.
1155 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1156 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1157 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1158 (input_node): Do not stream frequency.
1159 (input_edge): Do not stream frequency.
1160 (merge_profile_summaries): Scale only IPA profiles.
1161 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1162 * predict.c (drop_profile): Do not recompute frequency.
1163 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1164 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1165 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1166 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1167 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1168 * tree-inline.c (copy_bb): Do not scale frequency.
1169 (expand_call_inline): Do not scale frequency.
1170 (tree_function_versioning): Do not scale frequency.
1171 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1173 2017-11-10 Julia Koval <julia.koval@intel.com>
1175 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1176 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1177 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1178 (_mm256_maskz_gf2p8affine_epi64_epi8)
1179 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1180 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1181 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1182 (__builtin_ia32_vgf2p8affineqb_v32qi)
1183 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1184 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1186 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1190 (arm_option_override): Refactor.
1191 (arm_option_reconfigure_globals): New.
1192 (arm_options_perform_arch_sanity_checks): New.
1193 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1195 (arm_options_perform_arch_sanity_checks): Likewise
1197 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1199 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1200 (power9-qpmul): New.
1201 * rs6000/rs6000.md ("type" attr): Add qmul.
1202 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1203 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1204 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1206 2017-11-10 Martin Sebor <msebor@redhat.com>
1209 * builtins.c (compute_objsize): Handle arrays that
1210 compute_builtin_object_size likes to fail for. Make extern.
1211 * builtins.h (compute_objsize): Declare.
1212 (check_strncpy_sizes): New function.
1213 (expand_builtin_strncpy): Call check_strncpy_sizes.
1214 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1215 -Wstringop-truncation.
1216 (gimple_fold_builtin_strncat): Same.
1217 * gimple.c (gimple_build_call_from_tree): Set call location.
1218 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1219 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1220 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1221 (handle_builtin_strlen): Use strlen_to_stridx.
1222 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1224 Use strlen_to_stridx.
1225 (pass_strlen::execute): Release strlen_to_stridx.
1226 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1227 (-Wstringop-truncation): Document new option.
1229 2017-11-10 Martin Liska <mliska@suse.cz>
1231 PR gcov-profile/82702
1232 * gcov.c (main): Handle intermediate files in a different
1234 (get_gcov_intermediate_filename): New function.
1235 (output_gcov_file): Remove support of intermediate files.
1236 (generate_results): Allocate intermediate file.
1237 (release_structures): Clean-up properly fn_end.
1238 (output_intermediate_file): Start iterating with line 1.
1240 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1242 PR tree-optimization/82929
1243 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1244 ops_swapped_p non-static data member.
1245 (store_immediate_info::store_immediate_info): Clear it.
1246 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1247 ops set ops_swapped_p.
1248 (count_multiple_uses): Handle ops_swapped_p.
1250 2017-11-10 Martin Liska <mliska@suse.cz>
1252 * coverage.c (coverage_init): Stream information about
1253 support of has_unexecuted_blocks.
1254 * doc/gcov.texi: Document that.
1255 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1256 * gcov.c (read_graph_file): Likewise.
1257 (output_line_beginning): Fix a small issue with
1260 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1262 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1263 reference of trivial component.
1265 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1268 * gimple-ssa-store-merging.c
1269 (pass_store_merging::terminate_all_aliasing_chains): For
1270 gimple_store_p stmts also call refs_output_dependent_p.
1272 PR rtl-optimization/82913
1273 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1276 2017-11-09 Jeff Law <law@redhat.com>
1278 * vr-values.h: New file with vr_values class.
1279 * tree-vrp.c: Include vr-values.h
1280 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1281 data objects into the vr_values class.
1282 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1283 (get_value_range): Make it a member function within vr_values class.
1284 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1285 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1286 (op_with_constant_singleton_value_range): Likewise.
1287 (extract_range_for_var_from_comparison_expr): Likewise.
1288 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1289 (extract_range_from_binary_expr): Likewise.
1290 (extract_range_from_unary_expr): Likewise.
1291 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1292 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1293 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1294 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1295 (compare_name_with_value, compare_names): Likewise.
1296 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1297 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1298 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1299 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1300 (extract_range_from_phi_node): Likewise.
1301 (simplify_truth_ops_using_ranges): Likewise.
1302 (simplify_div_or_mod_using_ranges): Likewise.
1303 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1304 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1305 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1306 (simplify_float_conversion_using_ranges): Likewise.
1307 (simplify_internal_call_using_ranges): Likewise.
1308 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1309 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1310 poitner via x_vr_values for calls into gimple folder.
1311 (vrp_initialize_lattice): Make this the vr_values ctor.
1312 (vrp_free_lattice): Make this the vr_values dtor.
1313 (set_vr_value): New function.
1314 (class vrp_prop): Add vr_values data member. Add various member
1315 functions as well as member functions that delegate to vr_values.
1316 (check_array_ref): Make a member function within vrp_prop class.
1317 (search_for_addr_array, vrp_initialize): Likewise.
1318 (vrp_finalize): Likewise. Revamp to avoid direct access to
1319 vr_value, values_propagated, etc.
1320 (check_array_bounds): Extract vrp_prop class instance pointer from
1321 walk info structure. Use it to call member functions.
1322 (check_all_array_refs): Make a member function within vrp_prop class.
1323 Smuggle class instance pointer via walk info structure.
1324 (x_vr_values): New local static.
1325 (vrp_valueize): Use x_vr_values to get class instance.
1326 (vr_valueize_1): Likewise.
1327 (class vrp_folder): Add vr_values data member. Add various member
1328 functions as well as member functions that delegate to vr_values.
1329 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1330 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1331 class instance from vr_values. Use it to call member functions.
1332 (vrp_dom_walker): Add vr_values data member.
1333 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1334 instance via x_vr_values.
1335 (identify_jump_threads): Accept vr_values as argument. Store
1336 it into the walker structure.
1337 (evrp_dom_walker): Add vr_values class data member. Add various
1339 (evrp_dom_walker::try_find_new_range): Use vr_values data
1340 member to access the memory allocator.
1341 (evrp_dom_walker::before_dom_children): Store vr_values class
1342 instance into the vrp_folder class.
1343 (evrp_dom_walker::push_value_range): Rework to avoid direct
1344 access to num_vr_values and vr_value.
1345 (evrp_dom_walker::pop_value_range): Likewise.
1346 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1347 Use vr_values to get to dump_all_value_ranges member function.
1348 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1349 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1350 Pass vr_values class instance down to identify_jump_threads.
1351 Remove call to vrp_free_lattice.
1352 (debug_all_value_ranges): Remove.
1354 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1355 (vrp_folder): Likewise.
1357 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1358 Get it from the existing bitmap instead.
1359 (vrp_intersect_ranges_1): Likewise.
1361 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1363 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1365 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1366 set bit_not_p to it.
1367 (imm_store_chain_info::coalesce_immediate_stores): Break group
1368 if bit_not_p is different.
1369 (count_multiple_uses, split_group,
1370 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1371 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1372 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1373 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1374 to store_immediate_info ctor.
1376 2017-11-09 Jim Wilson <jimw@sifive.com>
1378 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1379 (scan_prog_file): Likewise.
1381 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1383 * bb-reorder.c (max_entry_frequency): Remove.
1384 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1385 connect_traces, push_to_next_round_p): Remove prototypes.
1386 (find_traces_1_round): Use counts only.
1387 (push_to_next_round_p): Likewise.
1388 (find_traces): Likewise.
1389 (rotate_loop): Likewise.
1390 (find_traces_1_round): Likewise.
1391 (connect_traces): Likewise.
1392 (edge_order): Likewise.
1394 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1396 * config/arm/arm.c (output_return_instruction): Add comments to
1397 indicate requirement for cmse_nonsecure_entry return to account
1398 for the size of clearing instruction output here.
1399 (thumb_exit): Likewise.
1400 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1401 return in hardfloat mode.
1403 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1405 * config/rs6000/rs6000.c (machine_function): Add a bool,
1406 "toc_is_wrapped_separately".
1407 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1408 if it wasn't explicitly set or unset, we are optimizing for speed, and
1409 doing separate shrink-wrapping.
1410 (rs6000_get_separate_components): Enable the TOC component if
1411 saving the TOC register in the prologue.
1412 (rs6000_components_for_bb): Handle the TOC component.
1413 (rs6000_emit_prologue_components): Store the TOC register where needed.
1414 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1415 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1417 2017-11-09 Martin Jambor <mjambor@suse.cz>
1419 * ipa-param-manipulation.c: New file.
1420 * ipa-param-manipulation.h: Likewise.
1421 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1422 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1423 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1424 (ipa_parm_adjustment): Likewise.
1425 (ipa_parm_adjustment_vec): Likewise.
1426 (ipa_get_vector_of_formal_parms): Moved declaration to
1427 ipa-param-manipulation.h.
1428 (ipa_get_vector_of_formal_parm_types): Likewise.
1429 (ipa_modify_formal_parameters): Likewise.
1430 (ipa_modify_call_arguments): Likewise.
1431 (ipa_combine_adjustments): Likewise.
1432 (ipa_dump_param_adjustments): Likewise.
1433 (ipa_modify_expr): Likewise.
1434 (ipa_get_adjustment_candidate): Likewise.
1435 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1436 ipa-param-manipulation.c.
1437 (ipa_get_vector_of_formal_parm_types): Likewise.
1438 (ipa_modify_formal_parameters): Likewise.
1439 (ipa_modify_call_arguments): Likewise.
1440 (ipa_modify_expr): Likewise.
1441 (get_ssa_base_param): Likewise.
1442 (ipa_get_adjustment_candidate): Likewise.
1443 (index_in_adjustments_multiple_times_p): Likewise.
1444 (ipa_combine_adjustments): Likewise.
1445 (ipa_dump_param_adjustments): Likewise.
1446 * tree-sra.c: Also include ipa-param-manipulation.h
1447 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1450 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1451 Alan Hayward <alan.hayward@arm.com>
1452 David Sherwood <david.sherwood@arm.com>
1454 * doc/sourcebuild.texi (vect_masked_store): Document.
1456 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1457 Alan Hayward <alan.hayward@arm.com>
1458 David Sherwood <david.sherwood@arm.com>
1460 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1462 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1463 Alan Hayward <alan.hayward@arm.com>
1464 David Sherwood <david.sherwood@arm.com>
1466 * doc/sourcebuild.texi (vect_variable_length): Document.
1468 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1469 Alan Hayward <alan.hayward@arm.com>
1470 David Sherwood <david.sherwood@arm.com>
1472 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1474 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1475 Alan Hayward <alan.hayward@arm.com>
1476 David Sherwood <david.sherwood@arm.com>
1478 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1480 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1485 previously undocumented selectors.
1486 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1488 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1490 * doc/rtl.texi (const_vector): Say that elements can be
1491 const_wide_ints too.
1492 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1493 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1494 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1495 * optabs.c (expand_vector_broadcast): Likewise.
1497 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1498 Alan Hayward <alan.hayward@arm.com>
1499 David Sherwood <david.sherwood@arm.com>
1501 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1502 scaled index even if the unscaled address was invalid.
1503 Don't increase the complexity of using a scale in that case.
1505 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1506 Alan Hayward <alan.hayward@arm.com>
1507 David Sherwood <david.sherwood@arm.com>
1509 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1510 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1511 UNITS_PER_WORD bytes.
1512 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1513 based on REGMODE_NATURAL_SIZE of the inner mode.
1514 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1515 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1516 * expmed.c (lowpart_bit_field_p): Divide the value up into
1517 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1518 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1519 whether something is likely to occupy more than one register.
1521 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1524 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1525 profile_count::adjust_for_ipa_scaling.
1526 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1527 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1529 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1531 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1533 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1534 (split_group): Add total_orig and total_new arguments, estimate the
1535 number of statements related to the store group without store merging
1536 and with store merging.
1537 (imm_store_chain_info::output_merged_store): Adjust split_group
1538 callers, punt if estimated number of statements with store merging
1539 is not smaller than estimated number of statements without it.
1541 (handled_load): Remove has_single_use checks.
1542 (pass_store_merging::process_store): Likewise.
1544 2017-11-09 Richard Biener <rguenther@suse.de>
1546 PR tree-optimization/82902
1547 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1549 2017-11-09 Martin Liska <mliska@suse.cz>
1552 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1555 2017-11-09 Martin Liska <mliska@suse.cz>
1557 PR tree-optimization/82669
1558 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1560 2017-11-09 Martin Liska <mliska@suse.cz>
1562 PR gcov-profile/48463
1563 * coverage.c (coverage_begin_function): Output also end locus
1564 of a function and information whether the function is
1566 * gcov-dump.c (tag_function): Parse and print the information.
1567 * gcov.c (INCLUDE_MAP): Add include.
1568 (INCLUDE_SET): Likewise.
1569 (struct line_info): Move earlier in the source file because
1570 of vector<line_info> in function_info structure.
1571 (line_info::line_info): Likewise.
1572 (line_info::has_block): Likewise.
1573 (struct source_info): Add new member index.
1574 (source_info::get_functions_at_location): New function.
1575 (function_info::group_line_p): New function.
1576 (output_intermediate_line): New function.
1577 (output_intermediate_file): Use the mentioned function.
1578 (struct function_start): New.
1579 (struct function_start_pair_hash): Likewise.
1580 (process_file): Add code that identifies group functions.
1581 Assign lines either to global or function scope.
1582 (generate_results): Skip artificial functions.
1583 (find_source): Assign index for each source file.
1584 (read_graph_file): Read new flag artificial and end_line.
1585 (add_line_counts): Assign it either to global of function scope.
1586 (accumulate_line_counts): Isolate core of the function to
1587 accumulate_line_info and call it for both function and global
1589 (accumulate_line_info): New function.
1590 (output_line_beginning): Fix GNU coding style.
1591 (print_source_line): New function.
1592 (output_line_details): Likewise.
1593 (output_function_details): Likewise.
1594 (output_lines): Iterate both source (global) scope and function
1596 (struct function_line_start_cmp): New class.
1597 * doc/gcov.texi: Reflect changes in documentation.
1599 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1602 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1603 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1604 and similarly for not instead of neg.
1606 2017-11-08 Andi Kleen <ak@linux.intel.com>
1608 * config/i386/i386.opt: Add -mforce-indirect-call.
1609 * config/i386/predicates.md: Check for flag_force_indirect_call.
1610 * doc/invoke.texi: Document -mforce-indirect-call
1612 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1614 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1616 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1617 config/riscv/riscv.c (predict.h): New include.
1618 (riscv_slow_unaligned_access_p): No longer static.
1619 (riscv_block_move_straight): Add require.
1620 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1622 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1625 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1626 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1627 nonimmediate_operand predicate for operand 1 instead of
1630 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1632 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1634 * config/aarch64/constraints.md (Uml): New constraint.
1635 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1638 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1640 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1641 of two vec_duplicates into a vec_concat.
1643 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1645 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1646 Simplify vec_merge of vec_duplicate and vec_concat.
1647 * config/aarch64/constraints.md (Utq): New constraint.
1648 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1651 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1653 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1654 Simplify vec_merge of vec_duplicate and const_vector.
1655 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1657 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1658 mode iterator. Update predicate on operand 1 to
1659 handle non-const_vec constants. Delete constraints.
1660 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1662 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1664 PR tree-optimization/78821
1665 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1667 (store_operand_info::store_operand_info): Initialize it to false.
1668 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1669 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1670 store in the group, and if chain_info is non-NULL, to ignore altogether
1672 (compatible_load_p): Fail if bit_not_p does not match.
1673 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1674 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1675 (pass_store_merging::process_store): Adjust
1676 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1677 call terminate_all_aliasing_chains newly when adding a store into
1678 a chain with non-NULL chain_info.
1680 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1682 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1684 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1686 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1688 (aarch64_layout_frame): Initialise emit_frame_chain.
1689 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1690 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1692 2017-11-08 Martin Liska <mliska@suse.cz>
1694 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1695 of gimple_call_internal_p.
1697 2017-11-07 Tom de Vries <tom@codesourcery.com>
1699 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1701 2017-11-07 Tom de Vries <tom@codesourcery.com>
1703 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1706 2017-11-08 Martin Liska <mliska@suse.cz>
1709 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1711 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1713 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1715 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1718 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1719 (*cmp<mode>_ccz_1): New insn with $k alternative.
1722 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1723 RTX_COMM_COMPARE as commutative as well.
1724 (ix86_binary_operator_ok): Formatting fix.
1725 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1726 *<code><mode>3<mask_name><round_saeonly_name>,
1727 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1728 *<s>mul<mode>3_highpart<mask_name>,
1729 *vec_widen_umult_even_v16si<mask_name>,
1730 *vec_widen_umult_even_v8si<mask_name>,
1731 *vec_widen_umult_even_v4si<mask_name>,
1732 *vec_widen_smult_even_v16si<mask_name>,
1733 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1734 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1735 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1736 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1737 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1738 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1739 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1740 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1741 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1742 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1743 ix86_binary_operator_ok. Formatting fixes.
1744 (*<plusminus_insn><mode>3<mask_name><round_name>,
1745 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1748 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1750 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1751 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1752 eq and ne if TARGET_ISEL.
1753 (cmp): New code_iterator.
1754 (UNS, UNSU_, UNSIK): New code_attrs.
1755 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1756 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1758 ("eq<mode>3"): ... this.
1759 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1761 ("ne<mode>3"): ... this.
1763 2017-11-07 Julia Koval <julia.koval@intel.com>
1766 * common/config/i386/i386-common.c
1767 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1768 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1769 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1770 * config/i386/i386.opt: Ditto.
1771 * config/i386/i386.c (ix86_target_string): Ditto.
1772 (ix86_option_override_internal): Ditto.
1773 (ix86_init_mpx_builtins): Move MPX to args2.
1774 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1775 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1776 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1777 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1778 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1779 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1780 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1782 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1785 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1786 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1787 (zero-extendsidi peephole2): Remove peephole.
1789 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1792 * stor-layout.c: Include attribs.h.
1793 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1794 explicit lookup of "aligned" attribute.
1796 2017-11-07 Andrew Waterman <andrew@sifive.com>
1798 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1799 (riscv_expand_block_move): Likewise.
1800 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1802 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1803 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1804 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1805 (riscv_adjust_block_mem): Likewise.
1806 (riscv_block_move_loop): Likewise.
1807 (riscv_expand_block_move): Likewise.
1808 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1810 2017-11-07 Michael Clark <michaeljclark@mac.com>
1812 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1813 (MUSL_DYNAMIC_LINKER): Likewise.
1815 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1817 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1818 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1821 2017-11-07 Richard Biener <rguenther@suse.de>
1823 * match.pd: Fix build.
1825 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1826 Jackson Woodruff <jackson.woodruff@arm.com>
1828 PR tree-optimization/71026
1829 * match.pd: Canonicalize negate in division.
1831 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1834 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1836 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1838 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1839 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1841 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1843 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1844 non-scalar integral types.
1845 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1846 (-(A-B), -(~A)): New transformations.
1848 2017-11-07 Tom de Vries <tom@codesourcery.com>
1850 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1851 semicolon after "do {} while (0)".
1852 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1853 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1854 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1855 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1856 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1858 2017-11-07 Tom de Vries <tom@codesourcery.com>
1860 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1861 after "do {} while (0)".
1862 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1863 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1864 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1865 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1866 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1868 2017-11-07 Tom de Vries <tom@codesourcery.com>
1871 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1873 (arm_rtx_costs_internal): Add missing semicolon after
1874 HANDLE_NARROW_SHIFT_ARITH call.
1876 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1878 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1879 disable isel if it was not set explicitly.
1881 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1883 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1885 (add_type_attribute) likewise.
1887 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1889 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1890 of struct ix86_frame.
1891 (ix86_initial_elimination_offset): Likewise.
1892 (ix86_expand_split_stack_prologue): Likewise.
1894 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1896 * tree-vrp.h (enum value_range_type): Update stale comment.
1898 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1899 Alan Hayward <alan.hayward@arm.com>
1900 David Sherwood <david.sherwood@arm.com>
1902 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1903 (aarch64_expand_vec_perm_const): Take the number of units too.
1904 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1905 (aarch64_expand_vec_perm_const): Likewise.
1906 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1907 (vec_perm<mode>): Update accordingly.
1909 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1910 Alan Hayward <alan.hayward@arm.com>
1911 David Sherwood <david.sherwood@arm.com>
1913 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1914 Take the number of units too.
1915 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1916 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1917 but check for a vector mode before rather than after the call.
1918 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1919 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1920 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1921 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1922 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1923 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1924 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1925 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1926 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1927 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1928 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1929 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1930 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1931 (aarch64_sqdmull2_n<mode>): Update accordingly.
1933 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1934 Alan Hayward <alan.hayward@arm.com>
1935 David Sherwood <david.sherwood@arm.com>
1937 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1938 the number of units too.
1939 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1940 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1941 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1942 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1943 (vec_store_lanesxi<mode>): Update accordingly.
1945 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1946 Alan Hayward <alan.hayward@arm.com>
1947 David Sherwood <david.sherwood@arm.com>
1949 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1950 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1951 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1952 of units rather than the mode.
1953 * config/aarch64/iterators.md (nunits): New mode attribute.
1954 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1955 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1956 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1957 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1958 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1959 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1960 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1961 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1962 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1963 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1964 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1965 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1966 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1967 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1968 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1969 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1970 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1971 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1972 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1973 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1974 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1975 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1976 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1977 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1978 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1979 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1980 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1981 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1982 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1983 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1984 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1985 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1986 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1987 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1988 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1989 (aarch64_simd_vec_setv2di): Likewise.
1991 2017-11-06 Carl Love <cel@us.ibm.com>
1993 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1995 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1996 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1998 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2000 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2001 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2002 Add power 8 macro expansions.
2003 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2004 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2005 power 8 instructions. (VSX_XXBR): Add iterator.
2007 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2009 * config/arm/arm.md (predicable_short_it): Change default to "no",
2010 improve documentation, remove uses that are identical to the default.
2011 (enabled_for_depr_it): Rename to enabled_for_short_it.
2012 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2013 Remove default uses.
2014 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2015 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2016 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2017 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2019 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2022 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2023 float128 helper macros, which are no longer used after deleting
2024 the old 'q' built-in functions, and moving the round to odd
2025 built-in functions to being special built-in functions.
2026 (BU_FLOAT128_2): Likewise.
2027 (BU_FLOAT128_1_HW): Likewise.
2028 (BU_FLOAT128_2_HW): Likewise.
2029 (BU_FLOAT128_3_HW): Likewise.
2030 (FABSQ): Delete old 'q' built-in functions.
2031 (COPYSIGNQ): Likewise.
2032 (SQRTF128_ODD): Move round to odd built-in functions to be
2033 special built-in functions, so that we can handle
2034 -mabi=ieeelongdouble.
2035 (TRUNCF128_ODD): Likewise.
2036 (ADDF128_ODD): Likewise.
2037 (SUBF128_ODD): Likewise.
2038 (MULF128_ODD): Likewise.
2039 (DIVF128_ODD): Likewise.
2040 (FMAF128_ODD): Likewise.
2041 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2042 built-in names to 'f128'.
2043 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2044 old 'q' built-in functions, as the machine independent code for
2045 'f128' built-in functions handles this.
2046 (rs6000_expand_builtin): Add expansion for float128 round to odd
2047 functions, keying off on -mabi=ieeelongdouble of whether to use
2048 the KFmode or TFmode variant.
2049 (rs6000_init_builtins): Initialize the _Float128 round to odd
2051 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2052 _Float128 'q' built-in functions are now mapped into the new
2053 'f128' built-in functions.
2055 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2057 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2058 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2060 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2063 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2064 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2065 * toplev.c: Include "ipa-fnsummary.h".
2066 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2068 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2070 PR tree-optimization/82838
2071 * gimple-ssa-store-merging.c
2072 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2073 on a separate gimple_seq which is then appended to seq.
2075 2017-11-06 Jeff Law <law@redhat.com>
2078 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2079 (get_probe_interval): New functions.
2080 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2081 (ix86_adjust_stack_and_probe): Likewise.
2082 (output_adjust_stack_and_probe): Likewise.
2083 (ix86_emit_probe_stack_range): Likewise.
2084 (ix86_expand_prologue): Likewise.
2086 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2088 PR tree-optimization/82816
2089 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2090 if the modes of the two types are the same.
2091 (convert_plusminus_to_widen): Likewise.
2093 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2095 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2097 (usadv16qi): New define_expand.
2098 (usadv8hi): New define_expand.
2100 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2103 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2105 (inline_transform): Likewise.
2106 * predict.c (counts_to_freqs): Remove useless conditional.
2107 * profile-count.h (profile_count::apply_scale): Move sanity check.
2108 * tree-inline.c (copy_bb): Always scale.
2109 (copy_cfg_body): Likewise.
2111 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2114 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2116 (*cmp_ite0): Add enabled_for_depr_it attribute.
2117 (*cmp_ite1): Likewise.
2119 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2121 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2124 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2126 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2127 return the value by pointer.
2128 (extract_range_from_multiplicative_op_1): Update accordingly.
2129 Return as soon as an operation fails.
2131 2017-11-05 Tom de Vries <tom@codesourcery.com>
2134 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2135 (DEF_SANITIZER_BUILTIN): ... here.
2136 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2137 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2139 2017-11-05 Tom de Vries <tom@codesourcery.com>
2142 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2144 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2145 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2146 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2148 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2149 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2150 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2152 2017-11-05 Tom de Vries <tom@codesourcery.com>
2155 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2158 2017-11-04 Michael Clark <michaeljclark@mac.com>
2160 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2161 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2163 (*addsi3_extended): Likewise.
2164 (*addsi3_extended2): Likewise.
2165 (<optab>si3): Likewise.
2166 (<optab>di3): Likewise.
2167 (<optab><mode>3): Likewise.
2168 (<*optabe>si3_internal): Likewise.
2169 (zero_extendqi<SUPERQI:mode>2): Likewise.
2170 (*add<mode>hi3): Likewise.
2171 (*xor<mode>hi3): Likewise.
2172 (<optab>di3): Likewise.
2173 (*<optab>si3_extend): Likewise.
2174 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2175 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2176 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2178 2017-11-04 Andrew Waterman <andrew@sifive.com>
2180 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2181 TARGET_STRICT_ALIGN based upon -mtune argument.
2183 2017-11-04 Andrew Waterman <andrew@sifive.com>
2185 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2187 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2189 config/i386/i386.c (choose_basereg): Use optional scratch
2190 register and add assertion.
2191 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2192 needed, and don't allocate stack.
2193 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2194 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2195 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2197 2017-11-03 Jeff Law <law@redhat.com>
2199 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2200 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2201 to probe at the start of a noreturn function.
2203 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2205 PR tree-optimization/78821
2206 * gimple-ssa-store-merging.c: Update the file comment.
2207 (MAX_STORE_ALIAS_CHECKS): Define.
2208 (struct store_operand_info): New type.
2209 (store_operand_info::store_operand_info): New constructor.
2210 (struct store_immediate_info): Add rhs_code and ops data members.
2211 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2212 and op1r arguments to the ctor, initialize corresponding data members.
2213 (struct merged_store_group): Add load_align_base and load_align
2215 (merged_store_group::merged_store_group): Initialize them.
2216 (merged_store_group::do_merge): Update them.
2217 (merged_store_group::apply_stores): Pick the constant for
2218 encode_tree_to_bitpos from one of the two operands, or skip
2219 encode_tree_to_bitpos if neither operand is a constant.
2220 (class pass_store_merging): Add process_store method decl. Remove
2221 bool argument from terminate_all_aliasing_chains method decl.
2222 (pass_store_merging::terminate_all_aliasing_chains): Remove
2223 var_offset_p argument and corresponding handling.
2224 (stmts_may_clobber_ref_p): New function.
2225 (compatible_load_p): New function.
2226 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2227 if there is overlap and rhs_code is not INTEGER_CST. For
2228 non-overlapping stores terminate group if rhs is not mergeable.
2229 (get_alias_type_for_stmts): Change first argument from
2230 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2231 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2232 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2234 (get_location_for_stmts): Change first argument from
2235 auto_vec<gimple *> & to vec<gimple *> &.
2236 (struct split_store): Remove orig_stmts data member, add orig_stores.
2237 (split_store::split_store): Create orig_stores rather than orig_stmts.
2238 (find_constituent_stmts): Renamed to ...
2239 (find_constituent_stores): ... this. Change second argument from
2240 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2241 to info structures rather than the statements.
2242 (split_group): Rename ALLOW_UNALIGNED argument to
2243 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2244 it. Adjust find_constituent_stores caller.
2245 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2246 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2247 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2248 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2249 (mem_valid_for_store_merging): New function.
2250 (handled_load): New function.
2251 (pass_store_merging::process_store): New method.
2252 (pass_store_merging::execute): Use process_store method. Adjust
2253 terminate_all_aliasing_chains caller.
2255 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2257 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2258 Return true for more constants, symbols and label references.
2259 (aarch64_valid_floating_const): Remove unused function.
2261 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2264 * config/aarch64/aarch64.c (aarch64_layout_frame):
2265 Undo forcing of LR at bottom of frame.
2267 2017-11-03 Jeff Law <law@redhat.com>
2270 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2271 for int_registers_saved.
2273 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2274 extracted from tree-ssa-dom.c.
2275 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2276 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2277 (record_equivalences_from_incoming_edge): Add additional argument
2278 to single_pred_edge_ignoring_loop_edges call.
2279 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2280 (uncprop_dom_walker::before_dom_children): Add additional argument
2281 to single_pred_edge_ignoring_loop_edges call.
2282 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2283 single_pred_edge_ignoring_loop_edges rather than open coding.
2284 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2286 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2288 * match.pd (-(-A)): Rewrite.
2290 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2292 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2293 (rs6000_emit_int_cmove): New declaration.
2294 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2295 (rs6000_emit_sISEL): Delete.
2296 (rs6000_emit_int_cmove): Make non-static.
2297 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2298 instead of rs6000_emit_sISEL.
2300 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2302 * asan.c (create_cond_insert_point): Maintain profile.
2303 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2305 * basic-block.h (struct basic_block_def): Remove frequency.
2306 (EDGE_FREQUENCY): Use to_frequency
2307 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2309 (find_traces): Update to use to_frequency.
2310 (find_traces_1_round): Likewise; use only IPA counts.
2311 (bb_to_key): Likewise.
2312 (connect_traces): Use IPA counts only.
2313 (copy_bb_p): Update to use to_frequency.
2314 (fix_up_crossing_landing_pad): Likewise.
2315 (sanitize_hot_paths): Likewise.
2316 * bt-load.c (basic_block_freq): Likewise.
2317 * cfg.c (init_flow): Set count_max to uninitialized.
2318 (check_bb_profile): Remove frequencies; check counts.
2319 (dump_bb_info): Do not dump frequencies.
2320 (update_bb_profile_for_threading): Update counts only.
2321 (scale_bbs_frequencies_int): Likewise.
2322 (MAX_SAFE_MULTIPLIER): Remove.
2323 (scale_bbs_frequencies_gcov_type): Update counts only.
2324 (scale_bbs_frequencies_profile_count): Update counts only.
2325 (scale_bbs_frequencies): Update counts only.
2326 * cfg.h (struct control_flow_graph): Add count-max.
2327 (update_bb_profile_for_threading): Update prototype.
2328 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2329 (find_many_sub_basic_blocks): Likewise.
2330 * cfgcleanup.c (try_forward_edges): Likewise.
2331 (try_crossjump_to_edge): Likewise.
2332 * cfgexpand.c (expand_gimple_cond): Likewise.
2333 (expand_gimple_tailcall): Likewise.
2334 (construct_init_block): Likewise.
2335 (construct_exit_block): Likewise.
2336 * cfghooks.c (verify_flow_info): Check consistency of counts.
2337 (dump_bb_for_graph): Do not dump frequencies.
2338 (split_block_1): Do not update frequencies.
2339 (split_edge): Do not update frequencies.
2340 (make_forwarder_block): Do not update frequencies.
2341 (duplicate_block): Do not update frequencies.
2342 (account_profile_record): Do not update frequencies.
2343 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2344 for global heuristics.
2345 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2346 (expected_loop_iterations_unbounded): Use counts only.
2347 * cfgloopmanip.c (scale_loop_profile): Simplify.
2348 (create_empty_loop_on_edge): Simplify
2350 (duplicate_loop_to_header_edge): Simplify
2351 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2352 (update_br_prob_note): Take care of removing note when profile
2354 (relink_block_chain): Do not dump frequency.
2355 (rtl_account_profile_record): Use to_frequency.
2356 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2357 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2358 (cgraph_update_edges_for_call_stmt_node): Likewise.
2359 (cgraph_edge::verify_count_and_frequency): Update.
2360 (cgraph_node::verify_node): Temporarily disable frequency verification.
2361 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2362 to_cgraph_frequency.
2363 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2364 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2366 (cgraph_node::expand_thunk): Update profile.
2367 * except.c (dw2_build_landing_pads): Do not update frequency.
2368 * final.c (compute_alignments): Use to_frequency.
2369 (dump_basic_block_info): Do not dump frequency.
2370 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2371 (dump_gimple_bb_header): Do not dump frequency.
2372 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2374 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2375 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2376 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2377 (init_before_recovery): Do not update frequency.
2378 (sched_create_recovery_edges): Do not update frequency.
2379 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2380 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2381 (ipa_cp_c_finalize): Set max_count to uninitialized.
2382 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2383 (param_change_prob): Use counts.
2384 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2386 * ipa-split.c (consider_split): Use to_frequency.
2387 (split_function): Use to_frequency.
2388 * ira-build.c (loop_compare_func): Likewise.
2389 (mark_loops_for_removal): Likewise.
2390 (mark_all_loops_for_removal): Likewise.
2391 * loop-doloop.c (doloop_modify): Do not update frequency.
2392 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2394 * lto-streamer-in.c (input_function): Update count_max.
2395 * omp-expand.c (expand_omp_taskreg): Update count_max.
2396 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2397 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2398 (maybe_hot_count_p): Use ipa counts only.
2399 (maybe_hot_bb_p): Simplify.
2400 (maybe_hot_edge_p): Simplify.
2401 (probably_never_executed): Do not take frequency argument.
2402 (probably_never_executed_bb_p): Do not pass frequency.
2403 (probably_never_executed_edge_p): Likewise.
2404 (combine_predictions_for_bb): Check that profile is nonzero.
2405 (propagate_freq): Do not set frequency.
2406 (drop_profile): Simplify.
2407 (counts_to_freqs): Simplify.
2408 (expensive_function_p): Use to_frequency.
2409 (propagate_unlikely_bbs_forward): Simplify.
2410 (determine_unlikely_bbs): Simplify.
2411 (estimate_bb_frequencies): Add hack to silence graphite issues.
2412 (compute_function_frequency): Use ipa counts.
2413 (pass_profile::execute): Update.
2414 (rebuild_frequencies): Use counts only.
2415 (force_edge_cold): Use counts only.
2416 * profile-count.c (profile_count::dump): Dump new count types.
2417 (profile_count::differs_from_p): Check compatiblity.
2418 (profile_count::to_frequency): New function.
2419 (profile_count::to_cgraph_frequency): New function.
2420 * profile-count.h (struct function): Declare.
2421 (enum profile_quality): Add profile_guessed_local and
2422 profile_guessed_global0.
2423 (class profile_proability): Decrease number of bits to 29;
2424 update from_reg_br_prob_note and to_reg_br_prob_note.
2425 (class profile_count: Update comment; decrease number of bits
2426 to 61. Check compatibility.
2427 (profile_count::compatible_p): New private member function.
2428 (profile_count::ipa_p): New member function.
2429 (profile_count::operator<): Handle global zero correctly.
2430 (profile_count::operator>): Handle global zero correctly.
2431 (profile_count::operator<=): Handle global zero correctly.
2432 (profile_count::operator>=): Handle global zero correctly.
2433 (profile_count::nonzero_p): New member function.
2434 (profile_count::force_nonzero): New member function.
2435 (profile_count::max): New member function.
2436 (profile_count::apply_scale): Handle IPA scalling.
2437 (profile_count::guessed_local): New member function.
2438 (profile_count::global0): New member function.
2439 (profile_count::ipa): New member function.
2440 (profile_count::to_frequency): Declare.
2441 (profile_count::to_cgraph_frequency): Declare.
2442 * profile.c (OVERLAP_BASE): Delete.
2443 (compute_frequency_overlap): Delete.
2444 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2445 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2446 * sched-ebb.c (rank): Use counts only.
2447 * shrink-wrap.c (handle_simple_exit): Use counts only.
2448 (try_shrink_wrapping): Use counts only.
2449 (place_prologue_for_one_component): Use counts only.
2450 * tracer.c (find_best_predecessor): Use to_frequency.
2451 (find_trace): Use to_frequency.
2452 (tail_duplicate): Use to_frequency.
2453 * trans-mem.c (expand_transaction): Do not update frequency.
2454 * tree-call-cdce.c: Do not update frequency.
2455 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2456 (gimple_merge_blocks): Likewise.
2457 (gimple_split_edge): Likewise.
2458 (gimple_duplicate_sese_region): Likewise.
2459 (gimple_duplicate_sese_tail): Likewise.
2460 (move_sese_region_to_fn): Likewise.
2461 (gimple_account_profile_record): Likewise.
2462 (insert_cond_bb): Likewise.
2463 * tree-complex.c (expand_complex_div_wide): Likewise.
2464 * tree-eh.c (lower_resx): Update profile.
2465 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2467 (initialize_cfun): Do not initialize frequencies
2468 (freqs_to_counts): Delete.
2469 (copy_cfg_body): Ignore count parameter.
2470 (copy_body): Update.
2471 (expand_call_inline): Update count_max.
2472 (optimize_inline_calls): Update count_max.
2473 (tree_function_versioning): Update count_max.
2474 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2475 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2477 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2478 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2479 (try_peel_loop): Likewise.
2480 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2482 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2483 (tree_transform_and_unroll_loop): Do not use frequencies
2484 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2485 Use reliable prediction only.
2486 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2487 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2488 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2489 probability scaling.
2490 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2491 not update frequency
2492 (any_remaining_duplicated_blocks): Likewise.
2493 (update_profile): Likewise.
2494 (estimated_freqs_path): Delete.
2495 (freqs_to_counts_path): Delete.
2496 (clear_counts_path): Delete.
2497 (ssa_fix_duplicate_block_edges): Likewise.
2498 (duplicate_thread_path): Likewise.
2499 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2500 * tree-tailcall.c (decrease_profile): Do not update frequency.
2501 (eliminate_tail_call): Likewise.
2502 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2503 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2504 (optimize_mask_stores): Likewise.
2505 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2506 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2507 (ubsan_expand_ptr_ifn): Update profile.
2508 * value-prof.c (gimple_ic): Simplify.
2509 * value-prof.h (gimple_ic): Update prototype.
2510 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2511 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2513 (want_inline_self_recursive_call_p): Likewise.
2514 (resolve_noninline_speculation): Only cummulate defined counts.
2515 (inline_small_functions): Use nonzero_p.
2516 (ipa_inline): Do not access freed node.
2518 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2520 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2521 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2523 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2525 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2526 non-legitimate address.
2528 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2530 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2531 (*lt0_<mode>di, *lt0_<mode>si): New.
2533 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2535 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2536 TARGET_PAIRED_FLOAT.
2538 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2539 Jim Wilson <jim.wilson@linaro.org>
2541 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2542 * config/aarch64/aarch64-tune.md: Regenerated.
2543 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2544 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2546 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2548 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2549 (arc_expand_prologue): Restore blink for millicode.
2550 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2552 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2555 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2556 gen_vec_duplicate after forcing the scalar into a register.
2558 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2560 * combine (try_combine): Print the insns input to try_combine to the
2563 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2566 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2567 Remove second argument from aarch64_process_target_attr call.
2568 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2570 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2572 (aarch64_handle_attr_arch): Remove second argument.
2573 (aarch64_handle_attr_cpu): Ditto.
2574 (aarch64_handle_attr_tune): Ditto.
2575 (aarch64_handle_attr_isa_flags): Ditto.
2576 (aarch64_process_one_target_attr): Ditto.
2577 (aarch64_process_target_attr): Ditto.
2578 (aarch64_option_valid_attribute_p): Remove second argument.
2579 on aarch64_process_target_attr call.
2581 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2583 * diagnostic.c: Include "selftest-diagnostic.h".
2584 (selftest::assert_location_text): New function.
2585 (selftest::test_diagnostic_get_location_text): New function.
2586 (selftest::diagnostic_c_tests): Call it.
2588 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2590 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2591 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2592 (class selftest::test_diagnostic_context): Move to...
2593 * selftest-diagnostic.c: New file.
2594 * selftest-diagnostic.h: New file.
2596 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2598 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2599 offset range for FT32B.
2600 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2601 * config/ft32/ft32.md: Add TARGET_NOPM.
2602 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2603 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2605 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2607 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2609 2017-11-02 Jeff Law <law@redhat.com>
2611 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2612 virtual keyword on FINAL OVERRIDE members.
2614 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2615 virtuals together. Add virtual destructor.
2616 (substitute_and_fold_engine): Similarly.
2618 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2620 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2622 2017-11-02 Richard Biener <rguenther@suse.de>
2624 PR tree-optimization/82795
2625 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2627 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2629 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2631 (gcc_GAS_CHECK_FEATURE): Remove.
2632 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2633 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2634 * configure: Regenerate.
2636 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2638 * config/arc/arc.c (hwloop_optimize): Account for empty
2641 2017-11-02 Richard Biener <rguenther@suse.de>
2644 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2645 Truncate ARRAY_REF index and element size.
2647 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2649 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2651 2017-11-01 Jeff Law <law@redhat.com>
2653 * tree-ssa-ccp.c (ccp_folder): New class derived from
2654 substitute_and_fold_engine.
2655 (ccp_folder::get_value): New member function.
2656 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2657 (ccp_fold_stmt): Remove prototype.
2658 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2659 * tree-ssa-copy.c (copy_folder): New class derived from
2660 substitute_and_fold_engine.
2661 (copy_folder::get_value): Renamed from get_value.
2662 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2663 * tree-vrp.c (vrp_folder): New class derived from
2664 substitute_and_fold_engine.
2665 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2666 (vrp_folder::get_value): New member function.
2667 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2668 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2669 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2670 provide a class interface to folder/substitute routines.
2671 (ssa_prop_fold_stmt_fn): Remove typedef.
2672 (ssa_prop_get_value_fn): Likewise.
2673 (subsitute_and_fold): Remove prototype.
2674 (replace_uses_in): Likewise.
2675 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2676 Renamed from replace_uses_in. Call the virtual member function
2677 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2678 (substitute_and_fold_dom_walker): Remove initialization of
2679 data member entries for calbacks. Add substitute_and_fold_engine
2680 member and initialize it.
2681 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2682 member functions for get_value, replace_phi_args_in c
2683 replace_uses_in, and fold_stmt calls.
2684 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2685 substitute_and_fold. Remove assert. Update ctor call.
2687 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2688 (ssa_prop_visit_phi_fn): Likewise.
2689 (class ssa_propagation_engine): New class to provide an interface
2691 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2693 (ssa_prop_visit_phi): Likewise.
2694 (ssa_propagation_engine::simulate_stmt): Moved into class.
2695 Call visit_phi/visit_stmt from the class rather than via
2696 file scoped static variables.
2697 (ssa_propagation_engine::simulate_block): Moved into class.
2698 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2699 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2700 set file scoped statics for the visit_stmt/visit_phi callbacks.
2701 * tree-complex.c (complex_propagate): New class derived from
2702 ssa_propagation_engine.
2703 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2704 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2705 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2707 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2708 ssa_propagation_engine.
2709 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2710 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2711 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2712 * tree-ssa-copy.c (copy_prop): New class derived from
2713 ssa_propagation_engine.
2714 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2715 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2716 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2717 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2718 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2719 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2720 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2722 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2724 PR rtl-optimization/82778
2725 PR rtl-optimization/82597
2726 * compare-elim.c (struct comparison): Add in_a_setter field.
2727 (find_comparison_dom_walker::before_dom_children): Remove killed
2728 bitmap and df_simulate_find_defs call, instead walk the defs.
2729 Compute last_setter and initialize in_a_setter. Merge definitions
2730 with first initialization for a few variables.
2731 (try_validate_parallel): Use insn_invalid_p instead of
2732 recog_memoized. Return insn rather than just the pattern.
2733 (try_merge_compare): Fix up comment. Don't uselessly test if
2734 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2736 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2739 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2740 Alan Hayward <alan.hayward@arm.com>
2741 David Sherwood <david.sherwood@arm.com>
2743 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2744 aarch64_hard_regno_nregs to get the number of registers
2747 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2748 Alan Hayward <alan.hayward@arm.com>
2749 David Sherwood <david.sherwood@arm.com>
2751 * config/aarch64/constraints.md (Upl): Rename to...
2753 * config/aarch64/aarch64.md
2754 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2757 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2758 Alan Hayward <alan.hayward@arm.com>
2759 David Sherwood <david.sherwood@arm.com>
2761 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2762 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2765 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2766 Alan Hayward <alan.hayward@arm.com>
2767 David Sherwood <david.sherwood@arm.com>
2769 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2770 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2771 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2773 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2774 of a single element.
2775 * config/aarch64/iterators.md: Add a comment above the permute
2776 unspecs to say that they are generated directly by
2777 aarch64_expand_vec_perm_const.
2778 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2780 2017-11-01 Nathan Sidwell <nathan@acm.org>
2782 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2784 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2786 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2787 and medany code models, and describe what they do.
2789 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2791 Revert accidental duplicate:
2793 * combine.c (can_change_dest_mode): Reject changes in
2794 REGMODE_NATURAL_SIZE.
2796 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2798 PR rtl-optimization/64682
2799 PR rtl-optimization/69567
2800 PR rtl-optimization/69737
2801 PR rtl-optimization/82683
2802 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2803 register mentioned in the note, drop the note, unless it came from I3,
2804 in which case it should go to I3 again.
2806 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2808 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2809 and return false if not.
2810 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2812 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2814 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2815 range is known to be empty.
2817 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2818 Alan Hayward <alan.hayward@arm.com>
2819 David Sherwood <david.sherwood@arm.com>
2821 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2822 and CONST_VECTOR_NUNITS instead of computing the number of units from
2823 the byte sizes of the vector and element.
2824 (simplify_binary_operation_1): Likewise.
2825 (simplify_const_binary_operation): Likewise.
2826 (simplify_ternary_operation): Likewise.
2828 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2829 Alan Hayward <alan.hayward@arm.com>
2830 David Sherwood <david.sherwood@arm.com>
2832 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2833 (int_mem_offset): ...this new function.
2834 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2835 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2836 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2839 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2840 Alan Hayward <alan.hayward@arm.com>
2841 David Sherwood <david.sherwood@arm.com>
2843 * lower-subreg.c (interesting_mode_p): New function.
2844 (compute_costs, find_decomposable_subregs, decompose_register)
2845 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2846 (resolve_clobber, dump_choices): Use it.
2848 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2849 Alan Hayward <alan.hayward@arm.com>
2850 David Sherwood <david.sherwood@arm.com>
2852 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2854 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2855 Alan Hayward <alan.hayward@arm.com>
2856 David Sherwood <david.sherwood@arm.com>
2858 * alias.c (find_base_value, find_base_term): Only process integer
2859 truncations. Check the precision rather than the size.
2861 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2862 Alan Hayward <alan.hayward@arm.com>
2863 David Sherwood <david.sherwood@arm.com>
2865 * machmode.h (is_narrower_int_mode): New function
2866 * optabs.c (expand_float, expand_fix): Use it.
2867 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2869 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2870 Alan Hayward <alan.hayward@arm.com>
2871 David Sherwood <david.sherwood@arm.com>
2873 * rtl.h (narrower_subreg_mode): New function.
2874 * ira-color.c (update_costs_from_allocno): Use it.
2876 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2877 Alan Hayward <alan.hayward@arm.com>
2878 David Sherwood <david.sherwood@arm.com>
2880 * optabs-query.h (convert_optab_p): New function, split out from...
2881 (convert_optab_handler): ...here.
2882 (widening_optab_handler): Delete.
2883 (find_widening_optab_handler): Remove permit_non_widening parameter.
2884 (find_widening_optab_handler_and_mode): Likewise. Provide an
2885 override that operates on mode class wrappers.
2886 * optabs-query.c (widening_optab_handler): Delete.
2887 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2888 parameter. Assert that the two modes are the same class and that
2889 the "from" mode is narrower than the "to" mode. Use
2890 convert_optab_handler instead of widening_optab_handler.
2891 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2892 instead of widening_optab_handler.
2893 * expr.c (expand_expr_real_2): Update calls to
2894 find_widening_optab_handler.
2895 * optabs.c (expand_widen_pattern_expr): Likewise.
2896 (expand_binop_directly): Take the insn_code as a parameter.
2897 (expand_binop): Only call find_widening_optab_handler for
2898 conversion optabs; use optab_handler otherwise. Update calls
2899 to find_widening_optab_handler and expand_binop_directly.
2900 Use convert_optab_handler instead of widening_optab_handler.
2901 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2902 find_widening_optab_handler and use scalar_mode rather than
2904 (convert_plusminus_to_widen): Likewise.
2906 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2907 Alan Hayward <alan.hayward@arm.com>
2908 David Sherwood <david.sherwood@arm.com>
2910 * machmode.h (fixed_size_mode): New class.
2911 * rtl.h (get_pool_mode): Return fixed_size_mode.
2912 * gengtype.c (main): Add fixed_size_mode.
2913 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2914 (get_raw_arg_mode): Likewise.
2915 * doc/tm.texi: Regenerate.
2916 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2917 * targhooks.c (default_get_reg_raw_mode): Likewise.
2918 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2919 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2920 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2921 (msp430_get_raw_result_mode): Likewise.
2922 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2923 * dbxout.c (dbxout_parms): Require fixed-size modes.
2924 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2925 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2926 * omp-low.c (lower_oacc_reductions): Likewise.
2927 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2928 (simplify_subreg): Update accordingly.
2929 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2930 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2931 that aren't fixed-size.
2932 (get_pool_mode): Return a fixed_size_mode.
2933 (output_constant_pool_2): Take a fixed_size_mode.
2935 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2936 Alan Hayward <alan.hayward@arm.com>
2937 David Sherwood <david.sherwood@arm.com>
2939 * doc/rtl.texi (vec_series): Document.
2940 (const): Say that the operand can be a vec_series.
2941 * rtl.def (VEC_SERIES): New rtx code.
2942 * rtl.h (const_vec_series_p_1): Declare.
2943 (const_vec_series_p): New function.
2944 * emit-rtl.h (gen_const_vec_series): Declare.
2945 (gen_vec_series): Likewise.
2946 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2947 (gen_vec_series): Likewise.
2948 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2949 * simplify-rtx.c (simplify_unary_operation): Handle negations
2951 (simplify_binary_operation_series): New function.
2952 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2953 (test_vector_ops_series): New function.
2954 (test_vector_ops): Call it.
2955 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2956 gen_const_vec_series.
2957 (altivec_lvsr): Likewise.
2958 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2960 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2961 Alan Hayward <alan.hayward@arm.com>
2962 David Sherwood <david.sherwood@arm.com>
2964 * doc/rtl.texi (const): Update description of address constants.
2965 Say that vector constants are allowed too.
2966 * common.md (E, F): Use CONSTANT_P instead of checking for
2968 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2969 checking for CONST_VECTOR.
2970 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2972 * expr.c (expand_expr_real_2): Check for vector modes instead
2973 of checking for CONST_VECTOR.
2974 * rtl.h (const_vec_p): New function.
2975 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2976 (unwrap_const_vec_duplicate): Handle them here too.
2978 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2979 David Malcolm <dmalcolm@redhat.com>
2980 Alan Hayward <alan.hayward@arm.com>
2981 David Sherwood <david.sherwood@arm.com>
2983 * rtl.h (vec_duplicate_p): New function.
2984 * selftest-rtl.c (assert_rtx_eq_at): New function.
2985 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2986 (assert_rtx_eq_at): Declare.
2987 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2988 * selftest-run-tests.c (selftest::run_tests): Call it.
2989 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2990 (simplify_unary_operation_1): Recursively handle vector duplicates.
2991 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2993 (simplify_subreg): Handle subregs of vector duplicates.
2994 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2995 (selftest::simplify_rtx_c_tests): New functions.
2997 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2998 Alan Hayward <alan.hayward@arm.com>
2999 David Sherwood <david.sherwood@arm.com>
3001 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3002 (gen_vec_duplicate): Likewise.
3003 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3005 (gen_const_vector): ...here.
3006 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3007 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3008 whose elements are all equal.
3009 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3010 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3011 (simplify_relational_operation): Likewise.
3012 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3014 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3015 (aarch64_expand_vector_init): Likewise.
3016 * config/arm/arm.c (neon_vdup_constant): Likewise.
3017 (neon_expand_vector_init): Likewise.
3018 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3019 (arm_block_set_unaligned_vect): Likewise.
3020 (arm_block_set_aligned_vect): Likewise.
3021 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3022 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3023 (expand_vec_perm_even_odd_pack): Likewise.
3024 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3025 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3026 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3027 gen_const_vec_duplicate.
3028 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3029 * config/mips/mips.c (mips_gen_const_int_vector): Use
3030 gen_const_vec_duplicate.
3031 (mips_expand_vector_init): Use CONST0_RTX.
3032 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3033 (define_split): Use gen_const_vec_duplicate.
3034 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3035 (define_split): Use gen_const_vec_duplicate.
3036 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3037 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3038 * config/spu/spu.c (spu_const): Likewise.
3040 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3041 Alan Hayward <alan.hayward@arm.com>
3042 David Sherwood <david.sherwood@arm.com>
3044 * combine.c (can_change_dest_mode): Reject changes in
3045 REGMODE_NATURAL_SIZE.
3047 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3049 * configure.ac (--enable-libssp): New.
3050 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3051 trying to determine target-specific default. Adjust indentation.
3052 * configure: Regenerated.
3053 * doc/install.texi (Configuration): Expand --disable-libssp
3056 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3058 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3061 2017-10-31 Martin Jambor <mjambor@suse.cz>
3064 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3066 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3068 * auto-profile.c (autofdo_source_profile::read): Use
3069 UNKNOWN_LOCATION rather than 0.
3070 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3071 (warning_at): ...this overload.
3072 (warning_at_rich_loc_n): Rename to...
3073 (warning_n): ...this overload.
3074 (error_at_rich_loc): Rename to...
3075 (error_at): ...this overload.
3076 (pedwarn_at_rich_loc): Rename to...
3077 (pedwarn): ...this overload.
3078 (permerror_at_rich_loc): Rename to...
3079 (permerror): ...this overload.
3080 (inform_at_rich_loc): Rename to...
3081 (inform): ...this overload.
3082 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3083 (diagnostic_n_impl_richloc): Rename to...
3084 (diagnostic_n_impl): ...this rich_location *-based decl.
3085 (inform_at_rich_loc): Rename to...
3086 (inform): ...this, and add an assertion.
3087 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3088 (warning_at_rich_loc): Rename to...
3089 (warning_at): ...this, and add an assertion.
3090 (warning_at_rich_loc_n): Rename to...
3091 (warning_n): ...this, and add an assertion.
3092 (warning_n): Update location_t-based implementation for removal of
3093 location_t-based diagnostic_n_impl.
3094 (pedwarn_at_rich_loc): Rename to...
3095 (pedwarn): ...this, and add an assertion.
3096 (permerror_at_rich_loc): Rename to...
3097 (permerror): ...this, and add an assertion.
3098 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3099 (error_at_rich_loc): Rename to...
3100 (error_at): ...this, and add an assertion.
3101 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3102 (driver::do_spec_on_infiles): Likewise.
3103 * substring-locations.c (format_warning_va): Update for renaming
3104 of inform_at_rich_loc.
3106 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3108 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3109 _Float<N>X built-in functions so that the variant without the
3110 "__builtin_" prefix is only enabled for the GNU C and Objective C
3111 languages when they are in non-strict ANSI/ISO mode.
3112 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3113 * target.def (floatn_builtin_p): Add a target hook to control
3114 whether _Float<N> and _Float<N>X built-in functions without the
3115 "__builtin_" prefix are enabled, and return true for C and
3116 Objective C in the default hook. Include langhooks.h in
3118 * targhooks.h (default_floatn_builtin_p): Likewise.
3119 * targhooks.c (default_floatn_builtin_p): Likewise.
3120 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3121 floatn_builtin_p target hook.
3122 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3124 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3125 Eric Botcazou <ebotcazou@adacore.com>
3127 PR rtl-optimization/81803
3128 * lra-constraints.c (curr_insn_transform): Also reload the whole
3129 register for a strict subreg no wider than a word if this is for
3130 a WORD_REGISTER_OPERATIONS target.
3132 2017-10-31 Jason Merrill <jason@redhat.com>
3134 * gdbinit.in: Skip over inlines from timevar.h.
3136 2017-10-31 Martin Liska <mliska@suse.cz>
3138 * doc/gcov.texi: Document new option.
3139 * gcov.c (print_usage): Likewise print it.
3140 (process_args): Support the argument.
3141 (format_count): New function.
3142 (format_gcov): Use the function.
3144 2017-10-31 Martin Liska <mliska@suse.cz>
3146 * gcov.c (struct name_map): do not use typedef.
3147 Define operator== and operator<.
3148 (name_search): Remove.
3149 (name_sort): Remove.
3150 (main): Do not allocate names.
3151 (process_file): Add vertical space.
3152 (generate_results): Use std::find.
3153 (release_structures): Do not release memory.
3154 (find_source): Use std::find.
3156 2017-10-31 Martin Liska <mliska@suse.cz>
3158 * gcov.c (struct line_info): Remove it's typedef.
3159 (line_info::line_info): Add proper ctor.
3160 (line_info::has_block): Do not use a typedef.
3161 (struct source_info): Do not use typedef.
3162 (circuit): Likewise.
3163 (get_cycles_count): Likewise.
3164 (output_intermediate_file): Iterate via vector iterator.
3165 (add_line_counts): Use std::vector methods.
3166 (accumulate_line_counts): Likewise.
3167 (output_lines): Likewise.
3169 2017-10-31 Martin Liska <mliska@suse.cz>
3171 * gcov.c (struct source_info): Remove typedef.
3172 (source_info::source_info): Add proper ctor.
3173 (accumulate_line_counts): Use struct, not it's typedef.
3174 (output_gcov_file): Likewise.
3175 (output_lines): Likewise.
3176 (main): Do not allocate an array.
3177 (output_intermediate_file): Use size of vector container.
3178 (process_file): Resize the vector.
3179 (generate_results): Do not preallocate, use newly added vector
3181 (release_structures): Do not release sources.
3182 (find_source): Use vector methods.
3183 (add_line_counts): Do not use typedef.
3185 2017-10-31 Martin Liska <mliska@suse.cz>
3187 * doc/gcov.texi: Document that.
3188 * gcov.c (add_line_counts): Mark lines with a non-executed
3190 (output_line_beginning): Handle such lines.
3191 (output_lines): Pass new argument.
3192 (output_intermediate_file): Print it in intermediate format.
3194 2017-10-31 Martin Liska <mliska@suse.cz>
3196 * color-macros.h: New file.
3197 * diagnostic-color.c: Factor out color related to macros to
3199 * doc/gcov.texi: Document -k option.
3200 * gcov.c (INCLUDE_STRING): Include string.h.
3201 (print_usage): Add -k option.
3202 (process_args): Parse it.
3203 (pad_count_string): New function.
3204 (output_line_beginning): Likewise.
3205 (DEFAULT_LINE_START): New macro.
3206 (output_lines): Support color output.
3208 2017-10-31 Martin Liska <mliska@suse.cz>
3210 PR gcov-profile/82633
3211 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3212 their interaction with GCOV infrastructure.
3213 * configure.ac: Add -fkeep-{inline,static}-functions to
3215 * configure: Regenerate.
3217 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3220 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3222 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3225 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3226 into a register if it does not fit into an immediate offset field.
3228 2017-10-31 Olivier Hainque <hainque@adacore.com>
3230 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3232 2017-10-31 Julia Koval <julia.koval@intel.com>
3234 * config.gcc: Add gfniintrin.h.
3235 * config/i386/gfniintrin.h: New.
3236 * config/i386/i386-builtin-types.def
3237 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3238 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3239 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3240 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3241 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3242 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3243 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3244 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3245 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3246 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3247 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3248 * config/i386/immintrin.h: Include gfniintrin.h.
3249 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3251 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3253 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3255 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3257 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3258 (arm_ashldi3_1bit): Remove pattern.
3259 (ashrdi3): Remove shift by 1 expansion.
3260 (arm_ashrdi3_1bit): Remove pattern.
3261 (lshrdi3): Remove shift by 1 expansion.
3262 (arm_lshrdi3_1bit): Remove pattern.
3263 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3264 cost of ashldi3 by 1.
3265 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3266 (<shift>di3_neon): Likewise.
3268 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3270 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3271 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3272 and (*aarch64_simd_mov<VQ:mode>).
3273 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3274 pattern alternative.
3275 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3276 attributes to match pattern alternative.
3278 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3280 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3281 * config/rs6000/emmintrin.h: New file.
3282 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3284 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3286 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3287 * (movdi_vfp_cortexa8): Remove pattern.
3289 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3291 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3293 (Specific, alpha*-dec-osf5.1): Remove.
3294 (Specific, mips-sgi-irix5): Remove.
3295 (Specific, mips-sgi-irix6): Remove.
3297 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3300 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3301 arguments to clear_bit_region_be.
3303 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3305 * gimplify.c: Include memmodel.h.
3307 2017-10-30 Martin Jambor <mjambor@suse.cz>
3309 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3310 condition whether loop should be executed at all.
3312 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3315 gimple folding of vec_madd() intrinsics.
3316 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3317 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3318 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3320 2017-10-30 Richard Biener <rguenther@suse.de>
3322 PR tree-optimization/82762
3324 2017-10-23 Richard Biener <rguenther@suse.de>
3326 PR tree-optimization/82129
3328 2017-08-01 Richard Biener <rguenther@suse.de>
3330 PR tree-optimization/81181
3331 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3332 (compute_antic): ... end of iteration here.
3334 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3336 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3337 -std=iso9899:2017 and -std=gnu17.
3338 * doc/standards.texi (C Language): Document C17 support.
3339 * doc/cpp.texi (Overview): Mention -std=c17.
3340 (Standard Predefined Macros): Document C11 and C17 values of
3341 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3342 * doc/extend.texi (Inline): Do not list individual options for
3343 standards newer than C99.
3344 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3346 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3349 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3351 * asan.c (asan_finish_file): Align asan globals array by shadow
3354 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3357 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3358 (struct store_immediate_info): Add bitregion_start and bitregion_end
3360 (store_immediate_info::store_immediate_info): Add brs and bre
3361 arguments and initialize bitregion_{start,end} from those.
3362 (struct merged_store_group): Add bitregion_start, bitregion_end,
3363 align_base and mask fields. Drop unnecessary struct keyword from
3364 struct store_immediate_info. Add do_merge method.
3365 (clear_bit_region_be): Use memset instead of loop storing zeros.
3366 (merged_store_group::do_merge): New method.
3367 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3368 stores as long as the surrounding bitregions have no gaps.
3369 (merged_store_group::merge_overlapping): Use do_merge.
3370 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3371 is byte aligned, rather than requiring that start and width are
3372 byte aligned. Drop unnecessary struct keyword from
3373 struct store_immediate_info. Allocate and populate also mask array.
3374 Make start of the arrays relative to bitregion_start rather than
3375 start and size them according to bitregion_{end,start} difference.
3376 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3377 struct store_immediate_info.
3378 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3379 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3380 struct keyword from struct store_immediate_info.
3381 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3382 between stores as long as the surrounding bitregions have no gaps.
3384 (struct split_store): Add orig non-static data member.
3385 (split_store::split_store): Initialize orig to false.
3386 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3387 if there is exactly a single original stmt. Change stmts argument
3388 to pointer from reference, if NULL, don't push anything to it. Add
3389 first argument, use it to optimize skipping over orig stmts that
3390 are known to be before bitpos already. Simplify.
3391 (split_group): Return unsigned int count how many stores are or
3392 would be needed rather than a bool. Add allow_unaligned argument.
3393 Change split_stores argument from reference to pointer, if NULL,
3394 only do a dry run computing how many stores would be produced.
3395 Rewritten algorithm to use both alignment and misalign if
3396 !allow_unaligned and handle bitfield stores with gaps.
3397 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3398 from bitregion_start instead of start. Compute allow_unaligned
3399 here, if true, do 2 split_group dry runs to compute which one
3400 produces fewer stores and prefer aligned if equal. Punt if
3401 new count is bigger or equal than original before emitting any
3402 statements, rather than during that. Remove no longer needed
3403 new_ssa_names tracking. Replace num_stmts with
3404 split_stores.length (). Use 32-bit stack allocated entries
3405 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3406 if possible. Handle bitfields with gaps.
3407 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3408 Compute bitregion_{start,end} for the stores and construct
3409 store_immediate_info with that. Formatting fixes.
3411 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3414 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3415 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3417 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3419 * gimplify.c: Include tm_p.h.
3421 * common.opt (gcoff): Re-add as ignored option.
3422 (gcoff1, gcoff2, gcoff3): Likewise.
3424 * Makefile.in (OBJS): Delete sdbout.o.
3425 (GTFILES): Delete $(srcdir)/sdbout.c.
3426 * debug.h: Delete sdb_debug_hooks.
3427 * final.c: Delete sdbout.h include.
3428 (final_scan_insn): Delete SDB_DEBUG check.
3429 (rest_of_clean_state): Likewise.
3430 * output.h: Delete sdb_begin_function_line.
3433 * toplev.c: Delete sdbout.h include.
3434 (process_options): Delete SDB_DEBUG check.
3435 * tree-core.h (tree_type_common): Delete pointer field of
3437 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3438 TYPE_SYMTAB_POINTER.
3439 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3440 (TYPE_SYMTAB_IS_POINTER): Delete.
3441 (TYPE_SYMTAB_IS_DIE): Renumber.
3442 * xcoffout.c: Refer to former sdbout.c file.
3443 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3445 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3446 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3447 (Debugging Options): Delete -gcoff.
3448 (-gstabs): Delete SDB reference.
3450 (-gcoff@var{level}): Delete.
3451 * doc/passes.texi (Debugging information output): Delete SDB and
3452 sdbout.c references.
3453 * doc/tm.texi: Regenerate.
3454 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3455 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3457 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3458 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3459 (SDB_DEBUGGING_INFO): Delete.
3460 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3461 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3462 * target.def (output_source_filename): Delete COFF reference.
3464 * common.opt (gcoff): Delete.
3465 (gxcoff+): Update Negative chain.
3466 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3468 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3469 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3470 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3472 (expand_function_start): Change sdb reference to past tense.
3473 (expand_function_end): Change sdb reference to past tense.
3474 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3475 * opts.c (debug_type_names): Delete coff entry.
3476 (common_handle_option): Delete OPT_gcoff case.
3477 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3479 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3480 * config/cris/cris.h: Delete SDB reference in comment.
3481 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3482 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3483 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3484 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3486 (ix86_expand_prologue): Likewise.
3487 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3488 * config/ia64/ia64.h: Likewise.
3489 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3490 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3492 * config/mmix/mmix.h: Likewise.
3493 * config/nds32/nds32.c: Likewise.
3494 * config/stormy/storym16.h: Likewise.
3495 * config/visium/visium.h: Likewise.
3496 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3498 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3500 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3501 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3502 FRAME_POINTER_REGNUM point at high end of local var area.
3504 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3506 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3507 Move comment around. Do not reset best_edge for a copiable
3508 destination if the copy would cause a partition change.
3509 (better_edge_p): Remove redundant check.
3511 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3513 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3515 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3517 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3518 for math functions that have _Float<N> and _Float<N>X variants.
3519 (mathfn_built_in_2): Add support for math functions that have
3520 _Float<N> and _Float<N>X variants.
3521 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3522 (expand_builtin_mathfn_ternary): Add support for fma with
3523 _Float<N> and _Float<N>X variants.
3524 (expand_builtin): Likewise.
3525 (fold_builtin_3): Likewise.
3526 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3527 create math function _Float<N> and _Float<N>X variants as external
3529 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3530 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3531 the __builtin_ prefix and if not strict ansi, without the prefix.
3532 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3533 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3534 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3535 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3536 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3537 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3538 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3539 function signatures for fma _Float<N> and _Float<N>X variants.
3540 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3541 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3542 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3543 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3544 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3545 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3546 * gencfn-macros.c (print_case_cfn): Add support for math functions
3547 that have _Float<N> and _Float<N>X variants.
3548 (print_define_operator_list): Likewise.
3549 (fltfn_suffixes): Likewise.
3551 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3552 for math functions that have _Float<N> and _Float<N>X variants.
3553 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3554 and _Float<N>X variants.
3555 (COPYSIGN): Likewise.
3558 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3559 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3561 (integer_valued_read_call_p): Likewise.
3562 * fold-const-call.c (fold_const_call_ss): Likewise.
3563 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3564 _Float<N> and _Float<N>X variants.
3565 (fold_const_call_ssss): Add support for fma _Float<N> and
3566 _Float<N>X variants.
3567 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3568 support for copysign and fma _Float<N> and _Float<N>X variants.
3569 (backprop::process_builtin_call_use): Likewise.
3570 * tree-call-cdce.c (can_test_argument_range); Add support for
3571 sqrt _Float<N> and _Float<N>X variants.
3572 (edom_only_function): Likewise.
3573 (get_no_error_domain): Likewise.
3574 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3575 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3576 copysign _Float<N> and _Float<N>X variants.
3577 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3578 handled by machine independent code.
3579 (FMAF128): Likewise.
3580 * doc/cpp.texi (Common Predefined Macros): Document defining
3581 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3582 fma _Float<N> and _Float<N>X variants.
3584 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3587 * config/i386/i386-modes.def (CCFPU): Remove definition.
3588 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3589 (ix86_cc_modes_compatible): Ditto.
3590 (ix86_expand_carry_flag_compare): Ditto.
3591 (ix86_expand_int_movcc): Ditto.
3592 (ix86_expand_int_addcc): Ditto.
3593 (ix86_reverse_condition): Ditto.
3594 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3595 Return true/false for unordered/ordered fp comparisons.
3596 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3597 (ix86_prepare_fp_compare_args): Update for rename.
3598 (ix86_expand_fp_compare): Update for rename. Generate unordered
3599 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3600 (ix86_expand_sse_compare_and_jump): Ditto.
3601 * config/i386/predicates.md (fcmov_comparison_operator):
3602 Remove CCFPU mode handling.
3603 (ix86_comparison_operator): Ditto.
3604 (ix86_carry_flag_operator): Ditto.
3605 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3606 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3607 (*cmpu<mode>_cc_i387): Ditto.
3608 (FPCMP): Remove mode iterator.
3609 (unord): Remove mode attribute.
3610 (unord_subst): New define_subst transformation
3611 (unord): New define_subst attribute.
3613 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3614 (*cmpi<unord>xf_i387): Ditto.
3615 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3616 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3617 using unord_subst transformation.
3618 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3619 (round_saeonly): Also handle CCFP mode.
3620 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3621 Remove UNSPEC_SAHF unspec handling.
3623 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3625 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3627 2017-10-27 Jeff Law <law@redhat.com>
3629 * gimple-ssa-sprintf.c: Include domwalk.h.
3630 (class sprintf_dom_walker): New class, derived from dom_walker.
3631 (sprintf_dom_walker::before_dom_children): New function.
3632 (struct call_info): Moved into sprintf_dom_walker class
3633 (compute_formath_length, handle_gimple_call): Likewise.
3634 (sprintf_length::execute): Call the dominator walker rather
3635 than walking the statements.
3637 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3638 gimple statement locations.
3639 (check_array_bounds): Corresponding changes. Get the statement's
3640 location directly from wi->stmt.
3642 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3645 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3647 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3649 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3650 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3652 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3655 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3656 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3657 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3658 maybe_get_pool_constant.
3659 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3662 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3664 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3665 2.26 caveat. Update gas and gld versions.
3666 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3669 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3671 * cgraph.h (set_malloc_flag): Declare.
3672 * cgraph.c (set_malloc_flag_1): New function.
3673 (set_malloc_flag): Likewise.
3674 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3675 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3677 (read_ipa_call_summary): Add support for reading is_return_callee.
3678 (write_ipa_call_summary): Stream is_return_callee.
3679 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3680 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3681 ipa-prop.h, ipa-fnsummary.h.
3682 (pure_const_names): Change to static.
3683 (malloc_state_e): Define.
3684 (malloc_state_names): Define.
3685 (funct_state_d): Add field malloc_state.
3686 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3687 (check_retval_uses): New function.
3688 (malloc_candidate_p): Likewise.
3689 (analyze_function): Add support for malloc attribute.
3690 (pure_const_write_summary): Stream malloc_state.
3691 (pure_const_read_summary): Add support for reading malloc_state.
3692 (dump_malloc_lattice): New function.
3693 (propagate_malloc): New function.
3694 (warn_function_malloc): New function.
3695 (ipa_pure_const::execute): Call propagate_malloc and
3696 ipa_free_fn_summary.
3697 (pass_local_pure_const::execute): Add support for malloc attribute.
3698 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3699 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3701 2017-10-27 Martin Liska <mliska@suse.cz>
3703 PR gcov-profile/82457
3704 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3705 for fork-like functions to be properly instrumented.
3707 2017-10-27 Richard Biener <rguenther@suse.de>
3710 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3711 info when we redirected EH.
3713 2017-10-26 Michael Collison <michael.collison@arm.com>
3715 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3717 (<optab>_trunchf<GPI:mode>2: New pattern.
3718 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3719 * config/aarch64/iterators.md (wv): New mode attribute.
3720 (vf, VF): New mode attributes.
3721 (vgp, VGP): New mode attributes.
3722 (s): Update attribute with SImode and DImode prefixes.
3724 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3726 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3727 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3728 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3729 (nios2_option_overide): Initialize it. Don't allow R0-relative
3730 addressing with PIC.
3731 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3732 (nios2_symbolic_constant_p): Likewise.
3733 (nios2_legitimate_address_p): Likewise.
3734 (nios2_r0rel_section_name_p): New.
3735 (nios2_symbol_ref_in_r0rel_data_p): New.
3736 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3737 (r0rel_constant_p): New.
3738 (nios2_print_operand_address): Handle r0rel_constant_p.
3739 (nios2_cdx_narrow_form_p): Likewise.
3740 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3741 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3742 (Nios II Options): Document -mr0rel-sec.
3744 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3746 * config/nios2/nios2.c: Include xregex.h.
3747 (nios2_gprel_sec_regex): New.
3748 (nios2_option_overide): Initialize it. Don't allow GP-relative
3749 addressing with PIC.
3750 (nios2_small_section_name_p): Check for regex match.
3751 * config/nios2/nios2.opt (mgprel-sec=): New option.
3752 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3753 (Nios II Options): Document -mgprel-sec.
3755 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3757 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3759 2017-10-26 Tom de Vries <tom@codesourcery.com>
3761 PR tree-optimization/82707
3762 * gimple.c (gimple_copy): Fix unsharing of
3763 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3765 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3767 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3768 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3769 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3770 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3771 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3772 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3773 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3774 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3775 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3776 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3777 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3778 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3779 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3780 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3781 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3782 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3783 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3785 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3787 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3789 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3790 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3791 warning to rs6000.c. Remove the Undocumented flag, since it has
3793 (-mabi=ibmlongdouble): Likewise.
3794 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3795 already set, set the default format for long double.
3796 (rs6000_debug_reg_global): Print whether long double is IBM or
3798 (rs6000_option_override_internal): Rework setting long double
3799 format. Only warn if the user is changing the long double default
3800 and they did not use -Wno-psabi.
3801 * doc/invoke.texi (PowerPC options): Update the documentation for
3802 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3804 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3805 Alan Hayward <alan.hayward@arm.com>
3806 David Sherwood <david.sherwood@arm.com>
3808 * rtl.h (wider_subreg_mode): New function.
3809 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3810 rather than an unsigned int *.
3811 * ira-color.c (regno_max_ref_width): Replace with...
3812 (regno_max_ref_mode): ...this new variable.
3813 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3814 Use wider_subreg_mode.
3815 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3816 rather than an unsigned int *.
3817 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3818 (process_alt_operands): Likewise.
3819 (invariant_p): Likewise.
3820 * lra-spills.c (assign_mem_slot): Likewise.
3821 (add_pseudo_to_slot): Likewise.
3822 * lra.c (collect_non_operand_hard_regs): Likewise.
3823 (add_regs_to_insn_regno_info): Likewise.
3824 * reload1.c (regno_max_ref_width): Replace with...
3825 (regno_max_ref_mode): ...this new variable.
3826 (reload): Update accordingly. Update call to
3827 ira_sort_regnos_for_alter_reg.
3828 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3829 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3830 (scan_paradoxical_subregs): Likewise.
3832 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3834 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3835 (aarch64_frame): Add emit_frame_chain boolean.
3836 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3837 Move eh_return case to aarch64_layout_frame.
3838 (aarch64_layout_frame): Initialize emit_frame_chain.
3839 (aarch64_expand_prologue): Use emit_frame_chain.
3841 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3843 * config/aarch64/aarch64.c (aarch64_layout_frame):
3844 Ensure LR is always stored at the bottom of the callee-saves.
3845 Remove rarely used frame layout which saves callee-saves at top of
3846 frame, so the store of LR can be used as a valid probe in all cases.
3848 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3850 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3851 Improve unaligned TImode/TFmode base/offset split.
3853 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3854 Alan Hayward <alan.hayward@arm.com>
3855 David Sherwood <david.sherwood@arm.com>
3857 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3858 * combine.c (find_single_use_1): Likewise.
3859 (expand_field_assignment): Likewise.
3860 (move_deaths): Likewise.
3861 * lra-constraints.c (simplify_operand_subreg): Likewise.
3862 (curr_insn_transform): Likewise.
3863 * lra.c (collect_non_operand_hard_regs): Likewise.
3864 (add_regs_to_insn_regno_info): Likewise.
3865 * rtlanal.c (reg_referenced_p): Likewise.
3866 (covers_regno_no_parallel_p): Likewise.
3868 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3870 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3871 Don't print any bits outside the precision of the value.
3872 * wide-int.cc (test_printing): Add some new tests.
3874 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3876 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3877 supports -xbrace_comment option.
3878 * configure: Regenerate.
3879 * config.in: Regenerate.
3880 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3881 (ASM_CPU_SPEC): Use it.
3883 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3885 * target.def (static_rtx_alignment): New hook.
3886 * targhooks.h (default_static_rtx_alignment): Declare.
3887 * targhooks.c (default_static_rtx_alignment): New function.
3888 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3889 * doc/tm.texi: Regenerate.
3890 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3891 instead of targetm.constant_alignment. Remove call to
3893 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3894 (cris_preferred_mininum_alignment): New function, split out from...
3895 (cris_constant_alignment): ...here.
3896 (cris_static_rtx_alignment): New function.
3897 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3899 (ix86_constant_alignment): ...here.
3900 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3901 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3902 (mmix_static_rtx_alignment): New function.
3903 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3904 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3906 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3909 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3910 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3912 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3914 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3915 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3916 mark as a sign-extending load.
3917 (local_pic_load_u): Define.
3919 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3922 * fold-const.c (operand_equal_for_comparison_p): Also return true
3923 if ARG0 is a simple variant of ARG1 with narrower precision.
3924 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3926 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3928 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3930 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3931 scatter_static, scatter_per_elt.
3932 * x86-tune-costs.h: Add new cost entries.
3934 2017-10-25 Richard Biener <rguenther@suse.de>
3936 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3937 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3938 class pass_fre): Move to ...
3939 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3940 class pass_fre): ... here and adjust for statistics.
3942 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3945 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3946 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3947 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3948 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3951 2017-10-25 Richard Biener <rguenther@suse.de>
3953 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3954 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3955 eliminate_push_avail, eliminate_insert): Move inside...
3956 (class eliminate_dom_walker): ... this class in preparation
3958 (fini_eliminate): Remove by merging with ...
3959 (eliminate): ... this function. Adjust for class changes.
3960 (pass_pre::execute): Remove fini_eliminate call.
3961 (pass_fre::execute): Likewise.
3963 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3966 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3967 (VPERMI2, VPERMI2I): New mode iterators.
3968 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3969 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3971 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3972 mode iterator. Remove 3 old define_insn patterns.
3973 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3974 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3975 VPERMI2 mode iterator, remove the other two expanders.
3976 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3977 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3978 instructions, remove the other two patterns.
3979 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3980 mode iterator, remove the other two patterns.
3981 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3982 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3983 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3984 and adjust argument order accordingly.
3985 (ix86_expand_vec_perm): Adjust caller.
3986 (expand_vec_perm_1): Likewise.
3987 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3988 (expand_vec_perm_vpermt2_vpshub2): ... this.
3989 (ix86_expand_vec_perm_const_1): Adjust caller.
3990 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3993 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3994 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3995 (vec_shl_<mode>): Remove unused expander.
3996 (avx512bw_<shift_insn><mode>3): New define_insn.
3997 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3998 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4000 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4003 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4006 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4008 PR rtl-optimization/82396
4009 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4010 (autopref_multipass_init): Simplify initialization.
4011 (autopref_rank_data): Simplify sort order.
4012 * gcc/sched-int.h (autopref_multipass_data_): Remove
4013 multi_mem_insn_p, min_offset and max_offset.
4015 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4018 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4019 Check special value of flag_omit_frame_pointer.
4020 (aarch64_can_eliminate): Likewise.
4021 (aarch64_override_options_after_change_1): Simplify handling of
4022 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4024 2017-10-24 Richard Biener <rguenther@suse.de>
4026 PR tree-optimization/82697
4027 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4028 zero for conditional load and unconditional store.
4030 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4032 * doc/install.texi: Document bootstrap-cet.
4034 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4037 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4038 ENDBR instruction at function entrance if function is only
4041 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4044 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4045 patterns to better describe from which operation the CF is computed.
4046 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4047 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4048 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4049 is 0, use _0 suffixed expanders instead of emitting a comparison
4052 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4054 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4055 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4057 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4060 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4061 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4062 * loop-iv.c (iv_get_reaching_def): Likewise.
4063 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4064 variable is promoted and the partition contains undefined values.
4066 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4068 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4070 (nios2_address_cost): Define.
4071 (nios2_legitimize_address): Recognize (exp + constant) directly.
4072 (TARGET_ADDRESS_COST): Define.
4074 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4076 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4077 (nios2_symbolic_memory_operand_p): Declare.
4078 (nios2_split_large_constant): Declare.
4079 (nios2_split_symbolic_memory_operand): Declare.
4080 * config/nios2/nios2.c: Adjust includes.
4081 (nios2_symbolic_constant_allowed): New.
4082 (nios2_symbolic_constant_p): New.
4083 (nios2_plus_symbolic_constant_p): New.
4084 (nios2_valid_addr_expr_p): Recognize addresses involving
4086 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4087 (nios2_symbolic_memory_operand_p): New.
4088 (nios2_large_constant_p): New.
4089 (nios2_split_large_constant): New.
4090 (nios2_split_plus_large_constant): New.
4091 (nios2_split_symbolic_memory_operand): New.
4092 (nios2_legitimize_address): Code refactoring. Handle addresses
4093 involving symbolic constants.
4094 (nios2_emit_move_sequence): Likewise.
4095 (nios2_print_operand): Improve error output.
4096 (nios2_print_operand_address): Handle LO_SUM.
4097 (nios2_cdx_narrow_form_p): Likewise.
4098 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4099 operands involving symbolic constants.
4100 (movhi_internal, movsi_internal): Likewise.
4101 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4102 (extendhisi2, extendqi<mode>2): Likewise.
4104 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4106 * tree-pass.h (PROP_rtl_split_insns): Define.
4107 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4109 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4111 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4113 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4116 * target.def (const_not_ok_for_debug_p): Default to
4117 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4118 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4119 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4120 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4121 which targetm.const_not_ok_for_debug_p returned true.
4122 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4124 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4126 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4127 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4128 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4130 (ix86_const_not_ok_for_debug_p): New function.
4131 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4132 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4134 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4137 * system.h: Conditionally include "unique-ptr.h" if
4138 INCLUDE_UNIQUE_PTR is defined.
4139 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4140 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4142 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4144 * config/rl78/rl78.md: New define_expand "subdi3".
4146 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4149 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4150 DF_REF_INSN if DF_REF_INSN_INFO is false.
4152 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4154 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4155 xmm_move instead of sse_move.
4156 (sse_store_index): New function.
4157 (ix86_register_move_cost): Be more sensible about mismatch stall;
4158 model AVX moves correctly; make difference between sse->integer and
4160 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4161 moves; make difference between SSE and AVX.
4162 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4163 and zmm_move. Increase size of sse load and store tables;
4164 add unaligned load and store tables; add ssemmx_to_integer.
4165 * x86-tune-costs.h: Update all entries according to real
4166 move latencies from Agner Fog's manual and chip documentation.
4168 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4171 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4172 * config/i386/constraints.md (Wf): New constraint.
4173 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4174 (cmp<dwi>_doubleword): Removed.
4175 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4176 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4177 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4178 expand with cmp<dwi>_doubleword. For LTU and GEU use
4179 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4181 * common.opt (gcolumn-info): Enable by default.
4182 * doc/invoke.texi (gcolumn-info): Document new default.
4184 2017-10-23 Richard Biener <rguenther@suse.de>
4186 PR tree-optimization/82672
4187 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4188 Fold the stmt if we propagated into it.
4190 2017-10-23 Richard Biener <rguenther@suse.de>
4192 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4193 (bitmap_remove_expr_from_set): ... this. All callers call this
4194 for non-constant values.
4195 (bitmap_set_subtract): Rename to...
4196 (bitmap_set_subtract_expressions): ... this. Adjust and
4198 (bitmap_set_contains_value): Remove superfluous check.
4199 (bitmap_set_replace_value): Inline into single caller ...
4200 (bitmap_value_replace_in_set): ... here and simplify.
4201 (dependent_clean): Merge into ...
4202 (clean): ... this using an overload. Adjust.
4203 (prune_clobbered_mems): Adjust.
4204 (compute_antic_aux): Likewise.
4205 (compute_partial_antic_aux): Likewise.
4207 2017-10-23 Richard Biener <rguenther@suse.de>
4209 PR tree-optimization/82129
4211 2017-08-01 Richard Biener <rguenther@suse.de>
4213 PR tree-optimization/81181
4214 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4215 (compute_antic): ... end of iteration here.
4217 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4219 * target.def (starting_frame_offset): New hook.
4220 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4221 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4222 * doc/tm.texi.in: Regenerate.
4223 * hooks.h (hook_hwi_void_0): Declare.
4224 * hooks.c (hook_hwi_void_0): New function.
4225 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4226 STARTING_FRAME_OFFSET.
4227 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4228 * reload1.c (reload): Likewise.
4229 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4230 instead of STARTING_FRAME_OFFSET.
4231 * function.c (try_fit_stack_local): Likewise.
4232 (assign_stack_local_1): Likewise
4233 (instantiate_virtual_regs): Likewise.
4234 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4235 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4236 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4237 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4238 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4239 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4240 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4241 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4242 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4243 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4244 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4245 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4246 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4247 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4248 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4249 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4250 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4251 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4252 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4253 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4254 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4255 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4256 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4257 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4258 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4259 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4260 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4261 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4262 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4263 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4264 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4265 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4266 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4267 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4268 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4269 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4270 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4271 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4272 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4273 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4274 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4275 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4276 return a HOST_WIDE_INT.
4277 (avr_builtin_setjmp_frame_value): Use it instead of
4278 STARTING_FRAME_OFFSET.
4279 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4280 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4281 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4283 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4284 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4285 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4286 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4287 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4288 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4289 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4290 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4291 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4292 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4293 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4294 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4296 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4297 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4298 * config/mips/mips.c (mips_compute_frame_info): Refer to
4299 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4300 (mips_starting_frame_offset): New function.
4301 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4302 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4303 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4304 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4305 and return a HOST_WIDE_INT.
4306 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4307 (mmix_initial_elimination_offset): Refer to
4308 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4309 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4310 * config/pa/pa.c (pa_starting_frame_offset): New function.
4311 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4312 (pa_expand_prologue): Likewise.
4313 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4314 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4315 !FRAME_GROWS_DOWNWARD handling to...
4316 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4317 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4318 !FRAME_GROWS_DOWNWARD handling to...
4319 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4320 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4321 !FRAME_GROWS_DOWNWARD handling to...
4322 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4323 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4325 (rs6000_starting_frame_offset): New function.
4326 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4327 !FRAME_GROWS_DOWNWARD handling to...
4328 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4329 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4330 !FRAME_GROWS_DOWNWARD handling to...
4331 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4332 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4333 !FRAME_GROWS_DOWNWARD handling to...
4334 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4335 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4336 (rs6000_starting_frame_offset): New function.
4337 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4338 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4339 * config/vax/vax.c (vax_starting_frame_offset): New function.
4340 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4341 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4342 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4343 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4344 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4345 * system.h (STARTING_FRAME_OFFSET): Poison.
4347 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4349 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4350 SCALAR_TYPE_MODE instead of TYPE_MODE.
4352 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4353 Alan Hayward <alan.hayward@arm.com>
4354 David Sherwood <david.sherwood@arm.com>
4356 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4358 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4359 Alan Hayward <alan.hayward@arm.com>
4360 David Sherwood <david.sherwood@arm.com>
4362 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4364 2017-10-23 Richard Biener <rguenther@suse.de>
4366 PR tree-optimization/82129
4367 * tree-ssa-pre.c (bitmap_set_and): Remove.
4368 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4369 canonicalizing expressions in the set to those with lowest
4370 ID rather than taking that from the first edge.
4372 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4374 * combine.c (rtx_equal_for_field_assignment_p): Use
4375 byte_lowpart_offset.
4377 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4378 Alan Hayward <alan.hayward@arm.com>
4379 David Sherwood <david.sherwood@arm.com>
4381 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4382 to a SUBREG_PROMOTED_VAR.
4384 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4385 Alan Hayward <alan.hayward@arm.com>
4386 David Sherwood <david.sherwood@arm.com>
4388 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4389 (expand_debug_source_expr): Likewise.
4390 * combine.c (combine_simplify_rtx): Likewise.
4391 * cse.c (fold_rtx): Likewise.
4392 * optabs.c (expand_float): Likewise.
4393 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4394 (simplify_binary_operation_1): Likewise.
4396 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4397 Alan Hayward <alan.hayward@arm.com>
4398 David Sherwood <david.sherwood@arm.com>
4400 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4401 (record_promoted_value): Likewise.
4402 * expr.c (expand_expr_real_2): Likewise.
4403 * ree.c (update_reg_equal_equiv_notes): Likewise.
4404 (combine_set_extension): Likewise.
4405 * rtlanal.c (low_bitmask_len): Likewise.
4406 * simplify-rtx.c (neg_const_int): Likewise.
4407 (simplify_binary_operation_1): Likewise.
4409 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4410 Alan Hayward <alan.hayward@arm.com>
4411 David Sherwood <david.sherwood@arm.com>
4413 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4414 * regcprop.c (maybe_mode_change): Likewise.
4415 * reload1.c (alter_reg): Likewise.
4417 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4419 * inchash.h (inchash::hash::add_wide_int): New function.
4420 * lto-streamer-out.c (hash_tree): Use it.
4422 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4424 * inchash.h (inchash::hash::add_wide_int): Rename to...
4425 (inchash::hash::add_hwi): ...this.
4426 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4427 (polymorphic_call_target_hasher::hash): Likewise.
4428 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4429 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4430 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4431 * lto-streamer-out.c (hash_tree): Likewise.
4432 * optc-save-gen.awk: Likewise.
4433 * tree.c (add_expr): Likewise.
4435 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4438 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4439 for ordered inequality comparisons even with TARGET_IEEE_FP.
4441 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4444 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4445 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4446 Expand with cmp<dwi>_doubleword.
4448 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4450 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4451 List CET intrinsics.
4452 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4453 specific to -fcf-protection option.
4455 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4457 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4458 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4459 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4460 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4461 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4462 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4463 (extra_objs): Add cet.o for Linux/x86 targets.
4464 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4465 * config/i386/cet.c: New file.
4466 * config/i386/cetintrin.h: Likewise.
4467 * config/i386/t-cet: Likewise.
4468 * config/i386/cpuid.h (bit_SHSTK): New.
4469 (bit_IBT): Likewise.
4470 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4471 pass IBT and SHSTK bits.
4472 * config/i386/i386-builtin-types.def
4473 (VOID_FTYPE_UNSIGNED_PVOID): New.
4474 (VOID_FTYPE_UINT64_PVOID): Likewise.
4475 * config/i386/i386-builtin.def: Add CET intrinsics.
4476 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4477 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4478 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4479 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4481 * config/i386/i386.c (rest_of_insert_endbranch): New.
4482 (pass_data_insert_endbranch): Likewise.
4483 (pass_insert_endbranch): Likewise.
4484 (make_pass_insert_endbranch): Likewise.
4485 (ix86_notrack_prefixed_insn_p): Likewise.
4486 (ix86_target_string): Add -mibt, -mshstk flags.
4487 (ix86_option_override_internal): Add flag_cf_protection
4489 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4490 (ix86_print_operand): Add 'notrack' prefix output.
4491 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4492 (ix86_expand_builtin): Expand CET intrinsics.
4493 (x86_output_mi_thunk): Add 'endbranch' instruction.
4494 * config/i386/i386.h (TARGET_IBT): New.
4495 (TARGET_IBT_P): Likewise.
4496 (TARGET_SHSTK): Likewise.
4497 (TARGET_SHSTK_P): Likewise.
4498 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4499 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4500 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4501 (builtin_setjmp_setup): New pattern.
4502 (builtin_longjmp): Likewise.
4503 (rdssp<mode>): Likewise.
4504 (incssp<mode>): Likewise.
4505 (saveprevssp): Likewise.
4506 (rstorssp): Likewise.
4507 (wrss<mode>): Likewise.
4508 (wruss<mode>): Likewise.
4509 (setssbsy): Likewise.
4510 (clrssbsy): Likewise.
4511 (nop_endbr): Likewise.
4512 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4514 * config/i386/immintrin.h: Include <cetintrin.h>.
4515 * config/i386/linux-common.h
4516 (file_end_indicate_exec_stack_and_cet): New prototype.
4517 (TARGET_ASM_FILE_END): New.
4519 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4521 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4522 latencies instead of having separate table; make difference between
4523 integer and float costs.
4524 * i386.h (processor_costs): Remove scalar_stmt_cost,
4525 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4526 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4528 * x86-tune-costs.h: Remove entries which has been removed in
4529 procesor_costs from all tables; make cond_taken_branch_cost
4530 and cond_not_taken_branch_cost COST_N_INSNS based.
4532 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4534 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4536 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4538 * config/i386/i386.md (isa): Remove fma_avx512f.
4539 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4540 <avx512>_fmadd_<mode>_mask3<round_name>,
4541 <avx512>_fmsub_<mode>_mask<round_name>,
4542 <avx512>_fmsub_<mode>_mask3<round_name>,
4543 <avx512>_fnmadd_<mode>_mask<round_name>,
4544 <avx512>_fnmadd_<mode>_mask3<round_name>,
4545 <avx512>_fnmsub_<mode>_mask<round_name>,
4546 <avx512>_fnmsub_<mode>_mask3<round_name>,
4547 <avx512>_fmaddsub_<mode>_mask<round_name>,
4548 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4549 <avx512>_fmsubadd_<mode>_mask<round_name>,
4550 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4551 (*vec_widen_umult_even_v16si<mask_name>,
4552 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4553 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4555 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4557 * extend.texi: Add 'nocf_check' documentation.
4558 * gimple.texi: Add second parameter to
4559 gimple_build_call_from_tree.
4560 * invoke.texi: Add -fcf-protection documentation.
4561 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4563 2017-10-20 Richard Biener <rguenther@suse.de>
4565 PR tree-optimization/82473
4566 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4567 the largest input type.
4569 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4571 * c-attribs.c (handle_nocf_check_attribute): New function.
4572 (c_common_attribute_table): Add 'nocf_check' handling.
4573 * gimple-parser.c: Add second argument NULL to
4574 gimple_build_call_from_tree.
4575 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4576 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4578 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4579 * common.opt: Add fcf-protection flag.
4580 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4581 * flag-types.h: Add enum cf_protection_level.
4582 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4583 Add 'nocf_check' attribute propagation to gimple call.
4584 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4585 (gimple_build_call_from_tree): Update prototype.
4586 (gimple_call_nocf_check_p): New function.
4587 (gimple_call_set_nocf_check): Likewise.
4588 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4589 * ipa-icf.c: Add nocf_check attribute in statement hash.
4590 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4591 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4592 * toplev.c (process_options): Add flag_cf_protection handling.
4594 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4596 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4598 2017-10-20 Richard Biener <rguenther@suse.de>
4600 PR tree-optimization/82603
4601 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4602 remove false predicated stores.
4604 2017-10-20 Richard Biener <rguenther@suse.de>
4606 * graphite-isl-ast-to-gimple.c
4607 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4608 Remove return value and simplify, dump copied stmt after lhs
4610 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4611 Reduce dump verbosity.
4612 (gsi_insert_earliest): Likewise.
4613 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4614 * graphite.c (print_global_statistics): Adjust dumping.
4615 (print_graphite_scop_statistics): Likewise.
4616 (print_graphite_statistics): Do not dump loops here.
4617 (graphite_transform_loops): But here.
4619 2017-10-20 Nicolas Roche <roche@adacore.com>
4621 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4622 * configure: Regenerate.
4624 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4627 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4628 functions when optimizing replace GIMPLE_RETURN stmts with
4629 calls to __builtin_unreachable ().
4632 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4633 for -fsanitize=thread link of executables.
4634 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4635 link of executables.
4638 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4640 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4641 define_insns for logical vector shifts to use VI248_AVX512BW
4642 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4643 condition, useless isa and prefix attributes. Change the first
4644 2 of these define_insns to ...
4645 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4646 define_insn for avx512vl.
4647 (<shift_insn><mode>3): ... and this, new define_insn without
4648 masking for non-avx512vl.
4651 * config/i386/sse.md (*andnot<mode>3,
4652 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4653 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4654 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4655 not applied use empty suffix even for TARGET_AVX512VL.
4656 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4657 is applied, supply evex,evex or evex,evex,evex instead of just
4660 2017-10-20 Julia Koval <julia.koval@intel.com>
4662 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4663 (OPTION_MASK_ISA_GFNI_UNSET): New.
4664 (ix86_handle_option): Handle OPT_mgfni.
4665 * config/i386/cpuid.h (bit_GFNI): New.
4666 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4667 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4668 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4669 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4670 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4671 * config/i386/i386.opt: Add mgfni.
4673 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4675 * config/msp430/msp430.c (msp430_option_override): Disable
4676 -fdelete-null-pointer-checks.
4677 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4679 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4681 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4682 of x87 and SSE instructions.
4684 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4686 * asan.c (create_cond_insert_point): Do not update edge count.
4687 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4688 (afdo_propagate_circuit): Likewise.
4689 (afdo_calculate_branch_prob): Likewise.
4690 (afdo_annotate_cfg): Likewise.
4691 * basic-block.h (struct edge_def): Remove count.
4692 (edge_def::count): New accessor.
4693 * bb-reorder.c (rotate_loop): Update.
4694 (find_traces_1_round): Update.
4695 (connect_traces): Update.
4696 (sanitize_hot_paths): Update.
4697 * cfg.c (unchecked_make_edge): Update.
4698 (make_single_succ_edge): Update.
4699 (check_bb_profile): Update.
4700 (dump_edge_info): Update.
4701 (update_bb_profile_for_threading): Update.
4702 (scale_bbs_frequencies_int): Update.
4703 (scale_bbs_frequencies_gcov_type): Update.
4704 (scale_bbs_frequencies_profile_count): Update.
4705 (scale_bbs_frequencies): Update.
4706 * cfganal.c (connect_infinite_loops_to_exit): Update.
4707 * cfgbuild.c (compute_outgoing_frequencies): Update.
4708 (find_many_sub_basic_blocks): Update.
4709 * cfgcleanup.c (try_forward_edges): Update.
4710 (try_crossjump_to_edge): Update
4711 * cfgexpand.c (expand_gimple_cond): Update
4712 (expand_gimple_tailcall): Update
4713 (construct_exit_block): Update
4714 * cfghooks.c (verify_flow_info): Update
4715 (redirect_edge_succ_nodup): Update
4716 (split_edge): Update
4717 (make_forwarder_block): Update
4718 (duplicate_block): Update
4719 (account_profile_record): Update
4720 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4721 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4722 * cfgloopmanip.c (scale_loop_profile): Update.
4724 (lv_adjust_loop_entry_edge): Update.
4725 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4726 (force_nonfallthru_and_redirect): Update.
4727 (purge_dead_edges): Update.
4728 (rtl_flow_call_edges_add): Update.
4729 * cgraphunit.c (init_lowered_empty_function): Update.
4730 (cgraph_node::expand_thunk): Update.
4731 * gimple-pretty-print.c (dump_probability): Update.
4732 (dump_edge_probability): Update.
4733 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4734 * haifa-sched.c (sched_create_recovery_edges): Update.
4735 * hsa-gen.c (convert_switch_statements): Update.
4736 * ifcvt.c (dead_or_predicable): Update.
4737 * ipa-inline-transform.c (inline_transform): Update.
4738 * ipa-split.c (split_function): Update.
4739 * ipa-utils.c (ipa_merge_profiles): Update.
4740 * loop-doloop.c (add_test): Update.
4741 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4742 * lto-streamer-in.c (input_cfg): Update.
4743 (input_function): Update.
4744 * lto-streamer-out.c (output_cfg): Update.
4745 * modulo-sched.c (sms_schedule): Update.
4746 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4747 * predict.c (maybe_hot_edge_p): Update.
4748 (unlikely_executed_edge_p): Update.
4749 (probably_never_executed_edge_p): Update.
4750 (dump_prediction): Update.
4751 (drop_profile): Update.
4752 (propagate_unlikely_bbs_forward): Update.
4753 (determine_unlikely_bbs): Update.
4754 (force_edge_cold): Update.
4755 * profile.c (compute_branch_probabilities): Update.
4756 * reg-stack.c (better_edge): Update.
4757 * shrink-wrap.c (handle_simple_exit): Update.
4758 * tracer.c (better_p): Update.
4759 * trans-mem.c (expand_transaction): Update.
4760 (split_bb_make_tm_edge): Update.
4761 * tree-call-cdce.c: Update.
4762 * tree-cfg.c (gimple_find_sub_bbs): Update.
4763 (gimple_split_edge): Update.
4764 (gimple_duplicate_sese_region): Update.
4765 (gimple_duplicate_sese_tail): Update.
4766 (gimple_flow_call_edges_add): Update.
4767 (insert_cond_bb): Update.
4768 (execute_fixup_cfg): Update.
4769 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4770 * tree-complex.c (expand_complex_div_wide): Update.
4771 * tree-eh.c (lower_resx): Update.
4772 (unsplit_eh): Update.
4773 (cleanup_empty_eh_move_lp): Update.
4774 * tree-inline.c (copy_edges_for_bb): Update.
4775 (freqs_to_counts): Update.
4776 (copy_cfg_body): Update.
4777 * tree-ssa-dce.c (remove_dead_stmt): Update.
4778 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4779 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4780 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4781 (unloop_loops): Update.
4782 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4783 * tree-ssa-loop-split.c (connect_loops): Update.
4784 (split_loop): Update.
4785 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4786 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4787 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4788 * tree-ssa-reassoc.c (branch_fixup): Update.
4789 * tree-ssa-tail-merge.c (replace_block_by): Update.
4790 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4791 (compute_path_counts): Update.
4792 (update_profile): Update.
4793 (recompute_probabilities): Update.
4794 (update_joiner_offpath_counts): Update.
4795 (estimated_freqs_path): Update.
4796 (freqs_to_counts_path): Update.
4797 (clear_counts_path): Update.
4798 (ssa_fix_duplicate_block_edges): Update.
4799 (duplicate_thread_path): Update.
4800 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4801 (case_bit_test_cmp): Update.
4802 (collect_switch_conv_info): Update.
4803 (gen_inbound_check): Update.
4804 (do_jump_if_equal): Update.
4805 (emit_cmp_and_jump_insns): Update.
4806 * tree-tailcall.c (decrease_profile): Update.
4807 (eliminate_tail_call): Update.
4808 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4809 (vect_do_peeling): Update.
4810 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4811 * ubsan.c (ubsan_expand_null_ifn): Update.
4812 (ubsan_expand_ptr_ifn): Update.
4813 * value-prof.c (gimple_divmod_fixed_value): Update.
4814 (gimple_mod_pow2): Update.
4815 (gimple_mod_subtract): Update.
4816 (gimple_ic): Update.
4817 (gimple_stringop_fixed_value): Update.
4819 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4822 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4824 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4826 PR rtl-optimization/82395
4827 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4828 based on non_spilled_static_chain_regno_p.
4830 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4832 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4833 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4834 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4836 2017-10-19 Martin Sebor <msebor@redhat.com>
4838 PR tree-optimization/82596
4839 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4841 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4843 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4844 (get_mem_refs_of_builtin_call): Likewise.
4845 * builtins.c (expand_builtin_apply): Adjust call to
4846 allocate_dynamic_stack_space.
4847 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4848 the third argument to allocate_dynamic_stack_space, otherwise -1.
4849 (expand_builtin): Deal with all alloca variants.
4850 (is_inexpensive_builtin): Likewise.
4851 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4852 * calls.c (special_function_p): Deal with all alloca variants.
4853 (initialize_argument_information): Adjust call to
4854 allocate_dynamic_stack_space.
4855 (expand_call): Likewise.
4856 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4857 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4858 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4859 use it for the stack usage computation.
4860 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4861 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4862 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4863 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4864 (in_loop_p): Remove first argument and useless check.
4865 (pass_walloca::execute): Remove useless test and adjust call to above.
4866 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4867 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4868 (gimplify_call_expr): Deal with all alloca variants.
4869 * hsa-gen.c (gen_hsa_alloca): Likewise.
4870 (gen_hsa_insns_for_call): Likewise.
4871 * ipa-pure-const.c (special_builtin_state): Likewise.
4872 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4873 * tree-object-size.c (alloc_object_size): Likewise.
4874 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4875 (call_may_clobber_ref_p_1): Likewise.
4876 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4877 (ccp_fold_stmt): Likewise.
4878 (optimize_stack_restore): Likewise.
4879 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4880 (mark_all_reaching_defs_necessary_1): Likewise.
4881 (propagate_necessity): Likewise.
4882 (eliminate_unnecessary_stmts): Likewise.
4883 * tree.c (build_common_builtin_nodes): Build
4884 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4885 (build_alloca_call_expr): New function.
4886 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4887 (CASE_BUILT_IN_ALLOCA): Likewise.
4888 (build_alloca_call_expr): Declare.
4889 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4891 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4894 * dwarf2out.c (new_die_raw): New static inline function.
4895 (new_die): Use it to create the DIE.
4896 (add_AT_external_die_ref): Likewise.
4897 (clone_die): Likewise.
4898 (clone_as_declaration): Likewise.
4899 (dwarf2out_vms_debug_main_pointer): Likewise.
4900 (base_type_die): Likewise. Remove early return for corner cases.
4901 Do not call add_pubtype on the DIE here.
4902 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4903 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4904 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4905 native order exists for base types, attach the DIE manually and call
4906 add_pubtype on it. Do not equate a reverse order DIE to the type.
4908 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4910 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4911 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4913 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4915 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4917 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4918 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4919 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4920 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4921 * opts.c (sanitizer_opts): Add builtin.
4922 * ubsan.c (instrument_builtin): New function.
4923 (pass_ubsan::execute): Call it.
4924 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4925 * doc/invoke.texi: Document -fsanitize=builtin.
4927 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4928 builtins, store max (log2 (align), 0) into uchar field instead of
4929 align into uptr field.
4930 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4931 store uchar 0 field instead of uptr 0 field.
4932 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4933 instead of passing one address of struct with 2 locations pass
4934 two addresses of structs with 1 location each.
4935 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4936 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4937 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4938 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4939 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4940 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4941 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4942 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4944 2017-10-19 Martin Liska <mliska@suse.cz>
4947 * file-find.c (remove_prefix): Remove.
4948 * file-find.h (remove_prefix): Likewise.
4949 * gcc-ar.c: Remove smartness of lookup.
4951 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4953 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4954 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4955 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4957 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4960 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4961 (setcc + and to xor + setcc): New peephole2.
4963 2017-10-19 Tom de Vries <tom@codesourcery.com>
4965 * doc/sourcebuild.texi (Test Directives, Variants of
4966 dg-require-support): Add dg-require-stack-size.
4968 2017-10-19 Martin Liska <mliska@suse.cz>
4971 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4972 that have a large alignment.
4973 (gimplify_target_expr): Likewise.
4975 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4977 PR rtl-optimization/82602
4978 * ira.c (rtx_moveable_p): Return false for volatile asm.
4980 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4983 * config/i386/i386-modes.def (CCGZ): New CC mode.
4984 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4985 * config/i386/predicates.md (ix86_comparison_operator):
4987 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4988 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4989 with double-word subtraction.
4990 (put_condition_code): Handle CCGZmode.
4992 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4994 * wide-int.cc (debug (const wide_int &)): New.
4995 (debug (const wide_int *)): New.
4996 (debug (const widest_int &)): New.
4997 (debug (const widest_int *)): New.
4999 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5002 * lra-constraints.c (curr_insn_transform): Use non-input operand
5003 instead of output one for matched reload.
5005 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5007 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5008 (tree-ssa-loop-ivopts.h): New header file.
5009 (struct builtin_info): New fields.
5010 (classify_builtin_1): Compute and record base and offset parts for
5011 memset builtin partition by calling strip_offset.
5012 (offset_cmp, fuse_memset_builtins): New functions.
5013 (finalize_partitions): Fuse adjacent memset partitions by calling
5015 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5016 Expose the interface.
5017 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5019 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5021 PR tree-optimization/82574
5022 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5023 that data reference must be executed exactly once per iteration
5024 against the outermost loop in nest.
5025 (classify_partition): Update call to above function.
5027 2017-10-18 Richard Biener <rguenther@suse.de>
5029 PR tree-optimization/82591
5030 * graphite.c (graphite_transform_loops): Move code gen message
5032 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5033 Here. Handle scop_to_isl_ast failing.
5034 (scop_to_isl_ast): Limit the number of ISL operations.
5036 2017-10-18 Richard Biener <rguenther@suse.de>
5038 * graphite-isl-ast-to-gimple.c
5039 (translate_isl_ast_to_gimple::set_rename): Simplify.
5040 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5041 (graphite_copy_stmts_from_block): ... here.
5042 (copy_bb_and_scalar_dependences): Simplify.
5043 (add_parameters_to_ivs_params): Canonicalize.
5044 (generate_entry_out_of_ssa_copies): Simplify.
5045 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5046 by passing in ISL dimension.
5047 (parameter_index_in_region_1): Rename to ...
5048 (parameter_index_in_region): ... this.
5049 (extract_affine): Adjust assert, pass down parameter index.
5050 (add_param_constraints): Use range-info when available.
5051 (build_scop_context): Adjust.
5052 * sese.c (new_sese_info): Adjust.
5053 (free_sese_info): Likewise.
5054 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5055 Remove unused typedefs.
5056 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5058 2017-10-18 Martin Liska <mliska@suse.cz>
5060 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5062 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5064 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5065 (s390_sched_init): Do not reset s390_sched_state if we entered the
5066 current basic block via a fallthru edge and all others are unlikely.
5068 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5070 * config/s390/s390.c (NUM_SIDES): New variable.
5071 (LONGRUNNING_THRESHOLD): New variable.
5072 (LATENCY_FACTOR): New variable.
5073 (s390_sched_score): Decrease score for long-running instructions on
5075 (s390_sched_variable_issue): Perform bookkeeping for long-running
5078 2017-10-18 Richard Biener <rguenther@suse.de>
5080 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5081 Simplify with removal of the parameter rename map.
5082 (set_rename): Likewise.
5083 (should_copy_to_new_region): Likewise.
5084 (graphite_copy_stmts_from_block): Likewise.
5085 (copy_bb_and_scalar_dependences): Remove initialization of
5086 unused copied_bb_map.
5088 (copy_internal_parameters): Likewise.
5089 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5090 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5091 Use INTEGRAL_TYPE_P.
5092 (parameter_index_in_region_1): Rename to ...
5093 (assign_parameter_index_in_region): ... this. Assert we have
5094 a parameter we handle.
5095 (scan_tree_for_params): Adjust.
5096 * sese.h (parameter_rename_map_t): Remove.
5097 (struct sese_info_t): Remove unused parameter_rename_map and
5098 copied_bb_map members.
5099 * sese.c (new_sese_info): Adjust.
5100 (free_sese_info): Likewise.
5102 2017-10-18 Martin Liska <mliska@suse.cz>
5105 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5106 on an abnormal edge.
5108 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5110 * doc/invoke.texi (ffunction-sections and fdata-sections):
5113 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5115 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5116 the use statement can throw internally.
5118 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5120 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5121 any RTX present on the RHS of a SET.
5122 * compare-elim.c (try_eliminate_compare): Restore comment.
5124 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5126 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5127 may be also called on tcc_type nodes.
5128 * langhooks.c (lhd_tree_size): Likewise.
5130 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5132 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5133 format_warning_at_substring.
5134 (maybe_warn): Convert source_range * param to a location_t. Pass
5135 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5136 (format_directive): Remove code to extract source_ranges and
5137 source_range * in favor of just a location_t.
5138 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5140 * substring-locations.c (format_warning_va): Convert
5141 source_range * param to a location_t.
5142 (format_warning_at_substring): Likewise.
5143 * substring-locations.h (format_warning_va): Likewise.
5144 (format_warning_at_substring): Likewise.
5146 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5148 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5150 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5151 and scatter/gather ops.
5153 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5154 vec_gather_load and vec_scatter_store.
5155 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5156 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5157 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5158 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5159 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5160 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5162 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5164 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5165 Detect FTST instruction and handle its register pops. Only pop
5166 second operand if can_pop_second_op is true.
5167 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5168 set can_pop_second_op to false in the compare_for_stack_reg call.
5170 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5171 output_fp_compare for stack register operands.
5172 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5173 instructions here. Do not emit stack register pops here. Assert
5174 that FCOMPP pops next to top stack register. Rewrite function.
5176 2017-10-17 Nathan Sidwell <nathan@acm.org>
5179 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5180 use DECL_ASSEMBLER_NAME_RAW.
5183 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5186 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5187 Wilco Dijkstra <wilco.dijkstra@arm.com>
5189 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5190 converstion to Pmode from the buf_addr.
5192 2017-10-17 Richard Biener <rguenther@suse.de>
5194 * graphite-dependences.c (scop_get_reads_and_writes): Change
5195 output parameters to references.
5197 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5199 PR 71026/tree-optimization
5200 * fold-const.c (distribute_real_division): Removed.
5201 (fold_binary_loc): Remove calls to distribute_real_divison.
5203 2017-10-17 Richard Biener <rguenther@suse.de>
5205 * graphite-scop-detection.c
5206 (scop_detection::stmt_has_simple_data_refs_p): Always use
5207 the full nest as region.
5208 (try_generate_gimple_bb): Likewise.
5209 * sese.c (scalar_evolution_in_region): Simplify now that
5210 SCEV can handle instantiation in regions.
5211 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5212 in the non-loop part of a function if requested.
5214 2017-10-17 Richard Biener <rguenther@suse.de>
5216 PR tree-optimization/82563
5217 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5219 (graphite_regenerate_ast_isl): Call it.
5220 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5222 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5224 PR tree-optimization/82549
5225 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5226 Formatting fixes. Instead of calling make_bit_field_ref with negative
5229 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5231 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5232 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5233 _mm_maskz_reduce_ss): New.
5234 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5235 __builtin_ia32_reducess_mask): Ditto..
5236 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5237 * config/i386/sse.md (reduces<mode>): Renamed to ...
5238 (reduces<mode><mask_scalar_name>): ... this.
5239 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5241 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5242 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5244 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5246 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5247 * selftest-run-tests.c (selftest::run_tests): Call
5248 selftest::unique_ptr_tests_cc_tests.
5249 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5250 * unique-ptr-tests.cc: New file.
5252 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5255 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5257 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5258 (make_hard_regno_born, make_hard_regno_dead): Update
5259 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5261 2017-10-16 Jeff Law <law@redhat.com>
5263 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5265 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5267 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5269 2017-10-16 Olivier Hainque <hainque@adacore.com>
5271 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5272 with_cpu if we were configured for an e500v2 target cpu name.
5274 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5276 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5277 * doc/invoke.texi: Document +nodsp as a valid extension for
5280 2017-10-16 Martin Liska <mliska@suse.cz>
5282 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5283 (test_set_range): Likewise.
5284 (test_range_functions): Rename to ...
5285 (test_bit_in_range): ... this.
5286 (sbitmap_c_tests): Add new test.
5288 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5290 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5292 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5293 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5295 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5297 * config/aarch64/aarch64-builtins.c
5298 (aarch64_types_quadopu_lane_qualifiers): New.
5299 (TYPES_QUADOPU_LANE): New.
5300 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5301 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5302 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5303 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5304 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5305 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5306 (Vdottype, DOTPROD): New.
5307 (sur): Add SDOT and UDOT.
5309 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5311 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5312 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5313 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5315 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5316 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5317 Enable TARGET_DOTPROD.
5318 (cortex-a75.cortex-a55): Likewise.
5319 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5321 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5323 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5324 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5326 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5328 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5329 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5330 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5331 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5332 * config/arm/types.md (neon_dot, neon_dot_q): New.
5333 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5335 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5337 * config/arm/arm.h (TARGET_DOTPROD): New.
5338 * config/arm/arm.c (arm_arch_dotprod): New.
5339 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5340 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5341 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5342 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5343 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5344 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5345 * doc/invoke.texi (armv8.2-a): Document dotprod
5347 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5349 * i386.c (ix86_vec_cost): New function.
5350 (ix86_rtx_costs): Handle vector operations better.
5351 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5352 * x86-tune-costs.h: Add new costs to all tables.
5354 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5356 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5358 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5359 divsd, sqrtss and sqrtsd
5360 * x86-tune-costs.h: Add new entries to all costs.
5361 (znver1_cost): Fix to match real instruction latencies.
5363 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5364 Michael Collison <michael.collison@arm.com>
5366 * compare-elim.c: Include emit-rtl.h.
5367 (can_merge_compare_into_arith): New function.
5368 (try_validate_parallel): Likewise.
5369 (try_merge_compare): Likewise.
5370 (try_eliminate_compare): Call the above when no previous clobber
5372 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5375 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5379 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5380 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5384 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5385 to be any operand_equal_p operands. For & (B - 1) require
5386 B to be power of 2. Recognize
5387 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5389 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5392 * optabs.c (expand_memory_blockage): Fix call of
5393 targetm.have_memory_blockage.
5395 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5398 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5399 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5400 objects to extra_objs instead of overwriting it.
5402 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5404 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5405 Use any_fp_register_operand as operand[3] predicate. Simplify
5406 equality test for operands[2] and operands[4] memory location.
5407 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5408 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5409 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5410 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5411 any_fp_register_operand as operand[1] predicate. Simplify
5412 equality test for operands[0] and operands[3] memory location.
5413 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5414 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5415 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5417 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5419 * target-insns.def: Add memory_blockage.
5420 * optabs.c (expand_memory_blockage): New function.
5421 (expand_asm_memory_barrier): Rename ...
5422 (expand_asm_memory_blockage): ... to this.
5423 (expand_mem_thread_fence): Call expand_memory_blockage
5424 instead of expand_asm_memory_barrier.
5425 (expand_mem_singnal_fence): Ditto.
5426 (expand_atomic_load): Ditto.
5427 (expand_atomic_store): Ditto.
5428 * doc/md.texi (Standard Pattern Names For Generation):
5429 Document memory_blockage instruction pattern.
5431 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5433 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5434 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5435 * config/rl78/rl78.md: New define_expand "adddi3".
5437 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5439 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5442 2017-10-13 Jeff Law <law@redhat.com>
5444 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5446 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5449 * internal-fn.c (expand_mul_overflow): If both operands have
5450 the same highpart of -1 or 0 and the topmost bit of lowpart
5451 is different, overflow is if res <= 0 rather than res < 0.
5453 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5455 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5456 TARGET_P9_VECTOR code for unaligned_load case.
5458 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5460 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5462 2017-10-13 Nathan Sidwell <nathan@acm.org>
5464 * tree-core.h (tree_contains_struct): Make bool.
5465 * tree.c (tree_contains_struct): Likewise.
5466 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5467 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5468 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5469 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5470 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5472 2017-10-13 Richard Biener <rguenther@suse.de>
5474 * graphite-isl-ast-to-gimple.c
5475 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5476 parameters and dominance check.
5477 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5478 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5479 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5480 Do not update SSA form here or do intermediate IL verification.
5481 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5482 (graphite_initialize): Remove check on the number of loops in
5483 the function and inline into graphite_transform_loops.
5484 (graphite_finalize): Inline into graphite_transform_loops.
5485 (graphite_transform_loops): Perform SSA update and IL verification
5487 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5489 2017-10-13 Richard Biener <rguenther@suse.de>
5491 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5492 graphite_expression_type_precision): Avoid global constructor
5494 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5495 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5496 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5497 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5498 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5499 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5501 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5504 * config/i386/i386.h (ix86_red_zone_size): New.
5505 * config/i386/i386.md (push peephole2s): Replace
5506 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5508 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5509 Alan Hayward <alan.hayward@arm.com>
5510 David Sherwood <david.sherwood@arm.com>
5512 * combine.c (can_change_dest_mode): Reject changes in
5513 REGMODE_NATURAL_SIZE.
5515 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5516 Alan Hayward <alan.hayward@arm.com>
5517 David Sherwood <david.sherwood@arm.com>
5519 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5520 (expand_debug_source_expr): Likewise.
5521 * combine.c (combine_simplify_rtx): Likewise.
5522 * cse.c (fold_rtx): Likewise.
5523 * fwprop.c (canonicalize_address): Likewise.
5524 * targhooks.c (default_shift_truncation_mask): Likewise.
5526 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5527 Alan Hayward <alan.hayward@arm.com>
5528 David Sherwood <david.sherwood@arm.com>
5530 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5531 (widened_mode): Likewise.
5532 (expand_unop): Likewise.
5533 * ree.c (transform_ifelse): Likewise.
5534 (merge_def_and_ext): Likewise.
5535 (combine_reaching_defs): Likewise.
5536 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5538 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5539 Alan Hayward <alan.hayward@arm.com>
5540 David Sherwood <david.sherwood@arm.com>
5542 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5543 * combine.c (gen_lowpart_for_combine): Likewise.
5544 * dwarf2out.c (rtl_for_decl_location): Likewise.
5545 * final.c (alter_subreg): Likewise.
5546 * rtlhooks.c (gen_lowpart_general): Likewise.
5547 (gen_lowpart_if_possible): Likewise.
5549 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5550 Alan Hayward <alan.hayward@arm.com>
5551 David Sherwood <david.sherwood@arm.com>
5553 * calls.c (expand_call): Use subreg_lowpart_offset.
5554 * cse.c (cse_insn): Likewise.
5555 * regcprop.c (copy_value): Likewise.
5556 (copyprop_hardreg_forward_1): Likewise.
5558 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5561 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5562 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5563 =Q constraints to +Q and into insn condition add check
5564 that operands[0] and operands[1] are equal.
5565 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5566 =Q constraints to +Q and into insn condition add check
5567 that operands[0] is equal to either operands[1] or operands[2].
5570 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5571 instead of handling MINUS_EXPR twice (once for each argument),
5572 canonicalize operand order and handle just once, use rtype where
5573 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5576 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5577 any values of __C while still being pattern recognizable as a simple
5580 2017-10-13 Richard Biener <rguenther@suse.de>
5582 PR tree-optimization/82451
5584 2017-10-02 Richard Biener <rguenther@suse.de>
5586 PR tree-optimization/82355
5587 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5588 a mapping for the enclosing loop but avoid generating one for
5590 (copy_bb_and_scalar_dependences): Remove premature codegen
5591 error on PHIs in blocks duplicated into multiple places.
5592 * graphite-scop-detection.c
5593 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5594 in the region use it as loop and nest to analyze the DR in.
5595 (try_generate_gimple_bb): Likewise.
5596 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5597 (add_loop_constraints): For blocks in a loop not in the region
5598 create a dimension with a single iteration.
5599 * sese.h (gbb_loop_at_index): Remove assert.
5601 * cfgloop.c (loop_preheader_edge): For the loop tree root
5602 return the single successor of the entry block.
5603 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5604 Reset the SCEV hashtable and niters.
5605 * graphite-scop-detection.c
5606 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5607 assert that we only have POLYNOMIAL_CHREC that vary in loops
5608 contained in the region.
5609 (scop_detection::graphite_can_represent_expr): Adjust.
5610 (scop_detection::stmt_has_simple_data_refs_p): For loops
5611 not in the region set loop to NULL. The nest is now the
5612 entry edge to the region.
5613 (try_generate_gimple_bb): Likewise.
5614 * sese.c (scalar_evolution_in_region): Adjust for
5615 instantiate_scev change.
5616 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5617 Make nest parameter the edge into the region.
5618 (create_data_ref): Likewise.
5619 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5620 entry edge into a region and adjust instantiate_scev calls.
5621 (create_data_ref): Likewise.
5622 (graphite_find_data_references_in_stmt): Likewise.
5623 (find_data_references_in_stmt): Pass the loop preheader edge
5624 from the nest argument.
5625 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5626 parameter the edge into the region.
5627 (instantiate_parameters): Use the loop preheader edge as entry.
5628 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5630 (get_instantiated_value_entry): Make instantiate_below parameter
5631 the edge into the region.
5632 (instantiate_scev_name): Likewise. Adjust dominance checks,
5633 when we cannot use loop-based instantiation instantiate by
5634 walking use-def chains.
5635 (instantiate_scev_poly): Adjust.
5636 (instantiate_scev_binary): Likewise.
5637 (instantiate_scev_convert): Likewise.
5638 (instantiate_scev_not): Likewise.
5639 (instantiate_array_ref): Remove.
5640 (instantiate_scev_3): Likewise.
5641 (instantiate_scev_2): Likewise.
5642 (instantiate_scev_1): Likewise.
5643 (instantiate_scev_r): Do not blindly handle N-operand trees.
5644 Do not instantiate array-refs. Handle all constants and invariants.
5645 (instantiate_scev): Make instantiate_below parameter
5646 the edge into the region.
5647 (resolve_mixers): Use the loop preheader edge for the region
5648 parameter to instantiate_scev_r.
5649 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5651 2017-10-13 Richard Biener <rguenther@suse.de>
5653 PR tree-optimization/82525
5654 * graphite-isl-ast-to-gimple.c
5655 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5657 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5658 Fail code generation when we cannot represent the isl integer.
5659 (binary_op_to_tree): Elide modulo operations that are no-ops
5660 in the type we code generate. Remove now superfluous code
5663 2017-10-13 Richard Biener <rguenther@suse.de>
5665 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5666 (scop_detection::harmful_loop_in_region): Remove premature
5667 IV type restriction.
5668 (scop_detection::graphite_can_represent_scev): We can handle
5669 pointer IVs just fine.
5671 2017-10-13 Alan Modra <amodra@gmail.com>
5673 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5674 "Clobbers and Scratch Registers". Add paragraph on
5675 alternative to clobbers for scratch registers and OpenBLAS
5678 2017-10-13 Alan Modra <amodra@gmail.com>
5680 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5681 example of a memory input for a string of known length. Move
5682 commentary out of table. Add a number of new examples
5683 covering array memory inputs.
5685 2017-10-12 Martin Liska <mliska@suse.cz>
5687 PR tree-optimization/82493
5688 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5689 (test_range_functions): New function.
5690 (sbitmap_c_tests): Likewise.
5691 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5692 * selftest.h (sbitmap_c_tests): New function.
5694 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5696 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5698 * config/rs6000/amo.h: Fix spacing issue.
5700 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5703 * config/i386/i386.md (*ashl<mode>3_mask_1,
5704 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5705 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5708 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5710 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5711 (profile_probability): Set max_probability
5712 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5714 * profile-count.c (profile_probability::differs_from_p): Do not
5715 rely on max_probaiblity == 10000
5717 2017-10-12 Jeff Law <law@redhat.com>
5719 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5722 2017-10-12 Martin Sebor <msebor@redhat.com>
5726 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5727 (handle_alias_pairs): Call it.
5728 * common.opt (-Wattribute-alias): New option.
5729 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5730 * doc/invoke.texi (-Wattribute-alias): Document.
5732 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5735 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5737 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5739 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5740 (make_hard_regno_born, make_hard_regno_dead): Update
5741 bb_killed_pseudos and bb_gen_pseudos.
5743 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5745 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5747 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5749 * config/alpha/alpha.c (alpha_split_conditional_move):
5750 Use std::swap instead of manually swapping.
5751 (alpha_stdarg_optimize_hook): Ditto.
5752 (alpha_canonicalize_comparison): Ditto.
5754 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5756 * tree-loop-distribution.c (struct builtin_info): New struct.
5757 (struct partition): Refactor fields into struct builtin_info.
5758 (partition_free): Free struct builtin_info.
5759 (build_size_arg_loc, build_addr_arg_loc): Delete.
5760 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5761 information from struct builtin_info.
5762 (find_single_drs): New function refactored from classify_partition.
5763 Also moved builtin validity checks to this function.
5764 (compute_access_range, alloc_builtin): New functions.
5765 (classify_builtin_st, classify_builtin_ldst): New functions.
5766 (classify_partition): Refactor code into functions find_single_drs,
5767 classify_builtin_st and classify_builtin_ldst.
5768 (distribute_loop): Don't do runtime alias check when distributing
5770 (find_seed_stmts_for_distribution): New function.
5771 (pass_loop_distribution::execute): Refactor code finding seed
5772 stmts into above function. Support distribution for the innermost
5773 two-level loop nest. Adjust dump information.
5775 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5777 * tree-loop-distribution.c: Adjust the general comment.
5778 (NUM_PARTITION_THRESHOLD): New macro.
5779 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5780 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5781 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5782 in call to build_partition_graph.
5783 (finalize_partitions): New parameter. Make loop distribution more
5784 conservative by fusing more partitions.
5785 (distribute_loop): Don't do runtime alias check in case of loop nest
5787 (find_seed_stmts_for_distribution): New function.
5788 (prepare_perfect_loop_nest): New function.
5789 (pass_loop_distribution::execute): Refactor code finding seed stmts
5790 and loop nest into above functions. Support loop nest distribution.
5791 Adjust dump information accordingly.
5793 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5795 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5796 and set PTYPE_SEQUENTIAL for merged partition.
5798 2017-10-12 Richard Biener <rguenther@suse.de>
5800 PR tree-optimization/69728
5802 2017-09-19 Richard Biener <rguenther@suse.de>
5804 PR tree-optimization/69728
5805 * graphite-sese-to-poly.c (schedule_error): New global.
5806 (add_loop_schedule): Handle empty domain by failing the
5808 (build_original_schedule): Handle schedule_error.
5810 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5811 domain by returning an unchanged schedule.
5813 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5815 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5816 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5818 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5820 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5823 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5826 * expr.c (store_field): Don't optimize away bitsize == 0 store
5827 from CALL_EXPR with addressable return type.
5829 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5831 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5832 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5833 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5834 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5835 TARGET_ISEL instead of TARGET_ISEL<sel>.
5837 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5839 * config/rs6000/rs6000.c
5840 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5842 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5844 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5846 (reg_or_cint_operand): Fix comment.
5847 (reg_or_zero_operand): New predicate.
5848 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5849 * config/rs6000/rs6000.c (output_isel): Delete.
5850 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5851 instead of reg_or_cint_operand. Output instruction directly (not via
5853 (isel_unsigned_<mode>): Ditto.
5854 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5855 gpc_reg_operand. Add an instruction alternative for this. Output
5856 instruction directly.
5857 (*isel_reversed_unsigned_<mode>): Ditto.
5859 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5861 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5862 (TARGET_CANONICALIZE_COMPARISON): Define.
5864 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5867 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5868 Check whether the dest is REG before adding REG_EQUIV note.
5870 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5873 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5875 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5876 (make_hard_regno_born, make_hard_regno_dead): Update
5877 bb_killed_pseudos and bb_gen_pseudos.
5879 2017-10-11 Nathan Sidwell <nathan@acm.org>
5881 * incpath.h (enum incpath_kind): Name enum, prefix values.
5882 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5883 * incpath.c (heads, tails): Use INC_MAX.
5884 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5885 (merge_include_chains, split_quote_chain,
5886 register_include_chains): Update incpath_kind names.
5887 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5888 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5890 (add_framework_path, darwin_register_objc_includes): Likewise.
5891 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5893 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5895 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5896 Do not use float_operator operator predicate.
5897 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5898 * config/i386/predicates.md (float_operator): Remove predicate.
5900 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5902 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5903 (*jccxf_i387): Ditto.
5904 (*jcc<mode>_i387): Ditto.
5905 (*jccu<mode>_i387): Ditto.
5906 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5907 (*jcc_*_i387 splitters): Remove.
5908 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5909 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5910 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5913 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5915 * profile-count.h (slow_safe_scale_64bit): New function.
5916 (safe_scale_64bit): New inline.
5917 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5918 * profile-count.c: Include wide-int.h
5919 (slow_safe_scale_64bit): New.
5921 2017-10-11 Nathan Sidwell <nathan@acm.org>
5923 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5924 HAS_DECL_ASSEMBLER_NAME_P.
5925 * gimple-expr.c (gimple_decl_printable_name: Check
5926 HAS_DECL_ASSEMBLER_NAME_P too.
5927 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5928 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5929 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5930 * passes.c (rest_of_decl_compilation): Check
5931 HAS_DECL_ASSEMBLER_NAME_P too.
5932 * recog.c (verify_changes): Likewise.
5933 * tree-pretty-print.c (dump_decl_name): Likewise.
5934 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5936 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5937 (SET_DECL_ASSEMBLER_NAME): Use it.
5938 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5939 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5940 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5942 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5944 * config.gcc (i386, x86_64): Add extra objects.
5945 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5946 (ix86_min_insn_size): Declare.
5947 (ix86_issue_rate): Declare.
5948 (ix86_adjust_cost): Declare.
5949 (ia32_multipass_dfa_lookahead): Declare.
5950 (ix86_macro_fusion_p): Declare.
5951 (ix86_macro_fusion_pair_p): Declare.
5952 (ix86_bd_has_dispatch): Declare.
5953 (ix86_bd_do_dispatch): Declare.
5954 (ix86_core2i7_init_hooks): Declare.
5955 (ix86_atom_sched_reorder): Declare.
5956 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5957 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5958 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5959 (rip_relative_addr_p): Rename to ...
5960 (ix86_rip_relative_addr_p): ... this one; export.
5961 (memory_address_length): Update.
5962 (ix86_issue_rate): Move to x86-tune-sched.c.
5963 (ix86_flags_dependent): Move to x86-tune-sched.c.
5964 (ix86_agi_dependent): Move to x86-tune-sched.c.
5965 (exact_dependency_1): Move to x86-tune-sched.c.
5966 (exact_store_load_dependency): Move to x86-tune-sched.c.
5967 (ix86_adjust_cost): Move to x86-tune-sched.c.
5968 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5969 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5970 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5971 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5972 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5973 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5974 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5975 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5976 (min_insn_size): Rename to ...
5977 (ix86_min_insn_size): ... this one; export.
5978 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5979 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5980 (core2i7_first_cycle_multipass_backtrack): Move to
5981 x86-tune-sched-core.c.
5982 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5983 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5984 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5985 (ix86_avoid_jump_mispredicts): Update.
5986 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5987 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5988 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5989 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5990 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5991 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5992 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5993 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5994 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5995 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5996 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5997 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5998 (BIG): Move to ix86-tune-sched-bd.c.
5999 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6000 (enum insn_path): Move to ix86-tune-sched-bd.c.
6001 (get_mem_group): Move to ix86-tune-sched-bd.c.
6002 (is_cmp): Move to ix86-tune-sched-bd.c.
6003 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6004 (is_branch): Move to ix86-tune-sched-bd.c.
6005 (is_prefetch): Move to ix86-tune-sched-bd.c.
6006 (init_window): Move to ix86-tune-sched-bd.c.
6007 (allocate_window): Move to ix86-tune-sched-bd.c.
6008 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6009 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6010 (process_end_window): Move to ix86-tune-sched-bd.c.
6011 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6012 (find_constant): Move to ix86-tune-sched-bd.c.
6013 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6014 (has_immediate): Move to ix86-tune-sched-bd.c.
6015 (get_insn_path): Move to ix86-tune-sched-bd.c.
6016 (get_insn_group): Move to ix86-tune-sched-bd.c.
6017 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6018 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6019 (add_insn_window): Move to ix86-tune-sched-bd.c.
6020 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6021 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6022 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6023 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6024 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6025 (do_dispatch): Move to ix86-tune-sched-bd.c.
6026 (has_dispatch): Move to ix86-tune-sched-bd.c.
6027 * config/i386/t-i386: Add new object files.
6028 * config/i386/x86-tune-costs.h: New file.
6029 * config/i386/x86-tune-sched-atom.c: New file.
6030 * config/i386/x86-tune-sched-bd.c: New file.
6031 * config/i386/x86-tune-sched-core.c: New file.
6032 * config/i386/x86-tune-sched.c: New file.
6034 2017-10-11 Liu Hao <lh_mouse@126.com>
6036 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6037 the generic version below instead.
6038 (should_colorize): Recognize Windows consoles as terminals
6040 * pretty-print.c [__MINGW32__] (write_all): New function.
6041 [__MINGW32__] (find_esc_head): Likewise.
6042 [__MINGW32__] (find_esc_terminator): Likewise.
6043 [__MINGW32__] (eat_esc_sequence): Likewise.
6044 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6046 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6049 2017-10-11 Richard Biener <rguenther@suse.de>
6051 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6052 Properly call analyze_scalar_evolution with the loop of the stmt.
6054 2017-10-11 Richard Biener <rguenther@suse.de>
6056 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6057 * tree-core.h (tree_base): Add chrec_var union member.
6058 * tree.h (CHREC_VAR): Remove.
6059 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6060 * tree-chrec.h (build_polynomial_chrec): Adjust.
6061 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6062 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6064 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6066 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6067 * match.pd: ... here.
6068 ((T) X == (T) Y): Relax condition.
6070 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6072 PR tree-optimization/82472
6073 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6075 (break_alias_scc_partitions): Update postorder number.
6077 2017-10-11 Martin Liska <mliska@suse.cz>
6080 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6082 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6084 2017-10-11 Martin Liska <mliska@suse.cz>
6086 * print-rtl.c (print_insn): Move declaration of idbuf
6087 to same scope as name.
6089 2017-10-11 Martin Liska <mliska@suse.cz>
6094 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6097 2017-10-11 Martin Liska <mliska@suse.cz>
6100 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6103 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6105 * config.gcc (default_gnu_indirect_function): Default to yes for
6106 arm*-*-linux* with glibc.
6108 2017-10-11 Richard Biener <rguenther@suse.de>
6110 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6111 default-defs and types we do not want to analyze.
6112 (interpret_loop_phi): Replace unreachable code with an assert.
6113 (compute_scalar_evolution_in_loop): Remove and inline ...
6114 (analyze_scalar_evolution_1): ... here, replacing condition with
6115 what makes the intent clearer. Remove handling of cases
6116 get_scalar_evolution now handles.
6118 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6120 PR rtl-optimization/81434
6121 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6122 comment for main loop. In sched_group_found if, also add checks for
6123 pass and min_cost_group.
6125 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6127 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6128 (rs6000_insn_cost): New function.
6129 * config/rs6000/rs6000.md (cost): New attribute.
6131 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6132 H.J. Lu <hongjiu.lu@intel.com>
6136 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6137 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6138 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6139 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6140 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6141 OPTION_MASK_ISA_MMX and some other bit set require both
6142 mmx and the ISAs without the mmx bit.
6143 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6144 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6145 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6146 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6147 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6148 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6149 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6150 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6151 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6152 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6153 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6154 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6155 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6156 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6157 Add OPTION_MASK_ISA_MMX.
6159 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6161 * config.gcc (armv7*-*-freebsd*): New target.
6162 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6164 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6166 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6167 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6170 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6172 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6173 a template parameter.
6174 (WIDE_INT_REF_FOR): Update accordingly.
6175 * tree.h (wi::int_traits <const_tree>): Delete.
6176 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6177 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6178 (wi::tree_to_wide_ref): New typedef.
6179 (wi::to_wide): New function.
6180 * calls.c (get_size_range): Use wi::to_wide when operating on
6182 * cgraph.c (cgraph_node::create_thunk): Likewise.
6183 * config/i386/i386.c (ix86_data_alignment): Likewise.
6184 (ix86_local_alignment): Likewise.
6185 * dbxout.c (stabstr_O): Likewise.
6186 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6187 * expr.c (const_vector_from_tree): Likewise.
6188 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6189 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6190 (fold_negate_expr_1, int_const_binop_1, const_binop)
6191 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6192 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6193 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6194 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6195 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6196 (fold_not_const, round_up_loc): Likewise.
6197 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6198 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6199 (alloca_call_type): Likewise.
6200 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6201 * godump.c (go_output_typedef): Likewise.
6202 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6203 * internal-fn.c (get_min_precision): Likewise.
6204 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6205 * ipa-polymorphic-call.c
6206 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6207 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6208 (ipa_modify_call_arguments): Likewise.
6209 * match.pd: Likewise.
6210 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6211 * print-tree.c (print_node_brief, print_node): Likewise.
6212 * stmt.c (expand_case): Likewise.
6213 * stor-layout.c (layout_type): Likewise.
6214 * tree-affine.c (tree_to_aff_combination): Likewise.
6215 * tree-cfg.c (group_case_labels_stmt): Likewise.
6216 * tree-data-ref.c (dr_analyze_indices): Likewise.
6217 (prune_runtime_alias_test_list): Likewise.
6218 * tree-dump.c (dequeue_and_dump): Likewise.
6219 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6220 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6221 * tree-pretty-print.c (dump_generic_node): Likewise.
6222 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6223 (simple_iv_with_niters): Likewise.
6224 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6225 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6226 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6227 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6228 (refine_value_range_using_guard, number_of_iterations_ne_max)
6229 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6230 (get_cst_init_from_scev, record_nonwrapping_iv)
6231 (scev_var_range_cant_overflow): Likewise.
6232 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6233 * tree-ssa-pre.c (compute_avail): Likewise.
6234 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6235 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6236 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6237 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6238 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6239 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6240 (dump_case_nodes, try_switch_expansion): Likewise.
6241 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6242 (vect_do_peeling): Likewise.
6243 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6244 * tree-vect-stmts.c (vectorizable_load): Likewise.
6245 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6246 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6247 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6248 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6249 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6250 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6251 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6252 (evrp_dom_walker::before_dom_children): Likewise.
6253 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6254 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6255 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6256 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6257 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6258 (get_range_pos_neg): Likewise.
6259 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6260 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6261 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6262 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6263 * config/avr/avr.c (avr_fold_builtin): Likewise.
6264 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6265 * config/msp430/msp430.c (msp430_attr): Likewise.
6266 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6267 * config/powerpcspe/powerpcspe-c.c
6268 (altivec_resolve_overloaded_builtin): Likewise.
6269 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6270 (rs6000_expand_ternop_builtin): Likewise.
6271 * config/rs6000/rs6000-c.c
6272 (altivec_resolve_overloaded_builtin): Likewise.
6273 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6274 (rs6000_expand_ternop_builtin): Likewise.
6275 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6277 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6279 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6280 when copying loop nest with only one inner loop.
6282 2017-10-10 Richard Biener <rguenther@suse.de>
6284 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6285 blocks if SCEV is active.
6286 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6288 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6289 (scev_initialize): Assert we are not yet initialized.
6291 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6293 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6294 inner loop's exit stmt by making it always exit the loop, otherwise
6295 we would generate an infinite empty loop.
6297 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6299 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6300 renaming variables in new preheader if it's deleted.
6302 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6304 * tree-loop-distribution.c (struct partition): Remove unused field
6305 loops of the structure.
6306 (partition_alloc, partition_free): Ditto.
6307 (build_rdg_partition_for_vertex): Ditto.
6309 2017-10-09 Jeff Law <law@redhat.com>
6311 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6312 return type to match prototype and documentation.
6314 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6316 * config/rs6000/rs6000.c (processor_costs): Move to ...
6317 * config/rs6000/rs6000.h: ... here.
6318 (rs6000_cost): Declare.
6320 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6322 * except.c (setjmp_fn): New global variable.
6323 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6324 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6325 if DONT_USE_BUILTIN_SETJMP is defined.
6327 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6329 * target.def (insn_cost): New hook.
6330 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6331 * doc/tm.texi: Regenerate.
6332 * rtlanal.c (insn_cost): Use the new hook.
6334 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6336 * combine.c (combine_validate_cost): Compute the new insn_cost,
6337 not just pattern_cost.
6338 (try_combine): Adjust comment.
6340 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6342 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6344 * combine.c (uid_insn_cost): Adjust comment.
6345 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6347 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6348 * dse.c (find_shift_sequence): Ditto.
6349 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6350 (bb_valid_for_noce_process_p): Use pattern_cost.
6351 * rtl.h (insn_rtx_cost): Delete.
6352 (pattern_cost): New prototype.
6353 (insn_cost): New prototype.
6354 * rtlanal.c (insn_rtx_cost): Rename to...
6355 (pattern_cost): ... this.
6358 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6360 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6361 (*jcc<mode>_0_r_i387): Ditto.
6362 (*jccxf_r_i387): Ditto.
6363 (*jcc<mode>_r_i387): Ditto.
6364 (*jccu<mode>_r_i387): Ditto.
6365 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6366 (*jcc): Rename from *jcc_1.
6368 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6370 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6371 deferred rescans after the lvx/stvx recombination pre-pass.
6373 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6375 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6376 memory operation instruction support.
6377 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6378 (rs6000-ibm-aix[789]*): Likewise.
6379 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6380 Document new functions.
6382 2017-10-09 Richard Biener <rguenther@suse.de>
6384 PR tree-optimization/82397
6385 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6386 equality only for semantically equal trees.
6388 2017-10-09 Richard Biener <rguenther@suse.de>
6390 PR tree-optimization/82449
6391 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6392 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6393 allow constant addresses.
6394 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6397 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6399 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6402 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6405 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6408 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6411 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6413 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6416 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6417 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6419 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6421 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6422 (WI_BINARY_PREDICATE_RESULT): Likewise.
6423 (wi::binary_traits::operator_result): New type.
6424 (wi::binary_traits::predicate_result): Likewise.
6425 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6426 (generic_wide_int::operator==, generic_wide_int::operator!=)
6427 (generic_wide_int::operator&, generic_wide_int::and_not)
6428 (generic_wide_int::operator|, generic_wide_int::or_not)
6429 (generic_wide_int::operator^, generic_wide_int::operator+
6430 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6432 (operator~, unary operator-, operator==, operator!=, operator&)
6433 (operator|, operator^, operator+, binary operator-, operator*): New
6435 * expr.c (get_inner_reference): Use wi::bit_and_not.
6436 * fold-const.c (fold_binary_loc): Likewise.
6437 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6438 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6439 (bit_value_binop): Likewise.
6440 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6441 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6442 (extract_range_from_binary_expr_1): Likewise.
6443 (masked_increment): Likewise.
6444 (simplify_bit_ops_using_ranges): Likewise.
6446 2017-10-09 Martin Jambor <mjambor@suse.cz>
6449 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6450 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6451 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6453 (hsa_fixup_mov_insn_type): New function.
6454 (hsa_op_with_type::get_in_type): Use it.
6455 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6456 immediates in an assert.
6457 (hsa_op_with_type::extend_int_to_32bit): New method.
6458 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6459 types. Convert to dest type if necessary.
6460 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6461 (reg_for_gimple_ssa): Pass false as min32int to
6462 hsa_type_for_scalar_tree_type.
6463 (gen_hsa_addr): Fixup type when creating addresable temporary.
6464 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6465 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6466 necessary. Call hsa_fixup_mov_insn_type.
6467 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6468 extend operands and convert to dest type if necessary.
6469 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6470 to dest type if necessary.
6471 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6472 if conversion nt necessary and size matches.
6473 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6474 to dest type if necessary.
6475 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6476 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6478 (gen_hsa_clrsb): Likewise.
6479 (gen_hsa_ffs): Likewise.
6480 (gen_hsa_divmod): Extend operands and convert to dest type if
6482 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6484 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6486 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6487 Remove empty default arguments. Use a brace block as output
6489 (conditional return): Ditto.
6491 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6492 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6494 (group_ending_nop): Ditto.
6495 (doloop_end): Ditto.
6496 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6497 (splitters for those): Ditto.
6499 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6501 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6502 a conditional jump (and the compare for it) so that pc_rtx is the
6504 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6505 for the deleted and renamed ctr<mode>_internal[234] patterns.
6506 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6507 Delete second conditional return pattern.
6508 (ctr<mode>_internal2): Delete this second bdnz pattern.
6509 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6510 (ctr<mode>_internal4): Delete this second bdz pattern.
6512 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6514 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6515 (always_initialized_rtx_for_ssa_name_p): New predicate.
6516 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6517 (finish_out_of_ssa): Free new field of SA.
6518 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6519 * tree-ssa-coalesce.c: Include tree-ssa.h.
6520 (get_parm_default_def_partitions): Remove extern keyword.
6521 (get_undefined_value_partitions): New function.
6522 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6523 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6525 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6527 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6529 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6531 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6533 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6534 for targets that preffer 128bit.
6536 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6538 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6540 2017-10-08 Olivier Hainque <hainque@adacore.com>
6542 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6543 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6545 (thumb_set_return_address): Likewise.
6547 2017-10-08 Olivier Hainque <hainque@adacore.com>
6549 * common/config/arm/arm-common.c (arm_except_unwind_info):
6550 Handle DWARF2_UNWIND_INFO.
6552 2017-10-07 Michael Collison <michael.collison@arm.com>
6554 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6557 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6559 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6560 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6561 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6562 defined, force the creation of a new block for a dispatch label.
6564 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6566 * invoke.texi (Wsuggest-attribute=cold): Document.
6567 * common.opt (Wsuggest-attribute=cold): New
6568 * ipa-pure-const.c (warn_function_cold): New function.
6569 * predict.c (compute_function_frequency): Use it.
6570 * predict.h (warn_function_cold): Declare.
6572 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6574 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6577 2017-10-06 Martin Liska <mliska@suse.cz>
6579 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6580 keyword for member functions.
6581 (struct sanopt_tree_couple): New struct.
6582 (struct sanopt_tree_couple_hash): New function.
6583 (struct sanopt_ctx): Add new hash_map.
6584 (has_dominating_ubsan_ptr_check): New function.
6585 (record_ubsan_ptr_check_stmt): Likewise.
6586 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6587 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6588 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6590 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6593 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6594 aarch64_simd_valid_immediate on CONST_VECTORs.
6595 (aarch64_reg_or_bic_imm): Likewise.
6597 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6599 PR rtl-optimization/82396
6600 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6602 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6604 * graphite-dependences.c (scop_get_reads): Move code to...
6605 (scop_get_must_writes): Move code to...
6606 (scop_get_may_writes): Move code to...
6607 (scop_get_reads_and_writes): ... here.
6608 (scop_get_dependences): Call scop_get_reads_and_writes.
6610 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6612 PR tree-optimization/82434
6613 * fold-const.h (can_native_encode_type_p,
6614 can_native_encode_string_p): Remove.
6615 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6616 don't encode anything, just return what would be otherwise returned.
6617 (native_encode_fixed, native_encode_complex, native_encode_vector):
6619 (native_encode_string): Likewise. Inline by hand
6620 can_native_encode_string_p.
6621 (can_native_encode_type_p): Remove.
6622 (can_native_encode_string_p): Remove.
6623 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6624 STRING_CSTs using can_native_encode_string_p, test all
6625 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6626 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6627 argument from native_encode_expr.
6628 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6629 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6632 2017-10-06 Richard Biener <rguenther@suse.de>
6634 PR tree-optimization/82397
6635 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6636 operand_equal_p but rely on data_ref_compare_tree for detecting
6638 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6639 to match up with dr_group_sort_cmp.
6641 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6644 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6646 * config/s390/s390-builtin-types.def: Regenerate.
6648 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6651 * config/s390/s390-builtin-types.def: Regenerate.
6652 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6653 Change flag from B_VXE to B_VX.
6654 (s390_vec_min_dbl): Remove B_VXE flag.
6656 2017-10-06 Richard Biener <rguenther@suse.de>
6658 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6659 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6660 translate_isl_ast_to_gimple::is_valid_rename,
6661 translate_isl_ast_to_gimple::get_rename,
6662 translate_isl_ast_to_gimple::get_def_bb_for_const,
6663 translate_isl_ast_to_gimple::get_new_name,
6664 translate_isl_ast_to_gimple::collect_all_ssa_names,
6665 translate_isl_ast_to_gimple::copy_loop_phi_args,
6666 translate_isl_ast_to_gimple::collect_all_ssa_names,
6667 translate_isl_ast_to_gimple::copy_loop_phi_args,
6668 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6669 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6670 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6671 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6672 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6673 translate_isl_ast_to_gimple::copy_cond_phi_args,
6674 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6675 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6676 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6677 translate_isl_ast_to_gimple::rename_uses,
6678 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6679 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6680 (set_rename_for_each_def): Likewise.
6681 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6682 here. Handle rewriting SCEV analyzable uses here.
6683 (copy_bb_and_scalar_dependences): Generate code for PHI
6685 (graphite_regenerate_ast_isl): Adjust.
6686 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6687 (add_write, add_read): New functions.
6688 (build_cross_bb_scalars_def): Use it and simplify.
6689 (build_cross_bb_scalars_use): Likewise.
6690 (graphite_find_cross_bb_scalar_vars): Inline into...
6691 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6692 simulating out-of-SSA. Compute liveout and add dependencies.
6693 (build_scops): Force an empty entry block.
6694 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6696 (sese_build_liveouts): Declare.
6697 (sese_trivially_empty_bb_p): Likewise.
6698 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6699 compute liveout and debug_liveout.
6700 (sese_bad_liveouts_use): Remove.
6701 (sese_reset_debug_liveouts_bb): Likewise.
6702 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6703 (sese_build_liveouts): Build liveout and debug_liveout and store
6705 (new_sese_info): Adjust.
6706 (free_sese_info): Likewise.
6707 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6708 do not build liveout here.
6709 (move_sese_in_condition): Adjust region entry.
6710 (scev_analyzable_p): Match up with chrec_apply requirements.
6711 (sese_trivially_empty_bb_p): New.
6712 * tree-into-ssa.c (get_reaching_def): Properly support generating
6713 default-defs for incremental rewrite of anonymous names.
6715 2017-10-06 Richard Biener <rguenther@suse.de>
6717 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6718 precision do not perform modulo reduction.
6720 2017-10-06 Richard Biener <rguenther@suse.de>
6722 PR tree-optimization/82436
6723 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6724 conservatively choose the vectorization factor when checking
6725 whether we can perform the required load permutation.
6726 (vect_transform_slp_perm_load): Assert when we may not fail.
6728 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6730 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6731 message for incompatible -msdata=* and -mcall-* options.
6733 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6735 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6736 rate for post-reload scheduling.
6738 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6740 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6742 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6744 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6745 to improve monte carlo in scimark.
6747 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6749 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6750 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6751 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6752 pentium4_cost, nocona_cost): Set reassociation width to 1.
6753 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6754 width to 2 for fp operations and 1 otherwise.
6755 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6757 (atom_cost): Set reassociation width to 2.
6758 (slm_cost, generic_cost): Set fp reassociation width
6759 to 2 and 1 otherwise.
6760 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6761 (core_cost): Set fp reassociation width to 4 and vector to 2.
6762 (ix86_reassociation_width): Rewrite using cost table; special case
6763 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6764 and TARGET_AVX128_OPTIMAL.
6765 * config/i386/i386.h (processor_costs): Add
6766 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6767 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6768 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6769 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6770 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6771 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6773 2017-10-05 Nathan Sidwell <nathan@acm.org>
6775 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6777 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6779 * config/arm/arm.c (arm_test_fpu_data): New.
6780 (arm_run_selftests): Call arm_test_fpu_data.
6782 2017-10-04 Nathan Sidwell <nathan@acm.org>
6784 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6786 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6788 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6790 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6791 check type for aarch64_simd_valid_immediate.
6792 (aarch64_output_simd_mov_immediate): Update prototype.
6793 (aarch64_simd_valid_immediate): Update prototype.
6794 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6795 support for ORR-immediate.
6796 (and<mode>3): modified pattern to add support for BIC-immediate.
6797 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6798 now checks for valid immediate for BIC and ORR based on new enum
6800 (aarch64_output_simd_mov_immediate): Function now used to output
6801 BIC/ORR imm as well based on new enum argument.
6802 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6804 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6805 (aarch64_reg_or_bic_imm): Likewise.
6807 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6809 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6810 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6812 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6816 PR rtl-optimization/82396
6817 * haifa-sched.c (autopref_multipass_init): Simplify
6819 (autopref_rank_data): Simplify sort order.
6820 * sched-int.h (autopref_multipass_data_): Remove
6821 multi_mem_insn_p, min_offset and max_offset.
6823 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6825 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6827 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6829 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6832 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6834 * doc/sourcebuild.texi: Document vect_long_mult.
6836 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6838 PR tree-optimization/82413
6839 * fold-const.c (build_range_check): Use widest_int when comparing
6840 the maximum ETYPE value with HIGH.
6842 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6844 PR rtl-optimization/82396
6845 * haifa-sched.c (autopref_multipass_init): Simplify
6847 (autopref_rank_data): Simplify sort order.
6848 * sched-int.h (autopref_multipass_data_): Remove
6849 multi_mem_insn_p, min_offset and max_offset.
6851 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6853 PR tree-optimization/82381
6854 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6855 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6858 PR tree-optimization/82374
6859 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6860 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6861 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6862 current_function_decl to the new decl.
6864 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6866 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6867 helper macro for IEEE float128 hardware built-in functions.
6868 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6870 (TRUNCF128_ODD): Likewise.
6871 (ADDF128_ODD): Likewise.
6872 (SUBF128_ODD): Likewise.
6873 (MULF128_ODD): Likewise.
6874 (DIVF128_ODD): Likewise.
6875 (FMAF128_ODD): Likewise.
6876 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6877 UNSPEC_TRUNC_ROUND_TO_ODD.
6878 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6879 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6880 floating point round to odd instructions.
6881 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6882 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6883 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6884 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6885 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6886 (trunc<mode>sf2_hw): Change the truncate with round to odd
6887 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6888 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6889 to odd hardware instructions.
6890 (sub<mode>3_odd): Likewise.
6891 (mul<mode>3_odd): Likewise.
6892 (div<mode>3_odd): Likewise.
6893 (sqrt<mode>2_odd): Likewise.
6894 (fma<mode>4_odd): Likewise.
6895 (fms<mode>4_odd): Likewise.
6896 (nfma<mode>4_odd): Likewise.
6897 (nfms<mode>4_odd): Likewise.
6898 (trunc<mode>df2_odd): Change the truncate with round to odd
6899 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6901 * doc/extend.texi (PowerPC built-in functions): Update documentation
6902 for existing IEEE float128-bit built-in functions. Add built-in
6903 functions that generate the IEEE 128-bit floating point round to
6906 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6908 PR rtl-optimization/77729
6909 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6910 to (X&(C1&~C2))|C2 transformations.
6912 2017-10-03 Martin Jambor <mjambor@suse.cz>
6914 PR tree-optimization/82363
6915 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6916 mismatch, mark lacc written regardless of racc.
6918 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6920 PR tree-optimization/82381
6921 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6922 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6923 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6924 fallthrough into reassoc_stmt_dominates_stmt_p.
6927 * combine.c (combine_instructions): Don't combine in unreachable
6930 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6933 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6934 function to not use the have_cpu variable. Do not set cpu_index,
6935 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6937 (rs6000_valid_attribute_p): Remove duplicate initializations of
6938 old_optimize and func_optimize.
6939 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6940 (rs6000_activate_target_options): Make global.
6941 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6944 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6946 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6947 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6948 Set *poffset to 0 and *psize and *pmax_size to -1 if
6949 *poffset + *psize overflows in HOST_WIDE_INT.
6951 PR tree-optimization/82387
6952 PR tree-optimization/82388
6953 PR tree-optimization/82389
6954 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6955 instead of live_bytes non-NULL.
6957 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6960 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6963 2017-10-02 Richard Biener <rguenther@suse.de>
6965 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6966 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6967 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6969 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6971 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6972 requested precision matches the type's.
6973 * calls.c (alloc_max_size): Calculate the new candidate size as
6974 a widest_int and use wi::to_widest when comparing it with the
6975 current candidate size.
6976 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6977 zero rather than integer_zero_node.
6978 * match.pd: Check for a no-op conversion before using wi::add
6979 rather than after. Use tree_to_uhwi when summing small shift
6980 counts into an unsigned int.
6982 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6983 Alan Hayward <alan.hayward@arm.com>
6984 David Sherwood <david.sherwood@arm.com>
6987 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6988 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6989 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6990 POINTER_AND_FP_REGS.
6992 2017-10-02 Richard Biener <rguenther@suse.de>
6994 PR tree-optimization/82355
6995 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6996 a mapping for the enclosing loop but avoid generating one for
6998 (copy_bb_and_scalar_dependences): Remove premature codegen
6999 error on PHIs in blocks duplicated into multiple places.
7000 * graphite-scop-detection.c
7001 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7002 in the region use it as loop and nest to analyze the DR in.
7003 (try_generate_gimple_bb): Likewise.
7004 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7005 (add_loop_constraints): For blocks in a loop not in the region
7006 create a dimension with a single iteration.
7007 * sese.h (gbb_loop_at_index): Remove assert.
7009 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7011 * omp-expand.c (adjust_context_scope): New function.
7012 (expand_parallel_call): Call adjust_context_scope.
7014 2017-10-01 Jeff Law <law@redhat.com>
7016 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7017 dom_opt_dom_walker class with direct access to private members.
7018 Add comments. Call test_for_singularity.
7019 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7020 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7021 m_dummy_cond anymore.
7022 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7024 (pass_dominator:execute): Build the dummy_cond here and pass it
7025 to the dom_opt_dom_walker ctor.
7026 (test_for_singularity): New function.
7028 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7029 Maya Rashish <coypu@sdf.org>
7031 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7032 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7033 (alpha*-*-netbsd*) Use nbsd_tm_file.
7034 (arm*-*-netbsdelf*) Likewise.
7035 (i[34567]86-*-netbsdelf*) Likewise.
7036 (x86_64-*-netbsd*) Likewise.
7037 (mips*-*-netbsd*) Likewise.
7038 (powerpc-*-netbsd*) Likewise.
7039 (sh*-*-netbsd*) Likewise.
7040 (sparc-*-netbsdelf*) Likewise.
7041 (sparc64-*-netbsd*) Likewise.
7042 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7044 (vax-*-netbsdelf*) Likewise.
7045 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7046 (UINT_FAST8_TYPE) Likewise.
7047 (INT_FAST16_TYPE) Check CHAR_FAST16.
7048 (UINT_FAST16_TYPE) Likewise.
7050 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7053 * config/i386/i386.md
7054 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7055 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7056 *divmodsi4_zext_2): New define_insn_and_split.
7057 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7058 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7059 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7060 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7061 New define_insn_and_split.
7062 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7063 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7064 operands[1] having DImode when mode is SImode.
7066 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7067 always SImode for DIV and MOD in REG_EQUAL notes.
7069 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7072 * match.pd: Fix handling of NaNs in pattern.
7074 2017-09-29 Jeff Law <law@redhat.com>
7076 * sbitmap.c (bitmap_bit_in_range_p): New function.
7077 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7078 * tree-ssa-dse.c (live_bytes_read): New function.
7079 (dse_classify_store): Ignore reads of dead bytes.
7081 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7082 typos and whitespace errors.
7083 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7084 * config/s390/s390.c (s390_emit_prologue): Likewise.
7086 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7089 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7090 with a symbol for LRA.
7092 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7094 PR rtl-optimization/82338
7095 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7097 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7099 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7100 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7101 (qsort_chk): Declare.
7102 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7103 (qsort_chk): New function.
7105 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7107 PR tree-optimization/82337
7108 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7109 phi definition if the PHI result appears in an abnormal PHI.
7110 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7111 basis appears in an abnormal PHI.
7113 2017-09-29 Richard Biener <rguenther@suse.de>
7115 * graphite-isl-ast-to-gimple.c
7116 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7117 (binary_op_to_tree): Use it.
7118 (get_rename_from_scev): Likewise.
7119 (copy_loop_phi_nodes): Likewise.
7120 (copy_bb_and_scalar_dependences): Likewise.
7121 (translate_pending_phi_nodes): Likewise.
7123 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7126 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7127 for movabsq $(i32 << shift), r64.
7129 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7131 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7132 index when encoding %esp as %rsp to avoid 0x67 prefix.
7134 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7136 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7137 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7139 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7141 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7142 Extensions with more than 16 double VFP registers.
7143 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7144 to_clear_mask and all code related to it. Replace the remaining
7145 entry by a sbitmap and adapt code accordingly.
7147 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7149 * brig-builtins.def: Change pure attributes to const.
7151 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7153 * config.gcc (default_gnu_indirect_function): Default to yes for
7154 sparc*-*-linux* with glibc.
7156 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7158 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7159 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7160 when creating .init_array and .fini_array sections with priority
7163 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7166 * config/aarch64/aarch64.c
7167 (aarch64_builtin_support_vector_misalignment): Always return false
7168 when misalignment is unknown.
7170 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7172 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7173 this function to return false if the definition used by the swap
7174 instruction is artificial, or if the memory address from which the
7175 constant value is loaded is not represented by a base address held
7176 in a register or if the base address register is a frame or stack
7177 pointer. Additionally, return false if the base address of the
7178 loaded constant is a SYMBOL_REF but is not considered to be a
7180 (replace_swapped_load_constant): New function.
7181 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7182 loaded constant vector with a load of a swapped constant vector.
7184 2017-09-27 Carl Love <cel@us.ibm.com>
7186 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7187 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7188 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7191 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7193 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7194 first, always call autopref_rank_data otherwise.
7196 2017-09-27 Richard Biener <rguenther@suse.de>
7198 * graphite-scop-detection.c (find_scop_parameters): Move
7199 loop bound handling ...
7200 (gather_bbs::before_dom_children): ... here, avoiding the need
7201 to build scop_info->loop_nest.
7202 (record_loop_in_sese): Remove.
7203 * sese.h (sese_info_t::loop_nest): Remove.
7204 * sese.c (new_sese_info): Do not allocate loop_nest.
7205 (free_sese_info): Do not free loop_nest.
7207 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7210 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7211 lhs from calls if the lhs has addressable type.
7213 2017-09-27 Richard Biener <rguenther@suse.de>
7215 * graphite.h (scop::max_alias_set): New member.
7216 * graphite-scop-detection.c: Remove references to non-existing
7217 --param in comments.
7218 (build_alias_sets): Record the maximum alias set used for drs.
7219 (build_scops): Support zero as unlimited for
7220 --param graphite-max-arrays-per-scop.
7221 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7223 (build_poly_sr_1): ... here. Compute alias set based on the
7224 maximum alias set used for drs rather than
7225 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7227 2017-09-27 Richard Biener <rguenther@suse.de>
7229 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7230 --param loop-block-tile-size=0 to disable tiling.
7232 2017-09-27 Richard Biener <rguenther@suse.de>
7234 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7235 (graphite-max-nb-scop-params): Document special value zero.
7236 * domwalk.h (dom_walker::STOP): New symbolical constant.
7237 (dom_walker::dom_walker): Add optional parameter for bb to
7239 (dom_walker::~dom_walker): Declare.
7240 (dom_walker::before_dom_children): Document STOP return value.
7241 (dom_walker::m_user_bb_to_rpo): New member.
7242 (dom_walker::m_bb_to_rpo): Likewise.
7243 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7244 mapping here if not provided by the user.
7245 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7246 provided by the user.
7247 (dom_walker::STOP): Define.
7248 (dom_walker::walk): Do not compute bb to RPO mapping here.
7249 Support STOP return value from before_dom_children to stop
7251 * graphite-optimize-isl.c (optimize_isl): If the schedule
7252 is the same still generate code if -fgraphite-identity
7253 or -floop-parallelize-all are given.
7254 * graphite-scop-detection.c: Include cfganal.h.
7255 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7257 (gather_bbs::before_dom_children): Return STOP for BBs
7259 (build_scops): Compute bb to RPO mapping and pass it to
7260 the domwalk. Treat --param graphite-max-nb-scop-params=0
7261 as not limiting the number of params.
7262 * graphite.c (graphite_initialize): Remove limit on the
7263 number of basic-blocks in a function.
7264 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7265 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7266 default value of 10.
7268 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7270 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7271 Adjust code to eliminate needing to do the shift right 32-bits
7272 operation after XSCVDPSPN.
7274 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7276 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7277 ((X / Y) != 0 -> X >= Y): Likewise.
7279 2017-09-26 Carl Love <cel@us.ibm.com>
7281 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7282 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7283 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7284 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7285 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7286 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7287 definitions and overloading.
7288 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7289 statement for P9V_BUILTIN_XST_LEN_R.
7290 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7291 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7292 define_expand and define_insn for the instructions and builtins.
7293 * doc/extend.texi: Update the built-in documentation file for the new
7295 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7296 define_insn for the instructions
7298 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7301 * gcc/config/netbsd-protos.h: New file.
7302 * gcc/config/netbsd.c: New file.
7303 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7304 * gcc/config/t-netbsd: New file.
7305 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7306 (tmake_file) Add t-netbsd.
7307 (extra_objs) Add netbsd.o.
7309 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7313 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7315 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7317 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7318 sign extension from a vector register to a GPR by doing a 32-bit
7319 direct move and then an EXTSW.
7320 (extendsi<mode>2 splitter): Likewise.
7321 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7322 right or vector extract after doing XSCVDPSPN. Use
7323 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7325 (movdi_from_sf_zero_ext): Likewise.
7326 (reload_gpr_from_vsxsf): Likewise.
7327 (p8_mfvsrd_4_disf): Delete, no longer used.
7328 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7329 then needing to move the SFmode to a GPR to use the XSCVDPSP
7330 instruction instead of FRSP and XSCVDPSPN.
7331 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7332 it is adjacent to the other XSCVSPDP insns.
7333 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7334 SFmode to be in traditional Altivec registers.
7335 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7336 (vsx_xscvspdpn): Likewise.
7337 (vsx_xscvspdpn_scalar): Likewise.
7339 2017-09-26 Martin Jambor <mjambor@suse.cz>
7341 * tree-sra.c (compare_access_positions): Put integral types first,
7342 stabilize sorting of integral types, remove conditions putting
7343 non-full-precision integers last.
7344 (sort_and_splice_var_accesses): Disable scalarization if a
7345 non-integert would be represented by a non-full-precision integer.
7347 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7349 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7350 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7351 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7352 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7353 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7354 conditionals inside the function instead of around it. Call
7355 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7356 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7358 2017-09-26 Richard Biener <rguenther@suse.de>
7360 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7362 (scop_detection::build_scop_breadth): ... this. Removed.
7363 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7364 (scop_detection::harmful_stmt_in_bb): Likewise.
7365 (scop_detection::graphite_can_represent_stmt): Likewise.
7366 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7367 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7368 (scop_detection::can_represent_loop_1): ... this. Removed.
7369 (scop_detection::harmful_loop_in_region): Simplify after inlining
7370 the above and remove more quadraticness.
7371 (build_scops): Adjust.
7372 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7375 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7378 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7379 REGNO (base) == SP_REG if base is a REG.
7382 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7383 if it is different SSA_NAME.
7384 (optimize_range_tests_cmp_bitwise): New function.
7385 (optimize_range_tests): Call it.
7387 2017-09-26 Richard Biener <rguenther@suse.de>
7389 PR tree-optimization/82321
7390 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7391 for the def being inside the loop.
7393 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7395 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7397 * config/s390/s390-builtins.def: Fix constraint on op4.
7399 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7401 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7402 independent expanders.
7403 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7404 ("vec_ordered", "vec_unordered"): New expanders.
7406 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7408 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7411 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7413 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7414 vec_unpacks_lo_v16qi.
7415 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7417 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7419 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7420 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7421 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7423 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7425 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7427 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7429 ("*vec_slb<mode>"): New insn pattern.
7430 ("vec_shr_<mode>"): New expander.
7431 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7432 and force the shift count operand to V16QImode.
7433 ("vec_srb<mode>"): Set shift count mode to V16QI.
7435 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7437 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7438 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7439 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7441 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7444 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7446 2017-09-26 Richard Biener <rguenther@suse.de>
7448 PR tree-optimization/82320
7449 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7452 2017-09-25 Jeff Law <law@redhat.com>
7454 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7455 prototype for new argument.
7456 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7457 mostly extracted from rs6000_emit_allocate_stack.
7458 (rs6000_emit_probe_stack_range_stack_clash): New function.
7459 (rs6000_emit_allocate_stack): Call
7460 rs6000_emit_probe_stack_range_stack_clash as needed.
7461 (rs6000_emit_probe_stack_range): Add additional argument
7462 to call to gen_probe_stack_range{si,di}.
7463 (output_probe_stack_range): New.
7464 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7465 (output_probe_stack_range_stack_clash): New.
7466 (rs6000_emit_prologue): Emit notes into dump file as requested.
7467 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7468 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7469 Add additional operand and pass it to output_probe_stack_range.
7471 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7473 PR tree-optimization/82163
7474 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7475 (checking_verify_loop_closed_ssa): New parameter.
7476 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7477 (check_loop_closed_ssa_stmt): Delete.
7478 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7479 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7480 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7483 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7485 * brig-builtins.def: Treat HSAIL barrier builtins as
7486 setjmp/longjump style functions.
7488 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7490 * target.def (constant_alignment): New hook.
7491 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7492 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7493 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7494 * doc/tm.texi: Regenerate.
7495 * targhooks.h (default_constant_alignment): Declare.
7496 (constant_alignment_word_strings): Likewise.
7497 * targhooks.c (default_constant_alignment): New function.
7498 (constant_alignment_word_strings): Likewise.
7499 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7500 instead of CONSTANT_ALIGNMENT.
7501 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7502 (force_const_mem): Likewise.
7503 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7504 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7505 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7506 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7507 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7509 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7510 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7511 constant_alignment_word_strings.
7512 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7513 (CONSTANT_ALIGNMENT): Likewise.
7514 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7515 (arm_constant_alignment): New function.
7516 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7517 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7518 constant_alignment_word_strings.
7519 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7520 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7521 constant_alignment_word_strings.
7522 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7523 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7524 (cris_constant_alignment): New function.
7525 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7526 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7527 (epiphany_constant_alignment): New function.
7528 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7529 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7530 constant_alignment_word_strings.
7531 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7532 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7533 constant_alignment_word_strings.
7534 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7535 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7536 constant_alignment_word_strings.
7537 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7538 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7539 * config/i386/i386.c (ix86_constant_alignment): Make static.
7540 Use the same interface as the target hook.
7541 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7542 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7543 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7544 constant_alignment_word_strings.
7545 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7546 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7547 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7548 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7549 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7550 constant_alignment_word_strings.
7551 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7552 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7553 constant_alignment_word_strings.
7554 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7555 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7556 constant_alignment_word_strings.
7557 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7558 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7560 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7561 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7562 * config/mips/mips.c (mips_constant_alignment): New function.
7563 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7564 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7565 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7566 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7567 (mmix_constant_alignment): Make static. Use the same interface
7569 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7570 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7571 constant_alignment_word_strings.
7572 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7573 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7574 constant_alignment_word_strings.
7575 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7576 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7577 constant_alignment_word_strings.
7578 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7579 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7580 (rs6000_constant_alignment): New function.
7581 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7582 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7583 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7584 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7585 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7586 (rs6000_constant_alignment): New function.
7587 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7588 * config/s390/s390.c (s390_constant_alignment): New function.
7589 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7590 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7591 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7592 constant_alignment_word_strings.
7593 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7594 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7595 (sparc_constant_alignment): New function.
7596 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7597 * config/spu/spu.c (spu_constant_alignment): New function.
7598 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7599 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7600 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7601 constant_alignment_word_strings.
7602 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7603 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7604 constant_alignment_word_strings.
7605 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7606 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7607 constant_alignment_word_strings.
7608 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7609 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7610 (visium_constant_alignment): New function.
7611 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7612 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7613 (xtensa_constant_alignment): New function.
7614 * system.h (CONSTANT_ALIGNMENT): Poison.
7616 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7618 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7619 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7620 (rs6000_builtin_valid_without_lhs): New helper function.
7621 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7622 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7624 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7626 * target.h (vec_perm_indices): Use unsigned short rather than
7628 (auto_vec_perm_indices): Likewise.
7629 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7630 Use unsigned int rather than unsigned char.
7631 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7633 2017-09-25 Richard Biener <rguenther@suse.de>
7635 * cfgloop.h (sort_sibling_loops): Declare.
7636 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7637 (sort_sibling_loops): New function sorting the sibling loop list
7639 * graphite.c (graphite_transform_loops): Sort sibling loops.
7641 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7643 * target.def (vec_perm_const_ok): Change sel parameter to
7645 * optabs-query.c (can_vec_perm_p): Update accordingly.
7646 * doc/tm.texi: Regenerate.
7647 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7648 auto_vec_perm_indices and remove separate nelt field.
7649 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7650 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7651 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7652 (aarch64_expand_vec_perm_const): Update accordingly.
7653 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7654 to vec_perm_indices.
7655 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7656 auto_vec_perm_indices and remove separate nelt field.
7657 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7658 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7659 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7661 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7662 to vec_perm_indices.
7663 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7664 sel to vec_perm_indices.
7665 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7666 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7667 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7669 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7672 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7675 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7676 on the FUNCTION_DECL function context if it has a DIE that is a
7679 2017-09-25 Richard Biener <rguenther@suse.de>
7681 PR tree-optimization/82285
7682 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7685 2017-09-25 Tom de Vries <tom@codesourcery.com>
7689 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7692 2017-09-25 Richard Biener <rguenther@suse.de>
7694 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7695 ISL errors other than isl_error_quota happen. Dump if the
7696 schedule is the same.
7697 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7698 errors instead of aborting inside ISL.
7700 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7703 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7704 of libgcc_eh for m64.
7705 * config/i386/darwin64.h: Likewise.
7707 2017-09-25 Richard Biener <rguenther@suse.de>
7710 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7711 attribute for incomplete types nor twice for complete ones.
7713 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7716 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7717 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7720 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7723 * config/i386/i386.opt (mprefer-avx256): Use
7724 ix86_target_flags variable.
7725 * config/i386/i386.c (ix86_target_string): Move
7726 -mprefer-avx256 to flag2_opts.
7728 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7731 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7732 and x != -1 | y != -1 into (x & y) != -1.
7734 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7736 * config.gcc: Add new case statement to set
7737 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7738 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7739 s390x-*-linux* case statements. Added aarch64 to the list of
7740 supported architectures.
7742 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7744 PR tree-optimization/82289
7745 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7746 STMT_VINFO_RELEVANT_P.
7748 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7749 Alan Hayward <alan.hayward@arm.com>
7750 David Sherwood <david.sherwood@arm.com>
7752 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7753 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7754 (extract_range_from_binary_expr_1): Don't call
7755 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7757 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7758 Alan Hayward <alan.hayward@arm.com>
7759 David Sherwood <david.sherwood@arm.com>
7761 * target.def (preferred_vector_alignment): New hook.
7762 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7764 * doc/tm.texi: Regenerate.
7765 * targhooks.h (default_preferred_vector_alignment): Declare.
7766 * targhooks.c (default_preferred_vector_alignment): New function.
7767 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7769 (DR_TARGET_ALIGNMENT): New macro.
7770 (aligned_access_p): Update commentary.
7771 (vect_known_alignment_in_bytes): New function.
7772 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7774 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7775 Calculate the misalignment based on the target alignment rather than
7777 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7778 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7779 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7780 the target alignment, rather than masking the element misalignment
7781 with the number of elements in a vector. Also use the target
7782 alignment when calculating the maximum number of peels.
7783 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7784 instead of TYPE_ALIGN_UNIT.
7785 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7786 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7787 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7788 (vect_create_data_ref_ptr): Likewise.
7789 (vect_setup_realignment): Realign by ANDing with
7790 -DR_TARGET_MISALIGNMENT.
7791 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7792 the number of peels based on DR_TARGET_ALIGNMENT.
7793 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7794 with the guaranteed alignment boundary when deciding whether
7796 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7797 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7798 (ensure_base_align): Remove stmt_info parameter. Get the
7799 target base alignment from DR_TARGET_ALIGNMENT.
7800 (vectorizable_store): Update call accordingly. Interpret
7801 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7803 (vectorizable_load): Likewise.
7805 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7806 Alan Hayward <alan.hayward@arm.com>
7807 David Sherwood <david.sherwood@arm.com>
7809 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7810 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7811 (vect_enhance_data_refs_alignment): Likewise.
7813 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7815 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7816 error. Only quit immediately if parsing is complete.
7817 (BEGIN): Initialize fatal_err and parse_done.
7818 (begin fpu, end fpu): Check number of arguments.
7819 (begin arch, end arch): Likewise.
7820 (begin cpu, end cpu): Likewise.
7821 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7822 (optalias): Likewise.
7824 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7826 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7827 * config/arm/arm-isa.h: Delete. Move definitions to ...
7828 * arm-cpus.in: ... here. Use new feature and fgroup values.
7829 * config/arm/arm.c (arm_option_override): Use lower case for feature
7831 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7832 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7833 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7835 (print_isa_bits_for): New function.
7836 (gen_isa): New function.
7837 (gen_comm_data): Use print_isa_bits_for.
7838 (define feature): New keyword.
7839 (define fgroup): New keyword.
7840 * config/arm/t-arm (TM_H): Remove.
7841 (GTM_H): Add arm-isa.h.
7842 (arm-isa.h): Add rule to generate file.
7843 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7844 case for feature bit names.
7846 2017-09-22 Richard Biener <rguenther@suse.de>
7848 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7850 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7851 print of no dependency loops ...
7852 * graphite.c (graphite_transform_loops): ... here.
7853 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7855 (same_close_phi_node, remove_duplicate_close_phi,
7856 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7857 (canonicalize_loop_closed_ssa): ... here and simplify.
7858 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7859 (optimize_isl): Use dump_printf_loc to tell when we stopped
7860 optimizing because of an ISL timeout.
7862 2017-09-22 Richard Biener <rguenther@suse.de>
7864 PR tree-optimization/82291
7865 * tree-if-conv.c (predicate_mem_writes): Make sure to
7866 remove writes in blocks predicated with false.
7868 2017-09-22 Richard Biener <rguenther@suse.de>
7870 * sese.c: Include cfganal.h.
7871 (if_region_set_false_region): Remove.
7872 (create_if_region_on_edge): Likewise.
7873 (move_sese_in_condition): Re-implement without destroying
7876 2017-09-22 Richard Biener <rguenther@suse.de>
7878 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7879 Verify both BBs contain loop PHI nodes before dispatching to
7881 (graphite_regenerate_ast_isl): Do not recompute dominators,
7882 do not verify three times. Restructure for clarity.
7883 * graphite-scop-detection.c (same_close_phi_node,
7884 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7885 defined_in_loop_p, canonicalize_loop_closed_ssa,
7886 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7887 checking and SSA rewrite, move to ...
7888 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7889 (graphite_initialize): Do not pass in ctx, do not reset the
7890 SCEV cache, compute only dominators.
7891 (graphite_transform_loops): Allocate ISL ctx after
7892 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7893 Maintain post-dominators only around build_scops.
7894 * sese.c (if_region_set_false_region): Make static. Free
7895 and recompute dominators.
7896 (move_sese_in_condition): Assert we don't get called with
7897 post-dominators computed.
7898 * sese.h (if_region_set_false_region): Remove.
7900 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7902 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7903 mode attribute for TARGET_AVX512VL.
7905 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7907 * config/i386/i386.opt (mprefer-avx256): New option.
7908 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7910 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7911 for TARGET_PREFER_AVX256.
7912 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7914 2017-09-21 Jeff Law <law@redhat.com>
7916 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7917 Fix dump output if the only stack space is for pushed registers.
7919 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7921 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7924 2017-09-21 Martin Sebor <msebor@redhat.com>
7927 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7928 code (in C++) or code that triggers warnings.
7930 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7932 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7934 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7936 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7937 * sched-rgn.c: Ditto.
7938 * sel-sched-ir.c: Ditto.
7940 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7942 * toplev.h (set_random_seed): Adjust return type.
7943 * toplev.c (init_local_tick): Move eager initialization of random_seed
7944 to get_random_seed. Adjust comment.
7945 (init_random_seed): Inline to get_random_seed, delete.
7946 (get_random_seed): Initialize random_seed lazily.
7947 (set_random_seed): Do not return previous value.
7948 (print_switch_value): Do not call get_random_seed.
7950 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7952 * cgraph.c (delete_function_version): New, broken out from...
7953 (cgraph_node::delete_function_version): ...here. Rename to
7954 cgraph_node::delete_function_version_by_decl. Update all uses.
7955 (cgraph_node::remove): Call delete_function_version.
7957 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7960 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7961 VAR_DECLs to which addressable non-volatile parameters are mapped
7962 and for id->retvar after the return value assignment. Clear
7963 id->retval and id->retbnd after inlining.
7965 2017-09-21 Richard Biener <rguenther@suse.de>
7967 PR tree-optimization/82276
7968 PR tree-optimization/82244
7969 * tree-vrp.c (build_assert_expr_for): Set
7970 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7972 (remove_range_assertions): Revert earlier change.
7974 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7977 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7979 2017-09-21 Richard Biener <rguenther@suse.de>
7981 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7982 Restore valid IL after code generation errors.
7983 * graphite.c (graphite_transform_loops): Diagnose code
7984 generation issues as MSG_MISSED_OPTIMIZATION and continue
7985 with processing SCOPs.
7987 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7988 Alan Hayward <alan.hayward@arm.com>
7989 David Sherwood <david.sherwood@arm.com>
7991 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7992 rather than choosing between plus_constant and gen_rtx_<CODE>.
7993 * expr.c (emit_push_insn): Likewise.
7994 (expand_expr_real_2): Likewise.
7996 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7997 Alan Hayward <alan.hayward@arm.com>
7998 David Sherwood <david.sherwood@arm.com>
8000 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8002 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8003 Alan Hayward <alan.hayward@arm.com>
8004 David Sherwood <david.sherwood@arm.com>
8006 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8007 calling tree_to_uhwi.
8009 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8010 Alan Hayward <alan.hayward@arm.com>
8011 David Sherwood <david.sherwood@arm.com>
8013 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8014 INTEGER_CST rather than a negative test for ADDR_EXPR.
8016 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8017 Alan Hayward <alan.hayward@arm.com>
8018 David Sherwood <david.sherwood@arm.com>
8020 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8021 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8023 2017-09-21 Richard Biener <rguenther@suse.de>
8025 PR tree-optimization/71351
8026 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8027 graphite_create_new_loop_guard): Remove, fold remaining parts
8029 (translate_isl_ast_node_for): ... here and simplify.
8031 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8034 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8035 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8036 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8037 alternative always use QI mode, for -Os imov (=R,R) alternative
8038 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8041 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8042 Jeff Law <law@redhat.com>
8044 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8045 (allocate_stack_space): New function, partially extracted from
8047 (s390_emit_prologue): Track offset to most recent stack probe.
8048 Code to allocate space moved into allocate_stack_space.
8049 Dump actions when no stack is allocated.
8050 (s390_prologue_plus_offset): New function.
8051 (s390_emit_stack_probe): Likewise.
8053 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8055 * common.opt (Wa, Wl, Wp, g, gz=): Add
8057 (gno-column-info): Remove.
8058 (gcolumn-info): Drop RejectNegative.
8060 (gno-record-gcc-switches): Remove.
8061 (grecord-gcc-switches): Drop RejectNegative.
8062 (gno-split-dwarf): Remove.
8063 (gsplit-dwarf): Drop RejectNegative.
8064 (gno-strict-dwarf): Remove.
8065 (gstrict-dwarf): Drop RejectNegative.
8066 * config/darwin.opt (gfull, gused): Add RejectNegative.
8067 * dwarf2out.c (gen_producer_string): Drop
8068 gno-record-gcc-switches handler.
8069 * optc-gen.awk: Add g to prefixes with negative forms.
8070 * opts-common.c (remapping_prefix_p): New.
8071 (find_opt): Check it.
8072 (generate_canonical_option): Test g prefix.
8073 (option_map): Add -gno- mapping.
8074 (add_misspelling_candidates): Check remapping_prefix_p.
8076 2017-09-20 Jeff Law <law@redhat.com>
8078 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8079 thinko in stack clash protection support.
8081 * explow.c (compute_stack_clash_protection_loop_data): Use
8082 CONST_INT_P instead of explicit test. Verify object is a
8083 CONST_INT_P before looking at INTVAL.
8084 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8085 instead of explicit test.
8087 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8090 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8091 address instead of to r1 and r11.
8093 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8095 * config.gcc: Support "knm".
8096 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8097 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8099 * config/i386/i386.c (m_KNM): Define.
8100 (processor_target_table): Add "knm".
8102 (ix86_option_override_internal): Add "knm".
8103 (ix86_issue_rate): Add PROCESSOR_KNM.
8104 (ix86_adjust_cost): Ditto.
8105 (ia32_multipass_dfa_lookahead): Ditto.
8106 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8107 (fold_builtin_cpu): Add M_INTEL_KNM.
8108 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8109 (processor_type): Add PROCESSOR_KNM.
8110 * config/i386/x86-tune.def: Add m_KNM.
8111 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8113 2017-09-20 Richard Biener <rguenther@suse.de>
8115 PR tree-optimization/80213
8116 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8117 are allowed in empty BBs as well.
8118 (canonicalize_loop_closed_ssa): Also look for other complex
8120 (scop_detection::get_sese): Include the loop-closed PHI block
8122 (scop_detection::merge_sese): Remove code adding extra blocks.
8123 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8124 (build_scops): Assert the final returned scop is invalid.
8126 2017-09-20 Richard Biener <rguenther@suse.de>
8128 PR tree-optimization/82264
8129 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8131 (vn_phi_lookup): Likewise.
8132 (vn_phi_insert): Likewise.
8134 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8136 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8137 that fits into uhwi or shwi, add DW_AT_const_value regardless
8138 of early_dwarf without going through RTL, using add_AT_unsigned
8141 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8142 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8143 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8144 (DEBUG_MACRO_SECTION): Likewise.
8145 (DEBUG_LTO_MACRO_SECTION): Likewise.
8146 (DEBUG_STR_DWO_SECTION): Likewise.
8147 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8148 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8149 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8150 (DEBUG_LTO_LINE_STR_SECTION): Define.
8151 (init_sections_and_labels): Initialize debug_line_str_section
8152 variable. Initialize debug_loc_section for -gdwarf-5 to
8153 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8155 2017-09-20 Richard Biener <rguenther@suse.de>
8157 * graphite-sese-to-poly.c (extract_affine): Properly handle
8158 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8160 2017-09-20 Richard Biener <rguenther@suse.de>
8162 PR tree-optimization/81373
8163 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8164 Force SESE live-out defs to be handled even if they are
8167 2017-09-19 Jeff Law <law@redhat.com>
8169 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8170 nothing for stack adjustments with REG_STACK_CHECK.
8171 * sched-deps.c (parse_add_or_inc): Reject insns with
8172 REG_STACK_CHECK from dependency breaking.
8173 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8174 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8175 * reg-notes.def (STACK_CHECK): New note.
8177 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8178 (ix86_expand_prologue): Dump stack clash info as needed.
8179 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8181 * function.c (dump_stack_clash_frame_info): New function.
8182 * function.h (dump_stack_clash_frame_info): Prototype.
8183 (enum stack_clash_probes): New enum.
8185 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8186 flag_stack_clash_protection.
8187 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8188 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8189 (arm_frame_pointer_required): Likewise.
8190 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8191 (ia64_expand_prologue): Likewise.
8192 * config/mips/mips.c (mips_expand_prologue): Likewise.
8193 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8194 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8195 (sparc_flat_expand_prologue): Likewise.
8196 * config/spu/spu.c (spu_expand_prologue): Likewise.
8198 * explow.c: Include "params.h".
8199 (anti_adjust_stack_and_probe_stack_clash): New function.
8200 (get_stack_check_protect): Likewise.
8201 (compute_stack_clash_protection_loop_data): Likewise.
8202 (emit_stack_clash_protection_loop_start): Likewise.
8203 (emit_stack_clash_protection_loop_end): Likewise.
8204 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8205 Use anti_adjust_stack_and_probe_stack_clash.
8206 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8207 (emit_stack_clash_protection_loop_start): Likewise.
8208 (emit_stack_clash_protection_loop_end): Likewise.
8209 * rtl.h (get_stack_check_protect): Prototype.
8210 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8211 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8212 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8214 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8216 * doc/tm.texi: Rebuilt.
8217 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8218 get_stack_check_protect.
8219 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8220 * config/arm/arm.c (arm_expand_prologue): Likewise.
8221 (arm_frame_pointer_required): Likewise.
8222 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8223 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8224 * config/mips/mips.c (mips_expand_prologue): Likewise.
8225 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8226 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8227 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8228 (sparc_flat_expand_prologue): Likewise.
8230 * common.opt (-fstack-clash-protection): New option.
8231 * flag-types.h (enum stack_check_type): Note difference between
8232 -fstack-check= and -fstack-clash-protection.
8233 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8234 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8235 * toplev.c (process_options): Issue warnings/errors for cases
8236 not handled with -fstack-clash-protection.
8237 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8238 (-fstack-check): Note additional problem with -fstack-check=generic.
8239 Note that -fstack-check is primarily for Ada and refer users
8240 to -fstack-clash-protection for stack-clash-protection.
8241 Document new params for stack clash protection.
8243 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8245 * config/i386/i386.c (ix86_split_long_move): Do not handle
8246 address used for LEA in a special way.
8248 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8250 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8252 2017-09-19 Martin Sebor <msebor@redhat.com>
8255 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8256 of incompatible types.
8258 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8260 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8261 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8262 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8263 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8265 2017-09-19 Richard Biener <rguenther@suse.de>
8267 PR tree-optimization/82244
8268 * tree-vrp.c (remove_range_assertions): Do not propagate
8269 a constant to abnormals but replace the assert with a copy.
8271 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8273 PR rtl-optimization/57878
8274 PR rtl-optimization/68988
8275 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8276 avoidance test involving non_reload_pseudos. Move frequency test
8277 below the general fragmentation avoidance test.
8279 2017-09-19 Richard Biener <rguenther@suse.de>
8281 PR tree-optimization/69728
8282 * graphite-sese-to-poly.c (schedule_error): New global.
8283 (add_loop_schedule): Handle empty domain by failing the
8285 (build_original_schedule): Handle schedule_error.
8287 2017-09-19 Richard Biener <rguenther@suse.de>
8289 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8290 Do not iterate to sibling loops but only to siblings of inner
8293 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8296 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8297 that the registers are different.
8299 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8301 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8302 to processor_model and "amdfam17h" to arch_names_table.
8303 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8305 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8308 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8310 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8311 Alan Hayward <alan.hayward@arm.com>
8312 David Sherwood <david.sherwood@arm.com>
8314 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8316 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8318 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8319 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8320 vect_schedule_slp_instance.
8321 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8322 Update call to vect_slp_analyze_node_operations. Simplify return
8324 (vect_slp_analyze_bb_1): Update call accordingly.
8325 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8326 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8327 (vect_schedule_slp): Update call accordingly.
8329 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8330 Alan Hayward <alan.hayward@arm.com>
8331 David Sherwood <david.sherwood@arm.com>
8333 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8334 with types that aren't in fact scalar.
8336 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8338 * tree-vect-slp.c (vect_record_max_nunits): New function,
8340 (vect_build_slp_tree_1): ...here.
8341 (vect_build_slp_tree_2): Call it for phis too.
8343 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8345 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8346 to vect_get_vec_def_for_operand when getting the mask operand.
8348 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8349 Alan Hayward <alan.hayward@arm.com>
8350 David Sherwood <david.sherwood@arm.com>
8352 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8355 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8356 Alan Hayward <alan.hayward@arm.com>
8357 David Sherwood <david.sherwood@arm.com>
8359 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8360 calculation for vector booleans.
8362 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8363 Alan Hayward <alan.hayward@arm.com>
8364 David Sherwood <david.sherwood@arm.com>
8366 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8368 (vect_transform_stmt): ...here.
8369 (vect_analyze_stmt): Use it instead of calling
8370 vectorizable_live_operation directly.
8372 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8374 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8375 non-SIMT targets in acc vector loops.
8377 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8379 * configure.ac: Add arc and check if assembler supports gdwarf2.
8380 * configure: Regenerate.
8382 2017-09-18 Richard Biener <rguenther@suse.de>
8384 PR tree-optimization/82220
8385 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8386 epilogue niters from the min_profitable_iters compute.
8388 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8391 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8392 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8393 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8395 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8398 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8399 switching to a new text section.
8401 2017-09-18 Richard Biener <rguenther@suse.de>
8403 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8405 (build_alias_set): Reject aliases with no access function.
8407 2017-09-18 Richard Biener <rguenther@suse.de>
8409 PR tree-optimization/79622
8410 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8412 (build_cross_bb_scalars_use): Likewise.
8414 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8416 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8418 2017-09-18 Alan Modra <amodra@gmail.com>
8421 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8422 stack_pointer_rtx for count 0. Update comments. Break up
8423 large rtl expression.
8425 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8427 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8428 Increase to 20 bytes.
8429 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8430 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8431 or avx version of the stub.
8433 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8436 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8437 compute the minimum stack alignment. Also update preferred stack
8438 boundary for leaf functions.
8440 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8442 PR tree-optimization/82228
8443 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8446 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8448 * common/config/nds32/nds32-common.c
8449 (nds32_option_optimization_table): Refine formatting.
8450 (nds32_option_optimization_table): Use -fsched-pressure and
8451 -fomit-frame-pointer for specific optimization level.
8453 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8455 * config/nds32/nds32.c: Refine formatting and comments.
8456 * config/nds32/nds32.h: Likewise.
8457 * config/nds32/nds32.md: Likewise.
8458 * config/nds32/nds32-cost.c: Likewise.
8459 * config/nds32/nds32-isr.c: Likewise.
8460 * config/nds32/nds32-md-auxiliary.c: Likewise.
8461 * config/nds32/nds32-multiple.md: Likewise.
8462 * config/nds32/nds32-predicates.c: Likewise.
8464 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8465 Jakub Jelinek <jakub@redhat.com>
8467 Add support for -std=c++2a.
8468 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8470 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8472 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8475 * doc/extend.texi (Common Function Attributes): Add
8476 references to ARM, AArch64, and S/390 specific attributes.
8477 (Function Specific Option Pragmas): Add AArch64 and S/390
8478 to list of back ends that support the target pragma.
8480 2017-09-15 Nathan Sidwell <nathan@acm.org>
8482 * doc/standards.texi: Fix C++17 description. Update URLs for
8485 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8487 * common.opt (Wcast-align=strict): New warning option.
8488 * doc/invoke.texi: Document -Wcast-align=strict.
8490 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8492 * cgraph.h (cgraph_thunk_info): Add comments.
8493 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8494 assert for VIRTUAL_* arguments stricter.
8496 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8498 PR tree-optimization/71026
8499 * match.pd: Move RDIV patterns from fold-const.c
8500 * fold-const.c (distribute_real_division): Removed.
8501 (fold_binary_loc): Remove calls to distribute_real_divison.
8503 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8505 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8506 c++1z and gnu++1z as deprecated. Change other references to
8507 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8508 Change -Wc++1z-compat to -Wc++17-compat.
8509 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8510 * dwarf2out.c (highest_c_language): Handle C++17.
8511 (gen_compile_unit_die): Likewise.
8513 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8515 PR rtl-optimization/82192
8516 * combine.c (make_extraction): Don't look through non-paradoxical
8517 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8520 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8521 Alan Hayward <alan.hayward@arm.com>
8522 David Sherwood <david.sherwood@arm.com>
8524 * target.def (function_arg_offset): New hook.
8525 * targhooks.h (default_function_arg_offset): Declare.
8526 * targhooks.c (default_function_arg_offset): New function.
8527 * function.c (locate_and_pad_parm): Use
8528 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8529 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8530 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8531 * doc/tm.texi: Regenerate.
8532 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8533 * config/spu/spu.c (spu_function_arg_offset): New function.
8534 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8535 * system.h (FUNCTION_ARG_OFFSET): Poison.
8537 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8538 Alan Hayard <alan.hayward@arm.com>
8539 David Sherwood <david.sherwood@arm.com>
8541 * target.def (truly_noop_truncation): New hook.
8542 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8543 than TRULY_NOOP_TRUNCATION.
8544 * hooks.h (hook_bool_uint_uint_true): Declare.
8545 * hooks.c (hook_bool_uint_uint_true): New function.
8546 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8547 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8548 * doc/tm.texi: Regenerate.
8549 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8550 rather than TRULY_NOOP_TRUNCATION in comments.
8551 (simplify_comparison): Likewise.
8552 (record_truncated_value): Likewise.
8553 * expmed.c (extract_bit_field_1): Likewise.
8554 (extract_split_bit_field): Likewise.
8555 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8556 instead of TRULY_NOOP_TRUNCATION.
8557 * function.c (assign_parm_setup_block): Likewise.
8558 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8559 * rtlhooks.c: Include target.h.
8560 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8561 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8562 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8563 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8564 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8565 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8566 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8567 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8568 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8569 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8570 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8571 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8572 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8573 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8574 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8575 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8576 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8577 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8578 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8579 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8580 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8581 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8582 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8583 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8584 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8585 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8586 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8587 rather than TRULY_NOOP_TRUNCATION in comments.
8588 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8589 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8590 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8591 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8592 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8593 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8594 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8595 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8596 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8597 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8598 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8599 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8600 rather than TRULY_NOOP_TRUNCATION in comments.
8601 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8602 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8603 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8604 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8605 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8606 TRULY_NOOP_TRUNCATION condition.
8607 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8608 (TRULY_NOOP_TRUNCATION): Delete.
8609 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8610 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8611 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8612 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8613 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8614 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8615 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8616 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8617 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8618 rather than TRULY_NOOP_TRUNCATION in comments.
8619 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8620 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8621 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8622 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8623 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8624 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8626 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8629 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8630 (*cmp_ior): Likewise.
8631 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8632 (*ior_scc_scc_cmp): Likewise.
8633 (*and_scc_scc): Likewise.
8634 (*and_scc_scc_cmp): Likewise.
8636 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8637 Alan Hayard <alan.hayward@arm.com>
8638 David Sherwood <david.sherwood@arm.com>
8640 * target.def (can_change_mode_class): New hook.
8641 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8642 (hard_regno_nregs): Likewise.
8643 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8644 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8645 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8646 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8647 (LOAD_EXTEND_OP): Update accordingly.
8648 * doc/tm.texi: Regenerate.
8649 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8650 CANNOT_CHANGE_MODE_CLASS.
8651 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8652 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8653 * combine.c (simplify_set): Update accordingly.
8654 * emit-rtl.c (validate_subreg): Likewise.
8655 * recog.c (general_operand): Likewise.
8656 * regcprop.c (mode_change_ok): Likewise.
8657 * reload1.c (choose_reload_regs): Likewise.
8658 (inherit_piecemeal_p): Likewise.
8659 * rtlanal.c (simplify_subreg_regno): Likewise.
8660 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8661 instead of CANNOT_CHANGE_MODE_CLASS.
8662 (reload_cse_simplify_operands): Likewise.
8663 * reload.c (push_reload): Use targetm.can_change_mode_class
8664 instead of CANNOT_CHANGE_MODE_CLASS.
8665 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8666 REG_CANNOT_CHANGE_MODE_P.
8667 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8668 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8669 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8670 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8671 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8672 (arm_can_change_mode_class): New function.
8673 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8674 than CANNOT_CHANGE_MODE_CLASS in comments.
8675 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8676 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8677 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8678 (ix86_can_change_mode_class): ...this new function, inverting the
8679 sense of the return value.
8680 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8681 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8682 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8683 (ia64_can_change_mode_class): New function.
8684 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8685 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8686 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8687 (m32c_can_change_mode_class): ...this new function, inverting the
8688 sense of the return value.
8689 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8690 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8691 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8692 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8693 (mips_can_change_mode_class): ...this new function, inverting the
8694 sense of the return value.
8695 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8696 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8697 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8698 (msp430_can_change_mode_class): New function.
8699 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8700 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8701 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8702 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8703 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8704 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8705 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8706 (pa_cannot_change_mode_class): Replace with...
8707 (pa_can_change_mode_class): ...this new function, inverting the
8708 sense of the return value.
8709 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8710 than CANNOT_CHANGE_MODE_CLASS in comments.
8711 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8712 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8713 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8714 (pdp11_cannot_change_mode_class): Replace with...
8715 (pdp11_can_change_mode_class): ...this new function, inverting the
8716 sense of the return value.
8717 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8718 * config/powerpcspe/powerpcspe-protos.h
8719 (rs6000_cannot_change_mode_class_ptr): Delete.
8720 * config/powerpcspe/powerpcspe.c
8721 (rs6000_cannot_change_mode_class_ptr): Delete.
8722 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8723 (rs6000_option_override_internal): Assign to
8724 targetm.can_change_mode_class instead of
8725 rs6000_cannot_change_mode_class_ptr.
8726 (rs6000_cannot_change_mode_class): Replace with...
8727 (rs6000_can_change_mode_class): ...this new function, inverting the
8728 sense of the return value.
8729 (rs6000_debug_cannot_change_mode_class): Replace with...
8730 (rs6000_debug_can_change_mode_class): ...this new function.
8731 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8732 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8733 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8734 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8735 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8737 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8738 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8739 (rs6000_option_override_internal): Assign to
8740 targetm.can_change_mode_class instead of
8741 rs6000_cannot_change_mode_class_ptr.
8742 (rs6000_cannot_change_mode_class): Replace with...
8743 (rs6000_can_change_mode_class): ...this new function, inverting the
8744 sense of the return value.
8745 (rs6000_debug_cannot_change_mode_class): Replace with...
8746 (rs6000_debug_can_change_mode_class): ...this new function.
8747 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8748 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8749 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8750 (s390_can_change_mode_class): ...this new function, inverting the
8751 sense of the return value.
8752 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8753 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8754 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8755 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8756 (sh_cannot_change_mode_class): Replace with...
8757 (sh_can_change_mode_class): ...this new function, inverting the
8758 sense of the return value.
8759 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8760 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8761 (sparc_can_change_mode_class): New function.
8762 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8763 * config/spu/spu.c (spu_can_change_mode_class): New function.
8764 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8765 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8766 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8767 (visium_can_change_mode_class): New function.
8768 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8770 2017-09-15 Richard Biener <rguenther@suse.de>
8772 PR tree-optimization/82217
8773 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8774 but not undefined case.
8776 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8779 * postreload.c (reload_cse_simplify_operands): Skip
8780 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8782 2017-09-15 Richard Biener <rguenther@suse.de>
8784 PR tree-optimization/68823
8785 * graphite-scop-detection.c (build_alias_set): If we have a
8786 possible dependence check whether we can handle them by just
8787 looking at the DRs DR_ACCESS_FNs.
8788 (build_scops): If build_alias_set fails, fail the SCOP.
8790 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8792 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8793 to support float128 built-in functions that require the ISA 3.0
8795 (BU_FLOAT128_3_HW): Likewise.
8796 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8798 (FMAF128): Likewise.
8800 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8801 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8802 floating point instructions.
8803 (rs6000_invalid_builtin): Likewise.
8804 (rs6000_builtin_mask_names): Likewise.
8805 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8806 (RS6000_BTM_FLOAT128_HW): Likewise.
8807 (RS6000_BTM_COMMON): Likewise.
8808 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8810 * doc/extend.texi (RS/6000 built-in functions): Document the
8811 IEEE 128-bit floating point square root and fused multiply-add
8814 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8816 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8817 reg (r2) isn't in the set of registers defined in the prologue.
8819 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8820 Alan Hayward <alan.hayward@arm.com>
8821 David Sherwood <david.sherwood@arm.com>
8823 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8824 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8825 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8826 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8827 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8829 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8830 max_vectorization_factor.
8831 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8833 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8834 Alan Hayward <alan.hayward@arm.com>
8835 David Sherwood <david.sherwood@arm.com>
8837 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8838 (vect_worthwhile_without_simd_p): Declare.
8839 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8840 (vectorizable_reduction): Use it.
8841 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8842 (vectorizable_operation): Likewise.
8844 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8845 Alan Hayward <alan.hayward@arm.com>
8846 David Sherwood <david.sherwood@arm.com>
8848 * tree-vectorizer.h (vect_get_num_copies): New function.
8849 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8850 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8851 (vectorizable_induction): Likewise.
8852 (vectorizable_live_operation): Likewise.
8853 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8854 (vectorizable_bswap): Likewise.
8855 (vectorizable_call): Likewise.
8856 (vectorizable_conversion): Likewise.
8857 (vectorizable_assignment): Likewise.
8858 (vectorizable_shift): Likewise.
8859 (vectorizable_operation): Likewise.
8860 (vectorizable_store): Likewise.
8861 (vectorizable_load): Likewise.
8862 (vectorizable_condition): Likewise.
8863 (vectorizable_comparison): Likewise.
8864 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8866 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8867 Alan Hayward <alan.hayward@arm.com>
8868 David Sherwood <david.sherwood@arm.com>
8870 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8871 of vect_init_vector.
8873 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8874 Alan Hayward <alan.hayward@arm.com>
8875 David Sherwood <david.sherwood@arm.com>
8877 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8878 an inline wrapper that provides a location.
8879 (gimple_build_vector): Likewise.
8880 * gimple-fold.c (gimple_build_vector_from_val): New function.
8881 (gimple_build_vector): Likewise.
8882 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8883 functions to build the initial value. Always return a gimple value.
8884 (get_initial_defs_for_reduction): Likewise. Only compute
8886 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8887 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8888 (vectorizable_induction): Use gimple_build_vector rather than
8891 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8892 Alan Hayward <alan.hayward@arm.com>
8893 David Sherwood <david.sherwood@arm.com>
8895 * target.h (vec_perm_indices): New typedef.
8896 (auto_vec_perm_indices): Likewise.
8897 * optabs-query.h: Include target.h
8898 (can_vec_perm_p): Take a vec_perm_indices *.
8899 * optabs-query.c (can_vec_perm_p): Likewise.
8900 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8901 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8902 * tree-vect-generic.c (lower_vec_perm): Likewise.
8903 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8904 (vect_grouped_load_supported): Likewise.
8905 (vect_shift_permute_load_chain): Likewise.
8906 (vect_permute_store_chain): Use auto_vec_perm_indices.
8907 (vect_permute_load_chain): Likewise.
8908 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8909 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8910 Update uses of can_vec_perm_p.
8911 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8912 mode with a number of elements. Take a vec_perm_indices *.
8913 (vect_create_epilog_for_reduction): Update accordingly.
8914 Use auto_vec_perm_indices.
8915 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8916 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8917 (vect_transform_slp_perm_load): Likewise.
8918 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8919 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8920 (vect_gen_perm_mask_checked): Likewise.
8921 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8922 (vect_gen_perm_mask_checked): Likewise.
8923 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8924 (vectorizable_store): Likewise.
8925 (vectorizable_load): Likewise.
8926 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8927 (vectorizable_bswap): Likewise.
8929 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8930 Alan Hayward <alan.hayward@arm.com>
8931 David Sherwood <david.sherwood@arm.com>
8933 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8934 * tree.c (build_vector): Likewise.
8935 (build_vector_from_ctor): Update accordingly.
8936 (build_vector_from_val): Likewise.
8937 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8938 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8939 * tree-vect-generic.c (add_rshift): Likewise.
8940 (expand_vector_divmod): Likewise.
8941 (optimize_vector_constructor): Likewise.
8942 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8943 (vect_transform_slp_perm_load): Likewise.
8944 (vect_schedule_slp_instance): Likewise.
8945 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8946 (vectorizable_call): Likewise.
8947 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8948 * expmed.c (make_tree): Likewise.
8949 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8950 a vector passed to build_vector.
8951 (fold_convert_const): Likewise.
8952 (exact_inverse): Likewise.
8953 (fold_ternary_loc): Likewise.
8954 (fold_relational_const): Likewise.
8955 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8956 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8957 (const_unop): Likewise. Store the reduction accumulator in a
8958 variable rather than an array.
8959 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8960 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8961 the new vector, rather than constructing it after the input arrays.
8962 (native_interpret_vector): Use auto_vec<tree> when building
8963 a vector passed to build_vector. Add elements in order.
8964 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8965 auto_vec<tree> when building a vector passed to build_vector.
8966 (vect_create_epilog_for_reduction): Likewise.
8967 (vectorizable_induction): Likewise.
8968 (get_initial_def_for_reduction): Likewise. Fix indentation of
8970 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8972 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8973 passed to build_vector.
8975 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8976 Alan Hayward <alan.hayward@arm.com>
8977 David Sherwood <david.sherwood@arm.com>
8979 * tree-core.h (tree_base::u): Add an "nelts" field.
8980 (tree_vector): Use VECTOR_CST_NELTS as the length.
8981 * tree.c (tree_size): Likewise.
8982 (make_vector): Initialize VECTOR_CST_NELTS.
8983 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8984 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8985 TYPE_VECTOR_SUBPARTS.
8986 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8987 as the number of units, setting it from VECTOR_CST_NELTS.
8988 (const_vector_from_tree): Likewise.
8989 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8990 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8991 (fold_negate_expr_1): Likewise.
8992 (fold_convert_const): Likewise.
8993 (const_binop): Likewise. Differentiate the number of output and
8995 (const_unop): Likewise.
8996 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8997 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8998 in cases that did the opposite.
9000 2017-09-14 Richard Biener <rguenther@suse.de>
9002 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9005 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9007 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9009 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9012 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9013 if and where to split a bb, except for splitting before debug insn
9014 sequences followed by non-label real insn. Delete debug insns
9015 in between basic blocks.
9017 * combine.c (make_compound_operation_int): Formatting fixes.
9019 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9020 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9021 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9022 * config/sol2.h (LINK_EH_SPEC): Likewise.
9023 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9024 * config/s390/linux.h (LINK_SPEC): Likewise.
9025 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9026 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9027 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9028 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9029 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9030 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9031 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9032 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9034 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9036 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9038 (ENDFILE_LINUX_SPEC): Likewise.
9039 (LINK_EH_SPEC): Likewise.
9040 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9041 (LINK_OS_LINUX_SPEC32): Likewise.
9042 (LINK_OS_LINUX_SPEC64): Likewise.
9043 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9044 (LINK_OS_LINUX_SPEC): Likewise.
9046 2017-09-13 Martin Liska <mliska@suse.cz>
9049 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9050 CASE_HIGH is NULL_TREE.
9052 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9053 Alan Hayward <alan.hayward@arm.com>
9054 David Sherwood <david.sherwood@arm.com>
9056 * target.def (secondary_memory_needed): New hook.
9057 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9058 instead of SECONDARY_MEMORY_NEEDED.
9059 (secondary_memory_needed_mode): Likewise.
9060 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9061 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9062 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9063 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9064 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9065 * doc/tm.texi: Regenerate.
9066 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9067 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9068 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9069 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9070 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9071 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9072 mode argument first and change the reg_class arguments to reg_class_t.
9073 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9074 Make static. Update the call to inline_secondary_memory_needed.
9075 (ix86_register_move_cost): Update the call to
9076 inline_secondary_memory_needed.
9077 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9078 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9080 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9081 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9083 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9084 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9085 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9086 and match hook interface. Add comment from mips.h.
9087 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9088 * config/mmix/mmix.md (truncdfsf2): Refer to
9089 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9091 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9092 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9093 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9094 (pa_secondary_memory_needed): New function.
9095 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9096 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9097 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9098 (pdp11_secondary_memory_needed): Make static and match hook interface.
9099 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9100 * config/powerpcspe/powerpcspe-protos.h
9101 (rs6000_secondary_memory_needed_ptr): Delete.
9102 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9104 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9105 (rs6000_option_override_internal): Assign to
9106 targetm.secondary_memory_needed rather than
9107 rs6000_secondary_memory_needed_ptr.
9108 (rs6000_secondary_memory_needed): Match hook interface.
9109 (rs6000_debug_secondary_memory_needed): Likewise.
9110 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9111 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9112 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9113 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9114 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9115 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9117 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9118 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9119 (rs6000_option_override_internal): Assign to
9120 targetm.secondary_memory_needed rather than
9121 rs6000_secondary_memory_needed_ptr.
9122 (rs6000_secondary_memory_needed): Match hook interface.
9123 (rs6000_debug_secondary_memory_needed): Likewise.
9124 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9125 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9126 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9127 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9128 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9129 (sparc_secondary_memory_needed): New function.
9130 * lra-constraints.c (check_and_process_move): Refer to
9131 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9133 (curr_insn_transform): Likewise.
9134 (process_alt_operands): Use targetm.secondary_memory_needed
9135 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9136 (check_secondary_memory_needed_p): Likewise.
9137 (choose_split_class): Likewise.
9138 * reload.c: Unconditionally include code that was previously
9139 conditional on SECONDARY_MEMORY_NEEDED.
9140 (push_secondary_reload): Use targetm.secondary_memory_needed
9141 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9142 (push_reload): Likewise.
9143 * reload1.c: Unconditionally include code that was previously
9144 conditional on SECONDARY_MEMORY_NEEDED.
9145 (choose_reload_regs): Use targetm.secondary_memory_needed
9146 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9147 (gen_reload): Likewise.
9148 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9150 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9151 Alan Hayward <alan.hayward@arm.com>
9152 David Sherwood <david.sherwood@arm.com>
9154 * target.def (secondary_memory_needed_mode): New hook:
9155 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9156 * targhooks.h (default_secondary_memory_needed_mode): New function.
9157 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9158 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9159 * doc/tm.texi: Regenerate.
9160 * lra-constraints.c (check_and_process_move): Use
9161 targetm.secondary_memory_needed_mode instead of
9162 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9163 (curr_insn_transform): Likewise.
9164 * reload.c (get_secondary_mem): Likewise.
9165 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9166 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9168 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9169 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9170 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9171 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9172 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9174 * config/powerpcspe/powerpcspe-protos.h
9175 (rs6000_secondary_memory_needed_mode): Delete.
9176 * config/powerpcspe/powerpcspe.c
9177 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9178 (rs6000_secondary_memory_needed_mode): Make static.
9179 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9180 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9182 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9184 (rs6000_secondary_memory_needed_mode): Make static.
9185 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9186 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9187 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9188 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9189 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9191 (sparc_secondary_memory_needed_mode): New function.
9192 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9194 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9196 * config/aarch64/constraints.md (Umq): New constraint.
9197 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9199 (mov<mode>): Update condition.
9201 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9203 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9204 when bitposition is the same.
9206 2017-09-13 Richard Biener <rguenther@suse.de>
9208 * dwarf2out.c (output_die_symbol): Remove.
9209 (output_die): Do not output a DIEs symbol.
9211 2017-09-13 Richard Biener <rguenther@suse.de>
9214 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9215 default-def to avoid breaking iterator update with the weird
9216 interaction with cgraph_update_edges_for_call_stmt_node.
9218 2017-09-13 Richard Biener <rguenther@suse.de>
9220 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9221 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9222 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9223 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9225 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9227 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9228 Disable pc relative literal load irrespective of
9229 TARGET_FIX_ERR_A53_84341 for default.
9231 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9233 * config/sparc/sparc.c (output_return): Output the source location of
9234 the insn in the delay slot, if any.
9235 (output_sibcall): Likewise.
9237 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9240 * common.opt (-static-pie): New alias.
9241 (shared): Negate static-pie.
9242 (-no-pie): Update help text.
9244 (static-pie): New option.
9245 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9246 -static-pie support.
9247 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9248 (LINK_EH_SPEC): Likewise.
9249 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9250 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9251 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9252 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9253 (init_gcc_specs): Likewise.
9254 (init_spec): Likewise.
9255 (display_help): Update help message for -pie.
9256 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9259 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9261 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9262 (movdi_aarch64): Likewise.
9263 (movti_aarch64): Likewise.
9265 2017-09-12 Simon Wright <simon@pushface.org>
9268 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9269 calculation of the minor version, always output as 0.
9271 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9274 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9275 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9276 on it early, rather than manual conversion late. For
9277 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9278 instead of performing manual conversion.
9280 2017-09-12 Carl Love <cel@us.ibm.com>
9282 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9283 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9285 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9286 VMULOSW): Add definitions.
9287 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9288 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9289 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9290 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9291 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9293 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9295 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9297 (movti_aarch64): Likewise.
9298 (movdf_aarch64): Likewise.
9299 (movtf_aarch64): Likewise.
9300 (load_pairdi): Likewise.
9301 (store_pairdi): Likewise.
9302 (load_pairdf): Likewise.
9303 (store_pairdf): Likewise.
9304 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9305 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9306 (ldr_got_small_<mode>): Likewise.
9307 (ldr_got_small_28k_<mode>): Likewise.
9308 (ldr_got_tiny): Likewise.
9309 * config/aarch64/iterators.md (ldst_sz): New.
9310 (ldpstp_sz): Likewise.
9311 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9313 (thunderx_load): Split load_8 to load_16.
9314 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9316 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9317 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9318 (xgene1_store_pair): Split store_8 to store_16.
9319 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9320 (falkor_st_0_st_sd): Split store_8 to store_16.
9322 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9324 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9325 and store1/2/3/4 to store_4/8/12/16.
9326 * config/aarch64/aarch64.md: Update for rename.
9327 * config/arm/arm.md: Likewise.: Likewise.
9328 * config/arm/arm.c: Likewise.
9329 * config/arm/thumb1.md: Likewise.
9330 * config/arm/thumb2.md: Likewise.
9331 * config/arm/vfp.md: Likewise.
9332 * config/arm/arm-generic.md: Likewise.
9333 * config/arm/arm1020e.md: Likewise.
9334 * config/arm/arm1026ejs.md: Likewise.
9335 * config/arm/arm1136jfs.md: Likewise.
9336 * config/arm/arm926ejs.md: Likewise.
9337 * config/arm/cortex-a15.md: Likewise.
9338 * config/arm/cortex-a17.md: Likewise.
9339 * config/arm/cortex-a5.md: Likewise.
9340 * config/arm/cortex-a53.md: Likewise.
9341 * config/arm/cortex-a57.md: Likewise.
9342 * config/arm/cortex-a7.md: Likewise.
9343 * config/arm/cortex-a8.md: Likewise.
9344 * config/arm/cortex-a9.md: Likewise.
9345 * config/arm/cortex-m4.md: Likewise.
9346 * config/arm/cortex-m7.md: Likewise.
9347 * config/arm/cortex-r4.md: Likewise.
9348 * config/arm/exynos-m1.md: Likewise.
9349 * config/arm/fa526.md: Likewise.
9350 * config/arm/fa606te.md: Likewise.
9351 * config/arm/fa626te.md: Likewise.
9352 * config/arm/fa726te.md: Likewise.
9353 * config/arm/fmp626.md: Likewise.
9354 * config/arm/iwmmxt.md: Likewise.
9355 * config/arm/ldmstm.md: Likewise.
9356 * config/arm/marvell-pj4.md: Likewise.
9357 * config/arm/xgene1.md: Likewise.
9358 * config/aarch64/thunderx.md: Likewise.
9359 * config/aarch64/thunderx2t99.md: Likewise.
9360 * config/aarch64/falkor.md: Likewise.
9362 2017-09-12 Martin Liska <mliska@suse.cz>
9364 * attribs.c (private_lookup_attribute): New function.
9365 * attribs.h (private_lookup_attribute): Declared here.
9366 (lookup_attribute): Called from this place.
9368 2017-09-12 Richard Biener <rguenther@suse.de>
9370 PR tree-optimization/82157
9371 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9372 stmts with side-effects.
9374 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9375 Alan Hayward <alan.hayward@arm.com>
9376 David Sherwood <david.sherwood@arm.com>
9378 * target.def (hard_regno_nregs): New hook.
9379 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9380 * targhooks.h (default_hard_regno_nregs): Declare.
9381 * targhooks.c (default_hard_regno_nregs): New function.
9382 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9383 (TARGET_HARD_REGNO_NREGS): ...this hook.
9384 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9385 (CLASS_MAX_NREGS): Likewise.
9386 * doc/tm.texi: Regenerate.
9387 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9388 instead of HARD_REGNO_NREGS.
9389 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9390 HARD_REGNO_NREGS in the comment.
9391 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9392 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9393 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9394 Return an unsigned int.
9395 (TARGET_HARD_REGNO_NREGS): Redefine.
9396 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9397 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9398 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9399 (arc_hard_regno_nregs): New function.
9400 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9401 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9402 (arm_hard_regno_nregs): New function.
9403 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9404 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9405 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9406 (TARGET_HARD_REGNO_NREGS): Redefine.
9407 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9408 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9409 (HARD_REGNO_NREGS): Delete.
9410 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9411 (cr16_hard_regno_nregs): New function.
9412 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9413 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9414 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9415 (cris_hard_regno_nregs): New function.
9416 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9417 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9418 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9419 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9420 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9421 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9422 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9423 (frv_hard_regno_nregs): Make static. Take and return an
9425 (frv_class_max_nregs): Remove outdated copy of documentation.
9426 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9427 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9428 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9429 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9430 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9431 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9432 (TARGET_HARD_REGNO_NREGS): Redefine.
9433 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9434 (CLASS_MAX_NREGS): Update comment.
9435 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9436 (ia64_hard_regno_nregs): New function.
9437 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9438 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9439 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9440 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9441 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9443 (m32c_hard_regno_nregs): Likewise. Make static.
9444 (TARGET_HARD_REGNO_NREGS): Redefine.
9445 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9446 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9447 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9448 (m68k_hard_regno_nregs): New function.
9449 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9450 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9451 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9452 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9453 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9454 Take and return an unsigned int.
9455 (TARGET_HARD_REGNO_NREGS): Redefine.
9456 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9457 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9458 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9459 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9460 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9461 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9462 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9463 (msp430_hard_regno_nregs): Make static. Take and return an
9465 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9466 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9467 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9468 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9469 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9470 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9471 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9472 (TARGET_HARD_REGNO_NREGS): Redefine.
9473 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9474 (PA_HARD_REGNO_NREGS): ...this.
9475 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9476 (PA_HARD_REGNO_NREGS): ...this.
9477 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9478 (pa_hard_regno_nregs): New function.
9479 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9480 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9481 (pdp11_hard_regno_nregs): New function.
9482 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9483 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9484 (rs6000_hard_regno_nregs_hook): New function.
9485 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9486 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9487 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9488 Take and return an unsigned int. Move earlier in file.
9489 (TARGET_HARD_REGNO_NREGS): Redefine.
9490 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9491 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9492 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9493 (rl78_hard_regno_nregs): Make static. Take and return an
9495 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9496 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9497 (rs6000_hard_regno_nregs_hook): New function.
9498 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9499 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9500 (TARGET_HARD_REGNO_NREGS): Redefine.
9501 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9502 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9503 instead of HARD_REGNO_NREGS.
9504 (s390_hard_regno_nregs): New function.
9505 (s390_hard_regno_mode_ok): Add comment from s390.h.
9506 (TARGET_HARD_REGNO_NREGS): Redefine.
9507 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9508 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9509 (sh_hard_regno_nregs): New function.
9510 (sh_pass_in_reg_p): Use it.
9511 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9512 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9513 (sparc_hard_regno_nregs): New function.
9514 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9515 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9516 (spu_function_arg_advance): Use it, supplying a valid register number.
9517 (TARGET_HARD_REGNO_NREGS): Redefine.
9518 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9519 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9520 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9521 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9522 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9523 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9524 (CLASS_MAX_NREGS): Remove copy of old documentation.
9525 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9526 (visium_hard_regno_nregs): New function.
9527 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9528 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9529 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9530 xtensa_hard_regno_nregs): New function.
9531 * system.h (HARD_REGNO_NREGS): Poison.
9533 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9535 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9536 hard_regno_nregs instead of HARD_REGNO_NREGS.
9537 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9538 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9539 (c6x_expand_epilogue): Likewise.
9540 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9541 (frv_read_iacc_argument): Likewise.
9542 * config/sh/sh.c: Include regs.h.
9543 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9544 (regs_used): Likewise.
9545 (output_stack_adjust): Likewise.
9546 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9547 * expmed.c: Include regs.h.
9548 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9549 * ree.c: Include regs.h.
9550 (combine_reaching_defs): Use hard_regno_nregs instead of
9552 (add_removable_extension): Likewise.
9554 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9556 * regs.h (hard_regno_nregs): Turn into a function.
9557 (end_hard_regno): Update accordingly.
9558 * caller-save.c (setup_save_areas): Likewise.
9559 (save_call_clobbered_regs): Likewise.
9560 (replace_reg_with_saved_mem): Likewise.
9561 (insert_restore): Likewise.
9562 (insert_save): Likewise.
9563 * combine.c (can_change_dest_mode): Likewise.
9564 (move_deaths): Likewise.
9565 (distribute_notes): Likewise.
9566 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9567 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9568 (rs6000_split_multireg_move): Likewise.
9569 (rs6000_register_move_cost): Likewise.
9570 (rs6000_memory_move_cost): Likewise.
9571 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9572 (rs6000_split_multireg_move): Likewise.
9573 (rs6000_register_move_cost): Likewise.
9574 (rs6000_memory_move_cost): Likewise.
9575 * cselib.c (cselib_reset_table): Likewise.
9576 (cselib_lookup_1): Likewise.
9577 * emit-rtl.c (set_mode_and_regno): Likewise.
9578 * function.c (aggregate_value_p): Likewise.
9579 * ira-color.c (setup_profitable_hard_regs): Likewise.
9580 (check_hard_reg_p): Likewise.
9581 (calculate_saved_nregs): Likewise.
9582 (assign_hard_reg): Likewise.
9583 (improve_allocation): Likewise.
9584 (calculate_spill_cost): Likewise.
9585 * ira-emit.c (modify_move_list): Likewise.
9586 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9587 (ira_hard_reg_in_set_p): Likewise.
9588 * ira.c (setup_reg_mode_hard_regset): Likewise.
9589 (clarify_prohibited_class_mode_regs): Likewise.
9590 (check_allocation): Likewise.
9591 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9592 (lra_setup_reg_renumber): Likewise.
9593 (setup_try_hard_regno_pseudos): Likewise.
9594 (spill_for): Likewise.
9595 (assign_hard_regno): Likewise.
9596 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9597 * lra-constraints.c (in_class_p): Likewise.
9598 (lra_constraint_offset): Likewise.
9599 (simplify_operand_subreg): Likewise.
9600 (lra_constraints): Likewise.
9601 (split_reg): Likewise.
9602 (split_if_necessary): Likewise.
9603 (invariant_p): Likewise.
9604 (inherit_in_ebb): Likewise.
9605 * lra-lives.c (process_bb_lives): Likewise.
9606 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9607 (get_hard_regs): Likewise.
9608 (do_remat): Likewise.
9609 * lra-spills.c (assign_spill_hard_regs): Likewise.
9610 * mode-switching.c (create_pre_exit): Likewise.
9611 * postreload.c (reload_combine_recognize_pattern): Likewise.
9612 * recog.c (peep2_find_free_register): Likewise.
9613 * regcprop.c (kill_value_regno): Likewise.
9614 (set_value_regno): Likewise.
9615 (copy_value): Likewise.
9616 (maybe_mode_change): Likewise.
9617 (find_oldest_value_reg): Likewise.
9618 (copyprop_hardreg_forward_1): Likewise.
9619 * regrename.c (check_new_reg_p): Likewise.
9620 (regrename_do_replace): Likewise.
9621 * reload.c (push_reload): Likewise.
9622 (combine_reloads): Likewise.
9623 (find_dummy_reload): Likewise.
9624 (operands_match_p): Likewise.
9625 (find_reloads): Likewise.
9626 (find_equiv_reg): Likewise.
9627 (reload_adjust_reg_for_mode): Likewise.
9628 * reload1.c (count_pseudo): Likewise.
9629 (count_spilled_pseudo): Likewise.
9630 (find_reg): Likewise.
9631 (clear_reload_reg_in_use): Likewise.
9632 (free_for_value_p): Likewise.
9633 (allocate_reload_reg): Likewise.
9634 (choose_reload_regs): Likewise.
9635 (reload_adjust_reg_for_temp): Likewise.
9636 (emit_reload_insns): Likewise.
9637 (delete_output_reload): Likewise.
9638 * rtlanal.c (subreg_get_info): Likewise.
9639 * sched-deps.c (sched_analyze_reg): Likewise.
9640 * sel-sched.c (init_regs_for_mode): Likewise.
9641 (mark_unavailable_hard_regs): Likewise.
9642 (choose_best_reg_1): Likewise.
9643 (verify_target_availability): Likewise.
9644 * valtrack.c (dead_debug_insert_temp): Likewise.
9645 * var-tracking.c (track_loc_p): Likewise.
9646 (emit_note_insn_var_location): Likewise.
9647 * varasm.c (make_decl_rtl): Likewise.
9648 * reginfo.c (choose_hard_reg_mode): Likewise.
9649 (init_reg_modes_target): Refer directly to
9650 this_target_regs->x_hard_regno_nregs.
9652 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9654 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9655 instead of hard_regno_nregs.
9657 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9659 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9660 end_hard_regno instead of hard_regno_nregs.
9661 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9662 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9663 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9664 * ira-color.c (improve_allocation): Likewise.
9665 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9666 * lra-lives.c (mark_regno_live): Likewise.
9667 (mark_regno_dead): Likewise.
9668 * lra-remat.c (operand_to_remat): Likewise.
9669 * lra.c (collect_non_operand_hard_regs): Likewise.
9670 * postreload.c (reload_combine_note_store): Likewise.
9671 (move2add_valid_value_p): Likewise.
9672 * reload.c (regno_clobbered_p): Likewise.
9674 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9676 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9678 * config/v850/v850.c (v850_reorg): Likewise.
9679 * reload.c (refers_to_regno_for_reload_p): Likewise.
9680 (find_equiv_reg): Likewise.
9681 * reload1.c (reload_reg_reaches_end_p): Likewise.
9683 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9685 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9687 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9688 * config/arm/arm.c (output_move_neon): Likewise.
9689 (arm_attr_length_move_neon): Likewise.
9690 (neon_split_vcombine): Likewise.
9691 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9692 (c6x_mark_reg_written): Likewise.
9693 (c6x_dwarf_register_span): Likewise.
9694 * config/i386/i386.c (ix86_save_reg): Likewise.
9695 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9696 (rws_access_reg): Likewise.
9697 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9698 * mode-switching.c (create_pre_exit): Likewise.
9699 * ree.c (combine_reaching_defs): Likewise.
9700 (add_removable_extension): Likewise.
9701 * regcprop.c (find_oldest_value_reg): Likewise.
9702 (copyprop_hardreg_forward_1): Likewise.
9703 * reload.c (reload_inner_reg_of_subreg): Likewise.
9704 (push_reload): Likewise.
9705 (combine_reloads): Likewise.
9706 (find_dummy_reload): Likewise.
9707 (reload_adjust_reg_for_mode): Likewise.
9708 * reload1.c (find_reload_regs): Likewise.
9709 (forget_old_reloads_1): Likewise.
9710 (reload_reg_free_for_value_p): Likewise.
9711 (reload_adjust_reg_for_temp): Likewise.
9712 (emit_reload_insns): Likewise.
9713 (delete_output_reload): Likewise.
9714 * sel-sched.c (choose_best_reg_1): Likewise.
9715 (choose_best_pseudo_reg): Likewise.
9717 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9718 Alan Hayward <alan.hayward@arm.com>
9719 David Sherwood <david.sherwood@arm.com>
9721 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9722 * target.def (slow_unaligned_access): New hook.
9723 * targhooks.h (default_slow_unaligned_access): Declare.
9724 * targhooks.c (default_slow_unaligned_access): New function.
9725 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9726 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9727 * doc/tm.texi: Regenerate.
9728 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9729 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9730 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9732 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9733 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9735 (rs6000_slow_unaligned_access): New function.
9736 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9737 (expand_block_compare): Likewise.
9738 (expand_strn_compare): Likewise.
9739 (rs6000_rtx_costs): Likewise.
9740 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9741 (riscv_slow_unaligned_access): Likewise.
9742 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9743 (riscv_slow_unaligned_access_p): ...this and make static.
9744 (riscv_option_override): Update accordingly.
9745 (riscv_slow_unaligned_access): New function.
9746 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9747 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9748 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9749 (rs6000_slow_unaligned_access): New function.
9750 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9751 (rs6000_rtx_costs): Likewise.
9752 * config/rs6000/rs6000-string.c (expand_block_compare)
9753 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9754 of SLOW_UNALIGNED_ACCESS.
9755 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9756 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9757 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9758 of SLOW_UNALIGNED_ACCESS.
9759 * expmed.c (simple_mem_bitfield_p): Likewise.
9760 * expr.c (alignment_for_piecewise_move): Likewise.
9761 (emit_group_load_1): Likewise.
9762 (emit_group_store): Likewise.
9763 (copy_blkmode_from_reg): Likewise.
9764 (emit_push_insn): Likewise.
9765 (expand_assignment): Likewise.
9766 (store_field): Likewise.
9767 (expand_expr_real_1): Likewise.
9768 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9769 * lra-constraints.c (simplify_operand_subreg): Likewise.
9770 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9771 * gimple-ssa-store-merging.c: Likewise in block comment at start
9773 * tree-ssa-strlen.c: Include target.h.
9774 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9775 of SLOW_UNALIGNED_ACCESS.
9776 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9778 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9780 PR rtl-optimization/82185
9781 * expmed.c (emit_store_flag_int): Only test tem if it has been
9784 2017-09-12 Richard Biener <rguenther@suse.de>
9787 * match.pd ((FTYPE) N CMP CST): Fix typo.
9789 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9791 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9793 (mips_near_type_p): Add 'short_call' attribute as a synonym
9795 * doc/extend.texi (short_call): Document new function attribute.
9797 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9800 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9801 assertion check that in the condition.
9802 (get_atomic_generic_size): Likewise. Before testing if parameter
9803 has pointer type, if it has array type, call for C++
9804 default_conversion to perform array-to-pointer conversion.
9806 2017-09-12 Richard Biener <rguenther@suse.de>
9808 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9809 for operations we cannot scalarize.
9811 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9813 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9814 vectors heap vectors. Clean up comments.
9815 Make visited_bbs a reference.
9816 (profitable_jump_thread_path): Make GC
9817 vectors heap vectors. Clean up comments.
9819 (convert_and_register_jump_thread_path): Make GC vectors heap
9821 (check_subpath_and_update_thread_path): Same. Clean up comments.
9822 Make visited_bbs a reference.
9823 (handle_phi): Abstract common code to to
9824 register_jump_thread_path_if_profitable.
9825 Rename VAR_BB to DEF_BB.
9827 Make GC vectors heap vectors.
9828 Make visited_bbs a reference.
9829 (handle_assignment): Same.
9830 (register_jump_thread_path_if_profitable): New.
9831 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9833 Make GC vectors heap vectors. Clean up comments.
9834 Make visited_bbs a reference.
9835 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9836 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9839 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9842 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9843 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9845 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9847 Revert r251800 and r251799.
9849 2017-09-11 Martin Jambor <mjambor@suse.cz>
9852 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9853 arguments in advance.
9854 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9855 use it to find predecessor edges.
9856 (naive_outof_ssa): Collect vector of predecessors.
9858 2017-09-08 Jason Merrill <jason@redhat.com>
9860 PR c++/70029 - ICE with ref-qualifier and -flto
9861 * langhooks.h (struct lang_hooks_for_types): Add
9862 copy_lang_qualifiers.
9863 * attribs.c (build_type_attribute_qual_variant): Use it.
9864 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9866 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9867 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9869 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9872 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9873 (*mulsi3_sp64): New instruction.
9874 (mulsi3): New expander.
9876 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9878 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9880 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9882 * sancov.c: Include memmodel.h.
9884 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9887 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9890 2017-09-07 Carl Love <cel@us.ibm.com>
9892 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9893 the sldi instruction.
9895 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9897 * sancov.c: Include tm_p.h.
9899 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9902 * output.h (switch_to_other_text_partition): New declaration.
9903 * varasm.c (switch_to_other_text_partition): New function.
9904 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9905 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9906 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9907 to the other text partition before emitting LCL label and switch back
9908 after emitting the word after it.
9910 2017-09-07 Richard Biener <rguenther@suse.de>
9912 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9913 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9914 critical edges here, after loop init.
9915 (pass_data_pre): Remove PROP_no_crit_edges flags.
9916 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9917 for valueization of call args to avoid leaking VN_TOP.
9918 (visit_use): Assert we do not visit default defs.
9919 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9920 Use error_mark_node to more easily detect leaking VN_TOP.
9921 All default-defs are varying, not VN_TOP. Mark them visited.
9922 (run_scc_vn): Make code match comment.
9924 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9926 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9927 OPTION_MASK_FLOAT128_KEYWORD.
9928 (POWERPC_MASKS): Likewise.
9929 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9930 support for the -mfloat128-type option, and make -mfloat128
9931 default on PowerPC Linux systems. Define or undefine
9932 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9933 Define __float128 to be __ieee128 if IEEE 128-bit support is
9934 enabled, or undefine it.
9935 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9936 Delete defining __FLOAT128_TYPE__.
9937 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9938 -mfloat128-type option and make -mfloat128 default on PowerPC
9940 (TARGET_FLOAT128_TYPE): Likewise.
9941 (-mfloat128-type): Likewise.
9942 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9943 Delete the -mfloat128-type option and make -mfloat128 default on
9944 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9945 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9946 128-bit floating point is enabled. Change tests from using
9947 -mfloat128-type to -mfloat128.
9948 (rs6000_mangle_type): Use the correct mangling for the __float128
9949 type even if normal long double is restricted to 64-bits.
9950 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9952 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9953 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9954 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9956 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9957 documentation for -mfloat128.
9959 2017-09-06 Olivier Hainque <hainque@adacore.com>
9961 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9963 2017-09-06 Wish Wu <wishwu007@gmail.com>
9964 Jakub Jelinek <jakub@redhat.com>
9966 * asan.c (initialize_sanitizer_builtins): Add
9967 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9968 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9969 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9970 BT_FN_VOID_UINT64_PTR variables.
9971 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9972 (BT_FN_VOID_UINT16_UINT16): Likewise.
9973 (BT_FN_VOID_UINT32_UINT32): Likewise.
9974 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9975 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9976 (BT_FN_VOID_UINT64_PTR): Likewise.
9977 * common.opt (flag_sanitize_coverage): New variable.
9978 (fsanitize-coverage=trace-pc): Remove.
9979 (fsanitize-coverage=): Add.
9980 * flag-types.h (enum sanitize_coverage_code): New enum.
9981 * fold-const.c (fold_range_test): Disable non-short-circuit
9982 optimization if flag_sanitize_coverage.
9983 (fold_truth_andor): Likewise.
9984 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9985 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9986 (coverage_sanitizer_opts): New array.
9987 (get_closest_sanitizer_option): Add OPTS argument, handle also
9988 OPT_fsanitize_coverage_.
9989 (parse_sanitizer_options): Adjusted to also handle
9990 OPT_fsanitize_coverage_.
9991 (common_handle_option): Add OPT_fsanitize_coverage_.
9992 * sancov.c (instrument_comparison, instrument_switch): New function.
9993 (sancov_pass): Add trace-cmp support.
9994 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9995 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9996 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9997 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9998 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9999 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10000 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10001 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10002 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10003 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10005 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10007 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10008 error. Only quit immediately if parsing is complete.
10009 (BEGIN): Initialize fatal_err and parse_done.
10010 (begin fpu, end fpu): Check number of arguments.
10011 (begin arch, end arch): Likewise.
10012 (begin cpu, end cpu): Likewise.
10013 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10014 (optalias): Likewise.
10016 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10018 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10019 * config/arm/arm-isa.h: Delete. Move definitions to ...
10020 * arm-cpus.in: ... here. Use new feature and fgroup values.
10021 * config/arm/arm.c (arm_option_override): Use lower case for feature
10023 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10024 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10025 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10027 (print_isa_bits_for): New function.
10028 (gen_isa): New function.
10029 (gen_comm_data): Use print_isa_bits_for.
10030 (define feature): New keyword.
10031 (define fgroup): New keyword.
10032 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10033 (arm-isa.h): Add rule to generate file.
10034 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10035 case for feature bit names.
10037 2017-09-06 Richard Biener <rguenther@suse.de>
10039 * tree-ssa-pre.c (NECESSARY): Remove.
10040 (create_expression_by_pieces): Do not touch pass-local flags.
10041 (insert_into_preds_of_block): Likewise.
10042 (do_pre_regular_insertion): Likewise.
10043 (eliminate_insert): Likewise.
10044 (eliminate_dom_walker::before_dom_children): Likewise.
10045 (fini_eliminate): Do not look at inserted_exprs.
10046 (mark_operand_necessary): Remove.
10047 (remove_dead_inserted_code): Replace with simple work-list
10048 algorithm based on inserted_exprs and SSA uses.
10049 (pass_pre::execute): Re-order fini_eliminate and
10050 remove_dead_inserted_code.
10052 2017-09-06 Olivier Hainque <hainque@adacore.com>
10054 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10055 for VxWorks 7. Adjust surrounding comments.
10057 2017-09-06 Richard Biener <rguenther@suse.de>
10059 * gimple-ssa-strength-reduction.c
10060 (find_candidates_dom_walker::before_dom_children): Also allow
10063 2017-09-06 Richard Biener <rguenther@suse.de>
10065 PR tree-optimization/82108
10066 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10067 for gap in the non-permutation SLP case.
10069 2017-09-06 Martin Jambor <mjambor@suse.cz>
10071 PR tree-optimization/82078
10072 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10073 add_access_to_work_queue...
10074 (build_accesses_from_assign): ...here.
10075 (propagate_all_subaccesses): Make sure racc is the group
10076 representative, if there is one.
10078 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10080 PR middle-end/82095
10081 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10084 2017-09-06 Richard Biener <rguenther@suse.de>
10086 * gimple-ssa-strength-reduction.c
10087 (find_candidates_dom_walker::before_doom_children): Use a
10088 type and not a mode check.
10090 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10093 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10095 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10097 2017-09-05 Jeff Law <law@redhat.com>
10099 PR tree-optimization/64910
10100 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10101 cases where we have 3 or more operands.
10103 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10105 PR middle-end/81768
10106 * omp-low.c (lower_omp_for): Recompute tree invariant if
10107 gimple_omp_for_initial/final is ADDR_EXPR.
10109 PR middle-end/81768
10110 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10111 into gimple val before gimplification fo the COND_EXPR.
10113 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10115 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10116 REGION_COPY argument.
10117 (thread_through_all_blocks): Remove unused argument to
10118 duplicate_thread_path.
10120 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10121 Alan Hayward <alan.hayward@arm.com>
10122 David Sherwood <david.sherwood@arm.com>
10124 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10125 Take a scalar_mode rather than a machine_mode.
10126 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10127 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10128 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10129 (aarch64_gen_adjusted_ldpstp): Likewise.
10130 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10132 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10133 Alan Hayward <alan.hayward@arm.com>
10134 David Sherwood <david.sherwood@arm.com>
10136 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10137 Take a scalar_int_mode instead of a machine_mode.
10138 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10139 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10140 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10141 (aarch64_simd_attr_length_rglist): Delete.
10142 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10143 a scalar_int_mode instead of a machine_mode.
10144 (aarch64_add_offset): Likewise.
10145 (aarch64_internal_mov_immediate): Likewise
10146 (aarch64_add_constant_internal): Likewise.
10147 (aarch64_add_constant): Likewise.
10148 (aarch64_movw_imm): Likewise.
10149 (aarch64_rtx_arith_op_extract_p): Likewise.
10150 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10151 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10152 Remove assert that the mode isn't a vector.
10153 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10154 (aarch64_expand_mov_immediate): Update calls after above changes.
10155 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10156 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10157 (aarch64_move_imm): Likewise.
10158 (aarch64_can_const_movi_rtx_p): Likewise.
10159 (aarch64_strip_extend): Likewise.
10160 (aarch64_extr_rtx_p): Likewise.
10161 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10162 a CONST_INT when the mode parameter is VOIDmode.
10163 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10165 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10167 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10168 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10169 (bitwise_type_for_mode): Update accordingly.
10171 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10173 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10174 * stor-layout.c (mode_for_size_tree): Likewise.
10175 (mode_for_array): Update accordingly.
10176 (layout_decl): Likewise.
10177 (compute_record_mode): Likewise. Only set the mode once.
10179 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10181 * target.def (get_mask_mode): Change return type to opt_mode.
10183 * doc/tm.texi: Regenerate.
10184 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10185 * targhooks.c (default_get_mask_mode): Likewise.
10186 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10187 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10188 targetm.get_mask_mode.
10189 * tree.c (build_truth_vector_type): Likewise.
10191 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10193 * machmode.h (mode_for_vector): Return an opt_mode.
10194 * stor-layout.c (mode_for_vector): Likewise.
10195 (mode_for_int_vector): Update accordingly.
10196 (layout_type): Likewise.
10197 * config/i386/i386.c (emit_memmov): Likewise.
10198 (ix86_expand_set_or_movmem): Likewise.
10199 (ix86_expand_vector_init): Likewise.
10200 (ix86_get_mask_mode): Likewise.
10201 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10203 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10204 * expmed.c (extract_bit_field_1): Likewise.
10205 * expr.c (expand_expr_real_2): Likewise.
10206 * optabs-query.c (can_vec_perm_p): Likewise.
10207 (can_vec_mask_load_store_p): Likewise.
10208 * optabs.c (expand_vec_perm): Likewise.
10209 * targhooks.c (default_get_mask_mode): Likewise.
10210 * tree-vect-stmts.c (vectorizable_store): Likewise.
10211 (vectorizable_load): Likewise.
10212 (get_vectype_for_scalar_type_and_size): Likewise.
10214 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10216 * machmode.h (mode_for_int_vector): New function.
10217 * stor-layout.c (mode_for_int_vector): Likewise.
10218 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10219 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10220 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10221 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10222 (s390_expand_vcond): Likewise.
10224 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10226 * machmode.h (opt_machine_mode): New type.
10227 (opt_mode<T>): Allow construction from anything that can be
10229 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10230 (mode_for_size): Return an opt_machine_mode.
10231 * stor-layout.c (mode_for_size): Likewise.
10232 (mode_for_size_tree): Update call accordingly.
10233 (bitwise_mode_for_mode): Likewise.
10234 (make_fract_type): Likewise.
10235 (make_accum_type): Likewise.
10236 * caller-save.c (replace_reg_with_saved_mem): Update call
10238 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10239 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10240 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10241 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10242 * expmed.c (extract_bit_field_1): Likewise.
10243 * reload.c (get_secondary_mem): Likewise.
10244 * varasm.c (assemble_integer): Likewise.
10245 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10248 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10250 * machmode.h (decimal_float_mode_for_size): New function.
10251 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10252 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10253 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10254 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10255 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10256 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10258 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10260 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10261 (get_builtin_sync_mode): Likewise.
10262 (expand_ifn_atomic_compare_exchange): Likewise.
10263 (expand_builtin_atomic_clear): Likewise.
10264 (expand_builtin_atomic_test_and_set): Likewise.
10265 (fold_builtin_atomic_always_lock_free): Likewise.
10266 * calls.c (compute_argument_addresses): Likewise.
10267 (emit_library_call_value_1): Likewise.
10268 (store_one_arg): Likewise.
10269 * combine.c (combine_instructions): Likewise.
10270 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10271 * config/arm/arm.c (arm_function_value): Likewise.
10272 (aapcs_allocate_return_reg): Likewise.
10273 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10274 * config/i386/i386.c (construct_container): Likewise.
10275 (ix86_gimplify_va_arg): Likewise.
10276 (ix86_expand_sse_cmp): Likewise.
10277 (emit_memmov): Likewise.
10278 (emit_memset): Likewise.
10279 (expand_small_movmem_or_setmem): Likewise.
10280 (ix86_expand_pextr): Likewise.
10281 (ix86_expand_pinsr): Likewise.
10282 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10283 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10285 * config/mips/mips.c (mips_function_value_1) Likewise.
10286 (mips_block_move_straight): Likewise.
10287 (mips_expand_ins_as_unaligned_store): Likewise.
10288 * config/powerpcspe/powerpcspe.c
10289 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10290 (rs6000_darwin64_record_arg_flush): Likewise.
10291 * config/rs6000/rs6000.c
10292 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10293 (rs6000_darwin64_record_arg_flush): Likewise.
10294 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10295 (sparc_function_value_1): Likewise.
10296 * config/spu/spu.c (adjust_operand): Likewise.
10297 (spu_emit_branch_or_set): Likewise.
10298 (arith_immediate_p): Likewise.
10299 * emit-rtl.c (gen_lowpart_common): Likewise.
10300 * expr.c (expand_expr_real_1): Likewise.
10301 * function.c (assign_parm_setup_block): Likewise.
10302 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10303 * reload1.c (alter_reg): Likewise.
10304 * stor-layout.c (mode_for_vector): Likewise.
10305 (layout_type): Likewise.
10307 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10309 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10310 (spu_convert_move): Likewise.
10311 * lower-subreg.c (resolve_simple_move): Likewise.
10313 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10316 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10317 define_insn to a define_expand.
10318 (altivec_vsum2sws_direct): New define_insn.
10319 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10321 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10323 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10326 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10331 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10332 realign stack if stack alignment needed is less than incoming
10335 2017-09-05 Marek Polacek <polacek@redhat.com>
10338 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10341 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10343 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10345 2017-09-05 Richard Biener <rguenther@suse.de>
10347 PR tree-optimization/82084
10348 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10350 2017-09-05 Richard Biener <rguenther@suse.de>
10352 PR tree-optimization/82102
10353 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10355 2017-09-05 Martin Liska <mliska@suse.cz>
10357 PR tree-optimization/82032
10358 * tree-cfg.c (generate_range_test): New function.
10359 * tree-cfg.h (generate_range_test): Declared here.
10360 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10361 (cleanup_control_expr_graph): Use it.
10362 * tree-switch-conversion.c (try_switch_expansion): Remove
10364 (emit_case_nodes): Use generate_range_test.
10366 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10369 * config/i386/i386.md (*<btsc><mode>_mask): Add
10370 TARGET_USE_BT to insn constraint.
10371 (*btr<mode>_mask): Ditto.
10373 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10375 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10376 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10378 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10381 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10382 TARGET_NEON and TARGET_IWMMXT.
10383 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10384 TARGET_NEON and TARGET_IWMMXT.
10385 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10387 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10389 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10390 (ix86_rewrite_tls_address): Ditto.
10391 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10392 (ix86_rewrite_tls_address_1): Ditto.
10393 (ix86_rewrite_tls_address): Ditto.
10394 * config/i386/predicates.md (tls_address_pattern): New predicate.
10395 * config/i386/i386.md (TLS address splitter): New splitter.
10397 2017-09-04 Richard Biener <rguenther@suse.de>
10399 PR tree-optimization/82084
10400 * fold-const.h (can_native_encode_string_p): Declare.
10401 * fold-const.c (can_native_encode_string_p): Factor out from ...
10402 (native_encode_string): ... here.
10403 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10404 vectorizing stores from constants we later cannot handle.
10406 2017-09-04 Marek Polacek <polacek@redhat.com>
10409 * doc/invoke.texi: Update -Wtautological-compare documentation.
10411 2017-09-04 Jeff Law <law@redhat.com>
10413 PR tree-optimization/64910
10414 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10415 swap the first and last operand if the last is a constant.
10417 2017-09-04 Marek Polacek <polacek@redhat.com>
10420 * convert.c (do_narrow): When sanitizing signed integer overflows,
10421 bail out for signed types.
10422 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10424 2017-09-04 Richard Biener <rguenther@suse.de>
10426 PR tree-optimization/82060
10427 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10428 Move devirtualization after stmt folding and before EH/AB/noreturn
10429 cleanup to get the stmt refs canonicalized. Use a bool instead
10430 of gimple_modified_p since that doesn't work for NOPs. Schedule
10431 NOPs generated by folding for removal.
10433 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10434 Alan Hayward <alan.hayward@arm.com>
10435 David Sherwood <david.sherwood@arm.com>
10437 * coretypes.h (pad_direction): New enum.
10438 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10439 (FUNCTION_ARG_PADDING): Likewise.
10440 * target.def (function_arg_padding): New hook.
10441 * targhooks.h (default_function_arg_padding): Declare.
10442 * targhooks.c (default_function_arg_padding): New function.
10443 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10444 (TARGET_FUNCTION_ARG_PADDING): ...this.
10445 * doc/tm.texi: Regenerate.
10446 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10447 instead of direction.
10448 (compute_argument_addresses): Likewise.
10449 (load_register_parameters): Likewise.
10450 (emit_library_call_value_1): Likewise.
10451 (store_one_arg): Use targetm.calls.function_arg_padding instead
10452 of FUNCTION_ARG_PADDING.
10453 (must_pass_in_stack_var_size_or_pad): Likewise.
10454 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10455 (emit_group_store): Likewise.
10456 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10457 instead of FUNCTION_ARG_PADDING.
10458 (emit_push_insn): Likewise, and propagate enum change throughout
10460 * function.h (direction): Delete.
10461 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10463 * function.c (assign_parm_find_stack_rtl): Likewise.
10464 (assign_parm_setup_block_p): Likewise.
10465 (assign_parm_setup_block): Likewise.
10466 (gimplify_parameters): Likewise.
10467 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10468 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10470 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10471 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10472 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10473 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10474 (aarch64_function_arg_padding): ...this new function.
10475 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10476 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10477 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10478 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10479 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10480 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10481 (arm_pad_arg_upward): Replace with...
10482 (arm_function_arg_padding): ...this new function.
10483 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10485 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10486 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10487 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10488 (ia64_hpux_function_arg_padding): Replace with...
10489 (ia64_function_arg_padding): ...this new function. Use pad_direction
10490 instead of direction. Check for TARGET_HPUX.
10491 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10492 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10493 (iq2000_function_arg_padding): New function.
10494 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10495 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10496 (mips_function_arg_padding): ...this new function.
10497 (mips_pad_reg_upward): Update accordingly.
10498 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10499 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10500 targetm.calls.function_arg_padding.
10501 (FUNCTION_ARG_PADDING): Delete.
10502 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10503 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10504 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10505 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10506 (nios2_block_reg_padding): Return pad_direction instead of direction.
10507 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10508 instead of direction.
10509 (nios2_function_arg_padding): Likewise. Make static.
10510 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10511 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10512 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10513 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10514 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10515 (pa_function_arg_padding): Make static. Return pad_direction instead
10517 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10518 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10519 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10520 instead of direction. Use targetm.calls.function_arg_padding.
10521 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10522 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10523 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10524 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10525 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10527 (function_arg_padding): Rename to...
10528 (rs6000_function_arg_padding): ...this. Make static. Return
10529 pad_direction instead of direction.
10530 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10531 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10532 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10533 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10534 instead of direction. Use targetm.calls.function_arg_padding.
10535 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10536 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10537 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10538 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10539 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10540 (function_arg_padding): Rename to...
10541 (rs6000_function_arg_padding): ...this. Make static. Return
10542 pad_direction instead of direction.
10543 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10544 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10545 * config/s390/s390.c (s390_function_arg_padding): New function.
10546 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10547 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10548 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10549 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10550 (function_arg_padding): Rename to...
10551 (sparc_function_arg_padding): ...this. Make static. Return
10552 pad_direction instead of direction.
10553 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10554 * config/spu/spu.c (spu_function_arg_padding): New function.
10555 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10556 * system.h (FUNCTION_ARG_PADDING): Poison.
10558 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10559 Alan Hayward <alan.hayward@arm.com>
10560 David Sherwood <david.sherwood@arm.com>
10562 * target.def (modes_tieable_p): New hook.
10563 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10564 (TARGET_MODES_TIEABLE_P): ...this.
10565 * doc/tm.texi.in: Regenerate.
10566 * hooks.h (hook_bool_mode_mode_true): Declare.
10567 * hooks.c (hook_bool_mode_mode_true): New function.
10568 * combine.c (subst): Use targetm.modes_tieable_p instead of
10570 * dse.c (find_shift_sequence): Likewise.
10571 * expmed.c (extract_low_bits): Likewise.
10572 * lower-subreg.c: Include target.h.
10573 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10575 * rtlanal.c (rtx_cost): Likewise.
10576 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10577 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10578 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10579 (TARGET_MODES_TIEABLE_P): Redefine.
10580 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10581 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10582 (TARGET_MODES_TIEABLE_P): Redefine.
10583 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10584 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10585 (arc_modes_tieable_p): New function.
10586 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10587 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10588 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10589 (arm_modes_tieable_p): Make static.
10590 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10591 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10592 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10593 (TARGET_MODES_TIEABLE_P): Redefine.
10594 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10595 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10596 (TARGET_MODES_TIEABLE_P): Redefine.
10597 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10598 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10599 (cr16_modes_tieable_p): New function.
10600 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10601 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10602 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10603 (TRULY_NOOP_TRUNCATION): Update comment.
10604 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10605 (TRULY_NOOP_TRUNCATION): Update comment.
10606 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10607 (frv_modes_tieable_p): New function.
10608 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10609 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10610 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10611 (TARGET_MODES_TIEABLE_P): Redefine.
10612 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10613 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10614 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10615 (TARGET_MODES_TIEABLE_P): Redefine.
10616 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10617 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10618 (ia64_modes_tieable_p): New function.
10619 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10620 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10621 (iq2000_modes_tieable_p): New function.
10622 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10623 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10624 (lm32_modes_tieable_p): New function.
10625 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10626 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10627 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10628 (TARGET_MODES_TIEABLE_P): Redefine.
10629 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10630 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10631 (m32r_modes_tieable_p): New function.
10632 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10633 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10634 (m68k_modes_tieable_p): New function.
10635 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10636 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10637 (mcore_modes_tieable_p): New function.
10638 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10639 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10641 (TARGET_MODES_TIEABLE_P): Redefine.
10642 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10643 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10644 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10645 (TARGET_MODES_TIEABLE_P): Redefine.
10646 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10647 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10648 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10649 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10650 (mn10300_modes_tieable_p): ...this and make static.
10651 (TARGET_MODES_TIEABLE_P): Redefine.
10652 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10653 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10654 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10655 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10656 (msp430_modes_tieable_p): Make static.
10657 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10658 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10659 (TARGET_MODES_TIEABLE_P): Redefine.
10660 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10661 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10662 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10663 (TARGET_MODES_TIEABLE_P): Redefine.
10664 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10665 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10666 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10667 (TARGET_MODES_TIEABLE_P): Redefine.
10668 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10669 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10670 (pdp11_modes_tieable_p): New function.
10671 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10672 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10673 (rs6000_modes_tieable_p): New function.
10674 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10675 * config/powerpcspe/powerpcspe.md: Update comment.
10676 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10677 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10678 (TARGET_MODES_TIEABLE_P): Redefine.
10679 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10680 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10681 (rl78_modes_tieable_p): New function.
10682 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10683 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10684 (rs6000_modes_tieable_p): New function.
10685 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10686 * config/rs6000/rs6000.md: Update comment.
10687 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10688 * config/rx/rx.c (rx_modes_tieable_p): New function.
10689 (TARGET_MODES_TIEABLE_P): Redefine.
10690 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10691 * config/s390/s390.c (s390_modes_tieable_p): New function.
10692 (TARGET_MODES_TIEABLE_P): Redefine.
10693 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10694 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10695 (sh_modes_tieable_p): New function.
10696 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10697 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10698 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10699 (sparc_modes_tieable_p): Make static.
10700 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10701 * config/spu/spu.c (spu_modes_tieable_p): New function.
10702 (TARGET_MODES_TIEABLE_P): Redefine.
10703 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10704 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10705 (TARGET_MODES_TIEABLE_P): Redefine.
10706 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10707 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10708 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10709 * config/v850/v850.c (v850_modes_tieable_p): New function.
10710 (TARGET_MODES_TIEABLE_P): Redefine.
10711 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10712 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10713 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10714 (visium_modes_tieable_p): New function.
10715 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10716 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10717 (xtensa_modes_tieable_p): New function.
10718 * system.h (MODES_TIEABLE_P): Poison.
10720 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10721 Alan Hayward <alan.hayward@arm.com>
10722 David Sherwood <david.sherwood@arm.com>
10724 * target.def (hard_regno_mode_ok): New hook.
10725 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10726 (TARGET_HARD_REGNO_MODE_OK): ...this.
10727 * doc/tm.texi.in: Regenerate.
10728 * hooks.h (hook_bool_uint_mode_true): Declare.
10729 * hooks.c (hook_bool_uint_mode_true): New function.
10730 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10731 HARD_REGNO_MODE_OK.
10732 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10733 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10734 instead of HARD_REGNO_MODE_OK.
10735 * caller-save.c: Include target.h.
10736 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10737 HARD_REGNO_MODE_OK.
10738 * combine.c (can_combine_p): Likewise.
10739 (combinable_i3pat): Likewise.
10740 (can_change_dest_mode): Likewise.
10741 * expr.c (init_expr_target): Likewise.
10742 (convert_move): Likewise.
10743 (convert_modes): Likewise.
10744 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10745 (setup_prohibited_mode_move_regs): Likewise.
10746 * ira.h (target_ira): Likewise.
10747 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10748 * lra-constraints.c (process_alt_operands): Likewise.
10749 (split_reg): Likewise.
10750 * recog.c (peep2_find_free_register): Likewise.
10751 * ree.c (combine_reaching_defs): Likewise.
10752 * regcprop.c (maybe_mode_change): Likewise.
10753 * reginfo.c (init_reg_sets_1): Likewise.
10754 (choose_hard_reg_mode): Likewise.
10755 (simplifiable_subregs): Likewise.
10756 * regrename.c (check_new_reg_p): Likewise.
10757 * reload.c (find_valid_class): Likewise.
10758 (find_valid_class_1): Likewise.
10759 (reload_inner_reg_of_subreg): Likewise.
10760 (push_reload): Likewise.
10761 (combine_reloads): Likewise.
10762 (find_dummy_reload): Likewise.
10763 (find_reloads): Likewise.
10764 * reload1.c (find_reg): Likewise.
10765 (set_reload_reg): Likewise.
10766 (allocate_reload_reg): Likewise.
10767 (choose_reload_regs): Likewise.
10768 (reload_adjust_reg_for_temp): Likewise.
10769 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10770 (simplify_subreg_regno): Likewise.
10771 * sel-sched.c (init_regs_for_mode): Likewise.
10772 * varasm.c (make_decl_rtl): Likewise.
10773 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10774 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10775 HARD_REGNO_MODE_OK.
10776 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10777 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10778 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10779 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10780 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10781 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10782 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10783 (arc_mode_class): Delete.
10784 (HARD_REGNO_MODE_OK): Delete.
10785 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10786 (arc_hard_regno_mode_ok): Rename old array to...
10787 (arc_hard_regno_mode_ok_modes): ...this.
10788 (arc_conditional_register_usage): Update accordingly.
10789 (arc_mode_class): Make static.
10790 (arc_hard_regno_mode_ok): New function.
10791 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10792 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10793 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10794 (arm_hard_regno_mode_ok): Make static.
10795 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10796 HARD_REGNO_MODE_OK.
10797 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10798 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10799 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10801 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10802 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10803 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10804 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10805 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10806 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10807 * config/bfin/predicates.md (valid_reg_operand): Use
10808 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10809 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10810 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10811 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10812 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10813 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10814 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10815 (cr16_hard_regno_mode_ok): Make static and return a bool.
10816 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10817 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10818 (cris_hard_regno_mode_ok): New function.
10819 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10820 (epiphany_mode_class): Delete.
10821 (HARD_REGNO_MODE_OK): Delete.
10822 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10823 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10824 (hard_regno_mode_ok): Rename to...
10825 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10826 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10827 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10828 HARD_REGNO_MODE_OK.
10829 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10830 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10831 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10832 (frv_hard_regno_mode_ok): Make static and return a bool.
10833 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10834 HARD_REGNO_MODE_OK.
10835 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10836 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10837 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10838 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10840 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10841 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10842 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10843 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10846 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10847 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10848 (ia64_hard_regno_mode_ok): New function.
10849 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10850 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10851 (iq2000_hard_regno_mode_ok): New function.
10852 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10853 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10854 (lm32_hard_regno_mode_ok): New function.
10855 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10856 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10857 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10858 instead of HARD_REGNO_MODE_OK.
10859 (m32c_hard_regno_ok): Rename to...
10860 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10861 (m32c_cannot_change_mode_class): Update accordingly.
10862 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10863 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10864 (m32r_mode_class): Delete.
10865 (HARD_REGNO_MODE_OK): Delete.
10866 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10867 (m32r_hard_regno_mode_ok): Rename to...
10868 (m32r_hard_regno_modes): ...this.
10869 (m32r_mode_class): Make static.
10870 (m32r_hard_regno_mode_ok): New function.
10871 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10872 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10873 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10874 (m68k_hard_regno_mode_ok): Make static.
10875 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10876 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10877 (mcore_hard_regno_mode_ok): New function.
10878 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10879 (HARD_REGNO_MODE_OK): Delete.
10880 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10882 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10883 (microblaze_hard_regno_mode_ok): New function.
10884 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10885 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10886 (mips_hard_regno_mode_ok): Delete.
10887 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10888 (mips_hard_regno_mode_ok_p): ...this and make static.
10889 (mips_hard_regno_mode_ok_p): Rename to...
10890 (mips_hard_regno_mode_ok_uncached): ...this.
10891 (mips_hard_regno_mode_ok): New function.
10892 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10893 of HARD_REGNO_MODE_OK.
10894 (mips_option_override): Update after above name changes.
10895 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10896 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10897 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10898 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10899 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10901 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10902 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10903 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10904 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10905 (msp430_hard_regno_mode_ok): Make static and return a bool.
10906 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10907 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10908 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10910 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10911 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10912 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10913 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10914 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10915 (PA_HARD_REGNO_MODE_OK): ...this
10916 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10917 (PA_HARD_REGNO_MODE_OK): ...this.
10918 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10919 (pa_hard_regno_mode_ok): New function.
10920 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10921 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10922 (pdp11_hard_regno_mode_ok): New function.
10923 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10924 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10926 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10928 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10929 (rs6000_hard_regno_mode_ok): Rename to...
10930 (rs6000_hard_regno_mode_ok_uncached): ...this.
10931 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10932 (rs6000_hard_regno_mode_ok): New function.
10933 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10934 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10935 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10936 (riscv_hard_regno_mode_ok): ...this and make static.
10937 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10938 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10939 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10940 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10941 (rl78_hard_regno_mode_ok): Make static and return bool.
10942 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10943 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10945 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10946 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10947 (rs6000_hard_regno_mode_ok): Rename to...
10948 (rs6000_hard_regno_mode_ok_uncached): ...this.
10949 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10950 (rs6000_hard_regno_mode_ok): New function.
10951 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10952 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10953 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10954 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10955 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10956 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10957 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10958 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10959 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10960 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10961 (sh_hard_regno_mode_ok): Make static.
10962 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10963 instead of HARD_REGNO_MODE_OK.
10964 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10965 (sparc_mode_class): Delete.
10966 (HARD_REGNO_MODE_OK): Delete.
10967 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10968 (hard_regno_mode_classes): Make static.
10969 (sparc_mode_class): Likewise.
10970 (sparc_hard_regno_mode_ok): New function.
10971 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10972 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10973 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10975 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10976 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10977 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10978 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10979 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10980 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10981 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10982 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10983 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10984 (visium_hard_regno_mode_ok): New function.
10985 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10986 instead of HARD_REGNO_MODE_OK.
10987 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10988 (HARD_REGNO_MODE_OK): Delete.
10989 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10990 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10991 (xtensa_option_override): Update accordingly.
10992 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10993 (xtensa_hard_regno_mode_ok): New function.
10994 * system.h (HARD_REGNO_MODE_OK): Poison.
10996 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10997 Alan Hayward <alan.hayward@arm.com>
10998 David Sherwood <david.sherwood@arm.com>
11000 * target.def (hard_regno_call_part_clobbered): New hook.
11001 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11002 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11003 * doc/tm.texi: Regenerate.
11004 * hooks.h (hook_bool_uint_mode_false): Declare.
11005 * hooks.c (hook_bool_uint_mode_false): New function.
11006 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11007 * cselib.c (cselib_process_insn): Use
11008 targetm.hard_regno_call_part_clobbered instead of
11009 HARD_REGNO_CALL_PART_CLOBBERED.
11010 * ira-conflicts.c (ira_build_conflicts): Likewise.
11011 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11012 * lra-constraints.c (need_for_call_save_p): Likewise.
11013 * lra-lives.c: Include target.h.
11014 (check_pseudos_live_through_calls): Use
11015 targetm.hard_regno_call_part_clobbered instead of
11016 HARD_REGNO_CALL_PART_CLOBBERED.
11017 * regcprop.c: Include target.h.
11018 (copyprop_hardreg_forward_1): Use
11019 targetm.hard_regno_call_part_clobbered instead of
11020 HARD_REGNO_CALL_PART_CLOBBERED.
11021 * reginfo.c (choose_hard_reg_mode): Likewise.
11022 * regrename.c (check_new_reg_p): Likewise.
11023 * reload.c (find_equiv_reg): Likewise.
11024 * reload1.c (emit_reload_insns): Likewise.
11025 * sched-deps.c (deps_analyze_insn): Likewise.
11026 * sel-sched.c (init_regs_for_mode): Likewise.
11027 (mark_unavailable_hard_regs): Likewise.
11028 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11029 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11030 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11032 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11033 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11034 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11036 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11038 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11039 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11040 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11042 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11043 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11044 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11046 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11047 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11049 * config/powerpcspe/powerpcspe.c
11050 (rs6000_hard_regno_call_part_clobbered): New function.
11051 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11052 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11053 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11055 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11056 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11057 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11059 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11060 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11061 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11063 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11064 Alan Hayward <alan.hayward@arm.com>
11065 David Sherwood <david.sherwood@arm.com>
11067 * rtl.h (subreg_memory_offset): Declare.
11068 * emit-rtl.c (subreg_memory_offset): New function.
11069 * expmed.c (store_bit_field_1): Use it.
11070 * expr.c (undefined_operand_subword_p): Likewise.
11071 * simplify-rtx.c (simplify_subreg): Likewise.
11073 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11075 PR rtl-optimization/57448
11078 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11079 using atomic_load pattern.
11080 (expand_atomic_store): Likewise.
11082 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11085 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11086 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11087 handling. Use replace_call_with_value with NULL instead of
11088 gsi_replace, unlink_stmt_vdef and release_defs.
11090 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11093 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11095 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11098 * rtl.h (emit_library_call_value_1): Declare.
11099 (emit_library_call): Replace declaration with a series of overloads.
11100 Remove the parameter count argument.
11101 (emit_library_call_value): Likewise.
11102 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11103 with an "rtx_mode_t *".
11104 (emit_library_call_value): Delete.
11105 (emit_library_call): Likewise.
11106 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11107 (asan_emit_allocas_unpoison): Likewise.
11108 * builtins.c (expand_builtin_powi): Likewise.
11109 (expand_asan_emit_allocas_unpoison): Likewise.
11110 * cfgexpand.c (expand_main_function): Likewise.
11111 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11112 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11113 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11114 * config/arm/arm.c (arm_trampoline_init): Likewise.
11115 (arm_call_tls_get_addr): Likewise.
11116 (arm_expand_divmod_libfunc): Likewise.
11117 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11118 (smulsi3_highpart): Likewise.
11119 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11120 (c6x_expand_compare): Likewise.
11121 (c6x_expand_movmem): Likewise.
11122 * config/frv/frv.c (frv_trampoline_init): Likewise.
11123 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11124 (ix86_expand_divmod_libfunc): Likewise.
11125 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11126 (ia64_expand_compare): Likewise.
11127 (ia64_profile_hook): Likewise.
11128 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11129 (nonlocal_goto): Likewise.
11130 (restore_stack_nonlocal): Likewise.
11131 * config/m32r/m32r.c (block_move_call): Likewise.
11132 (m32r_trampoline_init): Likewise.
11133 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11134 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11135 (m68k_call_m68k_read_tp): Likewise.
11136 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11137 (microblaze_expand_divide): Likewise.
11138 * config/mips/mips.h (mips_args): Likewise.
11139 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11140 (MIPS_ICACHE_SYNC): Likewise.
11141 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11142 (nios2_trampoline_init): Likewise.
11143 * config/pa/pa.c (hppa_tls_call): Likewise.
11144 (pa_trampoline_init): Likewise.
11145 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11146 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11147 (expand_strn_compare): Likewise.
11148 (rs6000_generate_compare): Likewise.
11149 (rs6000_expand_float128_convert): Likewise.
11150 (output_profile_hook): Likewise.
11151 (rs6000_trampoline_init): Likewise.
11152 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11153 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11154 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11155 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11156 (rs6000_generate_compare): Likewise.
11157 (rs6000_expand_float128_convert): Likewise.
11158 (output_profile_hook): Likewise.
11159 (rs6000_trampoline_init): Likewise.
11160 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11161 * config/sh/sh.c (sh_trampoline_init): Likewise.
11162 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11163 (sparc_emit_float_lib_cmp): Likewise.
11164 (sparc32_initialize_trampoline): Likewise.
11165 (sparc64_initialize_trampoline): Likewise.
11166 (sparc_profile_hook): Likewise.
11167 * config/spu/spu.c (ea_load_store): Likewise.
11168 * config/spu/spu.md (floatunssidf2): Likewise.
11169 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11170 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11171 * config/visium/visium.c (expand_block_move_4): Likewise.
11172 (expand_block_move_2): Likewise.
11173 (expand_block_move_1): Likewise.
11174 (expand_block_set_4): Likewise.
11175 (expand_block_set_2): Likewise.
11176 (expand_block_set_1): Likewise.
11177 (visium_trampoline_init): Likewise.
11178 (visium_profile_hook): Likewise.
11179 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11180 (xtensa_setup_frame_addresses): Likewise.
11181 (xtensa_trampoline_init): Likewise.
11182 * except.c (sjlj_emit_function_enter): Likewise.
11183 (sjlj_emit_function_exit): Likewise.
11184 * explow.c (allocate_dynamic_stack_space): Likewise.
11185 (probe_stack_range): Likewise.
11186 * expr.c (convert_mode_scalar): Likewise.
11187 * optabs.c (expand_binop): Likewise.
11188 (expand_twoval_binop_libfunc): Likewise.
11189 (expand_unop): Likewise.
11190 (prepare_cmp_insn): Likewise.
11191 (prepare_float_lib_cmp): Likewise.
11192 (expand_float): Likewise.
11193 (expand_fix): Likewise.
11194 (expand_fixed_convert): Likewise.
11195 (maybe_emit_sync_lock_test_and_set): Likewise.
11196 (expand_atomic_compare_and_swap): Likewise.
11197 (expand_mem_thread_fence): Likewise.
11198 (expand_atomic_fetch_op): Likewise.
11200 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11202 * doc/generic.texi (OpenACC): Adjust URL.
11203 * doc/invoke.texi (C Dialect Options): Ditto.
11205 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11207 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11208 predicate for operand 1. Add (m,<S>) constraint.
11209 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11210 Prevent memory operand 1 with register operand 2.
11212 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11214 PR rtl-optimization/82024
11215 * combine.c (try_combine): If the combination result is a PARALLEL,
11216 and we only need to retain the SET in there that would be placed
11217 at I2, check that we can place that at I3 instead, before doing so.
11219 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11222 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11224 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11225 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11228 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11229 Jeff Law <law@redhat.com>
11231 * varasm.c (bss_initializer_p): Do not put constants into .bss
11232 (categorize_decl_for_section): Handle bss_initializer_p returning
11233 false when DECL_INITIAL is NULL.
11235 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11238 * config/s390/s390.c (s390_can_inline_p): New function.
11240 2017-09-01 Jeff Law <law@redhat.com>
11242 PR tree-optimization/82052
11243 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11244 Always initialize the returned slot after a hash table miss
11245 when INSERT is true.
11247 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11249 * config/s390/s390.md (mem_signal_fence): Remove.
11250 * doc/md.texi (mem_signal_fence): Remove.
11251 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11253 * target-insns.def (mem_signal_fence): Remove.
11255 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11258 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11261 * asan.c (create_odr_indicator): Strip name encoding from assembler
11262 name before appending it after __odr_asan_.
11264 2017-09-01 Martin Liska <mliska@suse.cz>
11266 PR tree-optimization/82059
11267 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11268 frequency only when an edge is redirected.
11270 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11272 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11273 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11274 (arc_conditional_register_usage): Remove ARC600 lp_count
11276 (arc_file_start): Emit Tag_ARC_CPU_variation.
11277 (arc_can_use_doloop_p): New conditions to use ZOLs.
11278 (hwloop_fail): New function.
11279 (hwloop_optimize): Likewise.
11280 (hwloop_pattern_reg): Likewise.
11281 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11282 (arc_reorg_loops): New function, calls reorg_loops.
11283 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11284 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11286 (arc_loop_hazard): Remove function, functionality moved into
11288 (arc_hazard): Remove arc_loop_hazard call.
11289 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11290 into hwloop_optimize.
11291 (arc_label_align): Remove ZOL handling.
11292 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11293 * config/arc/arc.md (doloop_begin): Remove pattern.
11294 (doloop_begin_i): Likewise.
11295 (doloop_end_i): Likewise.
11296 (doloop_fallback): Likewise.
11297 (doloop_fallback_m): Likewise.
11298 (doloop_end): Reimplement expand.
11299 (arc_lp): New pattern for LP instruction.
11300 (loop_end): New pattern.
11301 (loop_fail): Likewise.
11302 (decrement_and_branch_until_zero): Likewise.
11303 * config/arc/arc.opt (mlpc-width): New option.
11304 * doc/invoke.texi (mlpc-width): Document option.
11306 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11308 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11309 (arc_ccfsm_advance): Fix checking for delay slots.
11310 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11312 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11314 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11316 (movhi_insn): Update store instruction constraint which are saving
11317 6-bit short immediates.
11318 (movsi_insn): Consider also short scaled load operations.
11319 (zero_extendhisi2_i): Use Usd constraint instead of T.
11320 (extendhisi2_i): Add q constraint.
11321 (arc_clzsi2): Add type and length attributes.
11322 (arc_ctzsi2): Likewise.
11323 * config/arc/constraints.md (Usc): Update constraint, the
11324 assembler can parse two relocations for a single instruction.
11326 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11328 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11329 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11331 2017-08-31 Olivier Hainque <hainque@adacore.com>
11333 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11334 match as powerpc-wrs-vxworks*.
11336 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11338 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11339 register constraint for by-element operand.
11340 (aarch64_mls_elt_merge<mode>): Likewise.
11342 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11344 * config/arc/arc.c (arc_can_follow_jump): Check for short
11347 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11349 * config.gcc: Use g.opt for arc.
11350 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11351 functionality moved to ...
11352 (legitimate_scaled_address_p): New function, ...here.
11353 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11354 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11355 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11357 (arc_override_options): Handle G option.
11358 (arc_output_pic_addr_const): Correct function definition.
11359 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11360 (arc_decl_anon_ns_mem_p): Delete.
11361 (arc_in_small_data_p): Overhaul this function to take into
11362 consideration the value given via G option.
11363 (arc_rewrite_small_data_1): Renamed and corrected old
11364 arc_rewrite_small_data function.
11365 (arc_rewrite_small_data): New function.
11366 (small_data_pattern): Don't use pic_offset_table_rtx.
11367 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11368 * config/arc/simdext.md (movmisalignv2hi): Use
11369 prepare_move_operands function.
11371 (movmisalign*): Likewise.
11372 * doc/invoke.texi (ARC options): Document -G option.
11374 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11376 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11378 * config/arc/arc.c (arc_print_operand): Output scalled address for
11379 sdata whenever is possible.
11380 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11381 load/stores are available.
11382 (compact_sda_memory_operand): Check for the alignment required by
11383 code density instructions.
11384 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11386 * config/arc/constraints.md (Usd): Update constraint.
11387 (Us0): New constraint.
11388 (Usc): Update constraint.
11390 2017-08-31 Richard Biener <rguenther@suse.de>
11392 PR middle-end/82054
11393 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11394 function only once.
11396 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11398 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11399 Resize type_signature.
11401 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11402 Alan Hayward <alan.hayward@arm.com>
11403 David Sherwood <david.sherwood@arm.com>
11405 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11406 subregs whose inner modes can be stored in GPRs.
11407 (aarch64_classify_index): Likewise.
11409 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11410 Alan Hayward <alan.hayward@arm.com>
11411 David Sherwood <david.sherwood@arm.com>
11413 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11414 (V_INT_EQUIV): ...this.
11415 (v_cmp_result): Rename to...
11416 (v_int_equiv): ...this.
11417 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11418 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11419 (copysign<mode>3): Likewise.
11420 (aarch64_simd_bsl<mode>_internal): Likewise.
11421 (aarch64_simd_bsl<mode>): Likewise.
11422 (vec_cmp<mode><mode>): Likewise.
11423 (vcond<mode><mode>): Likewise.
11424 (vcond<v_cmp_mixed><mode>): Likewise.
11425 (vcondu<mode><v_cmp_mixed>): Likewise.
11426 (aarch64_cm<optab><mode>): Likewise.
11427 (aarch64_cmtst<mode>): Likewise.
11428 (aarch64_fac<optab><mode>): Likewise.
11429 (vec_perm_const<mode>): Likewise.
11430 (vcond_mask_<mode><v_cmp_result>): Rename to...
11431 (vcond_mask_<mode><v_int_equiv>): ...this.
11432 (vec_cmp<mode><v_cmp_result>): Rename to...
11433 (vec_cmp<mode><v_int_equiv>): ...this.
11435 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11436 Alan Hayward <alan.hayward@arm.com>
11437 David Sherwood <david.sherwood@arm.com>
11439 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11441 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11442 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11443 (UNSPEC_LD4_DREG): New unspecs.
11444 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11445 (aarch64_ld2<mode>_dreg_be): Replace with...
11446 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11448 (aarch64_ld3<mode>_dreg_le)
11449 (aarch64_ld3<mode>_dreg_be): Replace with...
11450 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11452 (aarch64_ld4<mode>_dreg_le)
11453 (aarch64_ld4<mode>_dreg_be): Replace with...
11454 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11457 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11459 PR tree-optimization/81987
11460 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11461 insert an initializer in a location not dominated by the stride
11464 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11466 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11467 on the entire header of the finally block in the fallthru case.
11469 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11471 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11473 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11475 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11476 rs6000_emit_move_from_cr and call renamed function.
11477 (rs6000_emit_prologue): Call renamed functions.
11478 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11479 movesi_from_cr, remove volatile CRs.
11481 2017-08-30 Jon Beniston <jon@beniston.com>
11482 Richard Biener <rguenther@suse.de>
11484 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11485 of VECTOR_MODE_P check.
11486 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11487 element vector types.
11489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11491 * df.h (df_read_modify_subreg_p): Remove in favor of...
11492 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11493 const_rtx instead of an rtx.
11494 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11495 * df-problems.c (df_word_lr_mark_ref): Likewise.
11496 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11497 (mark_pseudo_reg_dead): Likewise.
11498 (mark_ref_dead): Likewise.
11499 * reginfo.c (init_subregs_of_mode): Likewise.
11500 * sched-deps.c (sched_analyze_1): Likewise.
11501 * df-scan.c (df_def_record_1): Likewise.
11502 (df_uses_record): Likewise.
11503 (df_read_modify_subreg_p): Remove in favor of...
11504 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11505 const_rtx instead of an rtx.
11507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11508 Alan Hayward <alan.hayward@arm.com>
11509 David Sherwood <david.sherwood@arm.com>
11511 * rtl.h (partial_subreg_p): New function.
11512 * caller-save.c (save_call_clobbered_regs): Use it.
11513 * calls.c (expand_call): Likewise.
11514 * combine.c (combinable_i3pat): Likewise.
11515 (simplify_set): Likewise.
11516 (make_extraction): Likewise.
11517 (make_compound_operation_int): Likewise.
11518 (gen_lowpart_or_truncate): Likewise.
11519 (force_to_mode): Likewise.
11520 (make_field_assignment): Likewise.
11521 (reg_truncated_to_mode): Likewise.
11522 (record_truncated_value): Likewise.
11523 (move_deaths): Likewise.
11524 * cse.c (record_jump_cond): Likewise.
11525 (cse_insn): Likewise.
11526 * cselib.c (cselib_lookup_1): Likewise.
11527 * expmed.c (extract_bit_field_using_extv): Likewise.
11528 * function.c (assign_parm_setup_reg): Likewise.
11529 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11530 * ira-build.c (create_insn_allocnos): Likewise.
11531 * lra-coalesce.c (merge_pseudos): Likewise.
11532 * lra-constraints.c (match_reload): Likewise.
11533 (simplify_operand_subreg): Likewise.
11534 (curr_insn_transform): Likewise.
11535 * lra-lives.c (process_bb_lives): Likewise.
11536 * lra.c (new_insn_reg): Likewise.
11537 (lra_substitute_pseudo): Likewise.
11538 * regcprop.c (mode_change_ok): Likewise.
11539 (maybe_mode_change): Likewise.
11540 (copyprop_hardreg_forward_1): Likewise.
11541 * reload.c (push_reload): Likewise.
11542 (find_reloads): Likewise.
11543 (find_reloads_subreg_address): Likewise.
11544 * reload1.c (alter_reg): Likewise.
11545 (eliminate_regs_1): Likewise.
11546 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11548 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11550 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11551 back to if statements, including unpack.
11553 2017-08-30 Martin Liska <mliska@suse.cz>
11555 PR inline-asm/82001
11556 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11558 (func_checker::compare_asm_inputs_outputs): ... this function.
11559 (func_checker::compare_gimple_asm): Use the function to compare
11560 also ASM constrains.
11561 * ipa-icf-gimple.h: Rename the function.
11563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11564 Alan Hayward <alan.hayward@arm.com>
11565 David Sherwood <david.sherwood@arm.com>
11567 * coretypes.h (complex_mode): New type.
11568 * gdbhooks.py (build_pretty_printer): Handle it.
11569 * machmode.h (complex_mode): New class.
11570 (complex_mode::includes_p): New function.
11571 (is_complex_int_mode): Likewise.
11572 (is_complex_float_mode): Likewise.
11573 * genmodes.c (get_mode_class): Handle complex mode classes.
11574 * function.c (expand_function_end): Use is_complex_int_mode.
11576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11577 Alan Hayward <alan.hayward@arm.com>
11578 David Sherwood <david.sherwood@arm.com>
11580 * coretypes.h (scalar_mode_pod): New typedef.
11581 * gdbhooks.py (build_pretty_printer): Handle it.
11582 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11583 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11584 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11585 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11586 as_a <scalar_mode>.
11588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11589 Alan Hayward <alan.hayward@arm.com>
11590 David Sherwood <david.sherwood@arm.com>
11592 * machmode.h (mode_for_vector): Take a scalar_mode instead
11594 * stor-layout.c (mode_for_vector): Likewise.
11595 * explow.c (promote_mode): Use as_a <scalar_mode>.
11596 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11599 Alan Hayward <alan.hayward@arm.com>
11600 David Sherwood <david.sherwood@arm.com>
11602 * target.def (preferred_simd_mode): Take a scalar_mode
11603 instead of a machine_mode.
11604 * targhooks.h (default_preferred_simd_mode): Likewise.
11605 * targhooks.c (default_preferred_simd_mode): Likewise.
11606 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11607 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11608 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11609 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11610 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11611 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11612 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11613 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11615 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11616 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11617 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11618 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11619 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11620 * doc/tm.texi: Regenerate.
11621 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11625 Alan Hayward <alan.hayward@arm.com>
11626 David Sherwood <david.sherwood@arm.com>
11628 * target.def (scalar_mode_supported_p): Take a scalar_mode
11629 instead of a machine_mode.
11630 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11631 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11632 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11633 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11634 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11635 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11636 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11637 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11638 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11639 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11640 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11641 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11642 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11643 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11645 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11646 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11647 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11648 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11649 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11651 * doc/tm.texi: Regenerate.
11653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11654 Alan Hayward <alan.hayward@arm.com>
11655 David Sherwood <david.sherwood@arm.com>
11657 * coretypes.h (opt_scalar_mode): New typedef.
11658 * gdbhooks.py (build_pretty_printers): Handle it.
11659 * machmode.h (mode_iterator::get_2xwider): Add overload for
11661 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11663 * expr.c (convert_mode_scalar): Likewise.
11664 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11665 * optabs.c (expand_float): Likewise.
11666 (expand_fix): Likewise.
11667 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11670 Alan Hayward <alan.hayward@arm.com>
11671 David Sherwood <david.sherwood@arm.com>
11673 * optabs.c (expand_float): Explicitly check for scalars before
11674 using a branching expansion.
11675 (expand_fix): Likewise.
11677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11678 Alan Hayward <alan.hayward@arm.com>
11679 David Sherwood <david.sherwood@arm.com>
11681 * expr.c (convert_mode): Split scalar handling out into...
11682 (convert_mode_scalar): ...this new function. Treat the modes
11685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11686 Alan Hayward <alan.hayward@arm.com>
11687 David Sherwood <david.sherwood@arm.com>
11689 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11691 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11694 Alan Hayward <alan.hayward@arm.com>
11695 David Sherwood <david.sherwood@arm.com>
11697 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11698 rather than a machine_mode.
11699 (fixed_from_string): Likewise.
11700 (fixed_convert): Likewise.
11701 (fixed_convert_from_int): Likewise.
11702 (fixed_convert_from_real): Likewise.
11703 (real_convert_from_fixed): Likewise.
11704 * fixed-value.c (fixed_from_double_int): Likewise.
11705 (fixed_from_string): Likewise.
11706 (fixed_convert): Likewise.
11707 (fixed_convert_from_int): Likewise.
11708 (fixed_convert_from_real): Likewise.
11709 (real_convert_from_fixed): Likewise.
11710 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11713 Alan Hayward <alan.hayward@arm.com>
11714 David Sherwood <david.sherwood@arm.com>
11716 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11717 of separate mode class checks. Do not allow vector modes here.
11718 (immed_wide_int_const): Use as_a <scalar_mode>.
11719 * explow.c (trunc_int_for_mode): Likewise.
11720 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11721 (wi::shwi): Likewise.
11722 (wi::min_value): Likewise.
11723 (wi::max_value): Likewise.
11724 * dwarf2out.c (loc_descriptor): Likewise.
11725 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11726 for CONST_WIDE_INT.
11728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11729 Alan Hayward <alan.hayward@arm.com>
11730 David Sherwood <david.sherwood@arm.com>
11732 * tree.h (SCALAR_TYPE_MODE): New macro.
11733 * expr.c (expand_expr_addr_expr_1): Use it.
11734 (expand_expr_real_2): Likewise.
11735 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11736 (fold_convert_const_fixed_from_int): Likewise.
11737 (fold_convert_const_fixed_from_real): Likewise.
11738 (native_encode_fixed): Likewise
11739 (native_encode_complex): Likewise
11740 (native_encode_vector): Likewise.
11741 (native_interpret_fixed): Likewise.
11742 (native_interpret_real): Likewise.
11743 (native_interpret_complex): Likewise.
11744 (native_interpret_vector): Likewise.
11745 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11746 (simd_clone_adjust_argument_types): Likewise.
11747 (simd_clone_init_simd_arrays): Likewise.
11748 (simd_clone_adjust): Likewise.
11749 * stor-layout.c (layout_type): Likewise.
11750 * tree.c (build_minus_one_cst): Likewise.
11751 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11752 * tree-inline.c (estimate_move_cost): Likewise.
11753 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11754 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11755 (vectorizable_reduction): Likewise.
11756 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11757 (vect_recog_mixed_size_cond_pattern): Likewise.
11758 (check_bool_pattern): Likewise.
11759 (adjust_bool_pattern): Likewise.
11760 (search_type_for_mask_1): Likewise.
11761 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11762 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11763 (vectorizable_load): Likewise.
11764 (vectorizable_store): Likewise.
11765 * ubsan.c (ubsan_encode_value): Likewise.
11766 * varasm.c (output_constant): Likewise.
11768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11769 Alan Hayward <alan.hayward@arm.com>
11770 David Sherwood <david.sherwood@arm.com>
11772 * coretypes.h (scalar_mode): New class.
11773 * machmode.h (scalar_mode): Likewise.
11774 (scalar_mode::includes_p): New function.
11775 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11776 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11777 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11778 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11779 * expmed.c (store_bit_field_1): Likewise.
11780 (extract_bit_field_1): Likewise.
11781 * expr.c (write_complex_part): Likewise.
11782 (read_complex_part): Likewise.
11783 (emit_move_complex_push): Likewise.
11784 (expand_expr_real_2): Likewise.
11785 * function.c (assign_parm_setup_reg): Likewise.
11786 (assign_parms_unsplit_complex): Likewise.
11787 * optabs.c (expand_binop): Likewise.
11788 * rtlanal.c (subreg_get_info): Likewise.
11789 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11790 * varasm.c (output_constant_pool_2): Likewise.
11792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11793 Alan Hayward <alan.hayward@arm.com>
11794 David Sherwood <david.sherwood@arm.com>
11796 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11798 (expmed_mult_highpart_optab): Likewise.
11799 (expmed_mult_highpart): Likewise.
11801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11802 Alan Hayward <alan.hayward@arm.com>
11803 David Sherwood <david.sherwood@arm.com>
11805 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11806 instead of a machine_mode.
11807 (builtin_memset_read_str): Likewise.
11808 * builtins.c (c_readstr): Likewise.
11809 (builtin_memcpy_read_str): Likewise.
11810 (builtin_strncpy_read_str): Likewise.
11811 (builtin_memset_read_str): Likewise.
11812 (builtin_memset_gen_str): Likewise.
11813 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11814 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11815 instead of a machine_mode.
11816 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11818 (make_extraction): Likewise.
11819 (try_widen_shift_mode): Take and return scalar_int_modes instead
11821 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11822 a scalar_int_mode instead of a machine_mode.
11823 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11824 (avr_addr_space_pointer_mode): Likewise.
11825 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11826 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11827 (msp430_unwind_word_mode): Likewise.
11828 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11829 (spu_addr_space_pointer_mode): Likewise.
11830 (spu_addr_space_address_mode): Likewise.
11831 (spu_libgcc_cmp_return_mode): Likewise.
11832 (spu_libgcc_shift_count_mode): Likewise.
11833 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11834 (rl78_addr_space_pointer_mode): Likewise.
11835 (fl78_unwind_word_mode): Likewise.
11836 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11838 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11839 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11840 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11841 (mips_valid_pointer_mode): Likewise.
11842 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11843 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11844 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11846 (ft32_addr_space_address_mode): Likewise.
11847 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11848 scalar_int_mode instead of a machine_mode.
11849 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11851 (m32c_addr_space_address_mode): Likewise.
11852 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11853 (rs6000_eh_return_filter_mode): Likewise.
11854 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11855 (rs6000_eh_return_filter_mode): Likewise.
11856 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11857 (s390_libgcc_shift_count_mode): Likewise.
11858 (s390_unwind_word_mode): Likewise.
11859 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11861 * target.def (mode_rep_extended): Likewise.
11862 (valid_pointer_mode): Likewise.
11863 (addr_space.valid_pointer_mode): Likewise.
11864 (eh_return_filter_mode): Return a scalar_int_mode rather than
11866 (libgcc_cmp_return_mode): Likewise.
11867 (libgcc_shift_count_mode): Likewise.
11868 (unwind_word_mode): Likewise.
11869 (addr_space.pointer_mode): Likewise.
11870 (addr_space.address_mode): Likewise.
11871 * doc/tm.texi: Regenerate.
11872 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11874 (do_jump): Use scalar_int_mode for local variables.
11875 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11876 rather than a machine_mode.
11877 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11878 (scompare_loc_descriptor_wide): Likewise.
11879 (scompare_loc_descriptor_narrow): Likewise.
11880 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11882 * except.c (sjlj_emit_dispatch_table): Likewise.
11883 (expand_builtin_eh_copy_values): Likewise.
11884 * explow.c (convert_memory_address_addr_space_1): Likewise.
11885 Take a scalar_int_mode rather than a machine_mode.
11886 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11887 than a machine_mode.
11888 (memory_address_addr_space): Use scalar_int_mode for local variables.
11889 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11890 rather than a machine_mode.
11891 * expmed.c (mask_rtx): Likewise.
11892 (init_expmed_one_conv): Likewise.
11893 (expand_mult_highpart_adjust): Likewise.
11894 (extract_high_half): Likewise.
11895 (expmed_mult_highpart_optab): Likewise.
11896 (expmed_mult_highpart): Likewise.
11897 (expand_smod_pow2): Likewise.
11898 (expand_sdiv_pow2): Likewise.
11899 (emit_store_flag_int): Likewise.
11900 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11902 (extract_low_bits): Likewise.
11903 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11905 * expr.c (pieces_addr::adjust): Likewise.
11906 (can_store_by_pieces): Likewise.
11907 (store_by_pieces): Likewise.
11908 (clear_by_pieces_1): Likewise.
11909 (expand_expr_addr_expr_1): Likewise.
11910 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11911 (expand_expr_real_1): Likewise.
11912 (try_casesi): Likewise.
11913 * final.c (shorten_branches): Likewise.
11914 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11915 type of "mode" to machine_mode.
11916 * internal-fn.c (expand_arith_overflow_result_store): Take a
11917 scalar_int_mode rather than a machine_mode.
11918 (expand_mul_overflow): Use scalar_int_mode for local variables.
11919 * loop-doloop.c (doloop_modify): Likewise.
11920 (doloop_optimize): Likewise.
11921 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11922 than a machine_mode.
11923 (expand_doubleword_shift_condmove): Likewise.
11924 (expand_doubleword_shift): Likewise.
11925 (expand_doubleword_clz): Likewise.
11926 (expand_doubleword_popcount): Likewise.
11927 (expand_doubleword_parity): Likewise.
11928 (expand_absneg_bit): Use scalar_int_mode for local variables.
11929 (prepare_float_lib_cmp): Likewise.
11930 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11931 rather than a machine_mode.
11932 (convert_memory_address_addr_space): Likewise.
11933 (get_mode_bounds): Likewise.
11934 (get_address_mode): Return a scalar_int_mode rather than a
11936 * rtlanal.c (get_address_mode): Likewise.
11937 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11938 than a machine_mode.
11939 * targhooks.c (default_mode_rep_extended): Likewise.
11940 (default_valid_pointer_mode): Likewise.
11941 (default_addr_space_valid_pointer_mode): Likewise.
11942 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11943 than a machine_mode.
11944 (default_libgcc_cmp_return_mode): Likewise.
11945 (default_libgcc_shift_count_mode): Likewise.
11946 (default_unwind_word_mode): Likewise.
11947 (default_addr_space_pointer_mode): Likewise.
11948 (default_addr_space_address_mode): Likewise.
11949 * targhooks.h (default_eh_return_filter_mode): Likewise.
11950 (default_libgcc_cmp_return_mode): Likewise.
11951 (default_libgcc_shift_count_mode): Likewise.
11952 (default_unwind_word_mode): Likewise.
11953 (default_addr_space_pointer_mode): Likewise.
11954 (default_addr_space_address_mode): Likewise.
11955 (default_mode_rep_extended): Take a scalar_int_mode rather than
11957 (default_valid_pointer_mode): Likewise.
11958 (default_addr_space_valid_pointer_mode): Likewise.
11959 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11961 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11962 rather than a machine_mode.
11963 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11964 for local variables.
11965 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11966 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11967 than a machine_mode.
11969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11970 Alan Hayward <alan.hayward@arm.com>
11971 David Sherwood <david.sherwood@arm.com>
11973 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11974 the mode argument to scalar_int_mode.
11975 (do_jump_by_parts_zero_rtx): Likewise.
11976 (do_jump_by_parts_equality_rtx): Likewise.
11977 (do_jump_by_parts_greater): Take a mode argument.
11978 (do_jump_by_parts_equality): Likewise.
11979 (do_jump_1): Update calls accordingly.
11981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11982 Alan Hayward <alan.hayward@arm.com>
11983 David Sherwood <david.sherwood@arm.com>
11985 * is-a.h (safe_dyn_cast): New function.
11986 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11987 (jump_table_for_label): Likewise.
11988 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11989 instead of an rtx_insn *.
11990 (shorten_branches): Use dyn_cast instead of LABEL_P and
11991 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11992 rtx_jump_table_data::get_data_mode.
11993 (final_scan_insn): Likewise.
11995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11996 Alan Hayward <alan.hayward@arm.com>
11997 David Sherwood <david.sherwood@arm.com>
11999 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12000 trying to combine a full-register integer set with a subreg
12003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12004 Alan Hayward <alan.hayward@arm.com>
12005 David Sherwood <david.sherwood@arm.com>
12007 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12008 that is always either address_mode or pointer_mode.
12010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12011 Alan Hayward <alan.hayward@arm.com>
12012 David Sherwood <david.sherwood@arm.com>
12014 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12015 when the two are known to be equal.
12017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12018 Alan Hayward <alan.hayward@arm.com>
12019 David Sherwood <david.sherwood@arm.com>
12021 * simplify-rtx.c (simplify_const_unary_operation): Use
12022 is_a <scalar_int_mode> instead of checking for a nonzero
12023 precision. Forcibly convert op_mode to a scalar_int_mode
12024 in that case. More clearly differentiate the operand and
12025 result modes and use the former when deciding what the value
12026 of a count-bits operation should be. Use is_int_mode instead
12027 of checking for a MODE_INT. Remove redundant check for whether
12028 this mode has a zero precision.
12030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12031 Alan Hayward <alan.hayward@arm.com>
12032 David Sherwood <david.sherwood@arm.com>
12034 * optabs.c (widen_leading): Change the type of the mode argument
12035 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12036 (widen_bswap): Likewise.
12037 (expand_parity): Likewise.
12038 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12039 (expand_ffs): Likewise.
12040 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12049 (expand_expr_real_1): Update call accordingly.
12051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12052 Alan Hayward <alan.hayward@arm.com>
12053 David Sherwood <david.sherwood@arm.com>
12055 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12056 value_mode arguments. Use scalar_int_mode internally.
12057 (store_bit_field_1): Rename the new integer mode from imode
12058 to op0_mode and use it instead of GET_MODE (op0). Update calls
12059 to store_split_bit_field, store_bit_field_using_insv and
12060 store_fixed_bit_field.
12061 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12062 Use scalar_int_mode internally. Use a bit count rather than a mode
12063 when calculating the largest bit size for get_best_mode.
12064 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12065 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12066 Remove assertion that OP0 has a scalar integer mode.
12067 (store_split_bit_field): Add op0_mode and value_mode arguments.
12068 Update calls to extract_fixed_bit_field.
12069 (extract_bit_field_using_extv): Add an op0_mode argument.
12070 Use scalar_int_mode internally.
12071 (extract_bit_field_1): Rename the new integer mode from imode to
12072 op0_mode and use it instead of GET_MODE (op0). Update calls to
12073 extract_split_bit_field, extract_bit_field_using_extv and
12074 extract_fixed_bit_field.
12075 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12076 to extract_split_bit_field and extract_fixed_bit_field_1.
12077 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12078 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12079 on the target mode.
12080 (extract_split_bit_field): Add an op0_mode argument. Update call
12081 to extract_fixed_bit_field.
12083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12084 Alan Hayward <alan.hayward@arm.com>
12085 David Sherwood <david.sherwood@arm.com>
12087 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12088 * explow.c (hard_function_value): Likewise.
12089 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12090 convert_to_mode call outside the loop.
12091 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12092 for the mode iterator. Require the mode specified by max_pieces
12094 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12096 (copy_blkmode_to_reg): Likewise.
12097 (set_storage_via_setmem): Likewise.
12098 * optabs.c (prepare_cmp_insn): Likewise.
12099 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12100 * stor-layout.c (finish_bitfield_representative): Likewise.
12102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12103 Alan Hayward <alan.hayward@arm.com>
12104 David Sherwood <david.sherwood@arm.com>
12106 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12107 * expr.c (convert_move): Use them.
12108 (convert_modes): Likewise.
12109 (store_expr_with_bounds): Likewise.
12111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12112 Alan Hayward <alan.hayward@arm.com>
12113 David Sherwood <david.sherwood@arm.com>
12115 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12116 parameter for the mode of "x". Remove the "known_x", "known_mode"
12117 and "known_ret" arguments. Change the type of the mode argument
12118 to scalar_int_mode.
12119 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12120 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12121 (reg_num_sign_bit_copies_for_combine): Likewise.
12122 * rtlanal.c (nonzero_bits1): Likewise.
12123 (num_sign_bit_copies1): Likewise.
12124 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12125 (reg_num_sign_bit_copies_general): Likewise.
12126 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12127 (reg_nonzero_bits_general): Likewise.
12129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12130 Alan Hayward <alan.hayward@arm.com>
12131 David Sherwood <david.sherwood@arm.com>
12133 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12134 than in subroutines. Return 1 for non-integer modes.
12135 (cached_num_sign_bit_copies): Change the type of the mode parameter
12136 to scalar_int_mode.
12137 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12138 classes. Handle CONST_INT_P first and then check whether X also
12139 has a scalar integer mode. Check the same thing for inner registers
12140 of a SUBREG and for values that are being extended or truncated.
12142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12143 Alan Hayward <alan.hayward@arm.com>
12144 David Sherwood <david.sherwood@arm.com>
12146 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12147 in subroutines. Return the mode mask for non-integer modes.
12148 (cached_nonzero_bits): Change the type of the mode parameter
12149 to scalar_int_mode.
12150 (nonzero_bits1): Likewise. Remove early exit for other mode
12151 classes. Handle CONST_INT_P first and then check whether X
12152 also has a scalar integer mode.
12154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12155 Alan Hayward <alan.hayward@arm.com>
12156 David Sherwood <david.sherwood@arm.com>
12158 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12159 Return a scalar_int_mode and assert that the size is greater than
12161 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12162 (op_by_pieces_d::op_by_pieces_d): Likewise.
12163 (op_by_pieces_d::run): Likewise.
12164 (can_store_by_pieces): Likewise.
12166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12167 Alan Hayward <alan.hayward@arm.com>
12168 David Sherwood <david.sherwood@arm.com>
12170 * combine.c (extract_left_shift): Add a mode argument and update
12172 (make_compound_operation_int): Change the type of the mode parameter
12173 to scalar_int_mode and update the call to extract_left_shift.
12175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12176 Alan Hayward <alan.hayward@arm.com>
12177 David Sherwood <david.sherwood@arm.com>
12179 * combine.c (simplify_and_const_int): Change the type of the mode
12180 parameter to scalar_int_mode.
12181 (simplify_and_const_int_1): Likewise. Update recursive call.
12183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12184 Alan Hayward <alan.hayward@arm.com>
12185 David Sherwood <david.sherwood@arm.com>
12187 * combine.c (simplify_compare_const): Check that the mode is a
12188 scalar_int_mode (rather than VOIDmode) before testing its
12190 (simplify_comparison): Move COMPARISON_P handling out of the
12191 loop and restrict the latter part of the loop to scalar_int_modes.
12192 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12193 and when considering SUBREG_REGs. Use is_int_mode instead of
12194 checking GET_MODE_CLASS against MODE_INT.
12196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12197 Alan Hayward <alan.hayward@arm.com>
12198 David Sherwood <david.sherwood@arm.com>
12200 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12201 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12202 shift_unit_mode and for modes involved in scalar shifts.
12204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12205 Alan Hayward <alan.hayward@arm.com>
12206 David Sherwood <david.sherwood@arm.com>
12208 * combine.c (force_int_to_mode): New function, split out from...
12209 (force_to_mode): ...here. Keep xmode up-to-date and use it
12210 instead of GET_MODE (x).
12212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12213 Alan Hayward <alan.hayward@arm.com>
12214 David Sherwood <david.sherwood@arm.com>
12216 * optabs-query.h (extraction_insn::struct_mode): Change type to
12217 opt_scalar_int_mode and update comment.
12218 (extraction_insn::field_mode): Change type to scalar_int_mode.
12219 (extraction_insn::pos_mode): Likewise.
12220 * combine.c (make_extraction): Update accordingly.
12221 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12222 (get_optab_extraction_insn): Likewise.
12223 * recog.c (simplify_while_replacing): Likewise.
12224 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12225 parameter to opt_scalar_int_mode.
12227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12228 Alan Hayward <alan.hayward@arm.com>
12229 David Sherwood <david.sherwood@arm.com>
12231 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12232 to a scalar_int_mode instead of a machine_mode.
12233 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12234 (get_best_mode): Return a boolean and use a pointer argument to store
12235 the selected mode. Replace the limit mode parameter with a bit limit.
12236 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12237 for the values returned by bit_field_mode_iterator::next_mode.
12238 (store_bit_field): Update call to get_best_mode.
12239 (store_fixed_bit_field): Likewise.
12240 (extract_fixed_bit_field): Likewise.
12241 * expr.c (optimize_bitfield_assignment_op): Likewise.
12242 * fold-const.c (optimize_bit_field_compare): Likewise.
12243 (fold_truth_andor_1): Likewise.
12244 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12245 Update for new type of m_mode.
12246 (get_best_mode): As above.
12248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12249 Alan Hayward <alan.hayward@arm.com>
12250 David Sherwood <david.sherwood@arm.com>
12252 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12253 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12254 (store_bit_field): Check is_a <scalar_int_mode> before calling
12255 strict_volatile_bitfield_p.
12256 (extract_bit_field): Likewise.
12258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12259 Alan Hayward <alan.hayward@arm.com>
12260 David Sherwood <david.sherwood@arm.com>
12262 * target.def (cstore_mode): Return a scalar_int_mode.
12263 * doc/tm.texi: Regenerate.
12264 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12265 * targhooks.h (default_cstore_mode): Likewise.
12266 * targhooks.c (default_cstore_mode): Likewise, using a forced
12268 * expmed.c (emit_cstore): Expect the target of the cstore to be
12271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12272 Alan Hayward <alan.hayward@arm.com>
12273 David Sherwood <david.sherwood@arm.com>
12275 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12277 (niter_desc): Likewise mode.
12278 (iv_analyze): Add a mode parameter.
12280 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12281 and change its type to scalar_int_mode.
12282 * loop-iv.c: Update commentary at head of file.
12283 (iv_constant): Pass the mode paraeter before the rtx it describes
12284 and change its type to scalar_int_mode. Remove VOIDmode handling.
12285 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12286 (iv_extend): Likewise.
12287 (shorten_into_mode): Likewise.
12288 (iv_add): Use scalar_int_mode.
12289 (iv_mult): Likewise.
12290 (iv_shift): Likewise.
12291 (canonicalize_iv_subregs): Likewise.
12292 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12293 it describes and change its mode to scalar_int_mode. Also change
12294 the type of the returned inner_mode to scalar_int_mode.
12295 (get_biv_step): Likewise, turning outer_mode from a pointer
12296 into a direct parameter. Update call to get_biv_step_1.
12297 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12298 iv_constant and get_biv_step.
12299 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12300 and change its type to scalar_int_mode. Don't initialise iv->mode
12301 to VOIDmode and remove later checks for its still being VOIDmode.
12302 Update calls to iv_analyze_op and iv_analyze_expr. Check
12303 is_a <scalar_int_mode> when changing the mode under consideration.
12304 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12305 Update call to iv_analyze_expr.
12306 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12307 inner register is not also a scalar_int_mode. Update call to
12309 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12310 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12311 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12312 separate mode class checks. Update calls to iv_analyze. Remove
12313 fix-up of VOIDmodes after iv_analyze_biv.
12314 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12315 don't have a scalar_int_mode. Update call to biv_p.
12317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12318 Alan Hayward <alan.hayward@arm.com>
12319 David Sherwood <david.sherwood@arm.com>
12321 * cfgexpand.c (convert_debug_memory_address): Use
12322 as_a <scalar_int_mode>.
12323 * combine.c (expand_compound_operation): Likewise.
12324 (make_extraction): Likewise.
12325 (change_zero_ext): Likewise.
12326 (simplify_comparison): Likewise.
12327 * cse.c (cse_insn): Likewise.
12328 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12329 (mem_loc_descriptor): Likewise.
12330 (loc_descriptor): Likewise.
12331 * expmed.c (init_expmed_one_mode): Likewise.
12332 (synth_mult): Likewise.
12333 (emit_store_flag_1): Likewise.
12334 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12335 of a comparison with size.
12336 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12337 (reduce_to_bit_field_precision): Likewise.
12338 * function.c (expand_function_end): Likewise.
12339 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12340 * loop-doloop.c (doloop_modify): Likewise.
12341 * optabs.c (expand_binop): Likewise.
12342 (expand_unop): Likewise.
12343 (expand_copysign_absneg): Likewise.
12344 (prepare_cmp_insn): Likewise.
12345 (maybe_legitimize_operand): Likewise.
12346 * recog.c (const_scalar_int_operand): Likewise.
12347 * rtlanal.c (get_address_mode): Likewise.
12348 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12349 (simplify_cond_clz_ctz): Likewise.
12350 * tree-nested.c (get_nl_goto_field): Likewise.
12351 * tree.c (build_vector_type_for_mode): Likewise.
12352 * var-tracking.c (use_narrower_mode): Likewise.
12354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12355 Alan Hayward <alan.hayward@arm.com>
12356 David Sherwood <david.sherwood@arm.com>
12358 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12359 * builtins.c (expand_builtin_signbit): Use it.
12360 * cfgexpand.c (expand_debug_expr): Likewise.
12361 * dojump.c (do_jump): Likewise.
12362 (do_compare_and_jump): Likewise.
12363 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12364 * expmed.c (make_tree): Likewise.
12365 * expr.c (expand_expr_real_2): Likewise.
12366 (expand_expr_real_1): Likewise.
12367 (try_casesi): Likewise.
12368 * fold-const-call.c (fold_const_call_ss): Likewise.
12369 * fold-const.c (unextend): Likewise.
12370 (extract_muldiv_1): Likewise.
12371 (fold_single_bit_test): Likewise.
12372 (native_encode_int): Likewise.
12373 (native_encode_string): Likewise.
12374 (native_interpret_int): Likewise.
12375 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12376 * internal-fn.c (expand_addsub_overflow): Likewise.
12377 (expand_neg_overflow): Likewise.
12378 (expand_mul_overflow): Likewise.
12379 (expand_arith_overflow): Likewise.
12380 * match.pd: Likewise.
12381 * stor-layout.c (layout_type): Likewise.
12382 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12383 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12384 * tree-ssanames.c (get_range_info): Likewise.
12385 * tree-switch-conversion.c (array_value_type) Likewise.
12386 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12387 (vect_recog_divmod_pattern): Likewise.
12388 (vect_recog_mixed_size_cond_pattern): Likewise.
12389 * tree-vrp.c (extract_range_basic): Likewise.
12390 (simplify_float_conversion_using_ranges): Likewise.
12391 * tree.c (int_fits_type_p): Likewise.
12392 * ubsan.c (instrument_bool_enum_load): Likewise.
12393 * varasm.c (mergeable_string_section): Likewise.
12394 (narrowing_initializer_constant_valid_p): Likewise.
12395 (output_constant): Likewise.
12397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12398 Alan Hayward <alan.hayward@arm.com>
12399 David Sherwood <david.sherwood@arm.com>
12401 * machmode.h (NARROWEST_INT_MODE): New macro.
12402 * expr.c (alignment_for_piecewise_move): Use it instead of
12403 GET_CLASS_NARROWEST_MODE (MODE_INT).
12404 (push_block): Likewise.
12405 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12407 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12410 Alan Hayward <alan.hayward@arm.com>
12411 David Sherwood <david.sherwood@arm.com>
12413 * postreload.c (move2add_valid_value_p): Change the type of the
12414 mode parameter to scalar_int_mode.
12415 (move2add_use_add2_insn): Add a mode parameter and use it instead
12417 (move2add_use_add3_insn): Likewise.
12418 (reload_cse_move2add): Update accordingly.
12420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12421 Alan Hayward <alan.hayward@arm.com>
12422 David Sherwood <david.sherwood@arm.com>
12424 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12426 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12427 * optabs.c (expand_unop): Likewise.
12429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12430 Alan Hayward <alan.hayward@arm.com>
12431 David Sherwood <david.sherwood@arm.com>
12433 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12434 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12435 (popcount_loc_descriptor): Likewise.
12436 (bswap_loc_descriptor): Likewise.
12437 (rotate_loc_descriptor): Likewise.
12438 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12439 calling the functions above.
12441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12442 Alan Hayward <alan.hayward@arm.com>
12443 David Sherwood <david.sherwood@arm.com>
12445 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12447 (try_combine): Likewise.
12448 (simplify_if_then_else): Likewise.
12449 * cse.c (cse_insn): Likewise.
12450 * dwarf2out.c (mem_loc_descriptor): Likewise.
12451 * emit-rtl.c (gen_lowpart_common): Likewise.
12452 * simplify-rtx.c (simplify_truncation): Likewise.
12453 (simplify_binary_operation_1): Likewise.
12454 (simplify_const_relational_operation): Likewise.
12455 (simplify_ternary_operation): Likewise.
12456 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12459 Alan Hayward <alan.hayward@arm.com>
12460 David Sherwood <david.sherwood@arm.com>
12462 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12463 * reload.c (push_reload): Likewise.
12464 (find_reloads): Likewise.
12466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12467 Alan Hayward <alan.hayward@arm.com>
12468 David Sherwood <david.sherwood@arm.com>
12470 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12471 (make_compound_operation_int): Likewise.
12472 (change_zero_ext): Likewise.
12473 * expr.c (convert_move): Likewise.
12474 (convert_modes): Likewise.
12475 * fwprop.c (forward_propagate_subreg): Likewise.
12476 * loop-iv.c (get_biv_step_1): Likewise.
12477 * optabs.c (widen_operand): Likewise.
12478 * postreload.c (move2add_valid_value_p): Likewise.
12479 * recog.c (simplify_while_replacing): Likewise.
12480 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12481 (simplify_binary_operation_1): Likewise. Remove redundant
12482 mode equality check.
12484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12485 Alan Hayward <alan.hayward@arm.com>
12486 David Sherwood <david.sherwood@arm.com>
12488 * combine.c (combine_simplify_rtx): Add checks for
12489 is_a <scalar_int_mode>.
12490 (simplify_if_then_else): Likewise.
12491 (make_field_assignment): Likewise.
12492 (simplify_comparison): Likewise.
12493 * ifcvt.c (noce_try_bitop): Likewise.
12494 * loop-invariant.c (canonicalize_address_mult): Likewise.
12495 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12498 Alan Hayward <alan.hayward@arm.com>
12499 David Sherwood <david.sherwood@arm.com>
12501 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12502 is_a <scalar_int_mode> instead of != BLKmode.
12504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12505 Alan Hayward <alan.hayward@arm.com>
12506 David Sherwood <david.sherwood@arm.com>
12508 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12509 instead of != VOIDmode.
12510 * combine.c (if_then_else_cond): Likewise.
12511 (change_zero_ext): Likewise.
12512 * dwarf2out.c (mem_loc_descriptor): Likewise.
12513 (loc_descriptor): Likewise.
12514 * rtlanal.c (canonicalize_condition): Likewise.
12515 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12518 Alan Hayward <alan.hayward@arm.com>
12519 David Sherwood <david.sherwood@arm.com>
12521 * simplify-rtx.c (simplify_binary_operation_1): Use
12522 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12525 Alan Hayward <alan.hayward@arm.com>
12526 David Sherwood <david.sherwood@arm.com>
12528 * wide-int.h (int_traits<unsigned char>) New class.
12529 (int_traits<unsigned short>) Likewise.
12530 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12531 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12533 * combine.c (set_nonzero_bits_and_sign_copies): Use
12534 is_a <scalar_int_mode>.
12535 (find_split_point): Likewise.
12536 (combine_simplify_rtx): Likewise.
12537 (simplify_logical): Likewise.
12538 (expand_compound_operation): Likewise.
12539 (expand_field_assignment): Likewise.
12540 (make_compound_operation): Likewise.
12541 (extended_count): Likewise.
12542 (change_zero_ext): Likewise.
12543 (simplify_comparison): Likewise.
12544 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12545 (ucompare_loc_descriptor): Likewise.
12546 (minmax_loc_descriptor): Likewise.
12547 (mem_loc_descriptor): Likewise.
12548 (loc_descriptor): Likewise.
12549 * expmed.c (init_expmed_one_mode): Likewise.
12550 * lra-constraints.c (lra_constraint_offset): Likewise.
12551 * optabs.c (prepare_libcall_arg): Likewise.
12552 * postreload.c (move2add_note_store): Likewise.
12553 * reload.c (operands_match_p): Likewise.
12554 * rtl.h (load_extend_op): Likewise.
12555 * rtlhooks.c (gen_lowpart_general): Likewise.
12556 * simplify-rtx.c (simplify_truncation): Likewise.
12557 (simplify_unary_operation_1): Likewise.
12558 (simplify_binary_operation_1): Likewise.
12559 (simplify_const_binary_operation): Likewise.
12560 (simplify_const_relational_operation): Likewise.
12561 (simplify_subreg): Likewise.
12562 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12563 * var-tracking.c (adjust_mems): Likewise.
12564 (prepare_call_arguments): Likewise.
12566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12567 Alan Hayward <alan.hayward@arm.com>
12568 David Sherwood <david.sherwood@arm.com>
12570 * machmode.h (is_int_mode): New fuction.
12571 * combine.c (find_split_point): Use it.
12572 (combine_simplify_rtx): Likewise.
12573 (simplify_if_then_else): Likewise.
12574 (simplify_set): Likewise.
12575 (simplify_shift_const_1): Likewise.
12576 (simplify_comparison): Likewise.
12577 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12578 * cse.c (notreg_cost): Likewise.
12579 (cse_insn): Likewise.
12580 * cselib.c (cselib_lookup_1): Likewise.
12581 * dojump.c (do_jump_1): Likewise.
12582 (do_compare_rtx_and_jump): Likewise.
12583 * dse.c (get_call_args): Likewise.
12584 * dwarf2out.c (rtl_for_decl_init): Likewise.
12585 (native_encode_initializer): Likewise.
12586 * expmed.c (emit_store_flag_1): Likewise.
12587 (emit_store_flag): Likewise.
12588 * expr.c (convert_modes): Likewise.
12589 (store_field): Likewise.
12590 (expand_expr_real_1): Likewise.
12591 * fold-const.c (fold_read_from_constant_string): Likewise.
12592 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12593 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12594 * optabs.c (expand_binop): Likewise.
12595 (expand_unop): Likewise.
12596 (expand_abs_nojump): Likewise.
12597 (expand_one_cmpl_abs_nojump): Likewise.
12598 * simplify-rtx.c (mode_signbit_p): Likewise.
12599 (val_signbit_p): Likewise.
12600 (val_signbit_known_set_p): Likewise.
12601 (val_signbit_known_clear_p): Likewise.
12602 (simplify_relational_operation_1): Likewise.
12603 * tree.c (vector_type_mode): Likewise.
12605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12606 Alan Hayward <alan.hayward@arm.com>
12607 David Sherwood <david.sherwood@arm.com>
12609 * machmode.h (smallest_mode_for_size): Fix formatting.
12610 (smallest_int_mode_for_size): New function.
12611 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12612 instead of smallest_mode_for_size.
12613 * combine.c (make_extraction): Likewise.
12614 * config/arc/arc.c (arc_expand_movmem): Likewise.
12615 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12616 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12617 * config/s390/s390.c (s390_expand_insv): Likewise.
12618 * config/sparc/sparc.c (assign_int_registers): Likewise.
12619 * config/spu/spu.c (spu_function_value): Likewise.
12620 (spu_function_arg): Likewise.
12621 * coverage.c (get_gcov_type): Likewise.
12622 (get_gcov_unsigned_t): Likewise.
12623 * dse.c (find_shift_sequence): Likewise.
12624 * expmed.c (store_bit_field_1): Likewise.
12625 * expr.c (convert_move): Likewise.
12626 (store_field): Likewise.
12627 * internal-fn.c (expand_arith_overflow): Likewise.
12628 * optabs-query.c (get_best_extraction_insn): Likewise.
12629 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12630 * stor-layout.c (layout_type): Likewise.
12631 (initialize_sizetypes): Likewise.
12632 * targhooks.c (default_get_mask_mode): Likewise.
12633 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12636 Alan Hayward <alan.hayward@arm.com>
12637 David Sherwood <david.sherwood@arm.com>
12639 * machmode.h (opt_mode::else_blk): New function.
12640 (int_mode_for_mode): Declare.
12641 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12642 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12644 * cfgexpand.c (expand_debug_expr): Likewise.
12645 * combine.c (gen_lowpart_or_truncate): Likewise.
12646 (gen_lowpart_for_combine): Likewise.
12647 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12648 * config/avr/avr.c (avr_to_int_mode): Likewise.
12649 (avr_out_plus_1): Likewise.
12650 (avr_out_plus): Likewise.
12651 (avr_out_round): Likewise.
12652 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12653 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12654 (s390_expand_vcond): Likewise.
12655 * config/spu/spu.c (spu_split_immediate): Likewise.
12656 (spu_expand_mov): Likewise.
12657 * dse.c (get_stored_val): Likewise.
12658 * expmed.c (store_bit_field_1): Likewise.
12659 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12661 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12662 (extract_low_bits): Likewise.
12663 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12664 handling rather than repeating the check.
12665 (emit_group_store): Likewise.
12666 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12667 * optabs.c (expand_absneg_bit): Likewise.
12668 (expand_copysign_absneg): Likewise.
12669 (expand_copysign_bit): Likewise.
12670 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12671 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12672 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12673 * var-tracking.c (prepare_call_arguments): Likewise.
12674 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12675 int_mode_for_mode instead of mode_for_size.
12676 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12679 Alan Hayward <alan.hayward@arm.com>
12680 David Sherwood <david.sherwood@arm.com>
12682 * machmode.h (int_mode_for_size): New function.
12683 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12684 instead of mode_for_size.
12685 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12687 * combine.c (expand_field_assignment): Use int_mode_for_size
12688 instead of mode_for_size.
12689 (make_extraction): Likewise.
12690 (simplify_shift_const_1): Likewise.
12691 (simplify_comparison): Likewise.
12692 * dojump.c (do_jump): Likewise.
12693 * dwarf2out.c (mem_loc_descriptor): Likewise.
12694 * emit-rtl.c (init_derived_machine_modes): Likewise.
12695 * expmed.c (flip_storage_order): Likewise.
12696 (convert_extracted_bit_field): Likewise.
12697 * expr.c (copy_blkmode_from_reg): Likewise.
12698 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12699 * internal-fn.c (expand_mul_overflow): Likewise.
12700 * lower-subreg.c (simple_move): Likewise.
12701 * optabs-libfuncs.c (init_optabs): Likewise.
12702 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12703 * tree.c (vector_type_mode): Likewise.
12704 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12705 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12706 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12707 * tree-vect-stmts.c (vectorizable_load): Likewise.
12708 (vectorizable_store): Likewise.
12710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12711 Alan Hayward <alan.hayward@arm.com>
12712 David Sherwood <david.sherwood@arm.com>
12714 * coretypes.h (pod_mode): New type.
12715 (scalar_int_mode_pod): New typedef.
12716 * machmode.h (pod_mode): New class.
12717 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12718 * genmodes.c (emit_mode_int_n): Update accordingly.
12719 * lower-subreg.h (target_lower_subreg): Change type to
12720 scalar_int_mode_pod.
12721 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12722 scalar_int_mode_pod.
12724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12725 Alan Hayward <alan.hayward@arm.com>
12726 David Sherwood <david.sherwood@arm.com>
12728 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12729 machine_mode to scalar_int_mode.
12730 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12731 (rs6000_option_override_internal): Remove cast to int.
12732 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12733 machine_mode to scalar_int_mode.
12734 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12735 (rs6000_option_override_internal): Remove cast to int.
12736 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12737 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12739 * config/s390/s390.c (s390_expand_builtin): Likewise.
12740 * coretypes.h (scalar_int_mode): New type.
12741 (opt_scalar_int_mode): New typedef.
12742 * machmode.h (scalar_int_mode): New class.
12743 (scalar_int_mode::includes_p): New function.
12744 (byte_mode): Change type to scalar_int_mode.
12745 (word_mode): Likewise.
12746 (ptr_mode): Likewise.
12747 * emit-rtl.c (byte_mode): Likewise.
12748 (word_mode): Likewise.
12749 (ptr_mode): Likewise.
12750 (init_derived_machine_modes): Update accordingly.
12751 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12752 and MODE_PARTIAL_INT.
12753 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12754 opt_scalar_int_mode.
12756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12757 Alan Hayward <alan.hayward@arm.com>
12758 David Sherwood <david.sherwood@arm.com>
12760 * target.def (libgcc_floating_mode_supported_p): Take a
12762 * doc/tm.texi: Regenerate.
12763 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12765 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12766 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12770 Alan Hayward <alan.hayward@arm.com>
12771 David Sherwood <david.sherwood@arm.com>
12773 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12774 * doc/tm.texi: Regenerate.
12775 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12776 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12777 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12778 * targhooks.h (default_floatn_mode): Likewise.
12779 * targhooks.c (default_floatn_mode): Likewise.
12780 * tree.c (build_common_tree_nodes): Update accordingly.
12782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12783 Alan Hayward <alan.hayward@arm.com>
12784 David Sherwood <david.sherwood@arm.com>
12786 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12787 (mode_iterator::iterate_p): Likewise.
12788 (mode_iterator::get_wider): Likewise.
12789 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12792 Alan Hayward <alan.hayward@arm.com>
12793 David Sherwood <david.sherwood@arm.com>
12795 * coretypes.h (opt_scalar_float_mode): New typedef.
12796 * machmode.h (float_mode_for_size): New function.
12797 * emit-rtl.c (double_mode): Delete.
12798 (init_emit_once): Use float_mode_for_size.
12799 * stor-layout.c (layout_type): Likewise.
12800 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12803 Alan Hayward <alan.hayward@arm.com>
12804 David Sherwood <david.sherwood@arm.com>
12806 * output.h (assemble_real): Take a scalar_float_mode.
12807 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12808 * config/arm/arm.md (consttable_4): Likewise.
12809 (consttable_8): Likewise.
12810 (consttable_16): Likewise.
12811 * config/mips/mips.md (consttable_float): Likewise.
12812 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12813 * varasm.c (assemble_real): Take a scalar_float_mode.
12814 (output_constant_pool_2): Update accordingly.
12815 (output_constant): Likewise.
12817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12818 Alan Hayward <alan.hayward@arm.com>
12819 David Sherwood <david.sherwood@arm.com>
12821 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12822 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12823 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12824 (native_encode_real): Likewise.
12825 (native_interpret_real): Likewise.
12826 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12827 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12830 Alan Hayward <alan.hayward@arm.com>
12831 David Sherwood <david.sherwood@arm.com>
12833 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12834 <scalar_float_mode>. Simplify.
12835 (gen_extend_conv_libfunc): Likewise.
12837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12838 Alan Hayward <alan.hayward@arm.com>
12839 David Sherwood <david.sherwood@arm.com>
12841 * coretypes.h (scalar_float_mode): New type.
12842 * machmode.h (mode_traits::from_int): Use machine_mode if
12843 USE_ENUM_MODES is defined.
12844 (is_a): New function.
12846 (dyn_cast): Likewise.
12847 (scalar_float_mode): New class.
12848 (scalar_float_mode::includes_p): New function.
12849 (is_float_mode): Likewise.
12850 * gdbhooks.py (MachineModePrinter): New class.
12851 (build_pretty_printer): Use it for scalar_float_mode.
12852 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12853 (format_helper::format_helper): Turn into a template.
12854 * genmodes.c (get_mode_class): New function.
12855 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12856 or machine_mode if none.
12857 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12858 as_a <scalar_float_mode>.
12859 * dwarf2out.c (mem_loc_descriptor): Likewise.
12860 (insert_float): Likewise.
12861 (add_const_value_attribute): Likewise.
12862 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12863 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12864 (expand_unop): Update accordingly.
12865 (expand_abs_nojump): Likewise.
12866 (expand_copysign_absneg): Take a scalar_float_mode.
12867 (expand_copysign_bit): Likewise.
12868 (expand_copysign): Update accordingly.
12870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12871 Alan Hayward <alan.hayward@arm.com>
12872 David Sherwood <david.sherwood@arm.com>
12874 * coretypes.h (opt_mode): New class.
12875 * machmode.h (opt_mode): Likewise.
12876 (opt_mode::else_void): New function.
12877 (opt_mode::require): Likewise.
12878 (opt_mode::exists): Likewise.
12879 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12880 (GET_MODE_2XWIDER_MODE): Likewise.
12881 (mode_iterator::get_wider): Update accordingly.
12882 (mode_iterator::get_2xwider): Likewise.
12883 (mode_iterator::get_known_wider): Likewise, turning into a template.
12884 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12885 forcing a wider mode to exist.
12886 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12887 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12888 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12889 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12890 * lower-subreg.c (init_lower_subreg): Likewise.
12891 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12892 on the final iteration.
12893 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12894 a wider mode exists before asking for a move pattern.
12895 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12896 forcing a wider mode to exist.
12897 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12898 returning false if no such mode exists.
12899 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12900 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12901 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12902 Avoid checking for a MODE_INT if we already know the mode is not a
12904 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12905 forcing a wider mode to exist.
12906 (expmed_mult_highpart_optab): Likewise.
12907 (expmed_mult_highpart): Likewise.
12908 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12910 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12911 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12912 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12913 * internal-fn.c (expand_mul_overflow): Update use of
12914 GET_MODE_2XWIDER_MODE.
12915 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12916 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12917 GET_MODE_WIDER_MODE.
12918 (convert_plusminus_to_widen): Likewise.
12919 * tree-switch-conversion.c (array_value_type): Likewise.
12920 * var-tracking.c (emit_note_insn_var_location): Likewise.
12921 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12922 Return false inside rather than outside the loop if no wider mode
12924 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12925 and GET_MODE_2XWIDER_MODE
12926 (can_compare_p): Use else_void.
12927 * gdbhooks.py (OptMachineModePrinter): New class.
12928 (build_pretty_printer): Use it for opt_mode.
12930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12931 Alan Hayward <alan.hayward@arm.com>
12932 David Sherwood <david.sherwood@arm.com>
12934 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12935 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12938 Alan Hayward <alan.hayward@arm.com>
12939 David Sherwood <david.sherwood@arm.com>
12941 * machmode.h (mode_traits): New structure.
12942 (get_narrowest_mode): New function.
12943 (mode_iterator::start): Likewise.
12944 (mode_iterator::iterate_p): Likewise.
12945 (mode_iterator::get_wider): Likewise.
12946 (mode_iterator::get_known_wider): Likewise.
12947 (mode_iterator::get_2xwider): Likewise.
12948 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12949 (FOR_EACH_MODE): Likewise.
12950 (FOR_EACH_MODE_FROM): Likewise.
12951 (FOR_EACH_MODE_UNTIL): Likewise.
12952 (FOR_EACH_WIDER_MODE): Likewise.
12953 (FOR_EACH_2XWIDER_MODE): Likewise.
12954 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12955 * combine.c (simplify_comparison): Likewise
12956 * config/i386/i386.c (type_natural_mode): Likewise.
12957 * cse.c (cse_insn): Likewise.
12958 * dse.c (find_shift_sequence): Likewise.
12959 * emit-rtl.c (init_derived_machine_modes): Likewise.
12960 (init_emit_once): Likewise.
12961 * explow.c (hard_function_value): Likewise.
12962 * expmed.c (extract_fixed_bit_field_1): Likewise.
12963 (extract_bit_field_1): Likewise.
12964 (expand_divmod): Likewise.
12965 (emit_store_flag_1): Likewise.
12966 * expr.c (init_expr_target): Likewise.
12967 (convert_move): Likewise.
12968 (alignment_for_piecewise_move): Likewise.
12969 (widest_int_mode_for_size): Likewise.
12970 (emit_block_move_via_movmem): Likewise.
12971 (copy_blkmode_to_reg): Likewise.
12972 (set_storage_via_setmem): Likewise.
12973 (compress_float_constant): Likewise.
12974 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12975 * optabs-query.c (get_best_extraction_insn): Likewise.
12976 * optabs.c (expand_binop): Likewise.
12977 (expand_twoval_unop): Likewise.
12978 (expand_twoval_binop): Likewise.
12979 (widen_leading): Likewise.
12980 (widen_bswap): Likewise.
12981 (expand_parity): Likewise.
12982 (expand_unop): Likewise.
12983 (prepare_cmp_insn): Likewise.
12984 (prepare_float_lib_cmp): Likewise.
12985 (expand_float): Likewise.
12986 (expand_fix): Likewise.
12987 (expand_sfix_optab): Likewise.
12988 * postreload.c (move2add_use_add2_insn): Likewise.
12989 * reg-stack.c (reg_to_stack): Likewise.
12990 * reginfo.c (choose_hard_reg_mode): Likewise.
12991 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12992 * stor-layout.c (mode_for_size): Likewise.
12993 (smallest_mode_for_size): Likewise.
12994 (mode_for_vector): Likewise.
12995 (finish_bitfield_representative): Likewise.
12996 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12997 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12998 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12999 * var-tracking.c (prepare_call_arguments): Likewise.
13001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13002 Alan Hayward <alan.hayward@arm.com>
13003 David Sherwood <david.sherwood@arm.com>
13005 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13006 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13007 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13008 * machmode.h (mode_size): Move earlier in file.
13009 (mode_precision): Likewise.
13010 (mode_inner): Likewise.
13011 (mode_nunits): Likewise.
13012 (mode_unit_size): Likewise.
13013 (unit_unit_precision): Likewise.
13014 (mode_wider): Likewise.
13015 (mode_2xwider): Likewise.
13016 (machine_mode): New class.
13017 (mode_to_bytes): New function.
13018 (mode_to_bits): Likewise.
13019 (mode_to_precision): Likewise.
13020 (mode_to_inner): Likewise.
13021 (mode_to_unit_size): Likewise.
13022 (mode_to_unit_precision): Likewise.
13023 (mode_to_nunits): Likewise.
13024 (GET_MODE_SIZE): Use mode_to_bytes.
13025 (GET_MODE_BITSIZE): Use mode_to_bits.
13026 (GET_MODE_PRECISION): Use mode_to_precision.
13027 (GET_MODE_INNER): Use mode_to_inner.
13028 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13029 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13030 (GET_MODE_NUNITS): Use mode_to_nunits.
13031 * system.h (ALWAYS_INLINE): New macro.
13032 * config/powerpcspe/powerpcspe-c.c
13033 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13034 int for arg1_mode and arg2_mode.
13036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13037 Alan Hayward <alan.hayward@arm.com>
13038 David Sherwood <david.sherwood@arm.com>
13040 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13041 Prefix mode names with E_ in case statements.
13042 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13043 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13044 (aarch64_split_simd_move): Likewise.
13045 (aarch64_gen_storewb_pair): Likewise.
13046 (aarch64_gen_loadwb_pair): Likewise.
13047 (aarch64_gen_store_pair): Likewise.
13048 (aarch64_gen_load_pair): Likewise.
13049 (aarch64_get_condition_code_1): Likewise.
13050 (aarch64_constant_pool_reload_icode): Likewise.
13051 (get_rsqrte_type): Likewise.
13052 (get_rsqrts_type): Likewise.
13053 (get_recpe_type): Likewise.
13054 (get_recps_type): Likewise.
13055 (aarch64_gimplify_va_arg_expr): Likewise.
13056 (aarch64_simd_container_mode): Likewise.
13057 (aarch64_emit_load_exclusive): Likewise.
13058 (aarch64_emit_store_exclusive): Likewise.
13059 (aarch64_expand_compare_and_swap): Likewise.
13060 (aarch64_gen_atomic_cas): Likewise.
13061 (aarch64_emit_bic): Likewise.
13062 (aarch64_emit_atomic_swap): Likewise.
13063 (aarch64_emit_atomic_load_op): Likewise.
13064 (aarch64_evpc_trn): Likewise.
13065 (aarch64_evpc_uzp): Likewise.
13066 (aarch64_evpc_zip): Likewise.
13067 (aarch64_evpc_ext): Likewise.
13068 (aarch64_evpc_rev): Likewise.
13069 (aarch64_evpc_dup): Likewise.
13070 (aarch64_gen_ccmp_first): Likewise.
13071 (aarch64_gen_ccmp_next): Likewise.
13072 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13073 (alpha_emit_xfloating_libcall): Likewise.
13074 (emit_insxl): Likewise.
13075 (alpha_arg_type): Likewise.
13076 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13077 (arc_preferred_simd_mode): Likewise.
13078 (arc_secondary_reload): Likewise.
13079 (get_arc_condition_code): Likewise.
13080 (arc_print_operand): Likewise.
13081 (arc_legitimate_constant_p): Likewise.
13082 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13083 * config/arc/arc.md (casesi_load): Likewise.
13084 (casesi_compact_jump): Likewise.
13085 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13086 (cc_use_register): Likewise.
13087 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13088 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13089 (arm_init_iwmmxt_builtins): Likewise.
13090 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13091 (neon_expand_vector_init): Likewise.
13092 (arm_attr_length_move_neon): Likewise.
13093 (maybe_get_arm_condition_code): Likewise.
13094 (arm_emit_vector_const): Likewise.
13095 (arm_preferred_simd_mode): Likewise.
13096 (arm_output_iwmmxt_tinsr): Likewise.
13097 (thumb1_output_casesi): Likewise.
13098 (thumb2_output_casesi): Likewise.
13099 (arm_emit_load_exclusive): Likewise.
13100 (arm_emit_store_exclusive): Likewise.
13101 (arm_expand_compare_and_swap): Likewise.
13102 (arm_evpc_neon_vuzp): Likewise.
13103 (arm_evpc_neon_vzip): Likewise.
13104 (arm_evpc_neon_vrev): Likewise.
13105 (arm_evpc_neon_vtrn): Likewise.
13106 (arm_evpc_neon_vext): Likewise.
13107 (arm_validize_comparison): Likewise.
13108 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13109 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13110 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13111 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13112 (c6x_preferred_simd_mode): Likewise.
13113 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13114 (epiphany_rtx_costs): Likewise.
13115 * config/epiphany/predicates.md (proper_comparison_operator):
13117 * config/frv/frv.c (condexec_memory_operand): Likewise.
13118 (frv_emit_move): Likewise.
13119 (output_move_single): Likewise.
13120 (output_condmove_single): Likewise.
13121 (frv_hard_regno_mode_ok): Likewise.
13122 (frv_matching_accg_mode): Likewise.
13123 * config/h8300/h8300.c (split_adds_subs): Likewise.
13124 (h8300_rtx_costs): Likewise.
13125 (h8300_print_operand): Likewise.
13126 (compute_mov_length): Likewise.
13127 (output_logical_op): Likewise.
13128 (compute_logical_op_length): Likewise.
13129 (compute_logical_op_cc): Likewise.
13130 (h8300_shift_needs_scratch_p): Likewise.
13131 (output_a_shift): Likewise.
13132 (compute_a_shift_length): Likewise.
13133 (compute_a_shift_cc): Likewise.
13134 (expand_a_rotate): Likewise.
13135 (output_a_rotate): Likewise.
13136 * config/i386/i386.c (classify_argument): Likewise.
13137 (function_arg_advance_32): Likewise.
13138 (function_arg_32): Likewise.
13139 (function_arg_64): Likewise.
13140 (function_value_64): Likewise.
13141 (ix86_gimplify_va_arg): Likewise.
13142 (ix86_legitimate_constant_p): Likewise.
13143 (put_condition_code): Likewise.
13144 (split_double_mode): Likewise.
13145 (ix86_avx256_split_vector_move_misalign): Likewise.
13146 (ix86_expand_vector_logical_operator): Likewise.
13147 (ix86_split_idivmod): Likewise.
13148 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13149 (ix86_build_const_vector): Likewise.
13150 (ix86_build_signbit_mask): Likewise.
13151 (ix86_match_ccmode): Likewise.
13152 (ix86_cc_modes_compatible): Likewise.
13153 (ix86_expand_branch): Likewise.
13154 (ix86_expand_sse_cmp): Likewise.
13155 (ix86_expand_sse_movcc): Likewise.
13156 (ix86_expand_int_sse_cmp): Likewise.
13157 (ix86_expand_vec_perm_vpermi2): Likewise.
13158 (ix86_expand_vec_perm): Likewise.
13159 (ix86_expand_sse_unpack): Likewise.
13160 (ix86_expand_int_addcc): Likewise.
13161 (ix86_split_to_parts): Likewise.
13162 (ix86_vectorize_builtin_gather): Likewise.
13163 (ix86_vectorize_builtin_scatter): Likewise.
13164 (avx_vpermilp_parallel): Likewise.
13165 (inline_memory_move_cost): Likewise.
13166 (ix86_tieable_integer_mode_p): Likewise.
13167 (x86_maybe_negate_const_int): Likewise.
13168 (ix86_expand_vector_init_duplicate): Likewise.
13169 (ix86_expand_vector_init_one_nonzero): Likewise.
13170 (ix86_expand_vector_init_one_var): Likewise.
13171 (ix86_expand_vector_init_concat): Likewise.
13172 (ix86_expand_vector_init_interleave): Likewise.
13173 (ix86_expand_vector_init_general): Likewise.
13174 (ix86_expand_vector_set): Likewise.
13175 (ix86_expand_vector_extract): Likewise.
13176 (emit_reduc_half): Likewise.
13177 (ix86_emit_i387_round): Likewise.
13178 (ix86_mangle_type): Likewise.
13179 (ix86_expand_round_sse4): Likewise.
13180 (expand_vec_perm_blend): Likewise.
13181 (canonicalize_vector_int_perm): Likewise.
13182 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13183 (expand_vec_perm_1): Likewise.
13184 (expand_vec_perm_interleave3): Likewise.
13185 (expand_vec_perm_even_odd_pack): Likewise.
13186 (expand_vec_perm_even_odd_1): Likewise.
13187 (expand_vec_perm_broadcast_1): Likewise.
13188 (ix86_vectorize_vec_perm_const_ok): Likewise.
13189 (ix86_expand_vecop_qihi): Likewise.
13190 (ix86_expand_mul_widen_hilo): Likewise.
13191 (ix86_expand_sse2_abs): Likewise.
13192 (ix86_expand_pextr): Likewise.
13193 (ix86_expand_pinsr): Likewise.
13194 (ix86_preferred_simd_mode): Likewise.
13195 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13196 * config/i386/sse.md (*andnot<mode>3): Likewise.
13197 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13198 (*<code><mode>3): Likewise.
13199 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13200 (ia64_expand_atomic_op): Likewise.
13201 (ia64_arg_type): Likewise.
13202 (ia64_mode_to_int): Likewise.
13203 (ia64_scalar_mode_supported_p): Likewise.
13204 (ia64_vector_mode_supported_p): Likewise.
13205 (expand_vec_perm_broadcast): Likewise.
13206 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13207 (iq2000_function_arg_advance): Likewise.
13208 (iq2000_function_arg): Likewise.
13209 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13210 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13211 (m68k_libcall_value): Likewise.
13212 (m68k_function_value): Likewise.
13213 (sched_attr_op_type): Likewise.
13214 * config/mcore/mcore.c (mcore_output_move): Likewise.
13215 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13217 (microblaze_function_arg): Likewise.
13218 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13219 (mips_print_operand): Likewise.
13220 (mips_mode_ok_for_mov_fmt_p): Likewise.
13221 (mips_vector_mode_supported_p): Likewise.
13222 (mips_preferred_simd_mode): Likewise.
13223 (mips_expand_vpc_loongson_even_odd): Likewise.
13224 (mips_expand_vec_unpack): Likewise.
13225 (mips_expand_vi_broadcast): Likewise.
13226 (mips_expand_vector_init): Likewise.
13227 (mips_expand_vec_reduc): Likewise.
13228 (mips_expand_msa_cmp): Likewise.
13229 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13230 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13231 (cc_flags_for_mode): Likewise.
13232 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13233 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13234 (nds32_output_casesi_pc_relative): Likewise.
13235 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13236 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13237 (nvptx_gen_unpack): Likewise.
13238 (nvptx_gen_pack): Likewise.
13239 (nvptx_gen_shuffle): Likewise.
13240 (nvptx_gen_wcast): Likewise.
13241 (nvptx_preferred_simd_mode): Likewise.
13242 * config/pa/pa.c (pa_secondary_reload): Likewise.
13243 * config/pa/predicates.md (base14_operand): Likewise.
13244 * config/powerpcspe/powerpcspe-c.c
13245 (altivec_resolve_overloaded_builtin): Likewise.
13246 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13248 (rs6000_preferred_simd_mode): Likewise.
13249 (output_vec_const_move): Likewise.
13250 (rs6000_expand_vector_extract): Likewise.
13251 (rs6000_split_vec_extract_var): Likewise.
13252 (reg_offset_addressing_ok_p): Likewise.
13253 (rs6000_legitimate_offset_address_p): Likewise.
13254 (rs6000_legitimize_address): Likewise.
13255 (rs6000_emit_set_const): Likewise.
13256 (rs6000_const_vec): Likewise.
13257 (rs6000_emit_move): Likewise.
13258 (spe_build_register_parallel): Likewise.
13259 (rs6000_darwin64_record_arg_recurse): Likewise.
13260 (swap_selector_for_mode): Likewise.
13261 (spe_init_builtins): Likewise.
13262 (paired_init_builtins): Likewise.
13263 (altivec_init_builtins): Likewise.
13264 (do_load_for_compare): Likewise.
13265 (rs6000_generate_compare): Likewise.
13266 (rs6000_expand_float128_convert): Likewise.
13267 (emit_load_locked): Likewise.
13268 (emit_store_conditional): Likewise.
13269 (rs6000_output_function_epilogue): Likewise.
13270 (rs6000_handle_altivec_attribute): Likewise.
13271 (rs6000_function_value): Likewise.
13272 (emit_fusion_gpr_load): Likewise.
13273 (emit_fusion_p9_load): Likewise.
13274 (emit_fusion_p9_store): Likewise.
13275 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13276 (fusion_gpr_mem_load): Likewise.
13277 (fusion_addis_mem_combo_load): Likewise.
13278 (fusion_addis_mem_combo_store): Likewise.
13279 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13280 (fusion_gpr_mem_load): Likewise.
13281 (fusion_addis_mem_combo_load): Likewise.
13282 (fusion_addis_mem_combo_store): Likewise.
13283 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13285 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13286 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13287 (rs6000_preferred_simd_mode): Likewise.
13288 (output_vec_const_move): Likewise.
13289 (rs6000_expand_vector_extract): Likewise.
13290 (rs6000_split_vec_extract_var): Likewise.
13291 (reg_offset_addressing_ok_p): Likewise.
13292 (rs6000_legitimate_offset_address_p): Likewise.
13293 (rs6000_legitimize_address): Likewise.
13294 (rs6000_emit_set_const): Likewise.
13295 (rs6000_const_vec): Likewise.
13296 (rs6000_emit_move): Likewise.
13297 (rs6000_darwin64_record_arg_recurse): Likewise.
13298 (swap_selector_for_mode): Likewise.
13299 (paired_init_builtins): Likewise.
13300 (altivec_init_builtins): Likewise.
13301 (rs6000_expand_float128_convert): Likewise.
13302 (emit_load_locked): Likewise.
13303 (emit_store_conditional): Likewise.
13304 (rs6000_output_function_epilogue): Likewise.
13305 (rs6000_handle_altivec_attribute): Likewise.
13306 (rs6000_function_value): Likewise.
13307 (emit_fusion_gpr_load): Likewise.
13308 (emit_fusion_p9_load): Likewise.
13309 (emit_fusion_p9_store): Likewise.
13310 * config/rx/rx.c (rx_gen_move_template): Likewise.
13311 (flags_from_mode): Likewise.
13312 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13313 (s390_slb_comparison): Likewise.
13314 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13315 (s390_vector_mode_supported_p): Likewise.
13316 (s390_cc_modes_compatible): Likewise.
13317 (s390_match_ccmode_set): Likewise.
13318 (s390_canonicalize_comparison): Likewise.
13319 (s390_emit_compare_and_swap): Likewise.
13320 (s390_branch_condition_mask): Likewise.
13321 (s390_rtx_costs): Likewise.
13322 (s390_secondary_reload): Likewise.
13323 (__SECONDARY_RELOAD_CASE): Likewise.
13324 (s390_expand_cs): Likewise.
13325 (s390_preferred_simd_mode): Likewise.
13326 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13327 * config/sh/sh.c (sh_print_operand): Likewise.
13328 (dump_table): Likewise.
13329 (sh_secondary_reload): Likewise.
13330 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13331 * config/sh/sh.md (casesi_worker_1): Likewise.
13332 (casesi_worker_2): Likewise.
13333 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13334 (fcc_comparison_operator): Likewise.
13335 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13336 (emit_soft_tfmode_cvt): Likewise.
13337 (sparc_preferred_simd_mode): Likewise.
13338 (output_cbranch): Likewise.
13339 (sparc_print_operand): Likewise.
13340 (sparc_expand_vec_perm_bmask): Likewise.
13341 (vector_init_bshuffle): Likewise.
13342 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13343 (spu_vector_mode_supported_p): Likewise.
13344 (spu_expand_insv): Likewise.
13345 (spu_emit_branch_or_set): Likewise.
13346 (spu_handle_vector_attribute): Likewise.
13347 (spu_builtin_splats): Likewise.
13348 (spu_builtin_extract): Likewise.
13349 (spu_builtin_promote): Likewise.
13350 (spu_expand_sign_extend): Likewise.
13351 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13352 (tilegx_simd_int): Likewise.
13353 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13354 (tilepro_simd_int): Likewise.
13355 * config/v850/v850.c (const_double_split): Likewise.
13356 (v850_print_operand): Likewise.
13357 (ep_memory_offset): Likewise.
13358 * config/vax/vax.c (vax_rtx_costs): Likewise.
13359 (vax_output_int_move): Likewise.
13360 (vax_output_int_add): Likewise.
13361 (vax_output_int_subtract): Likewise.
13362 * config/visium/predicates.md (visium_branch_operator): Likewise.
13363 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13364 (visium_print_operand_address): Likewise.
13365 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13366 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13367 (xtensa_expand_conditional_branch): Likewise.
13368 (xtensa_copy_incoming_a7): Likewise.
13369 (xtensa_output_literal): Likewise.
13370 * dfp.c (decimal_real_maxval): Likewise.
13371 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13374 Alan Hayward <alan.hayward@arm.com>
13375 David Sherwood <david.sherwood@arm.com>
13377 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13378 (mode_nunits_inline): Likewise.
13379 (mode_inner_inline): Likewise.
13380 (mode_unit_size_inline): Likewise.
13381 (mode_unit_precision_inline): Likewise.
13382 (emit_insn_modes_h): Likewise. Also emit a #define of the
13384 (emit_mode_wider): Add an E_ prefix to mode names.
13385 (emit_mode_complex): Likewise.
13386 (emit_mode_inner): Likewise.
13387 (emit_mode_adjustments): Likewise.
13388 (emit_mode_int_n): Likewise.
13389 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13390 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13391 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13392 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13393 (CRC32_BUILTIN, ENTRY): Likewise.
13394 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13395 (aarch64_pop_regs): Likewise.
13396 (aarch64_process_components): Likewise.
13397 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13398 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13399 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13400 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13401 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13402 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13404 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13405 (pcmpistr): Likewise.
13406 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13407 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13408 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13410 * config/rl78/rl78.c (mduc_regs): Likewise.
13411 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13412 (htm_expand_builtin): Likewise.
13413 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13414 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13415 * config/xtensa/xtensa.c (print_operand): Likewise.
13416 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13417 (NUM_MODE_VECTOR_INT): Likewise.
13418 * genoutput.c (null_operand): Likewise.
13419 (output_operand_data): Likewise.
13420 * genrecog.c (print_parameter_value): Likewise.
13421 * lra.c (debug_operand_data): Likewise.
13423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13425 * dwarf2out.c (scompare_loc_descriptor_wide)
13426 (scompare_loc_descriptor_narrow): New functions, split out from...
13427 (scompare_loc_descriptor): ...here.
13428 * expmed.c (emit_store_flag_int): New function, split out from...
13429 (emit_store_flag): ...here.
13431 2017-08-30 Richard Biener <rguenther@suse.de>
13433 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13434 (dwarf2out_early_finish): Move setting of AT_pubnames from
13435 early debug output to early finish.
13437 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13439 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13440 and -mdata-region to the assembler.
13442 2017-08-30 Richard Biener <rguenther@suse.de>
13444 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13446 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13447 (dwarf2out_early_global_decl): Only generate a DIE for the
13448 abstract origin if it doesn't already exist or is a declaration DIE.
13449 (resolve_addr): Do not add the linkage name twice when
13450 generating a stub DIE for the DW_TAG_GNU_call_site target.
13452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13454 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13455 Use machine_mode rather than int for arg1_mode.
13457 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13460 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13461 that the second argument of the built-in functions to unpack
13462 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13463 switch statement instead a lot of if statements.
13464 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13465 Allow 64-bit values to be in Altivec registers as well as
13466 traditional floating point registers.
13467 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13469 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13471 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13472 MAX_REGS_PER_ADDRESS == 1.
13474 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13476 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13477 * config/i386/i386.c (ix86_option_override_internal): Simplify
13478 setting of opts->x_flag_entry.
13480 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13481 Jakub Jelinek <jakub@redhat.com>
13482 Richard Biener <rguenther@suse.de>
13484 PR tree-optimization/81503
13485 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13486 folded constant fits in the target type; reorder tests for clarity.
13488 2017-08-29 Martin Liska <mliska@suse.cz>
13490 * passes.def: Include pass_lower_switch.
13491 * stmt.c (dump_case_nodes): Remove and move to
13492 tree-switch-conversion.
13493 (case_values_threshold): Likewise.
13494 (expand_switch_as_decision_tree_p): Likewise.
13495 (emit_case_decision_tree): Likewise.
13496 (expand_case): Likewise.
13497 (balance_case_nodes): Likewise.
13498 (node_has_low_bound): Likewise.
13499 (node_has_high_bound): Likewise.
13500 (node_is_bounded): Likewise.
13501 (emit_case_nodes): Likewise.
13502 (struct simple_case_node): New struct.
13503 (add_case_node): Remove.
13504 (emit_case_dispatch_table): Use vector instead of case_list.
13505 (reset_out_edges_aux): Remove.
13506 (compute_cases_per_edge): Likewise.
13507 (expand_case): Build list of simple_case_node.
13508 (expand_sjlj_dispatch_table): Use it.
13509 * tree-switch-conversion.c (struct case_node): Moved from
13510 stmt.c and adjusted.
13511 (emit_case_nodes): Likewise.
13512 (node_has_low_bound): Likewise.
13513 (node_has_high_bound): Likewise.
13514 (node_is_bounded): Likewise.
13515 (case_values_threshold): Likewise.
13516 (reset_out_edges_aux): Likewise.
13517 (compute_cases_per_edge): Likewise.
13518 (add_case_node): Likewise.
13519 (dump_case_nodes): Likewise.
13520 (balance_case_nodes): Likewise.
13521 (expand_switch_as_decision_tree_p): Likewise.
13522 (emit_jump): Likewise.
13523 (emit_case_decision_tree): Likewise.
13524 (try_switch_expansion): Likewise.
13525 (do_jump_if_equal): Likewise.
13526 (emit_cmp_and_jump_insns): Likewise.
13527 (fix_phi_operands_for_edge): New function.
13528 (record_phi_operand_mapping): Likewise.
13529 (class pass_lower_switch): New pass.
13530 (pass_lower_switch::execute): New function.
13531 (make_pass_lower_switch): Likewise.
13532 (conditional_probability):
13533 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13534 * tree-pass.h: Add make_pass_lower_switch.
13536 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13539 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13542 2017-08-29 Richard Biener <rguenther@suse.de>
13544 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13545 we do not add a DW_AT_inline attribute twice.
13546 (gen_subprogram_die): Remove code setting DW_AT_inline on
13547 DECL_ABSTRACT_P nodes.
13549 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13551 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13552 calls to internal functions.
13553 (gimplify_modify_expr): Likewise.
13554 * tree-call-cdce.c (use_internal_fn): Likewise.
13555 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13556 (convert_to_divmod): Set the nothrow flag.
13557 * tree-if-conv.c (predicate_mem_writes): Likewise.
13558 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13559 (vectorizable_call): Likewise.
13560 (vectorizable_store): Likewise.
13561 (vectorizable_load): Likewise.
13562 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13563 (vect_recog_mask_conversion_pattern): Likewise.
13565 2017-08-29 Martin Liska <mliska@suse.cz>
13568 * gcc.c (driver_handle_option): Add new argument.
13569 * opts-common.c (handle_option): Pass
13570 target_option_override_hook.
13571 * opts-global.c (lang_handle_option): Add new option.
13572 (set_default_handlers): Add new argument.
13573 (decode_options): Likewise.
13574 * opts.c (target_handle_option): Likewise.
13575 (common_handle_option): Call target_option_override_hook.
13576 * opts.h (struct cl_option_handler_func): Add hook for
13577 target option override.
13578 (struct cl_option_handlers): Likewise.
13579 (set_default_handlers): Add new argument.
13580 (decode_options): Likewise.
13581 (common_handle_option): Likewise.
13582 (target_handle_option): Likewise.
13583 * toplev.c (toplev::main): Pass targetm.target_option.override
13586 2017-08-29 Richard Biener <rguenther@suse.de>
13587 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13589 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13590 life to the active subtree.
13592 2017-08-28 Jeff Law <law@redhat.com>
13594 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13595 derive_equivalences.
13596 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13597 Code moved into....
13598 (edge_info::derive_equivalences): New private member function
13600 * tree-ssa-dom.c (class edge_info): Changed from a struct
13601 to a class. Add ctor/dtor, methods and data members.
13602 (edge_info::edge_info): Renamed from allocate_edge_info.
13603 Initialize additional members.
13604 (edge_info::~edge_info): New.
13605 (free_dom_edge_info): Delete the edge info.
13606 (record_edge_info): Use new class & associated member functions.
13607 Tighten forms for testing for edge equivalences.
13608 (record_temporary_equivalences): Iterate over the simple
13609 equivalences rather than assuming there's only one per edge.
13610 (cprop_into_successor_phis): Iterate over the simple
13611 equivalences rather than assuming there's only one per edge.
13612 (optimize_stmt): Use operand_equal_p rather than pointer
13613 equality for mini-DSE code.
13615 2017-08-28 Nathan Sidwell <nathan@acm.org>
13617 * gcc.c (execute): Fold SIGPIPE handling into switch
13618 statement. Adjust internal error message.
13620 2017-08-28 Richard Biener <rguenther@suse.de>
13623 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13624 Do nothing for removed DIEs.
13626 2017-08-28 Richard Biener <rguenther@suse.de>
13628 PR tree-optimization/81977
13629 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13632 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13635 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13636 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13637 using targetm.gen_mem_thread_fence.
13639 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13642 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13643 predicate to register_operand. Reorder operands.
13644 (*btr<mode>): Ditto.
13645 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13646 (*btr<mode>_mask): Ditto.
13648 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13650 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13651 * config/rs6000/xmmintrin.h: New file.
13652 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13654 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13657 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13658 parameter and_insn and return it.
13659 (recombine_lvx_pattern): Insert a copy to ensure availability of
13660 the base register of the copied masking operation at the point of
13661 the instruction replacement.
13662 (recombine_stvx_pattern): Likewise.
13664 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13666 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13667 undocumented switches.
13668 (-mpower9-dform-vector): Likewise.
13669 (-mpower9-dform): Likewise.
13670 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13671 comments to delete references to -mpower9-dform* switches.
13672 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13673 Delete reference to -mpower9-dform* switches, test for
13674 -mpower9-vector instead.
13675 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13676 (OTHER_P9_VECTOR_MASKS): Likewise.
13677 (POWERPC_MASKS): Likewise.
13678 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13679 tests against -mpower9-dform* to -mpower9-vector. Delete code
13680 that checked for -mpower9-dform* consistancy with other options.
13681 Add test for -mpower9-misc to enable other power9 switches.
13682 (rs6000_init_hard_regno_mode_ok): Likewise.
13683 (rs6000_option_override_internal): Likewise.
13684 (rs6000_emit_prologue): Likewise.
13685 (rs6000_emit_epilogue): Likewise.
13686 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13687 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13688 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13690 (emit_fusion_p9_store): Likewise.
13691 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13692 resetting these macros if the assembler does not support ISA 3.0
13694 (TARGET_P9_DFORM_VECTOR): Likewise.
13695 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13696 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13698 2017-08-25 Alan Modra <amodra@gmail.com>
13700 PR rtl-optimization/81747
13701 * cse.c (cse_extended_basic_block): Don't attempt to record
13702 equivalences for degenerate conditional branches that branch
13703 to their fall-through.
13705 2017-08-24 Martin Sebor <msebor@redhat.com>
13707 PR middle-end/81908
13708 * gimple-fold.c (size_must_be_zero_p): New function.
13709 (gimple_fold_builtin_memory_op): Call it.
13711 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13713 * config/rs6000/mm_malloc.h: New file.
13715 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13717 PR tree-optimization/81913
13718 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13719 analysis when either IVs in condition can wrap.
13721 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13723 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13724 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13726 2017-08-24 Richard Biener <rguenther@suse.de>
13729 * targhooks.c (default_target_can_inline_p): Properly
13730 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13731 is present and always compare.
13732 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13733 infer -mfpmath=sse from TARGET_SSE_P.
13734 (ix86_can_inline_p): Properly use target_option_default_node when
13735 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13737 2017-08-24 Richard Biener <rguenther@suse.de>
13740 * dwarf2out.c (output_die): Handle flag_generate_offload like
13742 (output_comp_unit): Likewise.
13743 (gen_array_type_die): Likewise.
13744 (dwarf2out_early_finish): Likewise.
13745 (note_variable_value_in_expr): Likewise.
13746 (dwarf2out_finish): Likewise. Adjust assert.
13747 * cgraphunit.c (symbol_table::compile): Move setting of
13748 flag_generate_offload earlier ...
13749 (symbol_table::finalize_compilation_unit): ... here, before
13750 early debug finalization.
13752 2017-08-24 Richard Biener <rguenther@suse.de>
13754 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13755 and ipa-fnsummary.h.
13756 (ix86_can_inline_p): When ix86_fpmath flags do not match
13757 check whether the callee uses FP math at all.
13759 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13761 PR middle-end/81931
13762 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13763 instead of TYPE_PRECISION.
13765 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13766 Alan Hayward <alan.hayward@arm.com>
13767 David Sherwood <david.sherwood@arm.com>
13769 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13771 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13773 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13775 (ix86_frame::stack_realign_allocate): New field.
13776 (struct machine_frame_state): Modify comments.
13777 (machine_frame_state::sp_realigned_fp_end): New field.
13778 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13779 layout calculation.
13780 (sp_valid_at): Add assertion to assure no attempt to access invalid
13781 offset of a realigned stack.
13782 (fp_valid_at): Likewise.
13783 (choose_baseaddr): Modify comments.
13784 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13785 ix86_expand_prologue.
13786 (ix86_expand_prologue): Modify stack realignment and allocation.
13787 (ix86_expand_epilogue): Modify comments.
13788 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13789 avx2_runtime, avx512f, and avx512f_runtime.
13791 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13793 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13794 (mstackrealign): Do not init to -1.
13795 * config/i386/i386.c (ix86_option_override_internal):
13796 Check opts_set, not opts when setting default value of
13797 opts->x_ix86_force_align_arg_pointer.
13799 2017-08-23 Richard Biener <rguenther@suse.de>
13801 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13802 lang_hooks.decl_printable_name.
13803 * print-rtl-function.c (print_rtx_function): Likewise.
13804 * tree-pretty-print.c (dump_function_header): Likewise.
13806 2017-08-23 Richard Biener <rguenther@suse.de>
13809 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13810 -g0 at compile-time.
13812 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13814 PR middle-end/19706
13815 * doc/sourcebuild.texi (Other hardware attributes):
13818 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13820 PR middle-end/19706
13821 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13822 Fix single-use check.
13824 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13826 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13828 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13830 * doc/install.texi: Modify to add more details on running selected
13833 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13835 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13836 is combined with -mabi=ms.
13837 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13840 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13842 PR tree-optimization/81488
13843 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13844 and cached_basis fields.
13845 (MAX_SPREAD): New constant.
13846 (alloc_cand_and_find_basis): Initialize new fields.
13847 (clear_visited): New function.
13848 (create_phi_basis_1): Rename from create_phi_basis, set visited
13849 and cached_basis fields.
13850 (create_phi_basis): New wrapper function.
13851 (phi_add_costs_1): Rename from phi_add_costs, add spread
13852 parameter, set visited field, short-circuit when limits reached.
13853 (phi_add_costs): New wrapper function.
13854 (record_phi_increments_1): Rename from record_phi_increments, set
13856 (record_phi_increments): New wrapper function.
13857 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13858 (phi_incr_cost): New wrapper function.
13859 (all_phi_incrs_profitable_1): Rename from
13860 all_phi_incrs_profitable, set visited field.
13861 (all_phi_incrs_profitable): New wrapper function.
13863 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13864 Alan Hayward <alan.hayward@arm.com>
13865 David Sherwood <david.sherwood@arm.com>
13867 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13868 that takes the outer and inner modes.
13869 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13870 comparison as the canonical test for a paradoxical subreg.
13871 * combine.c (simplify_set): Use paradoxical_subreg_p.
13872 (make_extraction): Likewise.
13873 (force_to_mode): Likewise.
13874 (rtx_equal_for_field_assignment_p): Likewise.
13875 (gen_lowpart_for_combine): Likewise.
13876 (simplify_comparison): Likewise.
13877 * cse.c (equiv_constant): Likewise.
13878 * expmed.c (store_bit_field_1): Likewise.
13879 * final.c (alter_subreg): Likewise.
13880 * fwprop.c (propagate_rtx): Likewise.
13881 (forward_propagate_subreg): Likewise.
13882 * ira-conflicts.c (ira_build_conflicts): Likewise.
13883 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13884 * lra-constraints.c (curr_insn_transform): Likewise.
13885 (split_reg): Likewise.
13886 * lra-eliminations.c (move_plus_up): Likewise.
13887 (lra_eliminate_regs_1): Likewise.
13888 * recog.c (general_operand): Likewise.
13889 * ree.c (combine_reaching_defs): Likewise.
13890 * reload.c (push_reload): Likewise.
13891 (find_reloads): Likewise.
13892 * reload1.c (elimination_effects): Likewise.
13893 (compute_reload_subreg_offset): Likewise.
13894 (choose_reload_regs): Likewise.
13895 * rtlanal.c (subreg_lsb_1): Likewise.
13896 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13897 (simplify_subreg): Likewise.
13898 * var-tracking.c (track_loc_p): Likewise.
13899 * emit-rtl.c (byte_lowpart_offset): Likewise.
13900 (paradoxical_subreg_p): Delete out-of-line definition.
13902 2017-08-22 Jeff Law <law@redhat.com>
13904 PR tree-optimization/81741
13905 PR tree-optimization/71947
13906 * tree-ssa-dom.c: Include tree-inline.h.
13907 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13908 equivalences if one is more expensive to compute than the other.
13909 * tree-ssa-scopedtables.h (class const_or_copies): Make
13910 record_const_or_copy_raw method private.
13911 (class avail_exprs_stack): New method simplify_binary_operation.
13912 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13913 avail_exprs_stack::simplify_binary_operation as needed.
13914 (avail_exprs_stack::simplify_binary_operation): New function.
13916 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13918 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13919 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13920 (DOT_SYMBOLS): Likewise.
13921 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13922 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13923 (RS6000_ABI_NAME): Likewise.
13924 (TARGET_CMODEL): Likewise.
13925 (TOC_SECTION_ASM_OP): Likewise.
13926 (SET_CMODEL): New macro.
13927 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13929 2017-08-22 Richard Biener <rguenther@suse.de>
13931 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13932 to allow for free-lang-data replacements similar to verify_type_variant.
13934 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13935 Alan Hayward <alan.hayward@arm.com>
13936 David Sherwood <david.sherwood@arm.com>
13938 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13939 VOIDmode for the LABEL_REF.
13941 2017-08-22 Richard Biener <rguenther@suse.de>
13943 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13945 2017-08-22 Richard Biener <rguenther@suse.de>
13947 * common.opt (feliminate-dwarf2-dups): Ignore.
13948 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13949 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13950 same_die_p_wrap, compute_section_prefix,
13951 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13952 (comdat_symbol_id, comdat_symbol_number): Likewise.
13953 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13955 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13956 (output_die): Mark unreachable path unreachable.
13957 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13958 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13959 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13960 (dwarf2out_early_finish): Likewise.
13962 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13964 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13966 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13969 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13970 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13971 (avr_attribute_table) <io, io_low, address>: Initialize
13972 .decl_required with true.
13974 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13976 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13977 undocumented debugging options.
13978 (-mvsx-scalar-double): Likewise.
13979 (-mallow-df-permute): Likewise.
13980 (-mvectorize-builtins): Likewise.
13981 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13982 (rs6000_builtin_vectorized_function): Likewise.
13983 (rs6000_builtin_md_vectorized_function): Likewise.
13984 (rs6000_opt_vars): Likewise.
13986 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13989 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13990 (*btrq_imm): Rename from *btrq.
13991 (*btcq_imm): Rename from *btcq.
13992 (btsc): New code attribute.
13993 (*<btsc><mode>): New insn pattern.
13994 (*btr<mode>): Ditto.
13995 (*<btsc><mode>_mask): New insn_and_split pattern.
13996 (*btr<mode>_mask): Ditto.
13998 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13999 Alan Hayward <alan.hayward@arm.com>
14000 David Sherwood <david.sherwood@arm.com>
14002 * function.c (pad_below): Simplify padding calculation.
14004 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14005 Alan Hayward <alan.hayward@arm.com>
14006 David Sherwood <david.sherwood@arm.com>
14008 * target.def (function_prologue): Remove frame size argument.
14009 (function_epilogue): Likewise.
14010 * doc/tm.texi: Regenerate.
14011 * final.c (final_start_function): Update call to function_prologue.
14012 (final_end_function): Update call to function_epilogue.
14013 (default_function_pro_epilogue): Remove frame size argument.
14014 * output.h (default_function_pro_epilogue): Likewise.
14015 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14016 (arm_output_function_prologue): Likewise.
14017 * config/frv/frv.c (frv_function_prologue): Likewise.
14018 (frv_function_epilogue): Likewise.
14019 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14020 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14021 (ia64_output_function_epilogue): Likewise.
14022 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14023 (m32r_output_function_epilogue): Likewise.
14024 * config/microblaze/microblaze.c (microblaze_function_prologue)
14025 (microblaze_function_epilogue): Likewise.
14026 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14027 (mips_output_function_epilogue): Likewise.
14028 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14029 (mmix_target_asm_function_epilogue): Likewise.
14030 * config/msp430/msp430.c (msp430_start_function): Likewise.
14031 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14032 (nds32_asm_function_epilogue): Likewise.
14033 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14034 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14035 (pa_output_function_epilogue): Likewise.
14036 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14037 (rs6000_output_function_epilogue): Likewise.
14038 * config/rl78/rl78.c (rl78_start_function): Likewise.
14039 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14040 (rs6000_output_function_epilogue): Likewise.
14041 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14042 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14043 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14044 (sparc_asm_function_epilogue): Likewise.
14046 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14048 * tree.h (type_has_mode_precision_p): New function.
14049 * convert.c (convert_to_integer_1): Use it.
14050 * expr.c (expand_expr_real_2): Likewise.
14051 (expand_expr_real_1): Likewise.
14052 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14053 * match.pd: Likewise.
14054 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14055 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14056 * tree-tailcall.c (process_assignment): Likewise.
14057 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14058 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14059 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14060 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14061 (vectorizable_assignment): Likewise.
14062 (vectorizable_shift): Likewise.
14063 (vectorizable_operation): Likewise.
14064 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14066 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14068 * match.pd: Add pow (C, x) simplification.
14070 2017-08-21 Richard Biener <rguenther@suse.de>
14072 PR tree-optimization/81900
14073 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14074 for blocks with abnormal predecessors.
14075 (compute_antic): Do not set visited flag prematurely.
14077 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14080 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14082 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14084 * stor-layout.h (vector_type_mode): Move to...
14085 * tree.h (vector_type_mode): ...here.
14086 * stor-layout.c (vector_type_mode): Move to...
14087 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14089 2017-08-21 Richard Biener <rguenther@suse.de>
14091 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14092 register_external_die hooks.
14093 (debug_false_tree_charstarstar_uhwistar): Declare.
14094 (debug_nothing_tree_charstar_uhwi): Likewise.
14095 * debug.c (do_nothing_debug_hooks): Adjust.
14096 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14097 (debug_nothing_tree_charstar_uhwi): Likewise.
14098 * dbxout.c (dbx_debug_hooks): Adjust.
14099 (xcoff_debug_hooks): Likewise.
14100 * sdbout.c (sdb_debug_hooks): Likewise.
14101 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14102 * dwarf2out.c (macinfo_label_base): New global.
14103 (dwarf2out_register_external_die): New function for the
14104 register_external_die hook.
14105 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14106 (dwarf2_debug_hooks): Use them.
14107 (dwarf2_lineno_debug_hooks): Adjust.
14108 (struct die_struct): Add with_offset flag.
14109 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14110 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14111 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14112 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14113 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14114 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14115 defining section names for the early LTO debug variants.
14116 (reset_indirect_string): New helper.
14117 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14118 (print_dw_val): Add support for offsetted symbol references.
14119 (get_ultimate_context): Split out from is_cxx.
14120 (is_cxx): Use get_ultimate_context.
14121 (is_fortran): Add decl overload.
14122 (compute_comp_unit_symbol): Split out worker from
14123 compute_section_prefix.
14124 (compute_section_prefix): Call compute_comp_unit_symbol and
14125 set comdat_type_p here.
14126 (output_die): Skip DIE symbol output for the LTO added one.
14127 Handle DIE symbol references with offset.
14128 (output_comp_unit): Guard section name mangling properly.
14129 For LTO debug sections emit a symbol at the section beginning
14130 which we use to refer to its DIEs.
14131 (add_abstract_origin_attribute): For DIEs registered via
14132 dwarf2out_register_external_die directly refer to the early
14133 DIE rather than indirectly through the shadow one we created.
14134 Remove obsolete call to dwarf2out_abstract_function for
14135 non-function/block origins.
14136 (gen_array_type_die): When generating early LTO debug do
14137 not emit DW_AT_string_length.
14138 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14139 late when in LTO. As suggested place a gcc_unreachable for
14140 the DECL_ABSTRACT_P case.
14141 (gen_subprogram_die): Avoid another specification DIE
14142 for early built declarations/definitions for the late LTO case.
14143 (gen_variable_die): Add type references for late duplicated VLA dies
14145 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14146 we have the abstract instance already.
14147 (process_scope_var): Adjust decl DIE contexts in LTO which
14148 first puts them in limbo.
14149 (gen_decl_die): Do not generate type DIEs late apart from
14150 types for VLAs or for decls we do not yet have a DIE. Do not
14151 call dwarf2out_abstract_function late.
14152 (dwarf2out_early_global_decl): Make sure to create DIEs
14153 for abstract instances of a decl first.
14154 (dwarf2out_late_global_decl): Adjust comment.
14155 (output_macinfo_op): With multiple macro sections use
14156 macinfo_label_base to distinguish labels.
14157 (output_macinfo): Likewise. Update macinfo_label_base.
14158 Pass in the line info label.
14159 (note_variable_value_in_expr): When generating LTO resolve
14160 all variable values here by generating DIEs as needed.
14161 (init_sections_and_labels): Add early LTO debug flag parameter
14162 and generate different sections and names if set. Add generation
14163 counter for the labels so we can have multiple of them.
14164 (reset_dies): Helper to allow DIEs to be output multiple times.
14165 (dwarf2out_finish): When outputting DIEs to the fat part of an
14166 LTO object first reset DIEs.
14167 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14168 (modified_type_die): Check for decl_ultimate_origin being self
14170 (gen_type_die_with_usage): Likewise.
14171 (gen_typedef_die): Allow decl_ultimate_origin being self.
14172 (set_decl_abstract_flags): Remove.
14173 (set_block_abstract_flags): Likewise.
14174 (dwarf2out_abstract_function): Treat the early generated DIEs
14175 as the abstract copy and only add DW_AT_inline and
14176 DW_AT_artificial here and call set_decl_origin_self.
14177 If the DIE has an abstract origin don't do anything.
14178 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14179 if we have none yet (Go fails to build one, PR78628).
14180 (variably_modified_type_p): Prevent endless recursion for Ada
14181 cyclic pointer types.
14182 * lto-streamer-in.c: Include debug.h.
14183 (dref_queue): New global.
14184 (lto_read_tree_1): Stream in DIE references.
14185 (lto_input_tree): Register DIE references.
14186 (input_function): Stream DECL_DEBUG_ARGS.
14187 * lto-streamer-out.c: Include debug.h.
14188 (lto_write_tree_1): Output DIE references.
14189 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14190 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14191 (output_function): Stream DECL_DEBUG_ARGS.
14192 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14193 Stream DECL_ABSTRACT_ORIGIN.
14194 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14195 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14196 DECL_CONTEXT for file-scope decls.
14197 * lto-streamer.h (struct dref_entry): Declare.
14198 (dref_queue): Likewise.
14199 * cfgexpand.c (pass_expand::execute): Do not call the
14200 outlining_inline_function hook here.
14201 * lto-wrapper.c (debug_obj): New global.
14202 (tool_cleanup): Unlink it if required.
14203 (debug_objcopy): New function.
14204 (run_gcc): Handle early debug sections in the IL files by
14205 extracting them to separate files, partially linkin them and
14206 feeding the result back as result to the linker.
14207 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14208 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14209 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14210 sections into a separate segment.
14211 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14213 (darwin_asm_dwarf_section): Likewise.
14214 (darwin_asm_output_dwarf_offset): Likewise.
14215 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14217 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14218 Alan Hayward <alan.hayward@arm.com>
14219 David Sherwood <david.sherwood@arm.com>
14221 * read-md.h (md_reader::record_potential_iterator_use): Replace
14222 pointer argument with an rtx and an index.
14223 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14224 (apply_mode_iterator): Likewise.
14225 (apply_code_iterator): Likewise.
14226 (apply_int_iterator): Likewise.
14227 (apply_subst_iterator): Likewise.
14228 (record_iterator_use): Likewise.
14229 (record_attribute_use): Likewise.
14230 (md_reader::record_potential_iterator_use): Likewise. Update calls
14231 to record_iterator_use and apply_iterator.
14232 (iterator_use): Replace ptr with x and index.
14233 (attribute_use): Likewise.
14234 (apply_attribute_uses): Update calls to apply_iterator.
14235 (apply_iterators): Likewise. Update initialization of iterator_use.
14236 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14237 and record_potential_iterator_use.
14238 (rtx_reader::read_rtx_operand): Likewise.
14240 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14241 Alan Hayward <alan.hayward@arm.com>
14242 David Sherwood <david.sherwood@arm.com>
14244 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14247 2017-08-21 Richard Biener <rguenther@suse.de>
14249 PR middle-end/81884
14250 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14251 at struct end conservatively when comparing common bases.
14253 2017-08-21 Richard Biener <rguenther@suse.de>
14255 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14256 (mem_ref_in_stmt): Remove.
14257 (determine_max_movement): Use ref index to get at the reference.
14258 (invariantness_dom_walker::before_dom_children): Deal with
14259 lim data already initialized.
14260 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14262 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14264 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14265 (TARGET_ISA_ROUND): Ditto.
14266 (TARGET_ROUND): Ditto.
14267 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14268 * config/i386/i386.md: Ditto.
14269 * config/i386/sse.md: Ditto.
14270 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14271 with OPTION_MASK_ISA_SSE4_1.
14273 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14276 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14277 __builtin_ia32_lzcnt_u16.
14279 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14282 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14283 (rs6000_set_current_function): Rewrite function to use it.
14285 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14288 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14289 and TYPE_WARN_IF_NOT_ALIGN.
14290 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14291 (handle_warn_if_not_align): New.
14292 (place_union_field): Call handle_warn_if_not_align.
14293 (place_field): Call handle_warn_if_not_align.
14294 Copy TYPE_WARN_IF_NOT_ALIGN.
14295 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14296 (layout_type): Likewise.
14297 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14299 (tree_decl_common): Add warn_if_not_align.
14300 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14301 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14302 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14303 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14304 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14305 * doc/extend.texi: Document warn_if_not_aligned attribute.
14306 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14308 2017-08-17 Martin Liska <mliska@suse.cz>
14311 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14312 (get_data_dependence): Use it as pointer type.
14313 (distribute_loop): Likewise.
14315 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14317 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14318 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14319 (p8_vmrgow_<mode>_direct): New define_insn.
14320 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14321 handle endianness for vmrgew and vmrgow permute patterns.
14323 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14325 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14326 * config/rs6000/rs6000-cpus.def: Remove comment.
14327 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14328 (POWERPC_MASKS): Likewise.
14329 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14330 use of TARGET_VSX_TIMODE.
14331 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14332 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14333 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14334 (rs6000_option_override_internal): Remove dead code.
14335 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14336 (rs6000_legitimize_reload_address): Likewise.
14337 (rs6000_legitimate_address_p): Likewise.
14338 (rs6000_opt_masks): Delete "vsx-timode".
14339 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14340 from function comment.
14341 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14342 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14343 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14345 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14346 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14347 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14348 (VSX_TI): Likewise.
14350 (define_peephole2): Likewise.
14352 2017-08-17 Martin Sebor <msebor@redhat.com>
14355 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14356 past the end of an array.
14357 (test_pp_format): Add test cases.
14359 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14361 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14362 missing ECF_NOTHROW flags.
14364 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14367 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14368 operands residing in integer registers.
14369 (*vsx_le_perm_load_<mode>): Likewise.
14370 (*vsx_le_perm_store_<mode>): Likewise.
14371 (define_peephole2): Add peepholes to optimize the above.
14373 2017-08-17 Marek Polacek <polacek@redhat.com>
14375 PR middle-end/81814
14376 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14377 to mimic what shorten_compare did. Change the return type to bool.
14378 (fold_cond_expr_with_comparison): Update call to
14379 operand_equal_for_comparison_p.
14380 (fold_ternary_loc): Likewise.
14382 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14384 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14386 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14388 2017-08-17 Richard Biener <rguenther@suse.de>
14390 * tree-ssa-structalias.c (solve_graph): When propagating
14391 to successors update the graphs succ edges and avoid duplicate work.
14393 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14396 * config/i386/i386.c (ix86_option_override_internal): Save target
14397 specific options after ix86_stack_protector_guard_reg was changed.
14399 2017-08-17 Richard Biener <rguenther@suse.de>
14401 PR tree-optimization/81827
14402 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14403 (new_var_info): Initialize it conservatively.
14404 (get_call_vi): Mark register vars.
14405 (new_scalar_tmp_constraint_exp): Likewise.
14406 (handle_rhs_call): Likewise.
14407 (handle_const_call): Likewise.
14408 (create_function_info_for): Likewise.
14409 (solve_constraints): Sort varinfos to separate register from
14410 non-register vars to pack points-to solution bitmaps during
14413 2017-08-17 Marek Polacek <polacek@redhat.com>
14415 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14417 2017-08-17 Richard Biener <rguenther@suse.de>
14419 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14420 to true when overflow is undefined and we saturated the result.
14422 2017-08-17 Alan Modra <amodra@gmail.com>
14425 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14426 Don't use store multiple if only one reg needs saving.
14427 (interesting_frame_related_regno): New function.
14428 (rs6000_frame_related): Don't emit frame info for regs that
14430 (rs6000_emit_epilogue): Likewise.
14432 2017-08-16 Nathan Sidwell <nathan@acm.org>
14434 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14435 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14436 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14437 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14438 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14439 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14440 (verify_type): Adjust for TYPE_BINFO move.
14441 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14442 process TYPE_BINFO directly.
14443 (hash_tree): Likewise.
14444 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14446 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14449 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14451 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14453 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14456 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14457 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14458 Add ix86_binary_operator_ok to insn constraint.
14459 (*iordi_1_bts): Ditto.
14460 (*xordi_1_btc): Ditto.
14461 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14462 Update corresponding peephole2 pattern.
14466 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14468 PR tree-optimization/81832
14469 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14470 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14472 2017-08-16 Marek Polacek <polacek@redhat.com>
14475 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14476 regarding size_zero_node.
14478 2017-08-16 Martin Liska <mliska@suse.cz>
14481 * config.gcc: Respect previously set extra_objs in case
14484 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14486 PR tree-optimization/81835
14487 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14488 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14489 not depend on the phi.
14491 2017-08-16 Alan Modra <amodra@gmail.com>
14493 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14496 2017-08-16 Alan Modra <amodra@gmail.com>
14498 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14499 (save_reg_p): ..into this. Update all callers.
14500 (first_reg_to_save): Simplify.
14502 2017-08-16 Alan Modra <amodra@gmail.com>
14504 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14507 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14511 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14512 constant count if that count is less than 32.
14514 2017-08-15 Nathan Sidwell <nathan@acm.org>
14516 * gcc.c (execute): Emit friendlier message if inferior is killed
14517 by an external cause.
14519 2017-08-15 Richard Biener <rguenther@suse.de>
14521 PR tree-optimization/81790
14522 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14523 CONSTRUCTORs from simplifying and VN.
14525 2017-08-14 Martin Sebor <msebor@redhat.com>
14527 * builtin-attrs.def: Add comments.
14529 2017-08-14 Martin Sebor <msebor@redhat.com>
14532 * doc/extend.texi (attribute nonstring): Document new attribute.
14534 2017-08-14 Martin Sebor <msebor@redhat.com>
14537 * tree-diagnostic.c (default_tree_printer): Handle %G.
14538 * gimple-pretty-print.h (percent_G_format): Declare new function.
14539 * gimple-pretty-print.c (percent_G_format): Define.
14540 * tree-pretty-print.c (percent_K_format): Add argument.
14542 2017-08-14 Martin Sebor <msebor@redhat.com>
14544 PR translation/79998
14545 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14546 Remove a stray space.
14548 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14551 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14552 (*iordi_1_bts): Ditto.
14553 (*xordi_1_btc): Ditto.
14555 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14558 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14559 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14561 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14562 quoted strings, and make more translator-friendly.
14563 (darwin_rs6000_override_options): Likewise.
14564 (rs6000_option_override_internal): Likewise.
14565 (rs6000_return_in_memory): Fix overlong line.
14566 (init_cmulative_args): Use quoted strings, and make more
14567 translator-friendly.
14568 (rs6000_pass_by_reference): Fix overlong line.
14569 (def_builtin): Use quoted strings.
14570 (altivec_expand_predicate_builtin): Use quoted strings, and make
14571 more translator-friendly.
14572 (htm_expand_builtin): Use quoted strings.
14573 (cpu_expand_builtin): Use quoted strings, and make more
14574 translator-friendly.
14575 (altivec_expand_builtin): Likewise.
14576 (paired_expand_predicate_builtin): Likewise.
14577 (rs6000_invalid_builtin): Likewise.
14578 (builtin_function_type): Use quoted strings.
14579 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14580 more translator-friendly.
14581 (rs6000_trampoline_init): Likewise.
14582 (rs6000_handle_altivec_attribute): Likewise.
14583 (rs6000_inner_target_options): Use quoted strings.
14584 (rs6000_disable_incompatible_switches): Likewise.
14585 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14586 strings, and make more translator-friendly.
14587 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14589 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14591 PR tree-optimization/81799
14592 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14593 cond_expr to simple gimple operand.
14595 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14597 PR middle-end/46932
14598 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14600 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14604 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14605 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14606 TARGET_GASISR_PROLOGUES.
14607 * config/avr/avr.c (avr_option_override): Same.
14608 (avr_pass_pre_proep::execute): Same.
14610 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14613 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14614 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14616 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14618 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14619 define_insn_and_split. Split to a memory load from 0 in
14620 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14621 using PTR mode iterator.
14622 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14623 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14624 (*add_tp_<mode>): Redefine as define_insn_and_split.
14625 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14626 address space. Merge with *add_tp_x32 using PTR mode iterator.
14627 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14628 Split to an add with a memory load from 0 in
14629 DEFAULT_TLS_SEG_REG address space.
14631 2017-08-12 Andrew Pinski <apinski@cavium.com>
14633 * config/aarch64/aarch64-option-extensions.def (rdma):
14634 Fix feature string to what Linux prints out in /proc/cpuinfo.
14636 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14639 * dwarf2out.c (modified_type_die): For C typedef types that have
14640 an ultimate origin, process the ultimate origin instead of the
14642 (gen_typedef_die): Assert that input DECLs have no ultimate
14644 (gen_type_die_with_usage): For typedef variants that have an
14645 ultimate origin, just call gen_decl_die on the original DECL.
14646 (process_scope_var): Avoid creating DIEs for local typedefs and
14647 concrete static variables.
14649 2017-08-12 Alan Modra <amodra@gmail.com>
14653 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14654 match gnu-user.h startfile.
14655 (ENDFILE_LINUX_SPEC): Similarly.
14657 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14660 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14662 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14664 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14665 * config/aarch64/aarch64.md (mov<mode>): Change.
14666 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14667 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14668 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14670 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14672 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14673 for storage order barriers.
14675 2017-08-11 Martin Liska <mliska@suse.cz>
14678 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14679 variables of void type.
14681 2017-08-11 Martin Liska <mliska@suse.cz>
14683 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14684 TARGET_SUPPORTS_ALIASES.
14685 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14686 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14687 (optimize_weakref): Likewise.
14688 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14689 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14690 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14692 2017-08-11 Martin Liska <mliska@suse.cz>
14695 * config/i386/i386.c (make_resolver_func): Do complete
14696 refactoring of the function.
14698 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14701 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14702 * config/i386/i386.c (ix86_stack_protect_guard): Use
14703 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14704 * doc/invoke.texi (x86 Options): Document
14705 -mstack-protector-guard-symbol= option.
14707 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14709 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14710 * config/i386/i386.c (ix86_split_stack_guard): New function.
14711 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14712 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14713 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14714 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14715 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14716 (split_stack_space_check): Call ix86_split_stack_guard.
14718 2017-08-10 Martin Sebor <msebor@redhat.com>
14720 * print-tree.c (print_node): Print location using the established
14722 Replace spaces with colons.
14723 (debug_raw, debug): Ditto.
14725 2017-08-10 Martin Sebor <msebor@redhat.com>
14728 * pretty-print.c (pp_format): Correct the handling of %s precision.
14730 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14733 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14735 (ix86_finalize_stack_frame_flags): This. Also clear
14736 frame_pointer_needed if -fno-omit-frame-pointer is used without
14738 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14739 with ix86_finalize_stack_frame_flags.
14740 (ix86_expand_epilogue): Likewise.
14741 (ix86_expand_split_stack_prologue): Likewise.
14742 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14744 2017-08-10 Martin Liska <mliska@suse.cz>
14747 * c-attribs.c (handle_target_attribute):
14748 Report warning for an empty string argument of target attribute.
14750 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14753 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14755 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14756 or DECL_NONLOCAL labels.
14757 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14758 or DECL_NONLOCAL labels here.
14760 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14762 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14763 to indicate when early gimple folding has been disabled.
14764 (rs6000_gimple_fold_builtin): Add debug content.
14765 (rs6000_invalid_builtin): Fix whitespace.
14766 (rs6000_expand_builtin): Fix whitespace.
14767 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14769 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14772 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14773 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14775 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14777 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14778 (qdf24xx): Likewise.
14779 * config/aarch64/aarch64.md: Include falkor.md.
14780 * config/aarch64/falkor.md: New.
14782 2017-08-09 Marek Polacek <polacek@redhat.com>
14785 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14786 * diagnostic.c (emit_diagnostic): Add a comment.
14787 (emit_diagnostic_valist): New function.
14789 2017-08-09 Marek Polacek <polacek@redhat.com>
14792 * input.c (make_location): New overload.
14793 * input.h (make_location): Declare.
14795 2017-08-08 Alan Modra <amodra@gmail.com>
14796 H.J. Lu <hongjiu.lu@intel.com>
14799 * gcc.c (NO_PIE_SPEC): Delete.
14800 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14802 (LINK_PIE_SPEC): ..to here.
14803 (LINK_COMMAND_SPEC): Support -no-pie.
14804 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14805 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14807 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14808 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14809 (ENDFILE_CRTEND_SPEC): Similarly.
14811 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14814 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14815 (mstack-protector-guard-offset=): Ditto.
14816 * config/i386/i386.c (ix86_option_override): Handle
14817 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14819 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14820 ix86_stack_protect_guard_offset variables.
14821 (TARGET_STACK_PROTECT_GUARD): Always define.
14822 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14823 and -mstack-protector-guard-offset= options.
14825 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14827 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14828 boundary case for the last candidate.
14830 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14832 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14833 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14835 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14837 PR middle-end/19706
14838 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14839 * config/aarch64/aarch64-builtins.c
14840 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14841 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14842 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14844 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14845 Andrew Pinski <pinskia@gmail.com>
14847 PR middle-end/19706
14848 * internal-fn.def (XORSIGN): New.
14849 * optabs.def (xorsign_optab): New.
14850 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14851 (convert_expand_mult_copysign): New.
14852 (pass_optimize_widening_mul::execute): Call
14853 convert_expand_mult_copysign.
14855 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14857 PR tree-optimization/81354
14858 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14859 Insert on edges rather than explicitly creating landing pads.
14860 (analyze_candidates_and_replace): Commit edge inserts.
14862 2017-08-08 Richard Biener <rguenther@suse.de>
14864 PR middle-end/81719
14865 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14866 (expand_simple_operations): Also look through ADDR_EXPRs with
14867 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14869 2017-08-08 Richard Biener <rguenther@suse.de>
14871 PR tree-optimization/81723
14872 * tree-vect-slp.c (struct bst_traits): New hash traits.
14873 (bst_fail): New global.
14874 (vect_build_slp_tree_2): New worker, split out from ...
14875 (vect_build_slp_tree): ... this now wrapping it with using
14876 bst_fail set to cache SLP tree build fails. Properly handle
14878 (vect_analyze_slp_instance): Allocate and free bst_fail.
14880 2017-08-08 Martin Liska <mliska@suse.cz>
14883 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14884 LABEL_DECLs that can be from a different function.
14886 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14888 PR tree-optimization/81744
14889 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14890 loop's number of iterations.
14892 2017-08-08 Martin Liska <mliska@suse.cz>
14894 * asan.c: Include header files.
14895 * attribs.c (build_decl_attribute_variant): New function moved
14897 (build_type_attribute_qual_variant): Likewise.
14898 (cmp_attrib_identifiers): Likewise.
14899 (simple_cst_list_equal): Likewise.
14900 (omp_declare_simd_clauses_equal): Likewise.
14901 (attribute_value_equal): Likewise.
14902 (comp_type_attributes): Likewise.
14903 (build_type_attribute_variant): Likewise.
14904 (lookup_ident_attribute): Likewise.
14905 (remove_attribute): Likewise.
14906 (merge_attributes): Likewise.
14907 (merge_type_attributes): Likewise.
14908 (merge_decl_attributes): Likewise.
14909 (merge_dllimport_decl_attributes): Likewise.
14910 (handle_dll_attribute): Likewise.
14911 (attribute_list_equal): Likewise.
14912 (attribute_list_contained): Likewise.
14913 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14914 (lookup_attribute_by_prefix): Likewise.
14915 * bb-reorder.c: Include header files.
14916 * builtins.c: Likewise.
14917 * calls.c: Likewise.
14918 * cfgexpand.c: Likewise.
14919 * cgraph.c: Likewise.
14920 * cgraphunit.c: Likewise.
14921 * convert.c: Likewise.
14922 * dwarf2out.c: Likewise.
14923 * final.c: Likewise.
14924 * fold-const.c: Likewise.
14925 * function.c: Likewise.
14926 * gimple-expr.c: Likewise.
14927 * gimple-fold.c: Likewise.
14928 * gimple-pretty-print.c: Likewise.
14929 * gimple.c: Likewise.
14930 * gimplify.c: Likewise.
14931 * hsa-common.c: Likewise.
14932 * hsa-gen.c: Likewise.
14933 * internal-fn.c: Likewise.
14934 * ipa-chkp.c: Likewise.
14935 * ipa-cp.c: Likewise.
14936 * ipa-devirt.c: Likewise.
14937 * ipa-fnsummary.c: Likewise.
14938 * ipa-inline.c: Likewise.
14939 * ipa-visibility.c: Likewise.
14941 * lto-cgraph.c: Likewise.
14942 * omp-expand.c: Likewise.
14943 * omp-general.c: Likewise.
14944 * omp-low.c: Likewise.
14945 * omp-offload.c: Likewise.
14946 * omp-simd-clone.c: Likewise.
14947 * opts-global.c: Likewise.
14948 * passes.c: Likewise.
14949 * predict.c: Likewise.
14950 * sancov.c: Likewise.
14951 * sanopt.c: Likewise.
14952 * symtab.c: Likewise.
14953 * toplev.c: Likewise.
14954 * trans-mem.c: Likewise.
14955 * tree-chkp.c: Likewise.
14956 * tree-eh.c: Likewise.
14957 * tree-into-ssa.c: Likewise.
14958 * tree-object-size.c: Likewise.
14959 * tree-parloops.c: Likewise.
14960 * tree-profile.c: Likewise.
14961 * tree-ssa-ccp.c: Likewise.
14962 * tree-ssa-live.c: Likewise.
14963 * tree-ssa-loop.c: Likewise.
14964 * tree-ssa-sccvn.c: Likewise.
14965 * tree-ssa-structalias.c: Likewise.
14966 * tree-ssa.c: Likewise.
14967 * tree-streamer-in.c: Likewise.
14968 * tree-vectorizer.c: Likewise.
14969 * tree-vrp.c: Likewise.
14970 * tsan.c: Likewise.
14971 * ubsan.c: Likewise.
14972 * varasm.c: Likewise.
14973 * varpool.c: Likewise.
14974 * tree.c: Remove functions moved to attribs.[ch].
14975 * tree.h: Likewise.
14976 * config/aarch64/aarch64.c: Add attrs.h header file.
14977 * config/alpha/alpha.c: Likewise.
14978 * config/arc/arc.c: Likewise.
14979 * config/arm/arm.c: Likewise.
14980 * config/avr/avr.c: Likewise.
14981 * config/bfin/bfin.c: Likewise.
14982 * config/c6x/c6x.c: Likewise.
14983 * config/cr16/cr16.c: Likewise.
14984 * config/cris/cris.c: Likewise.
14985 * config/darwin.c: Likewise.
14986 * config/epiphany/epiphany.c: Likewise.
14987 * config/fr30/fr30.c: Likewise.
14988 * config/frv/frv.c: Likewise.
14989 * config/ft32/ft32.c: Likewise.
14990 * config/h8300/h8300.c: Likewise.
14991 * config/i386/winnt.c: Likewise.
14992 * config/ia64/ia64.c: Likewise.
14993 * config/iq2000/iq2000.c: Likewise.
14994 * config/lm32/lm32.c: Likewise.
14995 * config/m32c/m32c.c: Likewise.
14996 * config/m32r/m32r.c: Likewise.
14997 * config/m68k/m68k.c: Likewise.
14998 * config/mcore/mcore.c: Likewise.
14999 * config/microblaze/microblaze.c: Likewise.
15000 * config/mips/mips.c: Likewise.
15001 * config/mmix/mmix.c: Likewise.
15002 * config/mn10300/mn10300.c: Likewise.
15003 * config/moxie/moxie.c: Likewise.
15004 * config/msp430/msp430.c: Likewise.
15005 * config/nds32/nds32-isr.c: Likewise.
15006 * config/nds32/nds32.c: Likewise.
15007 * config/nios2/nios2.c: Likewise.
15008 * config/nvptx/nvptx.c: Likewise.
15009 * config/pa/pa.c: Likewise.
15010 * config/pdp11/pdp11.c: Likewise.
15011 * config/powerpcspe/powerpcspe.c: Likewise.
15012 * config/riscv/riscv.c: Likewise.
15013 * config/rl78/rl78.c: Likewise.
15014 * config/rx/rx.c: Likewise.
15015 * config/s390/s390.c: Likewise.
15016 * config/sh/sh.c: Likewise.
15017 * config/sol2.c: Likewise.
15018 * config/sparc/sparc.c: Likewise.
15019 * config/spu/spu.c: Likewise.
15020 * config/stormy16/stormy16.c: Likewise.
15021 * config/tilegx/tilegx.c: Likewise.
15022 * config/tilepro/tilepro.c: Likewise.
15023 * config/v850/v850.c: Likewise.
15024 * config/vax/vax.c: Likewise.
15025 * config/visium/visium.c: Likewise.
15026 * config/xtensa/xtensa.c: Likewise.
15028 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15031 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15032 constraints since the -mupper-regs-* switches have been
15034 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15035 into a vector from a double word element that was extracted from
15036 another vector, and eliminate extra XXPERMDI instructions.
15037 (vsx_concat_<mode>_2): Likewise.
15038 (vsx_concat_<mode>_3): Likewise.
15039 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15040 concat to allow optimizing inserts from previous extracts.
15042 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15044 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15045 memory reference to a SSP offset in TLS address space.
15046 (ix86_print_operand) <case '@'>: Remove.
15047 (ix86_print_operand_punct_valid_p): Remove '@' code.
15048 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15049 UNSPEC_SP_TLS_TEST.
15050 (stack_tls_protect_set_<mode>): Remove.
15051 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15052 (stack_tls_protect_test_<mode>): Remove.
15053 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15055 2017-08-07 Olivier Hainque <hainque@adacore.com>
15058 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15060 2017-08-07 Douglas Rupp <rupp@adacore.com>
15062 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15063 variable was referenced as multidir in command.
15065 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15068 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15071 2017-08-07 Martin Liska <mliska@suse.cz>
15073 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15074 * config/rl78/rl78.c: Add include of attribs.h.
15075 * config/sh/sh.c: Likewise.
15076 * config/v850/v850.c: Likewise.
15078 2017-08-07 Tom de Vries <tom@codesourcery.com>
15080 PR middle-end/78266
15081 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15083 2017-08-07 Martin Liska <mliska@suse.cz>
15085 * config/mips/mips.c: Include attribs.h.
15087 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15090 * doc/invoke.texi: Document change in behvaior for -Ofast for
15093 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15095 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15097 (aarch64_pop_regs): Likewise.
15098 (aarch64_gen_load_pair): Likewise.
15099 (aarch64_save_callee_saves): Likewise.
15100 (aarch64_restore_callee_saves): Likewise.
15102 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15104 * config/i386/i386.c: Revert the last change.
15106 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15109 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15111 (ix86_finalize_stack_frame_flags): This. Also clear
15112 frame_pointer_needed if -fno-omit-frame-pointer is used without
15114 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15115 with ix86_finalize_stack_frame_flags.
15116 (ix86_expand_epilogue): Likewise.
15117 (ix86_expand_split_stack_prologue): Likewise.
15119 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15122 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15123 to P_AES for Westmere.
15125 2017-08-07 Jonathan Yong <10walls@gmail.com>
15127 * config/i386/mingw.opt (fset-stack-executable): Removed.
15128 * config/i386/cygming.opt (fset-stack-executable): Moved
15130 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15132 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15134 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15136 2017-08-07 Marek Polacek <polacek@redhat.com>
15138 PR middle-end/81737
15139 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15141 2017-08-07 Martin Liska <mliska@suse.cz>
15143 * attribs.h (canonicalize_attr_name): New function.
15144 (cmp_attribs): Move from c-format.c and adjusted.
15145 (is_attribute_p): Moved from tree.h.
15146 * tree-inline.c: Add new includes.
15147 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15148 (private_is_attribute_p): Remove.
15149 (private_lookup_attribute): Likewise.
15150 (private_lookup_attribute_by_prefix): Simplify.
15151 (remove_attribute): Use is_attribute_p.
15152 * tree.h: Remove removed declarations.
15154 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15156 PR middle-end/81698
15157 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15158 instead of computing it in the function. Formatting fix.
15159 (expand_case): Don't rely on default_edge being the first edge,
15160 clear it if removing it, pass default_edge to
15161 emit_case_dispatch_table.
15162 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15165 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15167 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15168 insn in the function, emit NOP after the insn.
15170 2017-08-06 Tom de Vries <tom@codesourcery.com>
15172 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15175 2017-08-06 Tom de Vries <tom@codesourcery.com>
15177 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15180 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15182 PR tree-optimization/57371
15183 * match.pd: New pattern.
15185 2017-08-04 Marek Polacek <polacek@redhat.com>
15187 PR middle-end/81695
15188 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15189 perform the computation in offset_int.
15191 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15193 PR tree-optimization/81136
15194 * tree-vectorizer.h: Include tree-hash-traits.h.
15195 (vec_base_alignments): New typedef.
15196 (vec_info): Add a base_alignments field.
15197 (vect_record_base_alignments): Declare.
15198 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15200 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15201 (create_data_ref): Add an is_conditional_in_stmt argument.
15202 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15203 the is_conditional_in_stmt field.
15204 (data_ref_loc): Add an is_conditional_in_stmt field.
15205 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15206 (find_data_references_in_stmt): Update call to create_data_ref.
15207 (graphite_find_data_references_in_stmt): Likewise.
15208 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15209 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15210 (vect_record_base_alignment): New function.
15211 (vect_record_base_alignments): Likewise.
15212 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15213 for nested statements even if we fail to compute a misalignment.
15214 Use pooled base alignments for unconditional references.
15215 (vect_find_same_alignment_drs): Compare base addresses instead
15217 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15218 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15220 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15222 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15223 Add an explicit name for the enum. Use auto_vec for slp_instances
15224 and grouped_stores.
15225 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15227 (_bb_vec_info): Add a constructor and destructor.
15228 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15229 (destroy_loop_vec_info): Delete.
15230 (vect_destroy_datarefs): Likewise.
15231 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15232 (vec_info::vec_info): New function.
15233 (vec_info::~vec_info): Likewise.
15234 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15235 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15236 destroy_loop_vec_info.
15237 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15238 (_loop_vec_info::_loop_vec_info): ...this.
15239 (destroy_loop_vec_info): Replace with...
15240 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15241 the stmt_vec_infos. Leave handling of vec_info information to its
15242 destructor. Remove explicit vector releases.
15243 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15244 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15245 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15246 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15247 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15248 (destroy_bb_vec_info): Replace with...
15249 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15250 information to its destructor.
15251 (vect_slp_analyze_bb_1): Use new and delete instead of
15252 new_bb_vec_info and destroy_bb_vec_info.
15253 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15256 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15258 * tree-data-ref.h (subscript): Add access_fn field.
15259 (data_dependence_relation): Add could_be_independent_p.
15260 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15261 (same_access_functions): Move to tree-data-ref.c.
15262 * tree-data-ref.c (ref_contains_union_access_p): New function.
15263 (access_fn_component_p): Likewise.
15264 (access_fn_components_comparable_p): Likewise.
15265 (dr_analyze_indices): Add a reference to access_fn_component_p.
15266 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15268 (constant_access_functions): Likewise.
15269 (add_other_self_distances): Likewise.
15270 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15271 (initialize_data_dependence_relation): Use XCNEW and remove
15272 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15273 of access functions that have the same type. Allow the
15274 subsequence to end with different bases in some circumstances.
15275 Record the chosen access functions in SUB_ACCESS_FN.
15276 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15277 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15278 (subscript_dependence_tester_1): Likewise dra and drb.
15279 (build_classic_dist_vector): Update calls accordingly.
15280 (subscript_dependence_tester): Likewise.
15281 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15282 DDR_COULD_BE_INDEPENDENT_P.
15283 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15284 comp_alias_ddrs instead of may_alias_ddrs.
15285 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15287 (vect_analyze_data_ref_dependence): Use it if
15288 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15289 distance vectors if that fails.
15290 (dependence_distance_ge_vf): New function.
15291 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15292 LOOP_VINFO_MAY_ALIAS_DDRS.
15294 2017-08-04 Richard Biener <rguenther@suse.de>
15296 PR middle-end/81705
15297 * fold-const.c (fold_binary_loc): Properly restrict
15298 minus_var0 && minus_var1 case when associating undefined overflow
15301 2017-08-04 Tom de Vries <tom@codesourcery.com>
15303 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15305 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15307 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15308 Don't start diagnostic messages with a capital letter.
15309 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15311 (rs6000_invalid_builtin): Likewise.
15312 (rs6000_trampoline_init): Likewise.
15314 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15317 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15318 after setting changeable df flags.
15320 2017-08-03 Richard Biener <rguenther@suse.de>
15322 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15323 up if the use is in USE - X.
15325 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15327 * toplev.c (dumpfile.h): New include.
15328 (internal_error_reentered): New static function. Use it...
15329 (internal_error_function): ...here to handle reentered internal_error.
15331 2017-08-03 Richard Biener <rguenther@suse.de>
15333 PR middle-end/81148
15334 * fold-const.c (split_tree): Add minus_var and minus_con
15335 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15336 here but always use minus_*.
15337 (associate_trees): Assert we never associate with MINUS_EXPR
15338 and NULL first operand. Do not recurse for PLUS_EXPR operands
15339 when associating as MINUS_EXPR either.
15340 (fold_binary_loc): Track minus_var and minus_con.
15342 2017-08-03 Tom de Vries <tom@codesourcery.com>
15345 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15346 ACCEL_COMPILER, apply finish_options on
15347 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15349 2017-08-03 Tom de Vries <tom@codesourcery.com>
15352 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15353 function_entry_patch_area_size > 0.
15355 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15358 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15359 instead of 10??LU, perform unit multiplication in wide_int,
15360 don't change alloc_object_size_limit if the limit is larger
15363 PR tree-optimization/81655
15364 PR tree-optimization/81588
15365 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15366 the case when ranges[i].low and high are 1 for unsigned type with
15369 PR middle-end/81052
15370 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15371 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15373 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15375 * tree-vrp.h: Add include guard.
15377 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15380 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15381 (ud2): New insn pattern.
15382 * config/i386/i386.c (ix86_expand_epilogue):
15383 For naked functions, generate ud2 instead of trap insn.
15385 2017-08-02 Marek Polacek <polacek@redhat.com>
15388 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15390 2017-08-02 Tom de Vries <tom@codesourcery.com>
15391 Cesar Philippidis <cesar@codesourcery.com>
15393 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15394 Add missing edge probabilities.
15396 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15398 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15399 Correct endianness.
15401 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15403 PR middle-end/79499
15404 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15405 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15406 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15407 prologue_seq sequences - if any.
15409 2017-08-02 Richard Biener <rguenther@suse.de>
15411 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15412 via vectors if supported, integer extracts via punning if supported
15413 or otherwise vector extracts.
15415 2017-08-02 Richard Biener <rguenther@suse.de>
15417 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15419 (bitmap_insert_into_set): ... this.
15421 2017-08-02 Richard Biener <rguenther@suse.de>
15423 PR tree-optimization/81633
15425 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15427 PR tree-optimization/71752
15428 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15430 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15432 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15433 (machine_function::call_ms2sysv_pad_out): Remove field.
15434 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15435 (ix86_compute_frame_layout): Likewise.
15437 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15440 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15441 attribute with interrupt attribute.
15443 2017-08-01 Andrew Pinski <apinski@cavium.com>
15445 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15446 BIT_INSERT_EXPR's operand 1
15447 to see if the types precision matches.
15449 2017-08-01 Martin Liska <mliska@suse.cz>
15451 PR middle-end/70140
15452 * builtins.c (expand_builtin_memcpy_args): Remove.
15453 (expand_builtin_memcpy): Call newly added function
15454 expand_builtin_memory_copy_args.
15455 (expand_builtin_memcpy_with_bounds): Likewise.
15456 (expand_builtin_mempcpy): Remove last argument.
15457 (expand_builtin_mempcpy_with_bounds): Likewise.
15458 (expand_builtin_memory_copy_args): New function created from
15459 expand_builtin_mempcpy_args with small modifications.
15460 (expand_builtin_mempcpy_args): Remove.
15461 (expand_builtin_stpcpy): Remove unused argument.
15462 (expand_builtin): Likewise.
15463 (expand_builtin_with_bounds): Likewise.
15465 2017-08-01 Martin Liska <mliska@suse.cz>
15468 Make mempcpy more optimal (PR middle-end/70140).
15470 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15473 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15474 __builtin_vec_cmpne verify both arguments are compatible vectors
15475 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15476 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15477 move computation of aligned to after checking the argument types.
15481 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15482 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15485 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15486 Doug Rupp <rupp@adacore.com>
15487 Olivier Hainque <hainque@adacore.com>
15489 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15490 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15492 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15493 for TARGET_OS_CPP_BUILTIN.
15494 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15495 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15497 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15498 passing required abi options to the assembler for EABI configurations.
15499 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15500 of .text.hot and .text.unlikely sections for kernel modules when
15501 using ARM style exceptions.
15502 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15503 options. Add EXTRA_CC1_SPEC.
15504 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15506 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15508 (ARM_TARGET2_DWARF_FORMAT): Define.
15509 * config/arm/t-vxworks: Adjust multilib control to removal of the
15510 Diab command line options.
15512 2017-08-01 Martin Liska <mliska@suse.cz>
15514 PR gcov-profile/81561
15515 * gcov.c (unblock): Make unblocking safe as we need to preserve
15516 index correspondence of blocks and block_lists.
15518 2017-08-01 Richard Biener <rguenther@suse.de>
15520 PR tree-optimization/81181
15521 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15522 (compute_antic): ... end of iteration here.
15524 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15526 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15527 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15528 (ftree-slp-vectorize): Likewise.
15529 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15530 can no longer be set independent of flag_tree_loop_vectorize.
15531 * omp-general.c (emp_max_vf): Likewise.
15532 * opts.c (enable_fdo_optimizations): Remove references to
15533 flag_tree_vectorize, these are now implicit.
15534 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15535 and leave it for the options machinery.
15537 2017-08-01 Martin Liska <mliska@suse.cz>
15539 PR middle-end/70140
15540 * builtins.c (expand_builtin_memcpy_args): Remove.
15541 (expand_builtin_memcpy): Call newly added function
15542 expand_builtin_memory_copy_args.
15543 (expand_builtin_memcpy_with_bounds): Likewise.
15544 (expand_builtin_mempcpy): Remove last argument.
15545 (expand_builtin_mempcpy_with_bounds): Likewise.
15546 (expand_builtin_memory_copy_args): New function created from
15547 expand_builtin_mempcpy_args with small modifications.
15548 (expand_builtin_mempcpy_args): Remove.
15549 (expand_builtin_stpcpy): Remove unused argument.
15550 (expand_builtin): Likewise.
15551 (expand_builtin_with_bounds): Likewise.
15553 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15556 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15557 print "ds:" only for immediates in generic address space.
15559 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15562 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15563 (ix86_function_ok_for_sibcall): Return false for naked functions.
15565 2017-08-01 Richard Biener <rguenther@suse.de>
15567 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15568 (compute_antic): Seed worklist with exit block predecessors.
15569 * cfganal.c (dfs_find_deadend): For a cycle return the source
15570 of the edge closing it.
15572 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15574 * config/aarch64/aarch64.c
15575 (aarch64_can_const_movi_rtx_p): Move 0 check.
15577 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15579 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15580 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15582 * match.pd: Ditto in address comparison pattern.
15584 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15586 PR tree-optimization/81627
15587 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15588 closed ssa form for store-store chain.
15590 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15592 PR tree-optimization/81620
15593 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15594 for store-store chain.
15596 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15598 PR tree-optimization/81588
15599 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15600 ranges[i].in_p, invert comparison code ccode. For >/>=,
15601 swap rhs1 and rhs2 and comparison code unconditionally,
15602 for </<= don't do that. Don't swap rhs1/rhs2 again if
15603 ranges[i].in_p, instead invert comparison code ccode if
15604 opcode or oe->rank is BIT_IOR_EXPR.
15607 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15608 a direct optab to conversion optab.
15609 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15610 with GET_MODE_INNER as last argument instead of optab_handler.
15611 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15612 vector extraction if possible and optab is available.
15613 * expr.c (store_constructor): Use convert_optab_handler instead
15614 of optab_handler. Use vector initialization from smaller
15615 vectors if possible and optab is available.
15616 * tree-vect-stmts.c (vectorizable_load): Likewise.
15617 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15618 now have two modes.
15619 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15620 of vec_init from half-sized vectors with the same element mode.
15621 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15622 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15623 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15624 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15625 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15626 after mode in gen_vec_extract* calls.
15627 (vec_extract<mode>): Renamed to ...
15628 (vec_extract<mode><ssescalarmodelower>): ... this.
15629 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15630 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15631 element mode after mode in gen_vec_init* calls.
15632 (VEC_INIT_HALF_MODE): New mode iterator.
15633 (vec_init<mode>): Renamed to ...
15634 (vec_init<mode><ssescalarmodelower>): ... this.
15635 (vec_init<mode><ssehalfvecmodelower>): New expander.
15636 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15637 (vec_extractv2sfsf): ... this.
15638 (vec_initv2sf): Renamed to ...
15639 (vec_initv2sfsf): ... this.
15640 (vec_extractv2si): Renamed to ...
15641 (vec_extractv2sisi): ... this.
15642 (vec_initv2si): Renamed to ...
15643 (vec_initv2sisi): ... this.
15644 (vec_extractv4hi): Renamed to ...
15645 (vec_extractv4hihi): ... this.
15646 (vec_initv4hi): Renamed to ...
15647 (vec_initv4hihi): ... this.
15648 (vec_extractv8qi): Renamed to ...
15649 (vec_extractv8qiqi): ... this.
15650 (vec_initv8qi): Renamed to ...
15651 (vec_initv8qiqi): ... this.
15652 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15653 (vec_init<mode>): Renamed to ...
15654 (vec_init<mode><VEC_base_l>): ... this.
15655 (vec_extract<mode>): Renamed to ...
15656 (vec_extract<mode><VEC_base_l>): ... this.
15657 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15658 (vec_initv2sfsf): ... this.
15659 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15660 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15661 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15662 element mode after mode in gen_vec_init* calls.
15663 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15664 (vec_init<mode><Vel>): ... this.
15665 (vec_extract<mode>): Renamed to ...
15666 (vec_extract<mode><Vel>): ... this.
15667 * config/aarch64/iterators.md (Vel): New mode attribute.
15668 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15669 Add element mode after mode in gen_vec_extract* calls.
15670 * config/s390/vector.md (non_vec_l): New mode attribute.
15671 (vec_extract<mode>): Renamed to ...
15672 (vec_extract<mode><non_vec_l>): ... this.
15673 (vec_init<mode>): Renamed to ...
15674 (vec_init<mode><non_vec_l>): ... this.
15675 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15676 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15678 * config/arm/iterators.md (V_elem_l): New mode attribute.
15679 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15680 (vec_extract<mode><V_elem_l>): ... this.
15681 (vec_extractv2di): Renamed to ...
15682 (vec_extractv2didi): ... this.
15683 (vec_init<mode>): Renamed to ...
15684 (vec_init<mode><V_elem_l>): ... this.
15685 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15686 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15687 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15688 Add element mode after gen_vec_extract* calls.
15689 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15690 (vec_init<mode><unitmode>): ... this.
15691 (vec_extract<mode>): Renamed to ...
15692 (vec_extract<mode><unitmode>): ... this.
15693 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15694 (vec_init<mode><unitmode>): ... this.
15695 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15696 (vec_initv2sfsf): ... this.
15697 (vec_extractv2sf): Renamed to ...
15698 (vec_extractv2sfsf): ... this.
15699 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15700 Add element mode after gen_vec_extract* calls.
15701 * config/mips/mips.md (unitmode): New mode iterator.
15702 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15703 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15704 * config/spu/spu.md (inner_l): New mode attribute.
15705 (vec_init<mode>): Renamed to ...
15706 (vec_init<mode><inner_l>): ... this.
15707 (vec_extract<mode>): Renamed to ...
15708 (vec_extract<mode><inner_l>): ... this.
15709 * config/sparc/sparc.md (veltmode): New mode iterator.
15710 (vec_init<VMALL:mode>): Renamed to ...
15711 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15712 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15713 (vec_initv2sisi): ... this.
15714 (vec_initv2sf): Renamed to ...
15715 (vec_initv2sfsf): ... this.
15716 (vec_extractv2sf): Renamed to ...
15717 (vec_extractv2sfsf): ... this.
15718 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15719 (vec_init<mode>): Renamed to ...
15720 (vec_init<mode><VEC_base_l>): ... this.
15721 (vec_extract<mode>): Renamed to ...
15722 (vec_extract<mode><VEC_base_l>): ... this.
15723 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15724 (vec_initv2sfsf): ... this.
15725 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15726 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15727 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15728 gen_vec_init* calls.
15730 2017-08-01 Richard Biener <rguenther@suse.de>
15732 PR tree-optimization/81297
15733 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15734 TREE_OVERFLOW from INTEGER_CSTs.
15736 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15738 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15740 2017-07-31 Carl Love <cel@us.ibm.com>
15742 * config/rs6000/rs6000-c: Add support for built-in functions
15743 vector signed char vec_xl_be (signed long long, signed char *);
15744 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15745 vector signed int vec_xl_be (signed long long, signed int *);
15746 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15747 vector signed long long vec_xl_be (signed long long, signed long long *);
15748 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15749 vector signed short vec_xl_be (signed long long, signed short *);
15750 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15751 vector double vec_xl_be (signed long long, double *);
15752 vector float vec_xl_be (signed long long, float *);
15753 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15754 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15755 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15757 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15758 (altivec_expand_builtin): Add switch statement to call
15759 altivec_expand_xl_be for each builtin.
15760 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15761 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15762 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15763 __builtin_vsx_le_be_v16qi.
15764 * doc/extend.texi: Update the built-in documentation file for the
15765 new built-in functions.
15767 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15770 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15772 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15774 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15776 * config.gcc: Add z14.
15777 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15778 CPU model numbers for z13s and z14.
15779 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15781 * config/s390/s390-opts.h (enum processor_type): Rename
15782 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15783 * config/s390/s390.c (processor_table): Add field for CPU name to
15784 be passed to Binutils.
15785 (s390_asm_output_machine_for_arch): Use the new field in
15786 processor_table for Binutils.
15787 (s390_expand_builtin): Replace arch12 with z14.
15788 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15789 (s390_get_sched_attrmask): Likewise.
15790 (s390_get_unit_mask): Likewise.
15791 * config/s390/s390.opt: Add z14 to processor_type enum.
15793 2017-07-31 Martin Jambor <mjambor@suse.cz>
15796 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15797 regardless of optimization level.
15799 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15800 Martin Liska <mliska@suse.cz>
15802 * predict.def: Remove old comment and adjust probability.
15803 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15804 PREDICT statements.
15806 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15809 * config/i386/i386.c (ix86_function_naked): New function.
15810 (ix86_can_use_return_insn_p): Return false for naked functions.
15811 (ix86_expand_prologue): Skip prologue for naked functions.
15812 (ix86_expand_epilogue): Skip epilogue for naked functions
15813 and emit trap instruction.
15814 (ix86_warn_func_return): New function.
15815 (ix86_attribute_table): Add "naked" attribute specification.
15816 (TARGET_WARN_FUNC_RETURN): Define.
15817 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15819 2017-07-31 Martin Liska <mliska@suse.cz>
15821 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15822 (dump_gimple_bb_header): Always dump BB info.
15823 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15825 2017-07-31 Martin Liska <mliska@suse.cz>
15828 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15829 also with current_function_decl non-null equality.
15831 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15834 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15835 change type to the element type, instead add eltype variable and
15836 use it where we are interested in the element type.
15838 PR tree-optimization/81603
15839 * ipa-polymorphic-call.c
15840 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15841 offset arithmetic in offset_int, bail out if the resulting bit offset
15842 doesn't fit into shwi.
15844 2017-07-31 Martin Liska <mliska@suse.cz>
15846 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15847 (gimplify_save_expr): Fix comment.
15849 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15852 * config/i386/i386.c (ix86_function_arg): Update arguments for
15854 (ix86_compute_frame_layout): Set the initial stack offset to
15855 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15856 INCOMING_FRAME_SP_OFFSET.
15857 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15858 stack before exception handler returns.
15859 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15860 the 'ERROR_CODE' for exception handler.
15862 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15864 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15865 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15866 (ASM_OUTPUT_REG_POP): Ditto.
15867 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15868 instead of asm_fprintf to output pure string.
15870 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15872 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15873 to imported_module_or_decl hook.
15874 (debug_nothing_tree_tree_tree_bool): Remove.
15875 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15876 * debug.c (do_nothing_debug_hooks): Use
15877 debug_nothing_tree_tree_tree_bool_bool instead of
15878 debug_nothing_tree_tree_tree_bool.
15879 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15880 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15881 * sdbout.c (sdb_debug_hooks): Likewise.
15882 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15883 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15885 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15886 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15887 attribute, don't add anything.
15889 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15891 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15892 (fold_build2_stat_loc): Likewise.
15893 (fold_build3_stat_loc): Likewise.
15894 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15895 (fold_build1_loc): Remove macro.
15896 (fold_build2_loc): Likewise.
15897 (fold_build3_loc): Likewise.
15899 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15901 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15902 (gimple_build_debug_bind_source_stat): Likewise.
15903 * gimple.h (gimple_build_debug_bind): Remove macro.
15904 (gimple_build_debug_bind_source): Likewise.
15906 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15908 * bitmap.c (bitmap_alloc): Adjust.
15909 (bitmap_gc_alloc): Likewise.
15910 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15912 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15914 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15915 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15916 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15917 (bitmap_gc_alloc_stat): Likewise.
15918 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15920 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15922 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15923 * rtl.h (shallow_copy_rtx): Remove macro.
15925 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15927 * emit-rtl.c (gen_raw_REG): Adjust.
15928 * gengenrtl.c (gendef): Likewise.
15929 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15930 * rtl.h (rtx_alloc): Remove macro.
15932 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15934 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15935 (build_tree_list_stat): Likewise.
15936 * tree.h (build_tree_list): Remove macro.
15937 (build_tree_list_vec): Likewise.
15939 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15941 * tree.c (make_vector_stat): Remove _stat from name.
15942 (build_vector_stat): Likewise.
15943 * tree.h (make_vector_stat): Remove macro.
15944 (build_vector_stat): Likewise.
15946 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15948 * tree.h (build_var_debug_value): Remove prototype.
15950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15952 * tree.c (tree_cons_stat): Remove _stat from name.
15953 * tree.h (tree_cons): Remove macro.
15955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15957 * tree.c (build_vl_exp_stat): Remove _stat from name.
15958 * tree.h (build_vl_exp): Remove macro.
15960 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15962 * tree.c (build_decl_stat): Remove _stat from name.
15963 * tree.h (build_decl): Remove macro.
15965 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15967 * gimple.c (gimple_build_with_ops_stat): Adjust.
15968 (gimple_alloc_stat): Remove _stat from name.
15969 * gimple.h (gimple_alloc): Remove macro.
15971 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15973 * tree.c (make_tree_vec_stat): Remove _stat from name.
15974 (grow_tree_vec_stat): Likewise.
15975 * tree.h (make_tree_vec_stat): Adjust prototype.
15976 (grow_tree_vec_stat): Likewise.
15977 (make_tree_vec): Remove macro.
15978 (grow_tree_vec): Likewise.
15980 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15982 * fold-const.c (fold_build1_stat_loc): Adjust.
15983 (fold_build2_stat_loc): Likewise.
15984 (fold_build3_stat_loc): Likewise.
15985 * tree.c (build0_stat): Remove _stat from name.
15986 (build1_stat): Likewise.
15987 (build2_stat): Likewise.
15988 (build3_stat): Likewise.
15989 (build4_stat): Likewise.
15990 (build5_stat): Likewise.
15991 * tree.h (build1_loc): Remove macro, and rename _stat function
15993 (build2_loc): Likewise.
15994 (build3_loc): Likewise.
15995 (build4_loc): Likewise.
15996 (build5_loc): Likewise.
15998 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16000 * tree.c (make_int_cst_stat): Remove _stat from name.
16001 * tree.h (make_int_cst_stat): Adjust prototype.
16002 (make_int_cst): Remove macro.
16004 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16006 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16007 * tree.h (make_tree_binfo_stat): Adjust prototype.
16008 (make_tree_binfo): Remove.
16010 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16012 * tree.c (copy_node_stat): Rename to copy_node.
16013 (build_distinct_type_copy): Adjust.
16014 * tree.h (copy_node_stat): Adjust prototype.
16015 (copy_node): Remove macro.
16017 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16019 * tree.c (make_node_stat): rename to make_node.
16020 (build_tree_list_stat): Adjust.
16021 (build0_stat): Likewise.
16022 (build2_stat): Likewise.
16023 (build3_stat): Likewise.
16024 (build4_stat): Likewise.
16025 (build5_stat): Likewise.
16026 (build_decl_stat): Likewise.
16027 * tree.h (make_node_stat): Adjust prototype.
16028 (make_node): remove macro.
16030 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16032 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16033 (PPC_FEATURE2_SCV): Likewise.
16034 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16036 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16038 * config/aarch64/aarch64.c
16039 (aarch64_internal_mov_immediate): Add new special pattern.
16040 * config/aarch64/aarch64.md (*movdi_aarch64):
16041 Add reg/32bit const mov case.
16043 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16044 Richard Sandiford <richard.sandiford@linaro.org>
16046 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16047 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16048 Add integer and movi cases.
16049 (movi-split-hf-df-sf split, fp16): New.
16050 (enabled): Added TARGET_FP_F16INST.
16051 * config/aarch64/iterators.md (GPF_HF): New.
16052 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16054 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16056 * config/aarch64/aarch64.c
16057 (aarch64_simd_container_mode): Add prototype.
16058 (aarch64_expand_mov_immediate): Add HI support.
16059 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16060 (aarch64_can_const_movi_rtx_p): New.
16061 (aarch64_preferred_reload_class):
16062 Remove restrictions of using FP registers for certain SIMD operations.
16063 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16064 (aarch64_valid_floating_const): Add integer move validation.
16065 (aarch64_simd_imm_scalar_p): Remove.
16066 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16067 (aarch64_legitimate_constant_p): Expand list of supported cases.
16068 * config/aarch64/aarch64-protos.h
16069 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16070 (aarch64_reinterpret_float_as_int): New.
16071 (aarch64_simd_imm_scalar_p): Remove.
16072 * config/aarch64/constraints.md (Uvi): New.
16073 (Dd): Split into Ds and new Dd.
16074 * config/aarch64/aarch64.md (*movsi_aarch64):
16076 (*movdi_aarch64): Add SIMD mov case.
16078 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16080 * tree-predcom.c: (struct chain): Handle store-store chain in which
16081 stores for elimination only store loop invariant values.
16082 (execute_pred_commoning_chain): Ditto.
16083 (prepare_initializers_chain_store_elim): Ditto.
16084 (prepare_finalizers): Ditto.
16085 (is_inv_store_elimination_chain): New function.
16086 (initialize_root_vars_store_elim_1): New function.
16088 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16090 * tree-predcom.c: Revise general description of the pass.
16091 (enum chain_type): New enum type for store elimination.
16092 (struct chain): New field supporting store elimination.
16093 (struct component): Ditto.
16094 (dump_chain): Dump store-stores chain.
16095 (release_chain): Release resources.
16096 (split_data_refs_to_components): Compute and create component
16097 contains only stores for elimination.
16098 (get_chain_last_ref_at): New function.
16099 (make_invariant_chain): Initialization.
16100 (make_rooted_chain): Specify chain type in parameter and record it.
16101 (add_looparound_copies): Skip for store-stores chain.
16102 (determine_roots_comp): Compute type of chain and pass it to
16104 (initialize_root_vars_store_elim_2): New function.
16105 (finalize_eliminated_stores): New function.
16106 (remove_stmt): Handle store for elimination.
16107 (execute_pred_commoning_chain): Execute predictive commoning on
16108 store-store chains.
16109 (determine_unroll_factor): Skip unroll for store-stores chain.
16110 (prepare_initializers_chain_store_elim): New function.
16111 (prepare_initializers_chain): Hanlde store-store chain.
16112 (prepare_finalizers_chain, prepare_finalizers): New function.
16113 (tree_predictive_commoning_loop): Return integer value indicating
16114 if loop is unrolled or lcssa form is corrupted.
16115 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16117 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16119 * tree-predcom.c (initialize_root): Delete.
16120 (execute_pred_commoning_chain): Initialize root vars and replace
16121 reference of non-combined chain directly, rather than call above
16124 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16126 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16127 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16129 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16131 * tree-predcom.c (struct chain): New field init_seq.
16132 (release_chain): Release init_seq.
16133 (prepare_initializers_chain): Record intialization stmts in above
16135 (insert_init_seqs): New function.
16136 (tree_predictive_commoning_loop): Call insert_init_seqs.
16138 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16140 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16142 2017-07-28 Richard Biener <rguenther@suse.de>
16144 * match.pd: Remove superfluous :c.
16145 * genmatch.c (simplify::id): Add member.
16146 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16148 (current_id): New global.
16149 (dt_node::parent): Move from ...
16150 (dt_operand::parent): ... here. Add for_id member.
16151 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16152 (decision_tree::find_node): Relax order requirement when
16153 merging DT_TRUE nodes to ones inbetween the current simplify
16154 and the one we try to merge with. Add diagnostic whenever
16155 we need to enforce pattern order by not merging.
16156 (decision_tree::insert): Set current_id.
16157 (decision_tree::print_node): Dump parent node and for_id.
16158 (parser::last_id): Add member.
16159 (parser::push_simplify): Assign unique id.
16160 (parser::parser): Initialize last_id.
16162 2017-07-28 Martin Liska <mliska@suse.cz>
16165 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16166 gimple_build_debug_bind.
16168 2017-07-28 Richard Biener <rguenther@suse.de>
16170 PR tree-optimization/81502
16171 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16173 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16175 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16176 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16177 for BIT_FIELD_REF args.
16178 * fold-const.c (make_bit_field_ref): Likewise.
16179 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16181 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16184 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16185 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16186 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16187 Or it into SANITIZER_UNDEFINED.
16188 * ubsan.c: Include gimple-fold.h and varasm.h.
16189 (ubsan_expand_ptr_ifn): New function.
16190 (instrument_pointer_overflow): New function.
16191 (maybe_instrument_pointer_overflow): New function.
16192 (instrument_object_size): Formatting fix.
16193 (pass_ubsan::execute): Call instrument_pointer_overflow
16194 and maybe_instrument_pointer_overflow.
16195 * internal-fn.c (expand_UBSAN_PTR): New function.
16196 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16197 * sanitizer.def (__ubsan_handle_pointer_overflow,
16198 __ubsan_handle_pointer_overflow_abort): New builtins.
16199 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16200 * internal-fn.def (UBSAN_PTR): New internal function.
16201 * opts.c (sanitizer_opts): Add pointer-overflow.
16202 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16203 * fold-const.c (build_range_check): Compute pointer range check in
16204 integral type if pointer arithmetics would be needed. Formatting
16207 2017-07-28 Martin Liska <mliska@suse.cz>
16210 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16211 parameters that are of a variable-length.
16213 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16215 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16217 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16218 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16219 (CRT_CALL_STATIC_FUNCTION): Likewise.
16220 (ASM_DEFAULT_SPEC): New define.
16221 (ASM_SPEC32): Likewise.
16222 (ASM_SPEC64): Likewise.
16223 (ASM_SPEC_COMMON): Likewise.
16224 (ASM_SPEC): Likewise.
16225 (INVALID_64BIT): Likewise.
16226 (LINK_OS_DEFAULT_SPEC): Likewise.
16227 (LINK_OS_SPEC32): Likewise.
16228 (LINK_OS_SPEC64): Likewise.
16229 (POWERPC_LINUX): Likewise.
16230 (PTRDIFF_TYPE): Likewise.
16231 (RESTORE_FP_PREFIX): Likewise.
16232 (RESTORE_FP_SUFFIX): Likewise.
16233 (SAVE_FP_PREFIX): Likewise.
16234 (SAVE_FP_SUFFIX): Likewise.
16235 (SIZE_TYPE): Likewise.
16236 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16237 (TARGET_64BIT): Likewise.
16238 (TARGET_64BIT): Likewise.
16239 (TARGET_AIX): Likewise.
16240 (WCHAR_TYPE_SIZE): Likewise.
16241 (WCHAR_TYPE): Undefine.
16242 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16243 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16244 (CPP_OS_RTEMS_SPEC): Delete.
16245 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16246 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16248 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16250 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16252 PR tree-optimization/81578
16253 * tree-parloops.c (build_new_reduction): Bail out if
16254 reduction_code isn't one of the standard OpenMP reductions.
16255 Move the details printing after that decision.
16257 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16259 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16260 related to reload_in_progress.
16261 (splat_input_operand): Likewise.
16262 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16264 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16266 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16267 (TARGET_INSTANTIATE_DECLS): Likewise.
16268 (legitimate_indexed_address_p): Delete reload_in_progress code.
16269 (rs6000_debug_legitimate_address_p): Likewise.
16270 (rs6000_eliminate_indexed_memrefs): Likewise.
16271 (rs6000_emit_le_vsx_store): Likewise.
16272 (rs6000_emit_move_si_sf_subreg): Likewise.
16273 (rs6000_emit_move): Likewise.
16274 (register_to_reg_type): Likewise.
16275 (rs6000_pre_atomic_barrier): Likewise.
16276 (rs6000_machopic_legitimize_pic_address): Likewise.
16277 (rs6000_allocate_stack_temp): Likewise.
16278 (rs6000_address_for_fpconvert): Likewise.
16279 (rs6000_address_for_altivec): Likewise.
16280 (rs6000_secondary_memory_needed_rtx): Delete function.
16281 (rs6000_check_sdmode): Likewise.
16282 (rs6000_alloc_sdmode_stack_slot): Likewise.
16283 (rs6000_instantiate_decls): Likewise.
16284 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16285 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16286 Delete reload_in_progress.
16287 (*vec_reload_and_plus_<mptrsize>): Likewise.
16288 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16289 (vsx_div_v2di): Likewise.
16290 (vsx_udiv_v2di): Likewise.
16292 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16294 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16295 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16296 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16297 (rs6000_debug_reg_global): Delete print of LRA status.
16298 (rs6000_option_override_internal): Delete dead LRA related code.
16299 (rs6000_lra_p): Delete function.
16300 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16302 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16304 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16305 * config/riscv/rtems.h: New file.
16307 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16308 Sudakshina Das <sudi.das@arm.com>
16310 * config/aarch64/aarch64.md
16311 (define_split for and<mode>3nr_compare): Move
16312 non aarch64_logical_operand to a register.
16313 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16314 register immediate operand to a register.
16315 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16317 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16319 PR middle-end/81564
16320 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16322 2017-07-27 Richard Biener <rguenther@suse.de>
16324 PR tree-optimization/81573
16325 PR tree-optimization/81494
16326 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16327 multi defuse cycle case.
16329 2017-07-27 Richard Biener <rguenther@suse.de>
16331 PR tree-optimization/81571
16332 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16335 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16337 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16338 earlier and only if MASK_FPU is set. Adjust formatting.
16340 2017-07-27 Martin Liska <mliska@suse.cz>
16342 * opt-functions.awk: Add validation of value of Init.
16343 * optc-gen.awk: Pass new argument.
16345 2017-07-27 Martin Liska <mliska@suse.cz>
16347 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16348 Fix wrong condition.
16350 2017-07-27 Martin Liska <mliska@suse.cz>
16352 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16353 BBs and edges seen by autoFDO.
16355 2017-07-27 Richard Biener <rguenther@suse.de>
16357 PR tree-optimization/81502
16358 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16359 with incompatible but same sized type.
16360 (execute_update_addresses_taken): Likewise.
16362 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16364 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16365 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16367 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16370 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16371 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16372 Change s_operand to memory_operand.
16374 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16376 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16377 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16378 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16379 Emit instructions rather than returning an expression. Handle TFmode
16380 and KFmode by casting to TImode.
16381 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16382 (rs6000_emit_le_vsx_store): Likewise.
16383 * config/rs6000/vsx.md (VSX_TI): New iterator.
16384 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16385 (*vsx_le_undo_permute_<mode>): Likewise.
16386 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16387 emit the split sequence.
16388 (*vsx_le_perm_store_<mode>): Likewise.
16390 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16392 PR tree-optimization/81555
16393 PR tree-optimization/81556
16394 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16395 if true, force CHANGED for the recursive invocation.
16396 (reassociate_bb): Remember original length of ops array, pass
16397 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16399 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16400 attributes for noipa attribute. For naked attribute use
16401 lookup_attribute first before lookup_attribute_spec.
16402 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16404 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16405 for functions with noipa attribute.
16406 (cgraph_externally_visible_p): Return true for functions with noipa
16408 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16409 for functions with noipa attribute.
16410 * doc/extend.texi: Document noipa function attribute.
16411 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16412 also for functions with noipa attribute.
16413 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16415 2017-07-26 Andrew Pinski <apinski@cavium.com>
16417 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16418 vec_unalign_load_cost and vec_unalign_store_cost.
16420 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16422 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16423 -mvsx-small-integer option.
16424 (ISA_3_0_MASKS_IEEE): Likewise.
16425 (OTHER_VSX_VECTOR_MASKS): Likewise.
16426 (POWERPC_MASKS): Likewise.
16427 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16428 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16429 code, only testing for DImode being allowed in non-VSX floating
16431 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16432 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16434 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16435 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16436 TARGET_P8_VECTOR test.
16437 (rs6000_secondary_reload_simple_move): Likewise.
16438 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16439 since TARGET_P9_VECTOR was already tested.
16440 (rs6000_opt_masks): Remove -mvsx-small-integer.
16441 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16442 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16444 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16445 test for TARGET_VEXTRACTUB was used, and that uses
16447 (p9 extract splitter): Likewise.
16448 (vsx_extract_<mode>_di_p9): Likewise.
16449 (vsx_extract_<mode>_store_p9): Likewise.
16450 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16451 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16452 the elimination of TARGET_VSX_SMALL_INTEGER.
16453 (vsx_extract_<mode>_p8): Likewise.
16454 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16455 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16456 (vsx_set_<mode>_p9): Likewise.
16457 (vsx_set_v4sf_p9): Likewise.
16458 (vsx_set_v4sf_p9_zero): Likewise.
16459 (vsx_insert_extract_v4sf_p9): Likewise.
16460 (vsx_insert_extract_v4sf_p9_2): Likewise.
16461 * config/rs6000/rs6000.md (sign extend splitter): Change
16462 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16463 (floatsi<mode>2_lfiwax_mem): Likewise.
16464 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16465 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16466 since a test for TARGET_P9_VECTOR was used.
16467 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16468 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16469 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16470 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16471 TARGET_P8_VECTOR test.
16472 (fix_trunc<mode>si2_stfiwx): Likewise.
16473 (fix_trunc<mode>si2_internal): Likewise.
16474 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16475 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16477 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16478 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16479 TARGET_P8_VECTOR test.
16480 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16481 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16482 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16484 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16485 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16486 since a test for TARGET_P9_VECTOR was used.
16487 (splitter for loading small constants): Likewise.
16489 2017-07-26 Andrew Pinski <apinski@cavium.com>
16491 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16494 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16497 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16498 (fp_valid_at): Likewise.
16500 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16502 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16503 (qdf24xx_addrcost_table): Likewise.
16504 (cortexa57_tunings): Update to use generic_branch_cost.
16505 (cortexa72_tunings): Likewise.
16506 (cortexa73_tunings): Likewise.
16507 (qdf24xx_tunings): Likewise.
16509 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16511 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16512 (thunderx2t99_branch_cost): Likewise.
16513 (cortexa35_tunings): Update to use generic_branch_cost.
16514 (cortexa53_tunings): Likewise.
16515 (cortexa57_tunings): Likewise.
16516 (cortexa72_tunings): Likewise.
16517 (cortexa73_tunings): Likewise.
16518 (thunderx2t99_tunings): Likewise.
16520 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16522 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16523 (sparc_option_override): Honour MASK_FSMULD.
16524 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16525 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16526 * config/sparc/sparc.opt (mfsmuld): New option.
16527 * doc/invoke.texi (mfsmuld): Document option.
16529 2017-07-26 Marek Polacek <polacek@redhat.com>
16531 PR middle-end/70992
16532 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16534 2017-07-26 Richard Biener <rguenther@suse.de>
16536 * gimple-match-head.c (do_valueize): Return OP if valueize
16538 (get_def): New helper to get at the def stmt of a SSA name
16539 if valueize allows.
16540 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16541 do_valueize to get at the def stmt.
16542 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16544 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16546 PR middle-end/46932
16547 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16549 2017-07-26 Martin Liska <mliska@suse.cz>
16552 * function.c (expand_function_start): Make expansion of
16553 nonlocal_goto_save_area after parm_birth_insn.
16555 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16557 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16558 from all CPU target flags enable members.
16560 2017-07-26 Richard Biener <rguenther@suse.de>
16562 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16563 (decision_tree::gen): Make 'type' const.
16564 (write_predicate): Likewise.
16566 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16568 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16569 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16570 (rs6000_option_override_internal): Likewise.
16571 (rs6000_expand_vector_set): Likewise.
16572 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16573 (TARGET_UPPER_REGS_SF): Likewise.
16574 (TARGET_UPPER_REGS_DI): Likewise.
16575 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16576 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16577 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16578 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16579 (Splitters for DI constants in Altivec registers): Likewise.
16580 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16581 (vsx_set_v4sf_p9): Likewise.
16582 (vsx_set_v4sf_p9_zero): Likewise.
16583 (vsx_insert_extract_v4sf_p9): Likewise.
16584 (vsx_insert_extract_v4sf_p9_2): Likewise.
16586 2017-07-25 Carl Love <cel@us.ibm.com>
16588 * doc/extend.texi: Update the built-in documentation file for the
16589 existing built-in functions
16590 vector signed char vec_cnttz (vector signed char);
16591 vector unsigned char vec_cnttz (vector unsigned char);
16592 vector signed short vec_cnttz (vector signed short);
16593 vector unsigned short vec_cnttz (vector unsigned short);
16594 vector signed int vec_cnttz (vector signed int);
16595 vector unsigned int vec_cnttz (vector unsigned int);
16596 vector signed long long vec_cnttz (vector signed long long);
16597 vector unsigned long long vec_cnttz (vector unsigned long long);
16599 2017-07-25 Andrew Pinski <apinski@cavium.com>
16601 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16602 accesses where the use is for the first operand of a BIT_INSERT.
16604 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16607 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16608 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16610 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16612 * config/i386/gstabs.h: Delete.
16613 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16615 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16617 * config/i386/i386.c (ix86_decompose_address): Do not check for
16618 register RTX when looking at index_reg or base_reg.
16619 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16621 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16623 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16624 to update EH info here.
16626 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16628 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16630 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16632 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16634 2017-07-25 Torsten Duwe <duwe@suse.de>
16636 * common.opt: Introduce -fpatchable-function-entry
16637 command line option, and its variables function_entry_patch_area_size
16638 and function_entry_patch_area_start.
16639 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16640 including a two-value parser.
16641 * target.def (print_patchable_function_entry): New target hook.
16642 * targhooks.h (default_print_patchable_function_entry): New function.
16643 * targhooks.c (default_print_patchable_function_entry): Likewise.
16644 * toplev.c (process_options): Switch off IPA-RA if
16645 patchable function entries are being generated.
16646 * varasm.c (assemble_start_function): Look at the
16647 patchable-function-entry command line switch and current
16648 function attributes and maybe generate NOP instructions by
16649 calling the print_patchable_function_entry hook.
16650 * doc/extend.texi: Document patchable_function_entry attribute.
16651 * doc/invoke.texi: Document -fpatchable_function_entry
16652 command line option.
16653 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16655 * doc/tm.texi: Re-generate.
16657 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16660 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16661 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16663 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16665 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16667 2017-07-25 Richard Biener <rguenther@suse.de>
16669 PR tree-optimization/81455
16670 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16671 not walk in cycles when looking for guards.
16673 2017-07-25 Richard Biener <rguenther@suse.de>
16675 PR tree-optimization/81529
16676 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16677 when optimizing backedge uses.
16679 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16681 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16683 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16684 to dl_section_ref. On AIX, append an expression to subtract
16685 the size of the section length to dl_section_ref.
16687 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16689 * configure.ac: If any of the config.* scripts fail, exit 1.
16690 * configure: Regenerate.
16692 2017-07-25 Richard Biener <rguenther@suse.de>
16694 PR middle-end/81546
16695 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16696 of immediate uses, be more verbose on errors.
16698 2017-07-25 Richard Biener <rguenther@suse.de>
16700 PR tree-optimization/81510
16701 * tree-vect-loop.c (vect_is_simple_reduction): When the
16702 reduction stmt is not inside the loop bail out.
16704 2017-07-25 Richard Biener <rguenther@suse.de>
16706 PR tree-optimization/81303
16707 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16708 profitability check against LOOP_VINFO_NITERSM1.
16710 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16712 * domwalk.c (cmp_bb_postorder): Simplify.
16713 (sort_bbs_postorder): New function. Use it...
16714 (dom_walker::walk): ...here to optimize common cases.
16716 2017-07-25 Martin Liska <mliska@suse.cz>
16719 * ipa-visibility.c (function_and_variable_visibility): Make the
16720 redirection just on target that supports aliasing.
16721 Fix GNU coding style.
16723 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16726 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16728 * config/arm/rtems.h: Likewise.
16729 * config/bfin/rtems.h: Likewise.
16730 * config/i386/rtemself.h: Likewise.
16731 * config/lm32/rtems.h: Likewise.
16732 * config/m32c/rtems.h: Likewise.
16733 * config/m68k/rtemself.h: Likewise.
16734 * config/microblaze/rtems.h: Likewise.
16735 * config/mips/rtems.h: Likewise.
16736 * config/moxie/rtems.h: Likewise.
16737 * config/nios2/rtems.h: Likewise.
16738 * config/powerpcspe/rtems.h: Likewise.
16739 * config/rs6000/rtems.h: Likewise.
16740 * config/rtems.h: Likewise.
16741 * config/sh/rtems.h: Likewise.
16742 * config/sh/rtemself.h: Likewise.
16743 * config/sparc/rtemself.h: Likewise.
16745 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16748 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16749 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16750 * tree-ssa-structalias.c (alias_get_name): Same.
16752 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16755 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16756 instructions if no du chain is found.
16758 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16760 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16762 2017-07-25 Richard Biener <rguenther@suse.de>
16764 PR middle-end/81505
16765 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16768 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16770 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16771 upper-regs options.
16772 (ISA_2_7_MASKS_SERVER): Likewise.
16773 (ISA_3_0_MASKS_IEEE): Likewise.
16774 (OTHER_P8_VECTOR_MASKS): Likewise.
16775 (OTHER_VSX_VECTOR_MASKS): Likewise.
16776 (POWERPC_MASKS): Likewise.
16777 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16778 duplicate list of options.
16779 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16780 explicit -mupper-regs options.
16781 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16782 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16783 alias for -mupper-regs-df.
16784 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16785 (rs6000_init_hard_regno_mode_ok): Likewise.
16786 (rs6000_option_override_internal): Likewise.
16787 (rs6000_opt_masks): Likewise.
16788 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16789 options in terms of whether -mvsx or -mpower8-vector was used.
16790 (TARGET_UPPER_REGS_DI): Likewise.
16791 (TARGET_UPPER_REGS_SF): Likewise.
16792 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16793 -mupper-regs-* options.
16795 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16797 * passes.c (emergency_dump_function): Print some empty lines and a
16798 header before the RTL dump.
16800 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16802 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16804 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16807 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16808 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16810 2017-07-24 Carl Love <cel@us.ibm.com>
16812 * config/rs6000/rs6000-c.c: Add support for built-in functions
16813 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16814 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16815 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16816 vec_extract_fp_from_shortl): Add defines for the two builtins.
16817 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16818 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16820 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16821 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16822 * doc/extend.texi: Update the built-in documentation file for the
16823 new built-in function.
16825 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16828 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16830 * doc/generic.texi: Likewise.
16831 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16832 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16834 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16836 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16837 (aarch64_mls_elt_merge<mode>): Likewise.
16839 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16841 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16842 having __cxa_atexit.
16844 2017-07-23 Michael Collison <michael.collison@arm.com>
16846 * config/arm/arm.c (arm_option_override): Deprecate
16847 use of -mstructure-size-boundary.
16848 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16849 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16851 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16854 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16855 Reduce cost estimate for direct moves.
16857 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16860 * config/i386/i386.c (ix86_option_override_internal): Disable
16861 BMI, BMI2 and TBM instructions for -m16.
16863 2017-07-21 Carl Love <cel@us.ibm.com>
16865 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16866 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16867 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16868 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16869 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16870 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16871 VMULOSW): New enum "unspec" values.
16872 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16873 altivec_vmulosw): New patterns.
16874 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16875 VMULOSW): Add definitions.
16877 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16879 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16880 (qdf24xx): Likewise.
16881 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16882 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16883 (AARCH64_FL_V8_1): Renumber.
16884 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16885 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16886 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16887 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16888 rdma to feature modifiers list.
16890 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16892 PR middle-end/56727
16893 * ipa-visibility (function_and_variable_visibility): Convert
16894 recursive PLT call to direct call if appropriate.
16896 2017-07-21 Andrew Pinski <apinski@cavium.com>
16898 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16899 operand 1 to see if the types precision matches.
16900 * fold-const.c (operand_equal_p): Likewise.
16902 2017-07-21 Richard Biener <rguenther@suse.de>
16904 PR tree-optimization/81303
16905 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16906 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16907 (vect_peeling_hash_get_lowest_cost): Adjust.
16908 (vect_enhance_data_refs_alignment): Likewise. Use
16909 vect_get_peeling_costs_all_drs to compute the penalty for no
16910 peeling to match up costs.
16912 2017-07-21 Richard Biener <rguenther@suse.de>
16914 PR tree-optimization/81500
16915 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16916 we didn't identify a reduction path.
16918 2017-07-21 Tom de Vries <tom@codesourcery.com>
16919 Cesar Philippidis <cesar@codesourcery.com>
16921 PR gcov-profile/81442
16922 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16925 2017-07-21 Tom de Vries <tom@codesourcery.com>
16928 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16930 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16931 nvptx_override_options_after_change.
16933 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16935 * dwarf2out.c (output_file_names): Avoid double testing for
16936 dwarf_version >= 5.
16938 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16940 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16942 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16944 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16946 (try_crossjump_to_edge): Do not punt on partitioned functions.
16948 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16950 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16951 Put all BBs reachable only via paths crossing cold region to cold
16953 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16955 2016-07-21 Richard Biener <rguenther@suse.de>
16957 PR tree-optimization/81303
16958 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16959 into account prologue and epilogue iterations when raising
16960 min_profitable_iters to sth at least covering one vector iteration.
16962 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16964 * config/arm/arm.c (arm_test_cpu_arch_dat):
16967 2017-07-20 Nathan Sidwell <nathan@acm.org>
16969 Remove TYPE_METHODS.
16970 * tree.h (TYPE_METHODS): Delete.
16971 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16972 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16973 (dbxout_type_methods): Scan TYPE_FIELDS.
16974 (dbxout_type): Don't check TYPE_METHODS here.
16975 * function.c (use_register_for_decl): Always ignore register for
16976 class types when not optimizing.
16977 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16978 * tree.c (free_lang_data_in_type): Stitch out member functions and
16979 templates from TYPE_FIELDS.
16980 (build_distinct_type_copy, verify_type_variant,
16981 verify_type): Member fns are on TYPE_FIELDS.
16982 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16983 * tree-pretty-print.c (dump_generic_node): Likewise.
16985 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16988 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16989 V2TImode and V4TImode.
16990 (ix86_expand_vector_extract): Likewise.
16991 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16992 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16993 (ssescalarmode): Handle V4TImode and V2TImode.
16994 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16995 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16996 (VEXTRACTI128_MODE): New mode iterator.
16997 (splitter for *vec_extractv?ti first element): New.
16998 (VEC_INIT_MODE): New mode iterator.
16999 (vec_init<mode>): Consolidate 3 expanders into one using
17000 VEC_INIT_MODE mode iterator.
17002 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17004 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17005 non_spilled_static_chain_regno_p.
17007 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17009 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17011 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17013 * bb-reorder.c (connect_traces): Allow copying of blocks within
17016 2017-07-20 Richard Biener <rguenther@suse.de>
17018 * gimple.h (gimple_phi_result): Add gphi * overload.
17019 (gimple_phi_result_ptr): Likewise.
17020 (gimple_phi_arg): Likewise. Adjust index assert to only
17021 allow actual argument accesses rather than all slots available
17023 (gimple_phi_arg_def): Add gphi * overload.
17024 * tree-phinodes.c (make_phi_node): Initialize only actual
17026 (resize_phi_node): Clear memory not covered by old node,
17027 do not initialize excess argument slots.
17028 (reserve_phi_args_for_new_edge): Initialize new argument slot
17031 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17033 PR tree-optimization/81388
17035 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17037 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17038 by removing computation of may_be_zero.
17040 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17041 Tom de Vries <tom@codesourcery.com>
17043 PR middle-end/81030
17044 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17045 when gimple level profile disagrees with what RTL expander did.
17047 2017-07-20 Richard Biener <rguenther@suse.de>
17049 PR tree-optimization/61171
17050 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17051 (vect_analyze_stmt): Add slp instance parameter.
17052 (vectorizable_reduction): Likewise.
17053 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17054 (vect_is_simple_reduction): Deal with chains not detected
17055 as SLP reduction chain, specifically not properly associated
17056 chains containing a mix of plus/minus.
17057 (get_reduction_op): Remove.
17058 (get_initial_defs_for_reduction): Simplify, pass in whether
17059 this is a reduction chain, pass in the SLP node for the PHIs.
17060 (vect_create_epilog_for_reduction): Get the SLP instance as
17062 (vectorizable_reduction): Get the SLP instance as arg.
17063 During analysis remember the SLP node with the PHIs in the
17064 instance. Simplify getting at the vectorized reduction PHIs.
17065 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17066 through SLP instance.
17067 (vect_slp_analyze_operations): Likewise.
17068 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17069 (vect_transform_stmt): Likewise.
17071 2017-07-20 Tom de Vries <tom@codesourcery.com>
17073 PR tree-optimization/81489
17074 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17075 read of phi arg location to before loop that modifies phi.
17077 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17079 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17082 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17084 PR middle-end/81331
17085 * except.c (execute): Fix ordering issue.
17087 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17089 PR rtl-optimization/81423
17090 * combine.c (make_compound_operation_int): Don't try to optimize
17091 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17093 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17095 PR rtl-optimization/81423
17096 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17097 with a constant that is -1 in the truncated to mode.
17099 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17101 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17102 (determine_unlikely_bbs): ... here.
17103 * predict.h (propagate_unlikely_bbs_forward): Declare.
17104 * cfgexpand.c (pass_expand::execute): Use it.
17105 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17107 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17108 propagate_unlikely_bbs_forward.
17110 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17112 PR middle-end/81331
17113 * except.c (maybe_add_nop_after_section_switch): New function.
17116 2017-07-19 Tom de Vries <tom@codesourcery.com>
17118 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17120 2017-07-19 Tom de Vries <tom@codesourcery.com>
17122 * gimple.h (gimple_phi_arg): Make assert more strict.
17124 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17126 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17127 * config/rs6000/mmintrin.h: New file.
17128 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17130 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17132 PR tree-optimization/81346
17133 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17135 2017-07-19 Tom de Vries <tom@codesourcery.com>
17137 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17139 2017-07-19 Tom de Vries <tom@codesourcery.com>
17141 * config/nvptx/nvptx-modes.def: Add V2DImode.
17142 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17143 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17144 (nvptx_output_mov_insn): Handle lack of mov.b128.
17145 (nvptx_print_operand): Handle 'H' and 'L' codes.
17146 (nvptx_vector_mode_supported): Allow V2DImode.
17147 (nvptx_preferred_simd_mode): New function.
17148 (nvptx_data_alignment): New function.
17149 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17150 nvptx_preferred_simd_mode.
17151 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17153 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17155 2017-07-19 Tom de Vries <tom@codesourcery.com>
17157 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17158 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17159 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17160 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17161 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17162 (mov<VECIM>_insn): New define_insn.
17163 (define_expand "mov<VECIM>): New define_expand.
17165 2017-07-19 Tom de Vries <tom@codesourcery.com>
17167 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17169 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17171 PR tree-optimization/81346
17172 * fold-const.h (fold_div_compare, range_check_type): Declare.
17173 * fold-const.c (range_check_type): New function.
17174 (build_range_check): Use range_check_type.
17175 (fold_div_compare): No longer static, rewritten into
17176 a match.pd helper function.
17177 (fold_comparison): Don't call fold_div_compare here.
17178 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17179 as helper function.
17181 2017-07-19 Nathan Sidwell <nathan@acm.org>
17183 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17184 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17185 * tree.c (find_decls_types_r, verify_type): Use
17186 TYPE_{MIN,MAX}_VALUE_RAW.
17187 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17188 (hash_tree): Likewise.
17189 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17191 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17194 2017-07-18 Tom de Vries <tom@codesourcery.com>
17196 PR middle-end/81464
17197 * omp-expand.c (expand_omp_for_static_chunk): Handle
17198 equal-argument loop exit phi.
17200 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17203 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17204 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17205 operand 2 predicate.
17206 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17207 operand 2 predicate.
17208 (ror,rol -> rorx splitters): Use const_int_operand as
17209 operand 2 predicate.
17211 2017-06-18 Richard Biener <rguenther@suse.de>
17213 PR tree-optimization/81410
17214 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17215 the gap in the ! slp_perm SLP case after each group.
17217 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17219 PR middle-end/81463
17220 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17223 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17225 PR middle-end/81462
17226 * predict.c (set_even_probabilities): Cleanup; do not affect
17227 probabilities that are already known.
17228 (combine_predictions_for_bb): Call even when count is set.
17230 2017-07-18 Nathan Sidwell <nathan@acm.org>
17232 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17235 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17238 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17239 optimization for loop niter analysis.
17241 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17244 * config/avr/avr.c (avr_optimize_casesi): Don't use
17245 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17247 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17249 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17250 body_cost_vec from _vect_peel_extended_info.
17251 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17252 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17255 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17257 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17259 2017-07-18 Richard Biener <rguenther@suse.de>
17261 PR tree-optimization/80620
17262 PR tree-optimization/81403
17263 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17264 info when re-using a VN table entry.
17266 2017-07-18 Richard Biener <rguenther@suse.de>
17268 PR tree-optimization/81418
17269 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17270 vectype_in. Verify that with lane-reducing reduction operations
17271 we have a single def-use cycle.
17273 2017-07-17 Carl Love <cel@us.ibm.com>
17275 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17277 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17278 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17279 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17280 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17281 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17282 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17283 VMULOSW): New enum "unspec" values.
17284 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17285 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17286 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17287 altivec_vmulosw): New patterns.
17288 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17289 VMULOSW): Add definitions.
17290 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17292 * config/alpha/alpha.c: Include predict.h.
17294 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17296 * tree-vrp.c (compare_assert_loc): Fix comparison function
17297 to return predictable results.
17299 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17301 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17303 (subdi3): Likewise.
17304 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17305 * doc/invoke.texi (mexpand-adddi): Update text.
17307 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17309 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17310 that also clobbers the CC register. The old expand code is moved
17312 (*arc_clzsi2): ... here.
17313 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17314 the CC register. The old expand code is moved to ...
17315 (arc_ctzsi2): ... here.
17317 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17319 * config/arc/arc.opt (mindexed-loads): Use initial value
17320 TARGET_INDEXED_LOADS_DEFAULT.
17321 (mauto-modify-reg): Use initial value
17322 TARGET_AUTO_MODIFY_REG_DEFAULT.
17323 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17324 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17325 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17326 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17328 2017-07-17 Martin Liska <mliska@suse.cz>
17331 * opts.c (finish_options): Do not allow -fgnu-tm
17332 w/ -fsanitize={kernel-,}address. Say sorry.
17334 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17337 * tree-loop-distribution.c (classify_partition): Only assert on
17338 numer of iterations.
17339 (merge_dep_scc_partitions): Delete prameter. Update function call.
17340 (distribute_loop): Remove code handling loop with unknown niters.
17341 (pass_loop_distribution::execute): Skip loop with unknown niters.
17343 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17346 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17347 function sort_partitions_by_post_order.
17349 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17351 PR tree-optimization/81374
17352 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17353 the max index of basic blocks, rather than number of basic blocks.
17355 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17357 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17359 (arc_legitimate_pic_operand_p): Likewise.
17360 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17362 (arc_needs_pcl_p): Likewise.
17363 (arc_legitimate_pc_offset_p): Likewise.
17364 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17365 function is also used in constrains.md.
17366 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17367 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17368 PLUS. Only return true/false in known cases, otherwise assert.
17369 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17370 is already called in arc_legitimate_constant_p.
17371 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17373 (LEGITIMATE_PIC_OPERAND_P): Use
17374 arc_raw_symbolic_reference_mentioned_p function.
17375 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17380 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17381 Andrew Burgess <andrew.burgess@embecosm.com>
17383 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17384 (arc_return_address_register): New function.
17385 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17386 (arc_handle_fndecl_attribute): Add naked attribute.
17387 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17388 (TARGET_WARN_FUNC_RETURN): Likewise.
17389 (arc_allocate_stack_slots_for_args): New function.
17390 (arc_warn_func_return): Likewise.
17391 (machine_function): Change type fn_type.
17392 (arc_compute_function_type): Consider new naked function type,
17393 change function return type.
17394 (arc_must_save_register): Adapt to handle new
17395 arc_compute_function_type's return type.
17396 (arc_expand_prologue): Likewise.
17397 (arc_expand_epilogue): Likewise.
17398 (arc_return_address_regs): Delete.
17399 (arc_return_address_register): New function.
17400 (arc_epilogue_uses): Use above function.
17401 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17402 (arc_function_type): Change encoding, add naked type.
17403 (ARC_INTERRUPT_P): Change to handle the new encoding.
17404 (ARC_FAST_INTERRUPT_P): Likewise.
17405 (ARC_NORMAL_P): Define.
17406 (ARC_NAKED_P): Likewise.
17407 (arc_compute_function_type): Delete prototype.
17408 * config/arc/arc.md (in_ret_delay_slot): Use
17409 arc_return_address_register function.
17410 (simple_return): Likewise.
17411 (p_return_i): Likewise.
17413 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17415 PR tree-optimization/81428
17416 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17417 can't be built for those types.
17419 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17421 Remove stuff dead since r239246.
17423 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17424 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17425 (avr_inform_devices): Remove dead stuff.
17427 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17429 * config/arm/arm_neon.h: Fix softp typo.
17431 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17433 PR tree-optimization/81365
17434 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17435 aggregate moves onto bb predecessor edges, make sure there are no
17436 loads that could alias the lhs in between the start of bb and the
17439 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17442 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17443 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17444 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17446 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17448 PR tree-optimization/81396
17449 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17450 (init_symbolic_number): Initialize it to 1.
17451 (perform_symbolic_merge): Add n_ops from both operands into the new
17453 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17454 without base_addr as useless if they need more than one operation.
17455 (bswap_replace): Handle !bswap case for NULL base_addr.
17457 2017-07-17 Tom de Vries <tom@codesourcery.com>
17460 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17463 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17465 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17466 conditional builtin define __FIX_LEON3FT_B2BST.
17468 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17470 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17471 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17474 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17476 PR rtl-optimization/81424
17477 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17478 to remove potential trapping from operands if -fnon-call-exceptions.
17480 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17482 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17483 profile_proability for scalling.
17484 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17486 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17488 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17490 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17492 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17493 fixpoint arithmetics.
17495 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17497 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17498 fixpoint arithmetics.
17500 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17502 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17503 fixpoint arithmetics.
17505 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17507 * profile-count.h (profile_probability::from_reg_br_prob_note,
17508 profile_probability::to_reg_br_prob_note): New functions.
17509 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17510 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17511 * predict.c (probability_reliable_p): Update.
17512 (edge_probability_reliable_p): Update.
17513 (br_prob_note_reliable_p): Update.
17514 (invert_br_probabilities): Update.
17515 (add_reg_br_prob_note): New function.
17516 (combine_predictions_for_insn): Update.
17517 * asan.c (asan_clear_shadow): Update.
17518 * cfgbuild.c (compute_outgoing_frequencies): Update.
17519 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17520 (update_br_prob_note): Update.
17521 (rtl_verify_edges): Update.
17522 (purge_dead_edges): Update.
17523 (fixup_reorder_chain): Update.
17524 * emit-rtl.c (try_split): Update.
17525 * ifcvt.c (cond_exec_process_insns): Update.
17526 (cond_exec_process_if_block): Update.
17527 (dead_or_predicable): Update.
17528 * internal-fn.c (expand_addsub_overflow): Update.
17529 (expand_neg_overflow): Update.
17530 (expand_mul_overflow): Update.
17531 * loop-doloop.c (doloop_modify): Update.
17532 * loop-unroll.c (compare_and_jump_seq): Update.
17533 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17534 * predict.h: Update.
17535 * reorg.c (mostly_true_jump): Update.
17537 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17538 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17539 * config/arc/arc.c: (emit_unlikely_jump): Update.
17540 * config/arm/arm.c: (emit_unlikely_jump): Update.
17541 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17542 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17543 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17544 (ix86_print_operand): Update.
17545 (ix86_split_fp_branch): Update.
17546 (predict_jump): Update.
17547 * config/ia64/ia64.c (ia64_print_operand): Update.
17548 * config/mmix/mmix.c (mmix_print_operand): Update.
17549 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17550 (rs6000_expand_split_stack_prologue): Update.
17551 * config/rs6000/rs6000.c: Update.
17552 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17553 (s390_expand_vec_movstr): Update.
17554 (s390_expand_cs_tdsi): Update.
17555 (s390_expand_split_stack_prologue): Update.
17556 * config/sh/sh.c (sh_print_operand): Update.
17557 (expand_cbranchsi4): Update.
17558 (expand_cbranchdi4): Update.
17559 * config/sparc/sparc.c (output_v9branch): Update.
17560 * config/spu/spu.c (get_branch_target): Update.
17561 (ea_load_store_inline): Update.
17562 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17563 * config/tilepro/tilepro.c: Update.
17565 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17567 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17568 (gimplify_save_expr): Likewise.
17570 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17572 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17574 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17576 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17578 * timevar.def (TV_IPA_FNSUMMARY): Define.
17580 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17582 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17583 to back store errata sensitive sequence from being generated.
17584 (sqrtdf2_fix): Likewise.
17586 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17588 * tree-ssa-threadupdate.c (compute_path_counts,
17589 update_joiner_offpath_counts): Use profile_probability.
17591 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17594 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17596 * config/arm/arm-c.c (arm_cpu_builtins): Define
17597 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17599 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17601 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17602 array entries to represent __ieee128 versions of the
17603 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17604 scalar_extract_sig, and scalar_insert_exp built-in functions.
17605 (altivec_resolve_overloaded_builtin): Add special case handling
17606 for the __builtin_scalar_insert_exp function, as represented by
17607 the P9V_BUILTIN_VEC_VSIEDP constant.
17608 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17609 exponent support for __ieee128 argument.
17610 (VSESQP): Add scalar extract signature support for __ieee128
17612 (VSTDCNQP): Add scalar test negative support for __ieee128
17614 (VSIEQP): Add scalar insert exponent support for __int128 argument
17615 with __ieee128 result.
17616 (VSIEQPF): Add scalar insert exponent support for __ieee128
17617 argument with __ieee128 result.
17618 (VSTDCQP): Add scalar test data class support for __ieee128
17620 (VSTDCNQP): Add overload support for scalar test negative with
17621 __ieee128 argument.
17622 (VSTDCQP): Add overload support for scalar test data class
17623 __ieee128 argument.
17624 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17625 UNSPEC_VSX_SXSIGDP.
17626 (UNSPEC_VSX_SIEXPQP): New constant.
17627 (xsxexpqp): New insn for VSX scalar extract exponent quad
17629 (xsxsigqp): New insn for VSX scalar extract significand quad
17631 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17632 precision with floating point argument.
17633 (xststdcqp): New expand for VSX scalar test data class quad
17635 (xststdcnegqp): New expand for VSX scalar test negative quad
17637 (xststdcqp): New insn to match expansions for VSX scalar test data
17638 class quad precision and VSX scalar test negative quad precision.
17639 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17640 special case operand checking to enforce that second operand of
17641 VSX scalar test data class with quad precision argument is a 7-bit
17643 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17644 prototypes and descriptions of __ieee128 versions of
17645 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17646 scalar_test_data_class, and scalar_test_neg built-in functions.
17648 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17650 PR tree-optimization/81162
17651 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17652 replace a negate with an add.
17654 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17656 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17658 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17660 * config/arm/arm-c.c (arm_cpu_builtins): Define
17661 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17663 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17665 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17666 (armv8-r): Set ARM Cortex-R52 as default CPU.
17667 * config/arm/arm-tables.opt: Regenerate.
17668 * config/arm/arm-tune.md: Regenerate.
17669 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17671 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17672 extension for -mcpu=cortex-r52.
17674 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17676 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17677 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17678 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17679 (fp-armv8): Define it as FP_ARMv8 only.
17680 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17681 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17683 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17684 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17685 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17686 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17687 than TARGET_FPU_ARMV8.
17688 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17689 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17690 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17692 * config/arm/neon.md (neon_vrint): Likewise.
17693 (neon_vcvt): Likewise.
17694 (neon_<fmaxmin_op><mode>): Likewise.
17695 (<fmaxmin><mode>3): Likewise.
17696 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17697 * config/arm/predicates.md (arm_cond_move_operator): Check against
17698 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17700 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17702 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17703 to top of function.
17705 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17707 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17708 loop in comment with memset.
17710 2017-07-14 Martin Liska <mliska@suse.cz>
17712 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17713 * dwarf2out.c (is_java): Remove the function.
17714 (output_pubname): Remove usage of the function.
17715 (lower_bound_default): Remove usage of DW_LANG_Java.
17716 (gen_compile_unit_die): Likewise.
17717 * gcc.c: Remove compiler defaults for .java and .zip files.
17718 * gimple-expr.c (remove_suffix): Change as there's no longer
17719 extension than 4-letter one.
17720 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17721 (gimplify_save_expr): Likewise.
17722 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17723 as it's possible even for other languages than Java.
17724 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17725 * lto-opts.c (lto_write_options): Remove reference to Java.
17726 * opts.c (strip_off_ending): Update file extension handling.
17727 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17728 * tree-eh.c (lower_resx): Likewise.
17729 * tree.c (free_lang_data_in_type): Remove dead code.
17730 (find_decls_types_r): Likewise.
17731 (build_common_builtin_nodes): Remove Java from a comment.
17732 (verify_type): Remove dead code.
17733 * varasm.c (assemble_external): Remove Java from a comment.
17735 2017-07-14 Martin Liska <mliska@suse.cz>
17737 * opts.c (finish_options): Add quotes.
17738 (common_handle_option): Likewise.
17740 2017-07-14 Martin Liska <mliska@suse.cz>
17742 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17743 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17744 Remove N_SO_PASCAL.
17745 * dwarf2out.c (lower_bound_default): Do not handle
17747 (gen_compile_unit_die): Likewise.
17748 * gcc.c: Remove default extension binding for GNU Pascal.
17749 * stmt.c: Remove Pascal language from a comment.
17750 * xcoffout.c: Likewise.
17752 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17755 * diagnostic-show-locus.c (fixit_cmp): New function.
17756 (layout::layout): Sort m_fixit_hints.
17757 (column_range::column_range): Assert that the values are valid.
17758 (struct char_span): New struct.
17759 (correction::overwrite): New method.
17760 (struct source_line): New struct.
17761 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17762 calls in terms of classes source_line and char_span, and
17763 correction::overwrite.
17764 (selftest::test_overlapped_fixit_printing_2): New function.
17765 (selftest::diagnostic_show_locus_c_tests): Call it.
17767 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17769 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17770 early if there is no lhs.
17772 2017-07-13 Martin Liska <mliska@suse.cz>
17774 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17775 (gen_reference_type_die): Likewise.
17776 * stor-layout.c: Remove Pascal-related comment.
17778 2017-07-13 Martin Liska <mliska@suse.cz>
17780 * opts.c (finish_options): Add quotes to error messages.
17781 (parse_sanitizer_options): Likewise.
17783 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17785 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17787 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17789 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17791 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17793 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17795 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17797 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17800 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17801 provides the hardware capability bits, define the macro
17802 __BUILTIN_CPU_SUPPORTS__.
17803 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17804 if GLIBC does not provide the hardware capability bits. Add a
17805 gcc_unreachable call if the built-in cpu function is neither
17806 __builtin_cpu_is nor __builtin_cpu_supports.
17807 (rs6000_get_function_versions_dispatcher): Change the warning
17808 that an old GLIBC is used which does not export the capability
17809 bits to be an error.
17810 * doc/extend.texi (target_clones attribute): Document the
17811 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17812 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17813 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17814 the macros defined by GCC if the newer GLIBC is available.
17816 2017-07-12 Jeff Law <law@redhat.com>
17818 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17819 remaining includes slightly.
17820 * config/riscv/riscv-builtins.c: Include profile-count.h.
17822 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17825 * config/avr/avr.c (avr_set_current_function): In diagnostic
17826 messages: Quote keywords and (parts of) identifiers.
17827 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17830 2017-07-12 Carl Love <cel@us.ibm.com>
17832 * config/rs6000/rs6000-c.c: Add support for built-in functions
17833 vector bool char vec_revb (vector bool char);
17834 vector bool short vec_revb (vector short char);
17835 vector bool int vec_revb (vector bool int);
17836 vector bool long long vec_revb (vector bool long long);
17837 * doc/extend.texi: Update the built-in documentation file for the
17838 new built-in functions.
17840 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17842 * config/s390/s390.md: Remove movcc splitter.
17844 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17846 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17847 load/store on condition.
17849 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17852 * config/avr/avr.c (avr_encode_section_info)
17853 [progmem && !TREE_READONLY]: Error if progmem object needs
17856 2017-07-11 Michael Collison <michael.collison@arm.com>
17858 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17861 2017-07-11 Carl Love <cel@us.ibm.com>
17863 * config/rs6000/rs6000-c.c: Add support for builtins
17864 vector unsigned int vec_parity_lsbb (vector signed int);
17865 vector unsigned int vec_parity_lsbb (vector unsigned int);
17866 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17867 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17868 vector unsigned long long vec_parity_lsbb (vector signed long long);
17869 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17870 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17871 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17872 * doc/extend.texi: Update the built-in documentation file for the
17873 new built-in functions.
17875 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17877 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17878 (layout::m_primary_loc): New field.
17879 (layout::layout): Initialize new field. Move location filtering
17880 logic from here to...
17881 (layout::maybe_add_location_range): ...this new method. Add
17882 support for filtering to just the lines already specified by other
17884 (layout::will_show_line_p): New method.
17885 (gcc_rich_location::add_location_if_nearby): New method.
17886 (selftest::test_add_location_if_nearby): New test function.
17887 (selftest::diagnostic_show_locus_c_tests): Call it.
17888 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17891 2017-07-11 Tom de Vries <tom@codesourcery.com>
17893 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17894 (bb_first_real_insn): New function.
17895 (nvptx_single): Add extra initialization of broadcasted condition
17898 2017-07-11 Nathan Sidwell <nathan@acm.org>
17900 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17902 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17904 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17905 Binutils doc as TEXI will mess them up.
17906 * doc/invoke.texi (AVR Options): Same here.
17908 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17910 * config/sparc/sparc.opt (mfix-ut700): New option.
17911 (mfix-gr712rc): Likewise.
17912 (sparc_fix_b2bst): New variable.
17913 * doc/invoke.texi (SPARC options): Document them.
17914 (ARM options): Fix warnings.
17915 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17916 instructions to prevent sequences that can trigger the store-store
17917 errata for certain LEON3FT processors.
17918 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17919 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17920 * config/sparc/sparc.md (fix_b2bst): New attribute.
17921 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17923 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17926 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17928 (*rsqrtsf2_sse): Ditto.
17930 (div<mode>3): Macroize insn from divdf3 and divsf3
17931 using MODEF mode iterator.
17933 2017-07-10 Martin Sebor <msebor@redhat.com>
17935 PR tree-optimization/80397
17936 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17937 instead of testing for equality to INTEGER_TYPE.
17939 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17941 * config.gcc: Remove uclibc from arc target spec.
17943 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17945 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17947 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17950 * lto-wrapper.c (remove_option): New function.
17951 (merge_and_complain): Merge PIC/PIE options more realistically.
17953 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17955 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17959 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17960 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17961 * config.in: Regenerate.
17962 * configure: Regenerate.
17963 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17964 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17965 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17966 (TARGET_GASISR_PROLOGUES): ...target mask.
17967 * common/config/avr/avr-common.c
17968 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17969 Set -mgas-isr-prologues.
17970 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17971 INSERT_PASS_BEFORE for it.
17972 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17973 * config/avr/avr.c (avr_option_override)
17974 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17975 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17976 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17977 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17978 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17980 (make_avr_pass_pre_proep): New function.
17981 (emit_push_sfr) <treg>: Add argument to function and use it
17982 instead of TMP_REG.
17983 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17984 and set machine->gasisr.yes.
17985 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17986 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17987 __gcc_isr.n_pushed to .L__stack_usage.
17988 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17989 (avr_asm_final_postscan_insn): ...this new static function.
17990 * config/avr/avr.h (machine_function)
17991 <is_no_gccisr, use_L__stack_usage>: New fields.
17992 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17993 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17994 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17995 (gasisr, *gasisr): New expander and insn.
17996 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17997 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17998 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18000 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18002 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18005 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18007 Move jump-tables out of .text again.
18010 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18011 (ASM_OUTPUT_ADDR_VEC): New function.
18012 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18013 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18014 INSN_ADDRESSes as asm comment.
18015 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18016 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18017 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18018 * config/avr/avr.md (*tablejump): Adjust comment.
18019 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18020 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18022 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18023 (avr_output_addr_vec): New proto.
18024 (avr_log_t) <insn_addresses>: New field.
18026 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18029 * config/i386/i386.c (ix86_function_arg_advance): Set
18030 outgoing_args_on_stack to true if there are outgoing arguments
18032 (ix86_function_arg): Likewise.
18033 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18034 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18035 * config/i386/i386.h (machine_function): Add
18036 outgoing_args_on_stack.
18038 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18040 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18041 supporting pthreds.
18042 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18044 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18046 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18047 (REAL_H): Remove $(MACHMODE_H).
18048 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18050 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18051 $(MACHMODE_H) and double-int.h.
18052 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18054 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18057 2017-07-07 Andrew Pinski <apinski@cavium.com>
18059 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18060 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18062 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18064 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18065 Add warning if GCC was not configured to link against a GLIBC that
18066 exports the hardware capability bits.
18067 (make_resolver_func): Make resolver function private and not a
18068 COMDAT function. Create the name with clone_function_name instead
18069 of make_unique_name.
18072 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18073 correct operand in doing the split.
18075 2017-07-07 Carl Love <cel@us.ibm.com>
18077 * config/rs6000/rs6000-c: Add support for built-in function
18078 vector unsigned short vec_pack_to_short_fp32 (vector float,
18080 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18081 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18082 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18083 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18084 (convert_4f32_8i16): Add define_expand.
18085 * doc/extend.texi: Update the built-in documentation file for the
18086 new built-in function.
18088 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18090 * config/sparc/m8.md: New file.
18091 * config/sparc/sparc.md: Include m8.md.
18093 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18095 * config/sparc/sparc.opt: New option -mvis4b.
18096 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18097 (sparc_option_override): Handle VIS4B.
18098 (enum sparc_builtins): Define
18099 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18100 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18101 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18102 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18103 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18104 (check_constant_argument): New function.
18105 (sparc_vis_init_builtins): Define builtins
18106 __builtin_vis_dictunpack{8,16,32},
18107 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18108 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18109 __builtin_vis_fpcmpde{8,16,32}shl and
18110 __builtin_vis_fpcmpur{8,16,32}shl.
18111 (sparc_expand_builtin): Check that the constant operands to
18112 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18113 constant and in range.
18114 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18116 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18117 (SPARC_IMM5_P): Likewise.
18118 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18119 (enabled): Handle vis4b.
18120 (UNSPEC_DICTUNPACK): New unspec.
18121 (UNSPEC_FPCMPSHL): Likewise.
18122 (UNSPEC_FPUCMPSHL): Likewise.
18123 (UNSPEC_FPCMPDESHL): Likewise.
18124 (UNSPEC_FPCMPURSHL): Likewise.
18125 (cpu_feature): New CPU feature `vis4b'.
18126 (dictunpack{8,16,32}): New insns.
18127 (FPCSMODE): New mode iterator.
18128 (fpcscond): New code iterator.
18129 (fpcsucond): Likewise.
18130 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18131 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18132 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18133 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18134 * config/sparc/constraints.md: Define constraints `q' for unsigned
18135 2-bit integer constants and `t' for unsigned 5-bit integer
18137 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18139 (imm5_operand_dictunpack16): Likewise.
18140 (imm5_operand_dictunpack32): Likewise.
18141 (imm2_operand): Likewise.
18142 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18143 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18144 ditunpack* and fpcmp*shl builtins.
18146 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18148 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18149 * config.in: Add HAVE_AS_SPARC6 define.
18150 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18152 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18154 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18155 (CPP_CPU_SPEC): Handle m8.
18156 (ASM_CPU_SPEC): Likewise.
18157 * config/sparc/sparc-opts.h (enum processor_type): Add
18159 * config/sparc/sparc.c (m8_costs): New struct.
18160 (sparc_option_override): Handle TARGET_CPU_m8.
18161 (sparc32_initialize_trampoline): Likewise.
18162 (sparc64_initialize_trampoline): Likewise.
18163 (sparc_issue_rate): Likewise.
18164 (sparc_register_move_cost): Likewise.
18165 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18166 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18167 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18168 (CPP_CPU_SPEC): Handle M8.
18169 (ASM_CPU_SPEC): Likewise.
18170 (AS_M8_FLAG): Define.
18171 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18172 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18173 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18175 * configure: Regenerate.
18176 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18179 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18181 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18183 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18184 ("*movdi_insn_sp32"): Do not set v3pipe.
18185 ("*movsi_insn"): Likewise.
18186 ("*movdi_insn_sp64"): Likewise.
18187 ("*movsf_insn"): Likewise.
18188 ("*movdf_insn_sp32"): Likewise.
18189 ("*movdf_insn_sp64"): Likewise.
18190 ("*zero_extendsidi2_insn_sp64"): Likewise.
18191 ("*sign_extendsidi2_insn"): Likewise.
18192 ("*mov<VM32:mode>_insn"): Likewise.
18193 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18194 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18195 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18196 ("<vlop:code><VL:mode>3"): Likewise.
18197 ("*not_<vlop:code><VL:mode>3"): Likewise.
18198 ("*nand<VL:mode>_vis"): Likewise.
18199 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18200 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18201 ("one_cmpl<VL:mode>2"): Likewise.
18202 ("faligndata<VM64:mode>_vis"): Likewise.
18203 ("alignaddrsi_vis"): Likewise.
18204 ("alignaddrdi_vis"): Likweise.
18205 ("alignaddrlsi_vis"): Likewise.
18206 ("alignaddrldi_vis"): Likewise.
18207 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18208 ("bmaskdi_vis"): Likewise.
18209 ("bmasksi_vis"): Likewise.
18210 ("bshuffle<VM64:mode>_vis"): Likewise.
18211 ("cmask8<P:mode>_vis"): Likewise.
18212 ("cmask16<P:mode>_vis"): Likewise.
18213 ("cmask32<P:mode>_vis"): Likewise.
18214 ("pdistn<P:mode>_vis"): Likewise.
18215 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18217 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18219 * config/sparc/sparc.md ("subtype"): New insn attribute.
18220 ("*wrgsr_sp64"): Set insn subtype.
18221 ("*rdgsr_sp64"): Likewise.
18222 ("alignaddrsi_vis"): Likewise.
18223 ("alignaddrdi_vis"): Likewise.
18224 ("alignaddrlsi_vis"): Likewise.
18225 ("alignaddrldi_vis"): Likewise.
18226 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18227 ("fexpand_vis"): Likewise.
18228 ("fpmerge_vis"): Likewise.
18229 ("faligndata<VM64:mode>_vis"): Likewise.
18230 ("bshuffle<VM64:mode>_vis"): Likewise.
18231 ("cmask8<P:mode>_vis"): Likewise.
18232 ("cmask16<P:mode>_vis"): Likewise.
18233 ("cmask32<P:mode>_vis"): Likewise.
18234 ("fchksm16_vis"): Likewise.
18235 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18236 ("fmean16_vis"): Likewise.
18237 ("fp<plusminus_insn>64_vis"): Likewise.
18238 ("<plusminus_insn>v8qi3"): Likewise.
18239 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18240 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18241 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18242 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18243 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18244 ("*movqi_insn"): Likewise.
18245 ("*movhi_insn"): Likewise.
18246 ("*movsi_insn"): Likewise.
18247 ("movsi_pic_gotdata_op"): Likewise.
18248 ("*movdi_insn_sp32"): Likewise.
18249 ("*movdi_insn_sp64"): Likewise.
18250 ("movdi_pic_gotdata_op"): Likewise.
18251 ("*movsf_insn"): Likewise.
18252 ("*movdf_insn_sp32"): Likewise.
18253 ("*movdf_insn_sp64"): Likewise.
18254 ("*zero_extendhisi2_insn"): Likewise.
18255 ("*zero_extendqihi2_insn"): Likewise.
18256 ("*zero_extendqisi2_insn"): Likewise.
18257 ("*zero_extendqidi2_insn"): Likewise.
18258 ("*zero_extendhidi2_insn"): Likewise.
18259 ("*zero_extendsidi2_insn_sp64"): Likewise.
18260 ("ldfsr"): Likewise.
18261 ("prefetch_64"): Likewise.
18262 ("prefetch_32"): Likewise.
18263 ("tie_ld32"): Likewise.
18264 ("tie_ld64"): Likewise.
18265 ("*tldo_ldub_sp32"): Likewise.
18266 ("*tldo_ldub1_sp32"): Likewise.
18267 ("*tldo_ldub2_sp32"): Likewise.
18268 ("*tldo_ldub_sp64"): Likewise.
18269 ("*tldo_ldub1_sp64"): Likewise.
18270 ("*tldo_ldub2_sp64"): Likewise.
18271 ("*tldo_ldub3_sp64"): Likewise.
18272 ("*tldo_lduh_sp32"): Likewise.
18273 ("*tldo_lduh1_sp32"): Likewise.
18274 ("*tldo_lduh_sp64"): Likewise.
18275 ("*tldo_lduh1_sp64"): Likewise.
18276 ("*tldo_lduh2_sp64"): Likewise.
18277 ("*tldo_lduw_sp32"): Likewise.
18278 ("*tldo_lduw_sp64"): Likewise.
18279 ("*tldo_lduw1_sp64"): Likewise.
18280 ("*tldo_ldx_sp64"): Likewise.
18281 ("*mov<VM32:mode>_insn"): Likewise.
18282 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18283 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18285 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18287 * config/sparc/sparc.md ("type"): New insn type viscmp.
18288 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18290 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18291 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18292 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18293 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18295 ("n7_vis_logical_11cycle"): Likewise.
18296 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18297 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18298 * config/sparc/niagara.md ("niag_vis"): Likewise.
18299 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18300 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18302 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18304 * config/sparc/sparc.md: New instruction type `bmask'.
18305 (bmaskdi_vis): Use the `bmask' type.
18306 (bmasksi_vis): Likewise.
18307 * config/sparc/ultra3.md (us3_array): Likewise.
18308 * config/sparc/niagara7.md (n7_array): Likewise.
18309 * config/sparc/niagara4.md (n4_array): Likewise.
18310 * config/sparc/niagara2.md (niag2_vis): Likewise.
18311 (niag3_vis): Likewise.
18312 * config/sparc/niagara.md (niag_vis): Likewise.
18314 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18316 * ipa-comdats.c: Remove optimize check from gate.
18317 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18318 for functions not optimized.
18319 (ipa_fn_summary_read): Skip optimize check.
18320 (ipa_fn_summary_write): Likewise.
18321 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18323 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18325 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18326 (check_callers): Give up on caller which is not optimized.
18327 (inline_small_functions): Likewise.
18328 (ipa_inline): Do not give up when not optimizing.
18329 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18330 away unoptimizes cdtors.
18331 (whole_program_function_and_variable_visibility): Do
18332 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18333 * ipa.c (process_references): Do not check optimize.
18334 (symbol_table::remove_unreachable_nodes): Update optimize check.
18335 (set_writeonly_bit): Update optimize check.
18336 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18337 (pass_ipa_single_use::gate): Remove.
18339 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18341 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18342 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18343 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18344 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18345 permute_load, permute_store, adjust_extract, adjust_splat,
18346 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18347 replace_swap_with_copy, dump_swap_insn_table,
18348 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18349 recombine_lvx_pattern, recombine_stvx_pattern,
18350 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18351 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18352 to file rs6000-p8swap.c.
18353 * config/rs6000/rs6000-p8swap.c: New file.
18354 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18355 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18356 and rs6000*-*-* targets.
18358 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18360 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18362 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18364 * lto-wrapper.c (merge_and_complain): Do not merge
18365 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18366 fsigned_zeros, ftrapping_math, fwrapv.
18367 (append_compiler_options): Do not track these options.
18368 (append_linker_options): Likewie
18370 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18372 * cgraphunit.c (cgraph_node::finalize_function): When
18373 !flag_toplevel_reorde set no_reorder flag.
18374 (varpool_node::finalize_decl): Likewise.
18375 (symbol_table::compile): Drop no toplevel reorder path.
18377 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18379 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18380 edges; zero probability is not better than uninitialized.
18382 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18384 * asan.h (asan_sanitize_allocas_p): Declare.
18385 * asan.c (asan_sanitize_allocas_p): New function.
18386 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18387 (handle_builtin_alloca): Likewise.
18388 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18389 if !asan_sanitize_allocas_p.
18390 * params.def (asan-instrument-allocas): Add new option.
18391 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18392 * opts.c (common_handle_option): Disable allocas sanitization for
18395 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18397 * asan.c: Include gimple-fold.h.
18398 (get_last_alloca_addr): New function.
18399 (handle_builtin_stackrestore): Likewise.
18400 (handle_builtin_alloca): Likewise.
18401 (asan_emit_allocas_unpoison): Likewise.
18402 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18403 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18404 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18405 (instrument_builtin_call): Pass gimple iterator to
18406 get_mem_refs_of_builtin_call.
18407 (last_alloca_addr): New global.
18408 * asan.h (asan_emit_allocas_unpoison): Declare.
18409 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18410 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18411 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18412 if function calls alloca.
18413 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18414 * gimple-fold.h (replace_call_with_value): Declare.
18415 * internal-fn.c: Include asan.h.
18416 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18417 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18419 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18421 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18422 (C_SELFTEST_FLAGS): New.
18423 (CPP_SELFTEST_FLAGS): New.
18424 (SELFTEST_DEPS): New, from deps of s-selftest.
18425 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18426 (CPP_SELFTEST_DEPS): New.
18427 (selftest): Add dependency on s-selftest-c++.
18428 (s-selftest): Rename to...
18429 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18430 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18431 than SELFTEST_FLAGS.
18432 (selftest-gdb): Rename to...
18433 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18435 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18436 (selftest-valgrind): Rename to...
18437 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18439 (selftest-valgrind): Reintroduce as an alias for
18440 selftest-c-valgrind.
18441 (s-selftest-c++): New.
18442 (selftest-c++-gdb): New.
18443 (selftest-c++-valgrind): New.
18445 2017-07-06 Olivier Hainque <hainque@adacore.com>
18447 * gcc.c (process_command): When deciding if undefined variables
18448 should be ignored when processing specs, accept "gcc -v" as well.
18450 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18452 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18453 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18455 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18457 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18458 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18459 * config/arm/arm-tables.opt: Regenerate.
18460 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18462 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18464 2017-07-06 Carl Love <cel@us.ibm.com>
18466 * ChangeLog: Clean up from mid air collision
18468 2017-07-06 Carl Love <cel@us.ibm.com>
18470 * config/rs6000/rs6000-c.c: Add support for built-in functions
18471 vector signed int vec_subc (vector signed int, vector signed int);
18472 vector signed __int128 vec_subc (vector signed __int128,
18473 vector signed __int128);
18474 vector unsigned __int128 vec_subc (vector unsigned __int128,
18475 vector unsigned __int128);
18476 vector signed int vec_sube (vector signed int, vector signed int,
18477 vector signed int);
18478 vector unsigned int vec_sube (vector unsigned int,
18479 vector unsigned int,
18480 vector unsigned int);
18481 vector signed __int128 vec_sube (vector signed __int128,
18482 vector signed __int128,
18483 vector signed__int128);
18484 vector unsigned __int128 vec_sube (vector unsigned __int128,
18485 vector unsigned __int128,
18486 vector unsigned __int128);
18487 vector signed int vec_subec (vector signed int, vector signed int,
18488 vector signed int);
18489 vector unsigned int vec_subec (vector unsigned int,
18490 vector unsigned int,
18491 vector unsigned int);
18492 vector signed __int128 vec_subec (vector signed __int128,
18493 vector signed __int128,
18494 vector signed__int128);
18495 vector unsigned __int128 vec_subec (vector unsigned __int128,
18496 vector unsigned __int128,
18497 vector unsigned __int128);
18498 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18499 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18500 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18501 BU_ALTIVEC_OVERLOAD_X definitions.
18502 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18503 * doc/extend.texi: Update the built-in documentation file for the new
18504 built-in functions.
18506 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18509 * diagnostic-show-locus.c (layout::layout): Use start and finish
18510 spelling location for the start and finish of each range.
18511 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18512 Add unused aspect param.
18513 * input.c (expand_location_1): Add "aspect" param, and use it
18514 to access the correct part of the location.
18515 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18517 (expand_location_to_spelling_point): Likewise.
18518 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18519 param, and pass it to expand_location_1.
18521 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18523 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18524 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18525 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18526 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18527 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18528 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18529 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18530 _mm_maskz_getmant_ss): New intrinsics.
18531 (__builtin_ia32_getexpss128_mask): Changed to ...
18532 __builtin_ia32_getexpss128_round ... this.
18533 (__builtin_ia32_getexpsd128_mask): Changed to ...
18534 __builtin_ia32_getexpsd128_round ... this.
18535 * config/i386/i386-builtin-types.def
18536 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18537 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18538 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18539 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18540 __builtin_ia32_getmantss_mask_round): New builtins.
18541 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18542 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18543 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18544 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18545 * config/i386/sse.md
18546 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18547 avx512f_sgetexp<mode><mask_scalar_name>
18548 <round_saeonly_scalar_name> ... this.
18549 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18550 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18551 vgetexp<ssescalarmodesuffix>
18552 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18553 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18554 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18555 avx512f_vgetmant<mode><mask_scalar_name>
18556 <round_saeonly_scalar_name> ... this.
18557 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18558 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18559 vgetmant<ssescalarmodesuffix>
18560 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18561 %0<mask_scalar_operand4>, %1, %2
18562 <round_saeonly_scalar_mask_op4>, %3} ... this.
18563 * config/i386/subst.md (mask_scalar_operand4,
18564 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18565 round_saeonly_scalar_nimm_predicate): New subst attributes.
18567 2017-07-06 Julia Koval <julia.koval@intel.com>
18569 * config/i386/i386.c (ix86_erase_embedded_rounding):
18570 Remove code for old rounding pattern.
18572 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18574 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18576 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18578 * doc/sourcebuild.texi (Test Directives, Variants of
18579 dg-require-support): Add documentation for dg-require-stack-check.
18581 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18583 * config/i386/subst.md (mask_scalar, round_scalar,
18584 round_saeonly_scalar): New meta-templates.
18585 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18586 round_scalar_mask_operand3, round_scalar_mask_op3,
18587 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18588 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18589 round_saeonly_scalar_constraint,
18590 round_saeonly_scalar_prefix): New subst attribute.
18591 * config/i386/sse.md
18592 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18593 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18594 <round_scalar_name> ... this.
18595 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18596 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18597 <round_scalar_name> ... this.
18598 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18599 <sse>_vm<code><mode>3<mask_scalar_name>
18600 <round_saeonly_scalar_name> ... this.
18601 (v<plusminus_mnemonic><ssescalarmodesuffix>
18602 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18603 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18604 v<plusminus_mnemonic><ssescalarmodesuffix>
18605 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18606 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18607 (v<multdiv_mnemonic><ssescalarmodesuffix>
18608 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18609 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18610 v<multdiv_mnemonic><ssescalarmodesuffix>
18611 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18612 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18613 (v<maxmin_float><ssescalarmodesuffix>
18614 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18615 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18616 v<maxmin_float><ssescalarmodesuffix>
18617 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18618 %0<mask_scalar_operand3>, %1, %<iptr>2
18619 <round_saeonly_scalar_mask_op3>} ... this.
18621 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18623 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18624 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18626 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18627 Alan Hayward <alan.hayward@arm.com>
18628 David Sherwood <david.sherwood@arm.com>
18630 * combine.c (simplify_if_then_else): Remove "enum" before
18632 * compare-elim.c (can_eliminate_compare): Likewise.
18633 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18635 (aarch64_lookup_simd_builtin_type): Likewise.
18636 (aarch64_simd_builtin_type): Likewise.
18637 (aarch64_init_simd_builtin_types): Likewise.
18638 (aarch64_simd_expand_args): Likewise.
18639 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18641 (aarch64_reverse_mask): Likewise.
18642 (aarch64_simd_emit_reg_reg_move): Likewise.
18643 (aarch64_gen_adjusted_ldpstp): Likewise.
18644 (aarch64_ccmp_mode_to_code): Likewise.
18645 (aarch64_operands_ok_for_ldpstp): Likewise.
18646 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18647 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18649 (aarch64_min_divisions_for_recip_mul): Likewise.
18650 (aarch64_reassociation_width): Likewise.
18651 (aarch64_get_condition_code_1): Likewise.
18652 (aarch64_simd_emit_reg_reg_move): Likewise.
18653 (aarch64_simd_attr_length_rglist): Likewise.
18654 (aarch64_reverse_mask): Likewise.
18655 (aarch64_operands_ok_for_ldpstp): Likewise.
18656 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18657 (aarch64_gen_adjusted_ldpstp): Likewise.
18658 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18660 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18661 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18662 (arm_lookup_simd_builtin_type): Likewise.
18663 (arm_simd_builtin_type): Likewise.
18664 (arm_init_simd_builtin_types): Likewise.
18665 (arm_expand_builtin_args): Likewise.
18666 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18667 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18668 (ft32_setup_incoming_varargs): Likewise.
18669 (ft32_function_arg): Likewise.
18670 (ft32_function_arg_advance): Likewise.
18671 (ft32_pass_by_reference): Likewise.
18672 (ft32_arg_partial_bytes): Likewise.
18673 (ft32_valid_pointer_mode): Likewise.
18674 (ft32_addr_space_pointer_mode): Likewise.
18675 (ft32_addr_space_legitimate_address_p): Likewise.
18676 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18678 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18679 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18680 (iamcu_alignment): Likewise.
18681 (canonicalize_vector_int_perm): Likewise.
18682 (ix86_noce_conversion_profitable_p): Likewise.
18683 (ix86_mpx_bound_mode): Likewise.
18684 (ix86_operands_ok_for_move_multiple): Likewise.
18685 * config/microblaze/microblaze-protos.h
18686 (microblaze_expand_conditional_branch_reg): Likewise.
18687 * config/microblaze/microblaze.c
18688 (microblaze_expand_conditional_branch_reg): Likewise.
18689 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18691 (rs6000_reassociation_width): Likewise.
18692 (rs6000_invalid_binary_op): Likewise.
18693 (fusion_p9_p): Likewise.
18694 (emit_fusion_p9_load): Likewise.
18695 (emit_fusion_p9_store): Likewise.
18696 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18698 (riscv_hard_regno_mode_ok_p): Likewise.
18699 (riscv_address_insns): Likewise.
18700 (riscv_split_symbol): Likewise.
18701 (riscv_legitimize_move): Likewise.
18702 (riscv_function_value): Likewise.
18703 (riscv_hard_regno_nregs): Likewise.
18704 (riscv_expand_builtin): Likewise.
18705 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18706 (riscv_build_integer): Likewise.
18707 (riscv_split_integer): Likewise.
18708 (riscv_legitimate_constant_p): Likewise.
18709 (riscv_cannot_force_const_mem): Likewise.
18710 (riscv_regno_mode_ok_for_base_p): Likewise.
18711 (riscv_valid_base_register_p): Likewise.
18712 (riscv_valid_offset_p): Likewise.
18713 (riscv_valid_lo_sum_p): Likewise.
18714 (riscv_classify_address): Likewise.
18715 (riscv_legitimate_address_p): Likewise.
18716 (riscv_address_insns): Likewise.
18717 (riscv_load_store_insns): Likewise.
18718 (riscv_force_binary): Likewise.
18719 (riscv_split_symbol): Likewise.
18720 (riscv_force_address): Likewise.
18721 (riscv_legitimize_address): Likewise.
18722 (riscv_move_integer): Likewise.
18723 (riscv_legitimize_const_move): Likewise.
18724 (riscv_legitimize_move): Likewise.
18725 (riscv_address_cost): Likewise.
18726 (riscv_subword): Likewise.
18727 (riscv_output_move): Likewise.
18728 (riscv_canonicalize_int_order_test): Likewise.
18729 (riscv_emit_int_order_test): Likewise.
18730 (riscv_function_arg_boundary): Likewise.
18731 (riscv_pass_mode_in_fpr_p): Likewise.
18732 (riscv_pass_fpr_single): Likewise.
18733 (riscv_pass_fpr_pair): Likewise.
18734 (riscv_get_arg_info): Likewise.
18735 (riscv_function_arg): Likewise.
18736 (riscv_function_arg_advance): Likewise.
18737 (riscv_arg_partial_bytes): Likewise.
18738 (riscv_function_value): Likewise.
18739 (riscv_pass_by_reference): Likewise.
18740 (riscv_setup_incoming_varargs): Likewise.
18741 (riscv_print_operand): Likewise.
18742 (riscv_elf_select_rtx_section): Likewise.
18743 (riscv_save_restore_reg): Likewise.
18744 (riscv_for_each_saved_reg): Likewise.
18745 (riscv_register_move_cost): Likewise.
18746 (riscv_hard_regno_mode_ok_p): Likewise.
18747 (riscv_hard_regno_nregs): Likewise.
18748 (riscv_class_max_nregs): Likewise.
18749 (riscv_memory_move_cost): Likewise.
18750 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18751 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18752 (rl78_addr_space_address_mode): Likewise.
18753 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18755 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18756 (rs6000_reassociation_width): Likewise.
18757 (rs6000_invalid_binary_op): Likewise.
18758 (fusion_p9_p): Likewise.
18759 (emit_fusion_p9_load): Likewise.
18760 (emit_fusion_p9_store): Likewise.
18761 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18762 (ok_for_simple_move_operands): Likewise.
18763 (ok_for_simple_move_strict_operands): Likewise.
18764 (ok_for_simple_arith_logic_operands): Likewise.
18765 (visium_legitimize_reload_address): Likewise.
18766 (visium_select_cc_mode): Likewise.
18767 (output_cbranch): Likewise.
18768 (visium_split_double_move): Likewise.
18769 (visium_expand_copysign): Likewise.
18770 (visium_expand_int_cstore): Likewise.
18771 (visium_expand_fp_cstore): Likewise.
18772 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18773 (visium_function_arg): Likewise.
18774 (visium_function_arg_advance): Likewise.
18775 (visium_libcall_value): Likewise.
18776 (visium_setup_incoming_varargs): Likewise.
18777 (visium_legitimate_constant_p): Likewise.
18778 (visium_legitimate_address_p): Likewise.
18779 (visium_legitimize_address): Likewise.
18780 (visium_secondary_reload): Likewise.
18781 (visium_register_move_cost): Likewise.
18782 (visium_memory_move_cost): Likewise.
18783 (prepare_move_operands): Likewise.
18784 (ok_for_simple_move_operands): Likewise.
18785 (ok_for_simple_move_strict_operands): Likewise.
18786 (ok_for_simple_arith_logic_operands): Likewise.
18787 (visium_function_value_1): Likewise.
18788 (rtx_ok_for_offset_p): Likewise.
18789 (visium_legitimize_reload_address): Likewise.
18790 (visium_split_double_move): Likewise.
18791 (visium_expand_copysign): Likewise.
18792 (visium_expand_int_cstore): Likewise.
18793 (visium_expand_fp_cstore): Likewise.
18794 (visium_split_cstore): Likewise.
18795 (visium_select_cc_mode): Likewise.
18796 (visium_split_cbranch): Likewise.
18797 (output_cbranch): Likewise.
18798 (visium_print_operand_address): Likewise.
18799 * expmed.c (flip_storage_order): Likewise.
18800 * expmed.h (emit_cstore): Likewise.
18801 (flip_storage_order): Likewise.
18802 * genrecog.c (validate_pattern): Likewise.
18803 * hsa-gen.c (gen_hsa_addr): Likewise.
18804 * internal-fn.c (expand_arith_overflow): Likewise.
18805 * ira-color.c (allocno_copy_cost_saving): Likewise.
18806 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18807 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18808 (process_invariant_for_inheritance): Likewise.
18809 * lra-eliminations.c (move_plus_up): Likewise.
18810 * omp-low.c (lower_oacc_reductions): Likewise.
18811 * simplify-rtx.c (simplify_subreg): Likewise.
18812 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18813 (TARGET_CHKP_BOUND_MODE): Likewise..
18814 * targhooks.c (default_chkp_bound_mode): Likewise.
18815 (default_setup_incoming_vararg_bounds): Likewise.
18816 * targhooks.h (default_chkp_bound_mode): Likewise.
18817 (default_setup_incoming_vararg_bounds): Likewise.
18818 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18819 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18820 (have_whole_vector_shift): Likewise.
18821 * tree-vect-stmts.c (vectorizable_load): Likewise.
18822 * doc/tm.texi: Regenerate.
18824 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18826 Graceful degrade if Binutils PR21472 is not available.
18829 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18830 .rodata in flash test fails.
18831 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18832 * confgure: Regenerate.
18833 * config.in: Regenerate.
18834 * config/avr/avr.c (avr_asm_named_section)
18835 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18836 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18837 (avr_asm_init_sections): Same.
18839 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18841 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18842 (fma<VH:mode>4_intrinsic): Likewise.
18843 (*fmsub<VCVTF:mode>4): Likewise.
18844 (*fmsub<VH:mode>4_intrinsic): Likewise.
18846 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18849 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18850 Don't depend on "optimize > 0".
18851 (out_movhi_r_mr, out_movqi_mr_r): Same.
18852 (out_movhi_mr_r, out_movqi_r_mr): Same.
18853 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18854 io_address_operand on "optimize > 0".
18856 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18858 * tree-loop-distribution.c: Add general explanantion on the pass.
18859 (generate_loops_for_partition): Mark distributed loop.
18860 (pg_add_dependence_edges): New parameter. Handle alias data
18861 dependence specially and record it in the parameter if asked.
18862 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18863 (init_partition_graph_vertices, add_partition_graph_edge): New.
18864 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18865 (free_partition_graph_vdata, build_partition_graph): New.
18866 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18867 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18868 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18869 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18870 (version_for_distribution_p, finalize_partitions): New.
18871 (distribute_loop): Handle alias data dependence specially. Factor
18872 out loop fusion code as functions and call these functions.
18874 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18876 * tree-loop-distribution.c (classify_partition): New parameter and
18877 better handle reduction statement.
18878 (rdg_build_partitions): Revise comment.
18879 (distribute_loop): Compute statements in all partitions and pass it
18880 to classify_partition.
18882 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18884 * tree-loop-distribution.c (enum partition_type): New.
18885 (struct partition): New field type.
18886 (partition_merge_into): Add parameter. Update partition type.
18887 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18888 (build_rdg_partition_for_vertex): Compute partition type.
18889 (rdg_build_partitions): Dump partition type.
18890 (distribute_loop): Update calls to partition_merge_into.
18892 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18894 * tree-loop-distribution.c (struct ddr_hasher): New.
18895 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18897 (classify_partition): Call get_data_dependence.
18898 (pg_add_dependence_edges): Ditto.
18899 (distribute_loop): Release data dependence hash table.
18901 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18903 * tree-loop-distribution.c (ref_base_address): Delete.
18904 (similar_memory_accesses): Rename ...
18905 (share_memory_accesses): ... to this. Check if partitions access
18906 the same memory reference.
18907 (distribute_loop): Call share_memory_accesses.
18909 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18911 * tree-loop-distribution.c (struct partition): New field recording
18912 its data reference.
18913 (partition_alloc, partition_free): Init and release data refs.
18914 (partition_merge_into): Merge data refs.
18915 (build_rdg_partition_for_vertex): Collect data refs for partition.
18916 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18918 (distribute_loop): Remve data refs from vertice data of partition
18921 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18923 * tree-loop-distribution.c (params.h): Include header file.
18924 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18925 (datarefs_vec): New global var.
18926 (create_rdg_vertices): Use datarefs_vec directly.
18927 (free_rdg): Don't free data references.
18928 (build_rdg): Update use. Don't free data references.
18929 (distribute_loop): Compute global variable for data references.
18930 Bail out if there are too many data references.
18932 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18934 * tree-loop-distribution.c (loop_nest): New global var.
18935 (build_rdg): Use loop directly, rather than loop nest.
18936 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18938 (distribute_loop): Compute global variable loop nest. Update use.
18940 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18942 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18943 (partition_merge_into): New parameter. Dump reason for fusion.
18944 (distribute_loop): Update use of partition_merge_into.
18946 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18948 * tree-loop-distribution.c (bb_top_order_index): New.
18949 (bb_top_order_index_size, bb_top_order_cmp): New.
18950 (stmts_from_loop): Use topological order.
18951 (pass_loop_distribution::execute): Compute and release topological
18952 order for basic blocks.
18954 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18956 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18959 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18961 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18962 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18963 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18964 * internal-fn.def (LOOP_DIST_ALIAS): New.
18965 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18966 (fold_loop_internal_call): ... this.
18967 (vect_loop_dist_alias_call): New function.
18968 (set_uid_loop_bbs): Call fold_loop_internal_call.
18969 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18972 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18975 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18976 Require dead FLAGS_REG at the beginning of a peephole.
18978 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18981 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18982 arguments in the call to __builtin_ia32_sbb_u32.
18983 (_subborrow_u64): Swap _X and _Y arguments in the call to
18984 __builtin_ia32_sbb_u64.
18986 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18989 * tree-vrp.c (compare_assert_loc): Turn into a function template
18990 with stable template parameter. Only test if a->e is NULL,
18991 !a->e == !b->e has been verified already. Use e == NULL or
18992 e != NULL instead of e or ! e tests. If stable is true, don't use
18993 iterative_hash_expr, on the other side allow a or b or both NULL
18994 and sort the NULLs last.
18995 (process_assert_insertions): Sort using compare_assert_loc<false>
18996 instead of compare_assert_loc, later sort using
18997 compare_assert_loc<true> before calling process_assert_insertions_for
18998 in a loop. Use break instead of continue once seen NULL pointer.
19000 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19002 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19003 Cortex-R7 and Cortex-R8 processors.
19005 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19007 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19008 uninitialized while src is not.
19010 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19012 * common/config/arm/arm-common.c: Adjust include path for
19014 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19015 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19016 (arm-cpu-data.h): Likewise.
19017 (arm-cpu-cdata.h): Likewise.
19018 * config/arm/arm-cpu.h: Delete.
19019 * config/arm/arm-cpu-cdata.h: Delete.
19020 * config/arm/arm-cpu-data.h: Delete.
19022 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19024 * config/arm/arm-cpus.in (cortex-a55): New.
19025 (cortex-a75): Likewise.
19026 (cortex-a75.cortex-a55): Likewise.
19027 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19029 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19030 * config/arm/arm-cpu-cdata.h: Regenerate.
19031 * config/arm/arm-cpu-data.h: Regenerate.
19032 * config/arm/arm-cpu.h: Regenerate.
19033 * config/arm/arm-tables.opt: Regenerate.
19034 * config/arm/arm-tune.md: Regenerate.
19036 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19038 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19040 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19042 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19045 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19047 PR tree-optimization/81292
19048 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19049 full_string_p, also call adjust_related_strinfos if the adjustment
19050 is simple, otherwise invalidate related strinfos.
19052 2017-07-04 Martin Liska <mliska@suse.cz>
19055 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19056 newly created variable as DECL_IGNORED_P.
19058 2017-07-04 Martin Liska <mliska@suse.cz>
19061 * ipa-inline.c (inline_small_functions):
19062 Use xstrdup_for_dump.
19064 2017-07-04 Tom de Vries <tom@codesourcery.com>
19066 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19068 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19071 * config/darwin.c (darwin_function_switched_text_sections):
19074 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19076 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19078 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19080 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19082 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19084 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19085 min_profitable_iters, and th as inclusive lower bounds.
19086 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19087 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19088 for min_profitable_iters and min_profitable_estimate.
19089 (vect_transform_loop): Treat th as an inclusive lower bound.
19090 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19092 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19095 * config/darwin.c (darwin_function_switched_text_sections):
19096 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19097 in two pieces, and suppress the use of buf.
19099 2017-07-03 Nathan Sidwell <nathan@acm.org>
19101 * hash-table.h (hash_table_mod1): Fix indentation.
19103 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19105 PR middle-end/81290
19106 * predict.c (force_edge_cold): Be more careful about propagation
19108 * profile-count.h (profile_probability::guessed,
19109 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19111 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19113 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19115 * doc/invoke.texi (rcpc architecture extension): Document it.
19117 2017-07-03 Richard Biener <rguenther@suse.de>
19119 PR tree-optimization/60510
19120 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19121 the scalar reduction PHI and use it.
19122 (vectorizable_reduction): Properly guard the single_defuse_cycle
19123 path for non-SLP reduction chains where we cannot use it.
19124 Rework reduc_def/index and vector type deduction. Rework
19125 vector operand gathering during reduction op code-gen.
19126 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19127 chains dissolve the chain and leave it to non-SLP reduction
19130 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19132 * tree-data-ref.h (dr_alignment): Declare.
19133 * tree-data-ref.c (dr_alignment): New function.
19134 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19135 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19137 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19139 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19141 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19142 and base_misalignment fields.
19143 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19144 * tree-data-ref.c: Include builtins.h.
19145 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19146 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19147 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19148 * tree-vect-data-refs.c: Include tree-cfg.h.
19149 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19150 fields instead of calculating an alignment here.
19151 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19152 innermost_loop_behavior fields.
19154 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19156 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19158 (DR_STEP_ALIGNMENT): New macro.
19159 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19160 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19161 (create_data_ref): Print it.
19162 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19163 to tell whether the step preserves vector (mis)alignment.
19164 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19165 Move the check for an integer step and generalise to all INTEGER_CST.
19166 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19167 Print the outer step alignment.
19169 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19171 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19172 with offset_alignment.
19173 (DR_ALIGNED_TO): Delete.
19174 (DR_OFFSET_ALIGNMENT): New macro.
19175 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19176 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19177 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19178 (create_data_ref): Likewise.
19179 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19180 (vect_analyze_data_refs): Likewise.
19181 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19182 creating dummy innermost behavior.
19184 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19186 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19187 with a "innermost_loop_behavior *" and refeence tree.
19188 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19189 (create_data_ref): Update call accordingly.
19190 * tree-predcom.c (find_looparound_phi): Likewise.
19192 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19194 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19195 fields with dr_wrt_vec_loop.
19196 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19197 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19198 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19199 (vect_dr_behavior): New function.
19200 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19201 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19202 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19203 track whether the step preserves the misalignment.
19204 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19205 Use vect_dr_behavior.
19206 (vect_setup_realignment): Update call accordingly.
19207 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19208 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19209 call to vect_create_addr_base_for_vector_ref.
19210 (vect_create_cond_for_align_checks): Likewise.
19211 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19212 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19213 (vect_recog_mask_conversion_pattern): Likewise.
19214 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19215 (new_stmt_vec_info): Remove redundant zeroing.
19217 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19219 * common/config/arm/arm-common.c (arm_be8_option): New function.
19220 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19221 (ISA_ARMv6): Add isa_bit_be8.
19222 * config/arm/arm.h (arm_be8_option): Add prototype.
19223 (BE8_SPEC_FUNCTION): New define.
19224 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19225 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19226 (mlittle-endian): Similarly.
19227 (mbe8, mbe32): New options.
19228 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19229 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19231 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19233 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19235 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19237 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19238 (cleanup_tree_cfg_bb): Use it.
19239 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19241 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19243 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19246 * loop-doloop.c (add_test): Update profile.
19248 2017-07-03 Martin Liska <mliska@suse.cz>
19251 * sanopt.c (rewrite_usage_of_param): New function.
19252 (sanitize_rewrite_addressable_params): Likewise.
19253 (pass_sanopt::execute): Call rewrite_usage_of_param.
19255 2017-07-03 Richard Biener <rguenther@suse.de>
19257 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19258 back to using VIEW_CONVERT_EXPR.
19260 2017-07-03 Martin Liska <mliska@suse.cz>
19263 * doc/extend.texi: Document when a resolver function is
19264 generated for target_clones.
19266 2017-07-03 Martin Liska <mliska@suse.cz>
19268 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19269 and shadow memory of auto variables which are subject of
19270 use-after-scope sanitization.
19271 (asan_expand_mark_ifn): Add do set only when is_poison.
19273 2016-07-03 Richard Biener <rguenther@suse.de>
19275 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19277 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19278 (vectorizable_reduction): Perform reduction PHI creation when
19279 visiting a reduction PHI and adjust and simplify code generation
19280 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19281 (vect_transform_loop): Visit reduction PHIs.
19282 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19283 defs into the SLP tree.
19284 (vect_build_slp_tree): Reduction defs terminate the recursion.
19285 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19287 (vect_get_vec_defs_for_stmt_copy): Export.
19288 (vect_get_vec_defs): Likewise.
19289 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19291 (vect_get_vec_defs_for_stmt_copy): Declare.
19292 (vect_get_vec_defs): Likewise.
19294 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19296 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19297 parameter with a "loop" parameter and use it instead of the
19298 loop containing DR_STMT. Don't check simple_iv when doing
19299 BB analysis. Describe the two analysis modes in the comment.
19301 2017-07-03 Tom de Vries <tom@codesourcery.com>
19303 PR tree-optimization/69468
19304 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19305 (find_same_succ_bb): Handle ignore_edge_flags.
19307 2017-07-03 Tom de Vries <tom@codesourcery.com>
19309 PR tree-optimization/81192
19310 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19312 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19314 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19316 2017-07-03 Tom de Vries <tom@codesourcery.com>
19318 PR tree-optimization/81192
19319 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19320 BB_SAME_SUCC (bb) == NULL.
19322 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19324 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19327 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19329 * dumpfile.c: Include profile-count.h
19330 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19332 (insert_cond_bb): Update profile.
19333 * tree-cfg.h (insert_cond_bb): Update prototype.
19334 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19335 * tree-dump.c: Do not include tree-cfg.
19337 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19339 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19341 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19343 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19346 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19348 * tree-complex.c (expand_complex_div_wide): update profile.
19350 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19351 Alan Hayward <alan.hayward@arm.com>
19352 David Sherwood <david.sherwood@arm.com>
19354 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19355 (CORETYPES_H): New define.
19356 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19357 (insn-modes-inline.h, s-modes-inline-h): New rules.
19358 (generated_files): Add insn-modes-inline.h.
19359 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19360 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19361 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19362 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19363 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19364 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19365 (build/gencodes.o, build/genconditions.o): Likewise.
19366 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19367 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19368 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19369 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19370 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19371 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19372 * coretypes.h: Include everything up to real.h for generators.
19373 Include insn-modes.h first. Include wide-int-print.h after
19374 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19375 * machmode.h: Don't include insn-modes.h here.
19376 * function-tests.c: Remove includes of signop.h, machmode.h,
19377 double-int.h and wide-int.h.
19379 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19381 * optc-save-gen.awk: Likewise.
19382 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19383 * godump.c: Remove include of wide-int-print.h.
19384 * pretty-print.h: Likewise.
19385 * wide-int-print.cc: Likewise.
19386 * wide-int.cc: Likewise.
19387 * hash-map-tests.c: Remove include of signop.h.
19388 * hash-set-tests.c: Likewise.
19389 * rtl-tests.c: Likewise.
19390 * mkconfig.sh: Remove include of machmode.h.
19391 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19393 (emit_insn_modes_inline_h): ...this new function. Emit the code
19394 into an insn-modes-inline.h header file, adding appropriate
19395 include guards and end comments.
19396 (emit_insn_modes_c_header): Remove include of machmode.h.
19397 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19399 (main): Handle -i flag and call emit_insn_modes_inline_h when
19402 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19404 * tree-ssa-strlen.c (strinfo): Rename the length field to
19405 nonzero_chars. Add a full_string_p field.
19406 (compare_nonzero_chars, zero_length_string_p): New functions.
19407 (get_addr_stridx): Add an offset_out parameter.
19408 Use compare_nonzero_chars.
19409 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19410 (new_strinfo): Update after above changes to strinfo.
19411 (set_endptr_and_length): Set full_string_p.
19412 (get_string_length): Update after above changes to strinfo.
19413 (unshare_strinfo): Update call to new_strinfo.
19414 (maybe_invalidate): Likewise.
19415 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19416 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19417 as a uhwi instead of an shwi. Update after above changes to
19418 strinfo and new_strinfo.
19419 (zero_length_string): Assert that chainsi contains full strings.
19420 Use zero_length_string_p. Update call to new_strinfo.
19421 (adjust_related_strinfos): Update after above changes to strinfo.
19422 Copy full_string_p from origsi.
19423 (adjust_last_stmt): Use zero_length_string_p.
19424 (handle_builtin_strlen): Update after above changes to strinfo and
19425 new_strinfo. Install the lhs as the string length if the previous
19426 entry didn't describe a full string.
19427 (handle_builtin_strchr): Update after above changes to strinfo
19429 (handle_builtin_strcpy): Likewise.
19430 (handle_builtin_strcat): Likewise.
19431 (handle_builtin_malloc): Likewise.
19432 (handle_pointer_plus): Likewise.
19433 (handle_builtin_memcpy): Likewise. Track nonzero characters
19434 that aren't necessarily followed by a nul terminator.
19435 (handle_char_store): Likewise.
19437 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19439 PR tree-optimization/80769
19440 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19441 for malloc and calloc. Document the new invariant that all related
19442 strinfos have delayed lengths or none do.
19443 (verify_related_strinfos): Move earlier in file.
19444 (set_endptr_and_length): New function, split out from...
19445 (get_string_length): ...here. Also set the lengths of related
19447 (zero_length_string): Assert that chainsi has known (rather than
19449 (adjust_related_strinfos): Likewise.
19451 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19453 PR tree-optimization/81136
19454 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19455 assert that two references with the same misalignment have the same
19456 compile-time misalignment if those compile-time misalignments
19459 2017-07-01 Andi Kleen <ak@linux.intel.com>
19461 * print-tree.c (print_node): Print all attributes.
19463 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19465 * cfg.c (scale_bbs_frequencies): New function.
19466 * cfg.h (scale_bbs_frequencies): Declare it.
19467 * cfgloopanal.c (single_likely_exit): Cleanup.
19468 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19470 (scale_loop_profile): Likewise.
19471 (loop_version): Likewise.
19472 (create_empty_loop_on_edge): Update.
19473 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19474 scale_loop_frequencies, scale_loop_profile, loopify,
19475 loop_version): Update prototypes.
19476 * modulo-sched.c (sms_schedule): Update.
19477 * predict.c (unlikely_executed_edge_p): Also check probability.
19478 (probably_never_executed_edge_p): Fix typo.
19479 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19480 * tree-parloops.c (gen_parallel_loop): Update.
19481 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19482 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19483 * tree-ssa-loop-split.c (split_loop): Update.
19484 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19485 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19486 (vect_loop_versioning): Update.
19487 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19489 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19491 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19493 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19495 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19496 to keep profile consistent.
19498 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19500 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19501 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19502 * profile-count.h (max_safe_multiplier): Make unsigned.
19503 (profile_count::guessed_zero): New.
19505 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19507 * bb-reorder.c (fix_up_crossing_landing_pad,
19508 fix_crossing_conditional_branches): Use make_single_succ_edge
19509 to keep profile consistent.
19511 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19513 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19516 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19519 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19520 the right scopes, make sure cond_jump isn't preserved between multiple
19521 iterations. Search for fallthru edge whenever there are 3+ edges and
19522 use find_fallthru_edge for it.
19524 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19526 Patch by Alexander Monakov <amonakov@ispras.ru>
19527 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19528 probabilities consistently.
19530 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19532 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19533 emit_cmp_and_jump_insns.
19535 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19538 * tree-inline.c (expand_call_inline): Combine profile statuses.
19540 2017-06-30 Andrew Pinski <apinski@cavium.com>
19542 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19543 fold_stmt returned true.
19545 2017-06-30 Nathan Sidwell <nathan@acm.org>
19547 * ggc.h (empty_string): Delete.
19548 * cfgexpand.c (expand_asm_stmt): Use plain "".
19549 * optabs.c (expand_asm_memory_barrier): Likewise.
19550 * stringpool.c (empty_string): Delete.
19551 (digit_vector, digit_string): Delete.
19552 (ggc_alloc_string): Use plain "", don't optimize single digit
19553 strings. Use ggc_alloc_atomic.
19555 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19557 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19558 comparison set and one other set, use the cost of the non-comparison
19561 2017-06-30 Nathan Sidwell <nathan@acm.org>
19563 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19566 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19568 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19569 loops. Remove now unneeded calls to gimple_switch_set_label() that
19570 just set removed labels to NULL_TREE.
19572 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19574 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19575 (set_range_info): ...here. Only call set_range_info_raw if domain
19577 (set_nonzero_bits): Call set_range_info_raw.
19578 * tree-ssanames.h (set_range_info_raw): New.
19580 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19583 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19584 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19585 of nonimmediate_operand and <store_mask_constraint> instead of m
19586 for the input operand. For V8FI iterator, always split if input
19587 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19588 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19589 <store_mask_predicate> instead of register_operand and
19590 <store_mask_constraint> instead of v for the input operand. Make
19591 sure both operands aren't MEMs for if not <mask_applied>.
19593 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19595 * lto-wrapper.c (copy_file) Close both file descriptors before
19598 2017-06-30 Martin Liska <mliska@suse.cz>
19601 * multiple_target.c (create_dispatcher_calls): Make ifunc
19602 also for function that don't have calls or are not referenced.
19604 2017-06-30 Richard Biener <rguenther@suse.de>
19606 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19607 analyze the first scalar stmt. Move vector type computation
19608 for the BB case here from ...
19609 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19610 live operation processing in the SLP case properly.
19612 2017-06-30 Richard Biener <rguenther@suse.de>
19614 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19616 2017-06-30 Martin Liska <mliska@suse.cz>
19619 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19620 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19622 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19624 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19627 2017-06-30 Richard Biener <rguenther@suse.de>
19629 PR tree-optimization/81249
19630 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19631 condition reduction result to original scalar type.
19633 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19635 * profile-count.h (enum profile_quality): Fix typos and whitespace
19638 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19640 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19641 type for branch probabilities.
19643 2017-06-29 Julian Brown <julian@codesourcery.com>
19644 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19646 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19647 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19648 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19649 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19651 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19653 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19654 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19655 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19656 CC usage from generic code to here.
19657 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19658 CC usage into the target macros.
19660 2017-06-29 Maya Rashish <coypu@sdf.org>
19662 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19665 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19667 * arm/arm-builtins.c: Include profile-count.h
19668 * except.c (sjlj_emit_function_enter): Use
19669 profile_probability::unlikely.
19671 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19673 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19674 and tocrel_offset be pointer args rather than implicitly using
19676 (legitimate_constant_pool_address_p, rs6000_emit_move,
19677 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19678 tocrel_offset and use in toc_relative_expr_p call.
19679 (print_operand, print_operand_address): Use static tocrel_base_oac
19680 and tocrel_offset_oac.
19681 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19684 2017-06-29 Maya Rashish <coypu@sdf.org>
19686 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19688 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19690 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19691 objects, take into account only the alignment of 'op0' and 'mode1' if
19694 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19696 * ccmp.c (ccmp_tree_comparison_p): New function.
19697 (ccmp_candidate_p): Update to use above function.
19698 (get_compare_parts): New function.
19699 (expand_ccmp_next): Update to use new functions.
19700 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19702 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19703 take mode as argument.
19704 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19705 * expr.c (expand_expr_real_1): Pass mode as argument.
19707 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19709 * combine.c (combine_instructions): Print insns to dump_file, together
19712 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19714 * asan.c (asan_emit_stack_protection): Update.
19715 (create_cond_insert_point): Update.
19716 * auto-profile.c (afdo_propagate_circuit): Update.
19717 * basic-block.h (struct edge_def): Turn probability to
19718 profile_probability.
19719 (EDGE_FREQUENCY): Update.
19720 * bb-reorder.c (find_traces_1_round): Update.
19721 (better_edge_p): Update.
19722 (sanitize_hot_paths): Update.
19723 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19724 (make_single_succ_edge): Update.
19725 (check_bb_profile): Update.
19726 (dump_edge_info): Update.
19727 (update_bb_profile_for_threading): Update.
19728 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19729 probabilitycount to 0.
19730 * cfgbuild.c (compute_outgoing_frequencies): Update.
19731 * cfgcleanup.c (try_forward_edges): Update.
19732 (outgoing_edges_match): Update.
19733 (try_crossjump_to_edge): Update.
19734 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19735 (expand_gimple_tailcall): Update.
19736 (construct_init_block): Use make_single_succ_edge.
19737 (construct_exit_block): Use make_single_succ_edge.
19738 * cfghooks.c (verify_flow_info): Update.
19739 (redirect_edge_succ_nodup): Update.
19740 (split_edge): Update.
19741 (account_profile_record): Update.
19742 * cfgloopanal.c (single_likely_exit): Update.
19743 * cfgloopmanip.c (scale_loop_profile): Update.
19744 (set_zero_probability): Remove.
19745 (duplicate_loop_to_header_edge): Update.
19746 * cfgloopmanip.h (loop_version): Update prototype.
19747 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19748 (force_nonfallthru_and_redirect): Update.
19749 (update_br_prob_note): Update.
19750 (rtl_verify_edges): Update.
19751 (purge_dead_edges): Update.
19752 (rtl_lv_add_condition_to_bb): Update.
19753 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19754 * cgraphunit.c (init_lowered_empty_function): Update.
19755 (cgraph_node::expand_thunk): Update.
19756 * cilk-common.c: Include profile-count.h
19757 * dojump.c (inv): Remove.
19758 (jumpifnot): Update.
19759 (jumpifnot_1): Update.
19760 (do_jump_1): Update.
19762 (do_jump_by_parts_greater_rtx): Update.
19763 (do_compare_rtx_and_jump): Update.
19764 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19765 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19766 * dwarf2cfi.c: Include profile-count.h
19767 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19768 (sjlj_emit_dispatch_table): Likewise.
19769 * explow.c: Include profile-count.h
19770 * expmed.c (emit_store_flag_force): Update.
19771 (do_cmp_and_jump): Update.
19772 * expr.c (compare_by_pieces_d::generate): Update.
19773 (compare_by_pieces_d::finish_mode): Update.
19774 (emit_block_move_via_loop): Update.
19775 (store_expr_with_bounds): Update.
19776 (store_constructor): Update.
19777 (expand_expr_real_2): Update.
19778 (expand_expr_real_1): Update.
19779 * expr.h (try_casesi, try_tablejump): Update prototypes.
19780 * gimple-pretty-print.c (dump_probability): Update.
19781 (dump_profile): New.
19782 (dump_gimple_label): Update.
19783 (dump_gimple_bb_header): Update.
19784 * graph.c (draw_cfg_node_succ_edges): Update.
19785 * hsa-gen.c (convert_switch_statements): Update.
19786 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19787 (find_if_case_1): Update.
19788 (find_if_case_2): Update.
19789 * internal-fn.c (expand_arith_overflow_result_store): Update.
19790 (expand_addsub_overflow): Update.
19791 (expand_neg_overflow): Update.
19792 (expand_mul_overflow): Update.
19793 (expand_vector_ubsan_overflow): Update.
19794 * ipa-cp.c (good_cloning_opportunity_p): Update.
19795 * ipa-split.c (split_function): Use make_single_succ_edge.
19796 * ipa-utils.c (ipa_merge_profiles): Update.
19797 * loop-doloop.c (add_test): Update.
19798 (doloop_modify): Update.
19799 * loop-unroll.c (compare_and_jump_seq): Update.
19800 (unroll_loop_runtime_iterations): Update.
19801 * lra-constraints.c (lra_inheritance): Update.
19802 * lto-streamer-in.c (input_cfg): Update.
19803 * lto-streamer-out.c (output_cfg): Update.
19804 * mcf.c (adjust_cfg_counts): Update.
19805 * modulo-sched.c (sms_schedule): Update.
19806 * omp-expand.c (expand_omp_for_init_counts): Update.
19807 (extract_omp_for_update_vars): Update.
19808 (expand_omp_ordered_sink): Update.
19809 (expand_omp_for_ordered_loops): Update.
19810 (expand_omp_for_generic): Update.
19811 (expand_omp_for_static_nochunk): Update.
19812 (expand_omp_for_static_chunk): Update.
19813 (expand_cilk_for): Update.
19814 (expand_omp_simd): Update.
19815 (expand_omp_taskloop_for_outer): Update.
19816 (expand_omp_taskloop_for_inner): Update.
19817 * omp-simd-clone.c (simd_clone_adjust): Update.
19818 * optabs.c (expand_doubleword_shift): Update.
19819 (expand_abs): Update.
19820 (emit_cmp_and_jump_insn_1): Update.
19821 (expand_compare_and_swap_loop): Update.
19822 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19823 * predict.c (predictable_edge_p): Update.
19824 (edge_probability_reliable_p): Update.
19825 (set_even_probabilities): Update.
19826 (combine_predictions_for_insn): Update.
19827 (combine_predictions_for_bb): Update.
19828 (propagate_freq): Update.
19829 (estimate_bb_frequencies): Update.
19830 (force_edge_cold): Update.
19831 * profile-count.c (profile_count::dump): Add missing space into dump.
19832 (profile_count::debug): Add newline.
19833 (profile_count::differs_from_p): Explicitly convert to unsigned.
19834 (profile_count::stream_in): Update.
19835 (profile_probability::dump): New member function.
19836 (profile_probability::debug): New member function.
19837 (profile_probability::differs_from_p): New member function.
19838 (profile_probability::differs_lot_from_p): New member function.
19839 (profile_probability::stream_in): New member function.
19840 (profile_probability::stream_out): New member function.
19841 * profile-count.h (profile_count_quality): Rename to ...
19842 (profile_quality): ... this one.
19843 (profile_probability): New.
19844 (profile_count): Update.
19845 * profile.c (compute_branch_probabilities): Update.
19846 * recog.c (peep2_attempt): Update.
19847 * sched-ebb.c (schedule_ebbs): Update.
19848 * sched-rgn.c (find_single_block_region): Update.
19849 (compute_dom_prob_ps): Update.
19850 (schedule_region): Update.
19851 * sel-sched-ir.c (compute_succs_info): Update.
19852 * stmt.c (struct case_node): Update.
19853 (do_jump_if_equal): Update.
19854 (get_outgoing_edge_probs): Update.
19855 (conditional_probability): Update.
19856 (emit_case_dispatch_table): Update.
19857 (expand_case): Update.
19858 (expand_sjlj_dispatch_table): Update.
19859 (emit_case_nodes): Update.
19860 * targhooks.c: Update.
19861 * tracer.c (better_p): Update.
19862 (find_best_successor): Update.
19863 * trans-mem.c (expand_transaction): Update.
19864 * tree-call-cdce.c: Update.
19865 * tree-cfg.c (gimple_split_edge): Upate.
19866 (move_sese_region_to_fn): Upate.
19867 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19868 * tree-eh.c (lower_resx): Upate.
19869 (cleanup_empty_eh_move_lp): Upate.
19870 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19871 * tree-inline.c (copy_edges_for_bb): Update.
19872 (copy_cfg_body): Update.
19873 * tree-parloops.c (gen_parallel_loop): Update.
19874 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19875 (gimple_gen_time_profiler): Update.
19876 * tree-ssa-dce.c (remove_dead_stmt): Update.
19877 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19878 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19879 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19880 (unloop_loops): Update.
19881 (try_peel_loop): Update.
19882 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19883 * tree-ssa-loop-split.c (connect_loops): Update.
19884 (split_loop): Update.
19885 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19886 (hoist_guard): Update.
19887 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19888 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19889 (value_replacement): Update.
19890 * tree-ssa-reassoc.c (branch_fixup): Update.
19891 * tree-ssa-tail-merge.c (replace_block_by): Update.
19892 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19893 (create_edge_and_update_destination_phis): Update.
19894 (compute_path_counts): Update.
19895 (recompute_probabilities): Update.
19896 (update_joiner_offpath_counts): Update.
19897 (freqs_to_counts_path): Update.
19898 (duplicate_thread_path): Update.
19899 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19900 (struct switch_conv_info): Update.
19901 (gen_inbound_check): Update.
19902 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19903 (vect_do_peeling): Update.
19904 (vect_loop_versioning): Update.
19905 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19906 (optimize_mask_stores): Update.
19907 * ubsan.c (ubsan_expand_null_ifn): Update.
19908 * value-prof.c (gimple_divmod_fixed_value): Update.
19909 (gimple_divmod_fixed_value_transform): Update.
19910 (gimple_mod_pow2): Update.
19911 (gimple_mod_pow2_value_transform): Update.
19912 (gimple_mod_subtract): Update.
19913 (gimple_mod_subtract_transform): Update.
19914 (gimple_ic): Update.
19915 (gimple_stringop_fixed_value): Update.
19916 (gimple_stringops_transform): Update.
19917 * value-prof.h: Update.
19919 2017-06-29 Carl Love <cel@us.ibm.com>
19921 * config/rs6000/rs6000-c.c: Add support for built-in functions
19922 vector signed int vec_signed (vector float);
19923 vector signed long long vec_signed (vector double);
19924 vector signed int vec_signed2 (vector double, vector double);
19925 vector signed int vec_signede (vector double);
19926 vector signed int vec_signedo (vector double);
19927 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19928 instruction generator.
19929 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19930 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19931 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19933 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19934 vunsignede_v2df): Add define_expands.
19935 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19936 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19937 VEC_UNSIGNEDO): Add definitions.
19938 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19939 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19940 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19941 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19942 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19943 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19944 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19945 vec_unsignede, vec_unsignedo): Add builtin defines.
19946 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19948 * doc/extend.texi: Update the built-in documentation file for the
19949 new built-in functions.
19951 2017-06-29 Richard Biener <rguenther@suse.de>
19953 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19954 reduction chains to LOOP_VINFO_REDUCTIONS.
19955 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19956 SLP reductions after processing reduction chains.
19958 2017-06-29 Nathan Sidwell <nathan@acm.org>
19960 * builtins.c (fold_builtin_FUNCTION): Use
19961 lang_hooks.decl_printable_name.
19963 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19965 PR middle-end/81194
19966 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19967 with only one label.
19968 * stmt.c (expand_case): Assert NCASES is greater than one.
19970 2017-06-29 Richard Biener <rguenther@suse.de>
19972 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19974 (group_case_labels): Likewise.
19975 (find_taken_edge): Push sanity checking on val to workers...
19976 (find_taken_edge_cond_expr): ... here
19977 (find_taken_edge_switch_expr): ... and here, handle cases
19978 with just a default label.
19979 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19980 (group_case_labels): Likewise.
19981 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19982 group_case_labels does anything cleanup the CFG again.
19984 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19986 PR tree-optimization/81196
19987 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19988 exit condition comparing two IVs.
19990 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19992 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19993 profile to the dummy entry at the end of the list of architectures.
19994 * config/arm/arm-cpu-cdata.h: Regenerated.
19996 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19997 Michael Collison <michael.collison@arm.com>
20000 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20002 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20003 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20004 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20005 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20006 of shift when the shift amount is masked with constant equal to
20007 the size of the mode.
20008 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20011 2017-06-29 Martin Liska <mliska@suse.cz>
20013 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20015 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20018 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20021 (edge_growth_cache_entry::edge_growth_cache_entry): New
20023 (reset_edge_growth_cache): Update to use constructor.
20025 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20027 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20028 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20029 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20031 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20033 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20034 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20036 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20038 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20039 (*-linux-uclibc*): Add t-uclibc tmake_file.
20040 * config/t-musl: New.
20041 * config/t-uclibc: New.
20043 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20045 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20047 (gen_comm_data): Emit architectural setting of arch_prof.
20048 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20050 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20051 (armv8-m.base, armv8-m.main): Likewise.
20052 * arm-protos.h (arm_build_target): Add profile field.
20053 (arch_option): Likewise.
20054 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20056 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20057 arm_active_target.profile.
20059 2017-06-28 Richard Biener <rguenther@suse.de>
20061 PR middle-end/81227
20062 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20063 TYPE_OVERFLOW_WRAPS.
20064 * match.pd (negate_expr_p): Likewise.
20065 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20066 fold_build2, not fold_binary.
20068 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20070 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20071 Convert memory address to Pmode.
20072 (aarch64_print_operand): Assert MEM operands are always Pmode.
20074 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20077 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20078 Remove redundant if.
20079 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20080 * config/arm/aarch-common-protos.h
20081 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20082 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20084 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20087 * multiple_target.c (create_dispatcher_calls): Set the default
20088 clone to be static, not public.
20090 2017-06-28 Richard Biener <rguenther@suse.de>
20092 * tree-vect-loop.c (vectorizable_reduction): Move special
20093 cond reduction IV var creation ...
20094 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20095 parameter. Use STMT_VINFO_VECTYPE.
20096 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20099 2017-06-28 Martin Liska <mliska@suse.cz>
20102 * ipa-visibility.c (non_local_p): Handle visibility.
20104 2017-06-28 Martin Liska <mliska@suse.cz>
20107 * common.opt: Add IntegerRange to various options.
20108 * opt-functions.awk (integer_range_info): New function.
20109 * optc-gen.awk: Add integer_range_info to cl_options struct.
20110 * opts-common.c (decode_cmdline_option): Handle
20111 CL_ERR_INT_RANGE_ARG.
20112 (cmdline_handle_error): Likewise.
20113 * opts.c (print_filtered_help): Show valid interval in
20114 when --help is provided.
20115 * opts.h (struct cl_option): Add range_min and range_max fields.
20116 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20118 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20120 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20121 (x * C EQ/NE y * C): New transformation.
20123 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20125 * genmultilib (combination_space): Accept '+' in option names.
20127 2017-06-28 Martin Liska <mliska@suse.cz>
20130 * asan.c (instrument_derefs): Bail out inner references
20131 that are hard register variables.
20133 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20136 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20137 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20139 2017-06-28 Richard Biener <rguenther@suse.de>
20141 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20142 (vect_get_slp_defs): Adjust.
20143 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20145 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20147 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20148 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20149 (vectorizable_reduction): Adjust.
20150 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20152 (vect_get_slp_defs): Likewise.
20153 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20154 (vectorizable_bswap): Adjust.
20155 (vectorizable_call): Likewise.
20156 (vectorizable_conversion): Likewise.
20157 (vectorizable_assignment): Likewise.
20158 (vectorizable_shift): Likewise.
20159 (vectorizable_operation): Likewise.
20160 (vectorizable_store): Likewise.
20161 (vectorizable_condition): Likewise.
20162 (vectorizable_comparison): Likewise.
20164 2017-06-28 Michael Collison <michael.collison@arm.com>
20167 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20169 (arm_gen_movmemqi): Removed unused variable 'i'.
20170 Convert 'for' loop into 'while' loop.
20171 (arm_expand_prologue): Remove last unnecessary set of insn.
20172 (thumb_pop): Remove unused variable 'pushed_words'.
20173 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20175 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20177 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20178 * config/s390/s390-protos.h: Add prototype of
20179 s390_rel_address_ok_p.
20180 * config/s390/s390.c (s390_got_symbol): New function.
20181 (s390_rel_address_ok_p): New function.
20182 (legitimize_pic_address): Use s390_rel_address_ok_p.
20183 (s390_load_got): Use s390_got_symbol.
20184 (s390_option_override): Issue error if
20185 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20186 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20188 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20190 2017-06-27 Andrew Pinski <apinski@cavium.com>
20192 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20193 (X * copysign (1.0, X)): New pattern.
20194 (X * copysign (1.0, -X)): New pattern.
20195 (copysign (-1.0, CST)): New pattern.
20197 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20199 * genmultilib (combination_space): Remove variable.
20200 Validate reuse rules against regular expression for any sequence
20201 of multilib options in any order.
20203 2017-06-27 Michael Collison <michael.collison@arm.com>
20205 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20206 call aarch64_split_simd_combine.
20207 * (aarch64_combine_internal<mode>): Delete pattern.
20208 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20209 Allow register and subreg operands.
20211 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20213 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20214 specific need, just fallback on defaults.
20215 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20217 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20218 Olivier Hainque <hainque@adacore.com>
20220 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20222 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20223 targets. Pick a default if no particular attempt applied.
20224 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20227 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20229 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20230 (x86_64-wrs-vxworks7): Likewise.
20232 2017-06-27 Marek Polacek <polacek@redhat.com>
20235 * ubsan.c (instrument_null): Check get_base_address's result for null.
20237 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20239 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20241 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20243 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20244 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20245 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20246 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20247 New function types.
20248 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20249 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20250 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20251 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20252 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20253 BUILT_IN_FEUPDATEENV): New builtins.
20254 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20255 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20256 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20257 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20259 (builtin_structptr_types): Adjust size.
20260 * tree.c (builtin_structptr_types): Add four entries.
20262 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20263 Olivier Hainque <hainque@adacore.com>
20265 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20266 (TLS_SYM): New local macro, forcing reference to __tls__ on
20267 link command lines for VxWorks 7 RTPs, triggering initialization
20269 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20270 OS features TLS support, true for RTPs on VxWorks 7.
20271 * config/vxworks.c (vxworks_override_options): Setup emutls
20274 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20276 * predict.c (test_prediction_value_range): Use -1U instead of -1
20277 to avoid narrowing conversion warning.
20278 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20279 to avoid narrowing conversion warning.
20280 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20282 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20284 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20286 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20287 64bit configurations.
20288 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20289 (SIZE_TYPE): Likewise.
20290 * config/vxworks.c (vxworks_emutls_var_fields): Use
20291 long_unsigned_type_node instead of unsigned_type_node as the offset
20292 field type, which is "pointer" mode in emutls.c.
20294 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20297 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20299 PR middle-end/81207
20300 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20301 gimple_vuse copying separately from gimple_vdef copying.
20303 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20305 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20308 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20309 Olivier Hainque <hainque@adacore.com>
20311 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20312 for all vxworks7 targets.
20313 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20314 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20315 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20316 variations for VX6/VX7 and 32/64bits later on in ...
20317 (VXWORKS_LIB_SPEC): Leverage new macros.
20318 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20319 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20321 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20323 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20324 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20326 2017-06-26 Carl Love <cel@us.ibm.com>
20328 * config/rs6000/rs6000-c.c: Add support for built-in functions
20329 vector bool char vec_reve (vector bool char);
20330 vector signed char vec_reve (vector signed char);
20331 vector unsigned char vec_reve (vector unsigned char);
20332 vector bool int vec_reve (vector bool int);
20333 vector signed int vec_reve (vector signed int);
20334 vector unsigned int vec_reve (vector unsigned int);
20335 vector bool long long vec_reve (vector bool long long);
20336 vector signed long long vec_reve (vector signed long long);
20337 vector unsigned long long vec_reve (vector unsigned long long);
20338 vector bool short vec_reve (vector bool short);
20339 vector signed short vec_reve (vector signed short);
20340 vector double vec_reve (vector double);
20341 vector float vec_reve (vector float);
20342 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20343 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20344 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20345 (altivec_vreve): New pattern.
20346 * config/rs6000/altivec.h (vec_reve): New define.
20347 * doc/extend.texi (vec_rev): Update the built-in documentation file
20348 for the new built-in functions.
20350 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20352 PR tree-optimization/71815
20353 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20355 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20357 (slsr_process_phi): Likewise.
20358 (replace_uncond_cands_and_profitable_phis): Don't replace a
20359 multiply candidate with a stride of 1 (copy or cast).
20360 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20362 (lowest_cost_path): Likewise.
20363 (total_savings): Likewise.
20365 2017-06-26 Richard Biener <rguenther@suse.de>
20368 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20369 Use def_builtin_pure for all gather builtins.
20371 2017-06-26 Richard Biener <rguenther@suse.de>
20373 PR tree-optimization/81203
20374 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20375 non-dominating BBs.
20377 2017-06-26 Marek Polacek <polacek@redhat.com>
20380 * doc/invoke.texi: Document -Wmultistatement-macros.
20382 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20384 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20385 arm_neon_ok_no_float_abi effective target.
20387 2017-06-26 Richard Biener <rguenther@suse.de>
20389 PR tree-optimization/80928
20390 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20391 (copy_bbs): Set BB_DUPLICATED flag early.
20392 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20394 (execute_on_shrinking_pred): Likewise.
20395 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20396 BB_DUPLICATED blocks.
20397 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20398 iterate over all PHIs considering removal of *gsi.
20400 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20402 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20405 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20407 * config/rs6000/rs6000-string.c: (expand_block_clear,
20408 do_load_for_compare, select_block_compare_mode,
20409 compute_current_alignment, expand_block_compare,
20410 expand_strncmp_align_check, expand_strn_compare,
20411 expand_block_move, rs6000_output_load_multiple)
20412 Move functions related to string/block move/compare
20413 to a separate file.
20414 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20415 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20416 for this function which is now used in two files.
20417 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20418 * config.gcc: Add rs6000-string.o to extra_objs for
20419 targets powerpc*-*-* and rs6000*-*-*.
20421 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20424 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20425 32-bit, since indexed is not valid for DImode.
20426 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20427 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20428 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20429 (define_peephole2 for Altivec d-form store): Likewise.
20432 * multiple_target.c (create_dispatcher_calls): Only create the
20433 dispatcher call if the function is the default clone of a
20434 versioned function.
20436 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20438 PR middle-end/80902
20439 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20440 a call, force the call to not be a tail call.
20442 2017-06-23 Jeff Law <law@redhat.com>
20444 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20447 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20449 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20450 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20451 handling for early expansion of vector shifts (sl,sr,sra,rl).
20452 (builtin_function_type): Add vector shift right instructions
20453 to the unsigned argument list.
20455 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20457 rtl-optimizatoin/79286
20458 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20459 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20460 trap. PIC register plus a const unspec without offset can never trap.
20462 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20464 * tree.h (builtin_structptr_type): New type.
20465 (builtin_structptr_types): Declare new array.
20466 * tree.c (builtin_structptr_types): New array.
20467 (free_lang_data, build_common_tree_nodes): Use it.
20469 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20472 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20475 2017-06-22 Matt Turner <mattst88@gmail.com>
20477 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20478 Lake models to skylake case. Assume skylake for unknown
20479 models with clflushopt.
20481 2017-06-22 Jeff Law <law@redhat.com>
20483 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20484 frame sizes that do not satisfy aarch64_uimm12_shift.
20486 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20488 * profile-count.h (apply_probability,
20489 apply_scale, probability_in): Fix checks for zero.
20491 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20493 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20494 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20496 2016-06-22 Richard Biener <rguenther@suse.de>
20498 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20499 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20500 REDUC_MAX_EXPR support.
20501 (vectorizable_reduction): Likewise.
20502 (vect_create_epilog_for_reduction): Likewise.
20504 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20506 * match.pd (A / (1 << B) -> A >> B): New.
20507 * generic-match-head.c: Include optabs-tree.h.
20508 * gimple-match-head.c: Likewise.
20509 * optabs-tree.h (target_supports_op_p): New.
20510 * optabs-tree.c (target_supports_op_p): New.
20512 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20514 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20515 $gcc_cv_ld --help output.
20516 (gcc_cv_ld_demangle): Likewise.
20517 (gcc_cv_ld_eh_frame_hdr): Likewise.
20518 (gcc_cv_ld_pie): Likewise.
20519 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20520 (gcc_cv_ld_buildid): Likewise.
20521 (gcc_cv_ld_sysroot): Likewise.
20522 (ld_bndplt_support): Likewise.
20523 (ld_pushpopstate_support): Likewise.
20524 * configure: Regenerate.
20525 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20527 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20530 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20531 operands indexes to avoid gap between operands and match_dups.
20533 2017-06-21 Andrew Pinski <apinski@cavium.com>
20535 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20536 Increment Arith_shift and Arith_shift_reg by 1.
20537 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20539 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20540 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20541 (aarch64_strip_extend): Add new argument and test for it.
20542 (aarch64_cheap_mult_shift_p): New function.
20543 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20544 add a cost if it is true.
20545 Update calls to aarch64_strip_extend.
20546 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20548 2017-06-21 Andrew Pinski <apinski@cavium.com>
20550 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20552 (thunderxt88): Likewise.
20553 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20554 (thunderx_prefetch_tune): New variable.
20555 (thunderx2t99_prefetch_tune): Update for the correct values.
20556 (thunderxt88_tunings): New variable.
20557 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20558 generic_prefetch_tune.
20559 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20561 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20563 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20564 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20565 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20566 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20567 (aarch64_atomic_cas<mode>, GPI): Likewise.
20569 2017-06-21 Martin Liska <mliska@suse.cz>
20571 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20572 statements on cold and hot labels.
20573 * predict.c (tree_estimate_probability_bb): Remove the
20574 prediction from this place.
20576 2017-06-21 Martin Liska <mliska@suse.cz>
20578 PR tree-optimization/79489
20579 * gimplify.c (maybe_add_early_return_predict_stmt): New
20581 (gimplify_return_expr): Call the function.
20582 * predict.c (tree_estimate_probability_bb): Remove handling
20584 * predict.def: Update comment about early return predictor.
20585 * gimple-predict.h (is_gimple_predict): New function.
20586 * predict.def: Change default value of early return to 66.
20587 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20589 * passes.def: Put pass_strip_predict_hints to the beginning of
20592 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20594 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20595 FUNCTION_DECL declarations.
20596 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20598 (dwaf2out_decl): Likewise.
20599 * godump.c (go_early_global_decl): Skip call to the real debug hook
20600 for FUNCTION_DECL declarations.
20601 * passes.c (rest_of_decl_compilation): Skip call to the
20602 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20603 -fdump-go-spec is passed.
20605 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20607 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20608 Reorder for compactness.
20609 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20610 (def_builtin_pure, def_builtin_pure2): New functions.
20611 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20613 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20615 * match.pd (nop_convert): New predicate.
20616 ((A +- CST1) +- CST2): Allow some NOP conversions.
20618 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20621 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20622 with ctors/dtors if GOVD_SHARED is set.
20624 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20626 * config/aarch64/aarch64.md (movti_aarch64):
20627 Emit mov rather than orr.
20628 (movtf_aarch64): Likewise.
20629 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20630 Emit mov rather than orr.
20632 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20634 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20635 Swap alternatives, make integer dup more expensive.
20637 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20639 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20640 Return true for non-tls symbols.
20642 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20644 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20645 (cortex-a75): Likewise.
20646 (cortex-a75.cortex-a55): Likewise.
20647 * config/aarch64/aarch64-tune.md: Regenerate.
20648 * doc/invoke.texi (-mtune): Document new values for -mtune.
20650 2017-06-21 Tom de Vries <tom@codesourcery.com>
20652 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20653 stack_size feature.
20654 (Effective-Target Keywords, Other attributes): Suggest using
20655 dg-add-options stack_size feature to get stack limit in stack_size
20656 effective target documentation.
20658 2017-06-21 Julian Brown <julian@codesourcery.com>
20659 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20661 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20662 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20663 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20665 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20666 attribute type list for neon_multiply.
20667 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20668 attribute type list for neon_multiply.
20669 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20670 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20671 attribute type list for neon_multiply.
20672 * config/arm/types.md (crypto_pmull): Add.
20673 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20674 attribute type list.
20676 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20678 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20681 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20683 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20684 to make sure not to dereference a NULL cost_classes_ptr pointer.
20686 2017-06-20 Carl Love <cel@us.ibm.com>
20688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20689 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20690 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20691 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20692 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20693 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20694 VMULOSW): New enum "unspec" values.
20695 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20696 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20697 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20698 altivec_vmulosw): New patterns.
20699 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20700 VMULOSW): Add definitions.
20702 2017-06-20 Julia Koval <julia.koval@intel.com>
20704 * config/i386/i386.c: Fix rounding expand for new pattern.
20705 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20707 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20709 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20710 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20712 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20714 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20717 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20719 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20720 architecture, then by implementer ID.
20721 * config/aarch64/aarch64-tune.md: Regenerate.
20723 2017-06-20 Richard Biener <rguenther@suse.de>
20725 PR middle-end/81097
20726 * fold-const.c (split_tree): Fold to type before negating.
20728 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20730 * diagnostic-show-locus.c
20731 (selftest::test_fixit_deletion_affecting_newline): New function.
20732 (selftest::diagnostic_show_locus_c_tests): Call it.
20734 2017-06-20 Andreas Schwab <schwab@suse.de>
20737 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20740 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20742 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20743 __ARM_FEATURE_COPROC according to support.
20745 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20747 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20748 Rewritten to avoid overflow for > 32-bit pointers.
20751 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20752 by removing enum keyword.
20753 (ubsan_type_descriptor): Likewise. Formatting fix.
20756 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20757 splitter): Require TARGET_SSE2 in the condition.
20759 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20762 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20763 for doing vector set of SFmode on ISA 3.0.
20764 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20765 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20767 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20768 SFmode value into a V4SF variable that was extracted from another
20769 V4SF variable without converting the element to double precision
20770 and back to single precision vector format.
20771 (vsx_insert_extract_v4sf_p9_2): Likewise.
20773 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20775 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20776 in UWHI to avoid undefined overflow.
20779 * ubsan.h (enum ubsan_encode_value_phase): New.
20780 (ubsan_encode_value): Change second argument to
20781 enum ubsan_encode_value_phase with default value of
20782 UBSAN_ENCODE_VALUE_GENERIC.
20783 * ubsan.c (ubsan_encode_value): Change second argument to
20784 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20785 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20786 create_tmp_var_raw instead of create_tmp_var and use a
20788 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20789 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20790 ubsan_encode_value callers.
20793 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20794 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20795 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20797 2017-06-19 Richard Biener <rguenther@suse.de>
20799 PR middle-end/81118
20800 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20801 estimates if we changed anything.
20803 2017-06-19 Richard Biener <rguenther@suse.de>
20805 PR tree-optimization/80887
20806 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20807 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20808 simplified lookups, then reset mprts_hook.
20809 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20811 (try_to_simplify): Likewise.
20813 2017-06-19 Martin Liska <mliska@suse.cz>
20816 * gimplify.c (gimplify_switch_expr):
20817 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20819 2017-06-19 Martin Liska <mliska@suse.cz>
20821 * doc/install.texi: Document that PGO runs in 4 stages.
20823 2017-06-19 Martin Liska <mliska@suse.cz>
20826 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20827 to dispatcher function name.
20828 * multiple_target.c (replace_function_decl): New function.
20829 (create_dispatcher_calls): Redirect both edges and references.
20831 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20833 * profile-count.c (profile_count::dump): Dump quality.
20834 (profile_count::differs_from_p): Update for unsigned val.
20835 * profile-count.h (profile_count_quality): New enum.
20836 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20838 2017-06-19 Richard Biener <rguenther@suse.de>
20840 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20841 struct function as arg.
20842 (estimate_numbers_of_iterations): Export overload with loop arg.
20843 (free_numbers_of_iterations_estimates_loop): Use an overload of
20844 free_numbers_of_iterations_estimates instead.
20845 * tree-cfg.c (remove_bb): Adjust.
20846 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20847 * tree-parloops.c (gen_parallel_loop): Likewise.
20848 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20850 (tree_unroll_loops_completely): Likewise.
20851 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20852 Use an overload instead and export.
20853 (estimated_loop_iterations): Adjust.
20854 (max_loop_iterations): Likewise.
20855 (likely_max_loop_iterations): Likewise.
20856 (estimate_numbers_of_iterations): Take struct function as arg
20858 (loop_exits_before_overflow): Adjust.
20859 (free_numbers_of_iterations_estimates_loop): Use an overload.
20860 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20861 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20863 2017-06-19 Richard Biener <rguenther@suse.de>
20866 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20867 RANGE_EXPR conservatively.
20869 2017-06-16 Carl Love <cel@us.ibm.com>
20871 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20872 definitions for vec_float, vec_float2, vec_floato,
20873 vec_floate built-ins.
20874 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20875 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20877 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20878 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20879 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20880 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20881 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20882 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20883 vec_floato): Add builtin defines.
20884 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20885 Update the built-in documentation file for the new built-in
20888 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20890 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20891 (mthumb): Mark as the negative of -marm.
20893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20895 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20897 (ARM Options, -mtune): Document that this accepts the same
20898 extension options as -mcpu.
20899 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20901 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20903 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20904 permitted extensions.
20906 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20908 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20909 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20910 (armv8-m.main): Add option +nodsp.
20911 * config/arm/arm-cpu-cdata.h: Regenerated.
20913 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20915 * config/arm/t-fuchsia: New file.
20916 * config.gcc (arm*-*-fuchsia*): Use it.
20918 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20920 * config/arm/t-symbian: Rewrite for new option infrastructure.
20922 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20924 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20925 (MULTILIB_REQUIRED): Likewise.
20927 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20929 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20930 (MULTILIB_RESUE): Likewise.
20931 (MULTILIB_MATCHES): Likewise.
20932 (MULTLIB_REQUIRED): Likewise.
20934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20936 * config/arm/t-rtems: Rewrite for new option framework.
20938 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20940 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20941 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20942 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20943 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20944 * config/arm/t-multilib: ... here.
20945 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20946 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20947 armv7-a and armv8*-a when A-profile libraries have not been built.
20948 * config/arm/t-rmprofile: Rewrite.
20950 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20952 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20953 with a backslash. Remove the backslash after substituting unescaped
20955 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20959 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20960 the driver to pass through the default setting of -mfloat-abi.
20961 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20963 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20964 (all_feat_combs): New rule.
20965 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20967 * config/arm/t-aprofile: Rewrite.
20969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20971 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20972 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20973 fpu is not specified by the user/command-line.
20974 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20975 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20976 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20977 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20978 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20979 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20981 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20983 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20984 * config/arm/t-arm-elf: Rewritten.
20986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20988 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20989 have some floating-point instructions.
20990 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20991 (TARGET_MAYBE_HARD_FLOAT): New macro.
20992 * config/arm/arm-builtins.c (arm_init_builtins): Use
20993 TARGET_MAYBE_HARD_FLOAT.
20994 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20998 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20999 (configargs.h): Include it.
21000 (arm_print_hint_for_fpu_option): New function.
21001 (arm_parse_fpu_option): New function.
21002 (candidate_extension): New class.
21003 (arm_canon_for_multilib): New function.
21004 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21005 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21006 (ARCH_CANONICAL_SPECS): New macro.
21007 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21009 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21011 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21012 are set after handling multilib fragments. Set target_cpu_default2
21015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21017 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21019 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21020 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21022 (check_arch): Likewise.
21023 * config/arm/arm.c (arm_configure_build_target): Handle
21024 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21025 options in the default.
21027 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21029 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21030 when an option is an alias of another.
21031 * config/arm/parsecpu.awk (optalias): New parser token.
21032 (gen_comm_data): Mark non-alias options as such. Emit entries
21033 for extension aliases.
21034 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21035 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21036 (armv6kz, armv6zk, armv6t2): Likewise.
21037 (armv7): Make vfpv3-d16 an alias.
21038 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21040 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21041 Sort in canonical order.
21042 (armv8-a): Sort in canonical order.
21043 (armv8.1-a, armv8.2-a): Likewise.
21044 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21046 (cortex-a9): Sort in canonical order.
21047 * config/arm/arm.c (selftests.h): Include it.
21048 (arm_test_cpu_arch_data): New function.
21049 (arm_run_self_tests): New function.
21050 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21051 (targetm): Move declaration to the end of the file.
21052 * arm-cpu-cdata.h: Regenerated.
21054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21056 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21057 call to target_mode_check describing the type of option passed.
21058 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21059 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21060 arm_parse_cpu_option_name to match parameters against list of
21062 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21063 arm_arch_core_flags data structure.
21064 * config/arm/arm-cpu_cdata.h: Regenerated.
21066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21068 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21070 (arm_print_hint_for_cpu_option): Likewise.
21071 (arm_print_hint_for_arch_option): Likewise.
21072 (arm_parse_cpu_option_name): Likewise.
21073 (arm_parse_arch_option_name): Likewise.
21074 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21075 of entries in the all_fpus list.
21076 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21077 (arm_parse_cpu_option_name): Declare.
21078 (arm_parse_arch_option_name): Declare.
21079 (arm_parse_option_features): Declare.
21080 (arm_intialize_isa): Declare.
21081 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21083 (gen_comm_data): ... here. Make definitions non-static.
21084 * config/arm/arm-cpu-data.h: Regenerated.
21085 * config/arm/arm-cpu-cdata.h: Regenerated.
21087 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21089 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21090 (cpu_arch_extension): New structure.
21091 (cpu_arch_option, arch_option, cpu_option): New structures.
21092 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21093 architecture types.
21094 (gen_data): Generate new format data tables.
21095 * config/arm/arm.c (cpu_tune): New structure.
21096 (cpu_option, processors): Delete.
21097 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21098 (arm_print_hint_for_cpu_option): ... this and ...
21099 (arm_print_hint_for_arch_option): ... this.
21100 (arm_parse_arch_cpu_name): Delete. Replace with ...
21101 (arm_parse_cpu_option_name): ... this and ...
21102 (arm_parse_arch_option_name): ... this.
21103 (arm_unrecognized_feature): Change type of target parameter to
21105 (arm_parse_arch_cpu_features): Delete. Replace with ...
21106 (arm_parse_option_features): ... this.
21107 (arm_configure_build_target): Rework to use new configuration data
21109 (arm_print_tune_info): Rework for new configuration data tables.
21110 * config/arm/arm-cpu-data.h: Regenerated.
21111 * config/arm/arm-cpu.h: Regenerated.
21113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21115 * Makefile.in (OBJS): Move sbitmap.o from here ...
21116 (OBJS-libcommon): ... to here.
21118 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21120 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21121 (ISA_ALL_CRYPTO): New macro.
21122 (ISA_ALL_SIMD): New macro
21123 (ISA_ALL_FP): New macro.
21124 * config/arm/arm.c (fpu_bitlist): Update initializer.
21125 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21127 (arm9e): Add fpu. Add option for nofp
21128 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21129 (arm926ej-s, arm1026ej-s): Likewise.
21130 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21131 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21132 neon-fp16, neon-vfpv4, nofp and nosimd.
21133 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21134 (cortex-a8): Add fpu. Add option for nofp.
21135 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21136 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21137 (cortex-r4f): Add fpu.
21138 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21139 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21141 (cortex-r8): Likewise.
21142 (cortex-m4): Add fpu. Add option for nofp.
21143 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21144 (cortex-a17.cortex-a7): Likewise.
21145 (cortex-a32): Add fpu. Add options for crypto and nofp.
21146 (cortex-a35, cortex-a53): Likewise.
21147 (cortex-a57): Add fpu. Add option for crypto.
21148 (cortex-a72, cortex-a73): Likewise.
21149 (exynos-m1): Likewise.
21150 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21151 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21152 (cortex-m33): Add fpu. Add option for nofp.
21153 * config/arm/arm-cpu-cdata.h: Regenerated
21154 * config/arm/arm-cpu-data.h: Regenerated.
21156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21158 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21159 (armv5te, armv5tej): Likewise.
21160 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21161 (armv7): Add options fp and vfpv3-d16.
21162 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21163 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21165 (armv7ve): Likewise.
21166 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21167 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21168 (armv8-a): Add nocrypto option.
21169 (armv8.1-a, armv8.2-a): Likewise.
21170 (armv8-m.main): add options fp, fp.dp and nofp.
21172 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21174 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21176 (armv8-a+crc): Delete.
21177 (armv8.1-a): Add options simd, crypto and nofp.
21178 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21179 (armv8.2-a+fp16): Delete.
21180 (armv8-m.main): Add option dsp.
21181 (armv8-m.main+dsp): Delete.
21182 (cortex-a8): Add fpu. Add nofp option.
21183 (cortex-a9): Add fpu. Add nofp and nosimd options.
21184 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21185 link to main cpu and architecture data structures.
21186 (gen_comm_data): Only put isa attributes from the main architecture
21188 (option): New statement for architecture and CPU entries.
21189 * arm.c (struct cpu_option): New structure.
21190 (struct processors): Add entry for options.
21191 (arm_unrecognized_feature): New function.
21192 (arm_parse_arch_cpu_name): Ignore any characters after the first
21194 (arm_parse_arch_cpu_feature): New function.
21195 (arm_configure_build_target): Separate out any CPU and architecture
21196 features and parse separately. Don't error out if -mfpu=auto is
21197 used with only an architecture string.
21198 (arm_print_asm_arch_directives): New function.
21199 (arm_file_start): Call it.
21200 * config/arm/arm-cpu-cdata.h: Regenerated.
21201 * config/arm/arm-cpu-data.h: Likewise.
21202 * config/arm/arm-tables.opt: Likewise.
21204 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21206 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21207 assembler when it is not -mfpu=auto.
21209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21211 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21212 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21213 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21214 (ASM_CPU_SPEC): Rewrite.
21215 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21216 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21217 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21218 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21219 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21220 copied string is NUL-terminated. Also strip any characters prefixed
21222 (arm_rewrite_selected_arch): New function.
21223 (arm_rewrite_march): New function.
21225 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21227 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21228 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21229 (march, mcpu, mtune): Convert to string-based options.
21230 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21231 (arm_parse_arch_cpu_name): New function.
21232 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21233 identify selected architecture or CPU.
21234 (arm_option_save): New function.
21235 (TARGET_OPTION_SAVE): Redefine.
21236 (arm_option_restore): Restore string options.
21237 (arm_option_print): Print string options.
21239 2017-06-16 Martin Sebor <msebor@redhat.com>
21241 PR tree-optimization/80933
21242 PR tree-optimization/80934
21243 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21244 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21245 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21246 (gimple_fold_builtin): Call them.
21248 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21250 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21251 as unlikely; update profile.
21253 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21255 * predict.c (force_edge_cold): Handle declaring edges impossible
21258 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21260 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21262 (try_unroll_loop_completely): Fix reporting.
21264 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21266 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21268 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21271 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21272 if given a non-constant argument for an intrinsic which requires a
21275 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21277 * profile.c (compare_freqs): New function.
21278 (branch_prob): Sort edge list.
21279 (find_spanning_tree): Assume that the list is priority sorted.
21281 2017-06-16 Richard Biener <rguenther@suse.de>
21283 PR tree-optimization/81090
21284 * passes.def (pass_record_bounds): Remove.
21285 * tree-pass.h (make_pass_record_bounds): Likewise.
21286 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21287 make_pass_record_bounds): Likewise.
21288 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21289 not free niter estimates at the beginning but at the end.
21290 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21292 2017-06-16 Richard Biener <rguenther@suse.de>
21294 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21295 initializer to workaround ICE in host GCC 4.8.
21297 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21299 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21301 (clone_inlined_nodes): Update.
21303 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21305 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21306 prefetch settings, and enable prefetching by default at -O3.
21308 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21310 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21311 Set flag_prefetch_loop_arrays according to tuning data.
21313 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21315 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21316 New tune structure.
21317 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21318 [Unrelated to main purpose of the patch] Place the pointer field last
21319 to enable type checking errors when tune structure are wrongly merged.
21320 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21321 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21322 (thunderx2t99_prefetch_tune): New tune constants.
21323 (tune_params *_tunings): Update all tunings (no functional change).
21324 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21325 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21326 from tunings structures.
21328 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21331 * ubsan.c (instrument_null): Add T argument, use it instead
21332 of computing it based on IS_LHS.
21333 (instrument_object_size): Likewise.
21334 (pass_ubsan::execute): Adjust instrument_null and
21335 instrument_object_size callers to pass gimple_get_lhs or
21336 gimple_assign_rhs1 result to it. Use instrument_null instead of
21337 calling get_base_address and instrument_mem_ref. Handle
21338 aggregate call arguments for object-size sanitization.
21340 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21342 PR tree-optimization/81089
21343 * tree-vrp.c (is_masked_range_test): Validate operands of
21346 2017-06-15 Martin Sebor <msebor@redhat.com>
21349 * dumpfile.c (dump_register): Avoid calling memset to initialize
21350 a class with a default ctor.
21351 * gcc.c (struct compiler): Remove const qualification.
21352 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21353 * hash-table.h: Ditto.
21354 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21356 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21357 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21359 * params.h (struct param_info): Make struct members non-const.
21360 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21361 with default initialization.
21362 * vec.h (vec_copy_construct, vec_default_construct): New helper
21364 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21365 with vec_copy_construct.
21366 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21367 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21368 * doc/invoke.texi (-Wclass-memaccess): Document.
21370 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21372 * emit-rtl.h (is_leaf): Update comment about local
21373 register allocator.
21375 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21378 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21379 for a variable to have a section before checking if the section has a
21381 Set section to.persistent if persistent attribute is set.
21382 Warn if .persistent attribute is used on an automatic variable.
21384 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21386 PR rtl-optimization/80474
21387 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21389 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21391 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21394 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21396 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21397 (casesi): Emit bounds checking as RTL.
21398 (casesi_internal_mips16_<mode>): Remove bounds checking.
21400 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21402 * config/xtensa/xtensa.c (xtensa_option_override): Append
21403 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21404 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21405 xtensa_doloop_hooks): Define unconditionally.
21406 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21408 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21409 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21410 for it in xtensa_option_override.
21411 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21412 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21414 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21416 * doc/cppopts.texi: Document '-' special value to -MF.
21418 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21420 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21421 (cortex_a53_fconst): Likewise.
21422 (cortex_a53_fpmul): Likewise.
21423 (cortex_a53_f_load_64): Likewise.
21424 (cortex_a53_f_load_many): Likewise.
21425 (cortex_a53_advsimd_alu): Likewise.
21426 (cortex_a53_advsimd_alu_q): Likewise.
21427 (cortex_a53_advsimd_mul): Likewise.
21428 (cortex_a53_advsimd_mul_q): Likewise.
21429 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21430 Add missing fmul, r2f_cvt and fconst cases.
21432 2017-06-14 Richard Biener <rguenther@suse.de>
21434 PR middle-end/81088
21435 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21437 (fold_binary_loc): When associating do not treat pre-existing
21438 TREE_OVERFLOW on literal constants as a reason to allow
21439 TREE_OVERFLOW on associated literal constants.
21441 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21443 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21444 (MASK_FEATURES): New macro.
21445 * config/sparc/sparc.c (sparc_option_override): Remove the special
21446 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21448 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21450 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21451 a division of 0 if non-call exceptions are enabled.
21453 2017-06-14 Andrew Pinski <apinski@cavium.com>
21454 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21457 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21458 Improve vector initialization code gen for only variable case.
21460 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21462 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21464 2017-06-14 Richard Biener <rguenther@suse.de>
21466 PR tree-optimization/81083
21467 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21470 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21472 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21473 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21474 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21475 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21476 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21477 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21479 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21481 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21482 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21484 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21486 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21488 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21490 * config/rs6000/t-rtems: Don't handle SPE.
21492 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21494 * config/rs6000/t-linux: Don't handle SPE.
21496 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21498 * config/rs6000/eabispe.h: Delete file.
21500 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21502 * config/rs6000/t-spe: Delete file.
21504 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21506 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21507 (rs6000_legitimate_offset_address_p): Return false for anything in
21508 V2SImode or V2SFmode.
21510 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21512 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21513 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21514 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21516 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21517 (rs6000_legitimate_offset_address_p): Ditto.
21518 (rs6000_emit_move): Ditto.
21519 (rs6000_init_builtins): Remove V4HI_type_node.
21521 2017-06-13 Martin Liska <mliska@suse.cz>
21524 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21525 (gate_asan): Likewise.
21526 * asan.h (asan_no_sanitize_address_p): Remove the function.
21527 (sanitize_flags_p): New function.
21528 * builtins.def: Fix coding style.
21529 * common.opt: Use renamed enum value.
21530 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21531 * doc/extend.texi: Document no_sanitize attribute.
21532 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21533 to SANITIZE_UNDEFINED_NONDEFAULT.
21534 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21535 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21536 Use sanitize_flags_p.
21537 * gimplify.c (gimplify_function_tree): Likewise.
21538 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21539 * opts.c (parse_no_sanitize_attribute): New function.
21540 (common_handle_option): Use renamed enum value.
21541 * opts.h (parse_no_sanitize_attribute): Declare.
21542 * tree.c (sanitize_flags_p): New function.
21543 * tree.h: Declared here.
21544 * tsan.c: Use sanitize_flags_p.
21545 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21546 (instrument_mem_ref): Likewise.
21547 (instrument_bool_enum_load): Likewise.
21548 (do_ubsan_in_current_function): Remove the function.
21549 (pass_ubsan::execute): Use sanitize_flags_p.
21550 * ubsan.h: Remove do_ubsan_in_current_function
21551 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21552 (dump_function_to_file): Use it here.
21554 2017-06-13 Martin Jambor <mjambor@suse.cz>
21556 PR tree-optimization/80803
21557 PR tree-optimization/81063
21558 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21559 (propagate_subaccesses_across_link): Enqueue subtree whenever
21560 necessary instead of relying on the caller.
21562 2017-06-13 Martin Jambor <mjambor@suse.cz>
21564 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21565 that have a first_link.
21566 (sort_and_splice_var_accesses): Do not check first_link before
21568 (subtree_mark_written_and_enqueue): Likewise.
21569 (propagate_all_subaccesses): Likewise and do not stop at first
21570 parent with a first_link.
21572 2017-06-13 Martin Jambor <mjambor@suse.cz>
21574 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21577 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21579 * match.pd: New pattern.
21581 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21583 * tree-vrp.c (is_masked_range_test): New function.
21584 (register_edge_assert_for): Determine ranges for
21587 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21589 PR tree-optimization/67328
21590 * fold-const.c (maskable_range_p): New function.
21591 (build_range_check): Generate bittests if possible.
21593 2017-06-13 Martin Liska <mliska@suse.cz>
21595 * gimple-pretty-print.c (dump_probability): Add new argument.
21596 (dump_edge_probability): Dump both probability and count.
21597 (dump_gimple_label): Likewise.
21598 (dump_gimple_bb_header): Likewise.
21600 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21603 * config/avr/avr-devices.c: Fix indentation.
21604 * config/avr/gen-avr-mmcu-specs.c: Dito.
21606 2017-06-13 Richard Biener <rguenther@suse.de>
21608 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21609 instead get vector type from stmt_info.
21610 (vectorizable_reduction): Adjust. Remove dead code.
21612 2017-06-13 Richard Biener <rguenther@suse.de>
21614 PR middle-end/81065
21615 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21616 case of C * (x * C2 + C3).
21617 (fold_addr_of_array_ref_difference): Properly fold index difference.
21619 2017-06-12 David S. Miller <davem@davemloft.net>
21622 * config/sparc/sparc.md (return expander): Emit frame blockage if
21623 function uses alloca.
21625 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21627 * combine.c (make_field_assignment): Check len rather than the mode
21628 precision when calling force_to_mode.
21630 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21632 Support multilibs and devices that see flash in RAM address range.
21635 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21636 (avr_mcu_t) <flash_pm_offset>: New field.
21637 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21638 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21639 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21640 (AVR_TINY_PM_OFFSET): Remove macro.
21641 * config/avr/avr.opt (-mshort-calls): New option.
21642 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21643 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21644 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21645 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21646 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21647 instead of avr_arch->have_jmp_call.
21648 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21649 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21650 avr_arch->flash_pm_offset to define.
21651 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21652 new field flash_pm_offset. Add entry for avrxmega3.
21653 (avr_texinfo): Add entry for avrxmega3.
21654 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21655 attiny212, attiny214,
21656 attiny412, attiny414, attiny416, attiny417,
21657 attiny814, attiny816, attiny817,
21658 attiny1614, attiny1616, attiny1617,
21659 attiny3214, attiny3216, attiny3217.
21660 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21661 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21662 (avr_print_operand_address) [AVR_TINY]: Same.
21663 (avr_asm_init_sections) <readonly_data_section>: Only patch
21664 callback if avr_arch->flash_pm_offset = 0.
21665 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21666 for rodata if avr_arch->flash_pm_offset != 0.
21667 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21668 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21669 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21670 (m_options): Append opt_rcall.
21671 (m_dirnames): Append dir_rcall.
21672 * config/avr/t-multilib: Regenerate.
21674 * configure.ac [target=avr]: Check whether avrxmega3 default
21675 linker description file works as needed.
21676 * configure: Regenerate.
21677 * doc/avr-mmcu.texi: Regenerate.
21678 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21679 <__AVR_ARCH__>: Document avrxmega3 and 103.
21680 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21681 <__AVR_SHORT_CALLS__>: Document it.
21682 <__AVR_PM_BASE_ADDRESS__>: Document it.
21683 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21684 (AVR Variable Attributes) <progmem>: Document this is
21685 not needed for avrxmega3.
21686 (AVR Named Address Spaces) <__flash>: Dito.
21688 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21690 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21692 2017-06-12 Doug Rupp <rupp@adacore.com>
21694 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21695 Append vxworks-stdint.h to the tm_file list.
21696 * config/vxworks-stdint.h: New file.
21698 2017-06-12 Martin Liska <mliska@suse.cz>
21700 PR tree-optimization/81041
21701 * tree-profile.c (gimple_gen_ic_func_profiler):
21702 Create an extra BB in profile-generate
21703 (gimple_gen_time_profiler): Likewise.
21705 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21707 PR tree-optimization/81003
21708 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21709 (update_range_test): Use it instead of force_gimple_operand_gsi.
21711 2017-06-12 Richard Biener <rguenther@suse.de>
21713 PR tree-optimization/81053
21714 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21715 with backedge value not defined in loop. Simplify def stmt
21718 2017-06-11 Tom de Vries <tom@codesourcery.com>
21721 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21723 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21724 nvptx_cannot_force_const_mem.
21726 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21728 * opts.c (finish_options): Move test for flag_split_stack after
21729 it has been initialized.
21731 2017-06-11 Jason Merrill <jason@redhat.com>
21733 * tree.h (id_equal): New.
21734 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21735 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21736 instead of strcmp of IDENTIFIER_POINTER.
21738 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21740 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21741 (mark_all_inlined_calls_cdtor): Fix formating.
21742 (inline_transform): Rescale profile before inlining.
21744 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21746 * cgraph.h (cgraph_edge::clone): Update prototype.
21747 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21748 (cgraph_node::create_clone): Update.
21749 (cgraph_node::create_version_clone): Update.
21750 * tree-inline.c (copy_bb): Update.
21751 (expand_call_inline): Update.
21753 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21755 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21756 factored out from ...
21757 (rs6000_emit_prologue): ... here.
21759 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21761 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21762 factored out from ...
21763 (rs6000_emit_prologue): ... here.
21765 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21767 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21769 (handle_missing_profiles): Fix computation of tp_first_run.
21770 (counts_to_freqs): Do not touch freqs when count is 0.
21772 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21774 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21777 2017-06-10 Tom de Vries <tom@codesourcery.com>
21779 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21780 attributes): Document signal effective target.
21782 2017-06-10 Tom de Vries <tom@codesourcery.com>
21784 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21785 Document effective target stack_size.
21787 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21789 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21790 to the edit_context if they can be auto-applied.
21792 2017-06-9 Ian Lance Taylor <iant@golang.org>
21794 * opts.c (finish_options): If -fsplit-stack, disable implicit
21795 -forder-blocks-and-partition.
21796 * doc/invoke.texi (Optimize Options): Document that when using
21797 -fsplit-stack -forder-blocks-and-partition is not implicitly
21800 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21802 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21803 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21804 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21805 * builtins.def (abort, trap, unreachable): Declare cold.
21806 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21807 * tree-core.h (ECF_COLD): New.
21808 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21809 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21811 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21813 * predict.c (unlikely_executed_stmt_p): Cleanup.
21815 2017-06-09 Richard Biener <rguenther@suse.de>
21817 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21818 model if the ref is always written to.
21820 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21822 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21824 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21826 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21829 2017-06-09 Tom de Vries <tom@codesourcery.com>
21832 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21833 "target cannot support label values" when encountering LABEL_REF.
21835 2017-06-09 Martin Liska <mliska@suse.cz>
21837 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21838 (gimple_gen_ic_func_profiler): Emit direct comparison
21839 of __gcov_indirect_call_callee with NULL.
21840 (gimple_gen_time_profiler): Change probability from
21841 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21843 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21845 * profile.c (edge_gcov_counts): Turn to pointer.
21846 (compute_branch_probabilities, compute_branch_probabilities): Update.
21847 (branch_prob): Do not clear edge_gcov_count.
21848 * profile.h (edge_gcov_counts): Turn to pointer.
21849 (edge_gcov_count): Update.
21851 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21853 * gimple.h (gimple_check_failed): Mark cold.
21855 2017-06-09 Richard Biener <rguenther@suse.de>
21857 PR tree-optimization/66623
21858 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21859 refactor check_reduction into two parts, properly computing
21860 whether we have to check reduction validity for outer loop
21863 2017-06-09 Richard Biener <rguenther@suse.de>
21865 PR tree-optimization/79483
21866 * graphite-scop-detection.c (order): New global.
21867 (get_order): Compute bb to order mapping that satisfies code
21868 generation constraints.
21869 (cmp_pbbs): New helper.
21870 (build_scops): Start domwalk at entry block, sort generated
21873 2017-06-09 Richard Biener <rguenther@suse.de>
21875 PR middle-end/81007
21876 * ipa-polymorphic-call.c
21877 (ipa_polymorphic_call_context::restrict_to_inner_class):
21878 Skip FIELD_DECLs with error_mark_node type.
21879 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21882 2017-06-09 Martin Liska <mliska@suse.cz>
21884 * predict.c (struct branch_predictor): New struct.
21885 (test_prediction_value_range): New test.
21886 (predict_c_tests): New function.
21887 * selftest-run-tests.c (selftest::run_tests): Run the function.
21888 * selftest.h: Declare new tests.
21890 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21893 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21894 gen_add3_insn did not fail.
21895 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21896 r0, construct that number in a temporary reg and add that reg to r0.
21897 If asked to put the result in r0 as well, fail.
21899 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21901 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21902 for early expansion of vec_eqv.
21904 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21906 PR middle-end/81005
21907 * ubsan.c (instrument_null): Avoid pointless code temporary.
21908 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21910 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21914 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21917 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21920 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21922 * predict.c (unlikely_executed_edge_p): Move ahead.
21923 (probably_never_executed_edge_p): Use it.
21925 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21927 PR middle-end/79988
21928 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21929 gimple_call_builtin_p call.
21931 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21933 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21934 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21935 rtl_check_failed_type2, rtl_check_failed_code1,
21936 rtl_check_failed_code2, rtl_check_failed_code_mode,
21937 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21938 rtvec_check_failed_bounds, rtl_check_failed_flag,
21939 _fatal_insn_not_found, _fatal_insn): Likewise.
21940 * tree.h (tree_contains_struct_check_failed,
21941 tree_check_failed, tree_not_check_failed,
21942 tree_class_check_failed, tree_range_check_failed,
21943 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21944 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21945 tree_operand_check_failed, omp_clause_check_failed,
21946 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21949 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21951 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21952 flag_branch_probabilities.
21953 * ipa-inline.c (edge_badness): Likewise.
21954 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21955 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21956 * predict.c (maybe_hot_frequency_p): Likewise.
21957 (probably_never_executed): Likewise.
21958 * sched-ebb.c (schedule_ebbs): Likewise.
21959 * sched-rgn.c (find_single_block_region): Likewise.
21960 * tracer.c (tail_duplicate): Likewise.
21962 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21964 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21965 longer requires x_flag_profile_use.
21967 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21969 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21970 instead of flag_reorder_blocks_and_partition.
21971 * dbxout.c (dbxout_function_end): Likewise.
21972 * dwarf2out.c (gen_subprogram_die): Likewise.
21973 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21974 * hw-doloop.c (reorg_loops): Likewise.
21975 * varasm.c (assemble_start_function,
21976 assemble_end_function): Likewise.
21977 (decide_function_section): Do not check for
21978 flag_reorder_blocks_and_partition.
21980 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21982 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21984 (chkp_get_hard_register_fake_addr_expr): Ditto.
21985 (chkp_build_addr_expr): Add check for hard reg case.
21986 (chkp_parse_array_and_component_ref): Ditto.
21987 (chkp_find_bounds_1): Ditto.
21988 (chkp_process_stmt): Don't generate bounds store for
21991 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21993 * predict.c (maybe_hot_bb_p): Do not check profile status.
21994 (maybe_hot_edge_p): Likewise.
21995 (probably_never_executed): Check for zero counts even if profile
21997 (unlikely_executed_edge_p): New function.
21998 (unlikely_executed_stmt_p): New function.
21999 (unlikely_executed_bb_p): New function.
22000 (set_even_probabilities): Use unlikely predicates.
22001 (combine_predictions_for_bb): Likewise.
22002 (predict_paths_for_bb): Likewise.
22003 (predict_paths_leading_to_edge): Likewise.
22004 (determine_unlikely_bbs): New function.
22005 (estimate_bb_frequencies): Use it.
22006 (compute_function_frequency): Use zero counts even if profile is
22008 * profile-count.h: Fix typo.
22010 2017-08-08 Julia Koval <julia.koval@intel.com>
22012 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22013 _mm512_mask_cvtsepi16_storeu_epi8,
22014 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22015 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22016 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22017 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22018 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22019 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22020 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22021 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22022 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22023 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22024 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22025 __builtin_ia32_pmovuswb256mem_mask,
22026 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22027 __builtin_ia32_pmovwb512mem_mask): New builtins.
22029 2017-08-08 Julia Koval <julia.koval@intel.com>
22031 PR target/73350,80862
22032 * config/i386/subst.md (round): Fix round pattern.
22033 * config/i386/i386.c (ix86_erase_embedded_rounding):
22034 Fix erasing rounding for the fixed pattern.
22036 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22038 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22040 2017-06-08 Martin Liska <mliska@suse.cz>
22042 PR gcov-profile/80911
22043 * gcov.c (block_info::block_info): New constructor.
22045 2017-06-07 Carl Love <cel@us.ibm.com>
22047 * config/rs6000/rs6000-c: The return type of the following
22048 built-in functions was implemented as int not long long. Fix sign
22049 of return value for the unsigned version of vec_mulo and vec_mule.
22050 vector unsigned long long vec_bperm (vector unsigned long long,
22051 vector unsigned char)
22052 vector signed long long vec_mule (vector signed int,
22054 vector unsigned long long vec_mule (vector unsigned int,
22055 vector unsigned int)
22056 vector signed long long vec_mulo (vector signed int,
22058 vector unsigned long long vec_mulo (vector unsigned int,
22059 vector unsigned int)
22060 * doc/extend.texi: Fix the documentation for the built-in
22063 2017-06-07 Carl Love <cel@us.ibm.com>
22066 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22069 2017-06-07 Carl Love <cel@us.ibm.com>
22071 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22072 support, Generate doublehv for signed int/float for BE case only.
22074 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22076 * doc/invoke.texi (mcx16): Rewrite.
22078 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22080 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22081 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22082 *mov<mode>_softfloat, and an anonymous splitter): Use
22083 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22085 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22087 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22089 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22090 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22091 (rs6000_debug_reg_global): Adjust.
22092 (rs6000_init_hard_regno_mode_ok): Adjust.
22093 (rs6000_dbx_register_number): Adjust.
22094 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22095 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22096 Remove SPE_ACC and SPEFSCR.
22097 (REG_ALLOC_ORDER): Ditto.
22098 (FRAME_POINTER_REGNUM): Change to 111.
22099 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22100 (REG_CLASS_NAMES): Ditto.
22101 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22102 (REGISTER_NAMES): Ditto.
22103 (ADDITIONAL_REG_NAMES): Ditto.
22104 (rs6000_reg_names): Ditto.
22105 * config/rs6000/rs6000.md: Renumber some register number
22108 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22110 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22112 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22113 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22115 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22116 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22117 Delete the SPE high registers.
22118 (REG_ALLOC_ORDER): Ditto.
22119 (enum reg_class): Remove SPE_HIGH_REGS.
22120 (REG_CLASS_NAMES): Ditto.
22121 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22122 (REGISTER_NAMES): Ditto.
22123 (rs6000_reg_names): Ditto.
22124 * doc/tm.texi.in: Remove SPE as example.
22125 * doc/tm.texi: Regenerate.
22127 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22129 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22130 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22131 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22132 * config/rs6000/rs6000.md (type): Remove "brinc".
22134 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22136 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22137 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22138 * config/rs6000/linuxspe.h: Delete file.
22139 * config/rs6000/rs6000.md: Don't include spe.md.
22140 * config/rs6000/spe.h: Delete file.
22141 * config/rs6000/spe.md: Delete file.
22142 * config/rs6000/t-rs6000: Remove spe.md.
22144 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22146 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22147 (reg_or_none500mem_operand): Delete.
22148 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22149 instead of reg_or_none500mem_operand.
22151 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22153 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22154 handling of SPE flags.
22155 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22157 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22159 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22161 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22162 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22163 paired_divv2sf3): Similar.
22164 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22165 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22166 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22168 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22169 Rename the paired_* instruction patterns.
22170 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22172 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22173 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22174 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22175 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22176 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22177 PAIRED_VECTOR_MODE.
22178 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22179 (spe_func_has_64bit_regs_p): Delete.
22180 (spe_expand_predicate_builtin): Delete.
22181 (spe_expand_evsel_builtin): Delete.
22182 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22183 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22184 (invalid_e500_subreg): Delete.
22185 (rs6000_legitimize_address): Always force_reg op2 as well, for
22186 paired single memory accesses.
22187 (rs6000_member_type_forces_blk): Delete.
22188 (rs6000_spe_function_arg): Delete.
22189 (rs6000_expand_unop_builtin): Delete SPE handling.
22190 (rs6000_expand_binop_builtin): Ditto.
22191 (spe_expand_stv_builtin): Delete.
22192 (bdesc_2arg_spe): Delete.
22193 (spe_expand_builtin): Delete.
22194 (spe_expand_predicate_builtin): Delete.
22195 (spe_expand_evsel_builtin): Delete.
22196 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22197 (spe_init_builtins): Delete.
22198 (spe_func_has_64bit_regs_p): Delete.
22199 (savres_routine_name): Delete "info" parameter. Adjust callers.
22200 (rs6000_emit_stack_reset): Ditto.
22201 (rs6000_dwarf_register_span): Delete.
22202 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22203 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22204 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22206 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22208 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22209 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22210 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22211 mulv2sf3, divv2sf3): Delete expanders.
22213 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22215 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22217 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22219 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22220 * config/rs6000/rs6000.c: Ditto.
22222 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22224 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22225 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22226 comparison_operator.
22228 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22230 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22231 * config/rs6000/rs6000.opt: Ditto.
22232 * config/rs6000/t-rtems: Ditto.
22234 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22236 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22237 TARGET_E500_SINGLE by 0, simplify.
22238 * config/rs6000/rs6000.c: Ditto.
22239 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22240 (spe_build_register_parallel): Delete.
22241 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22242 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22243 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22244 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22245 (E500_CONVERT): Delete.
22246 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22248 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22250 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22251 * config/rs6000/dfp.md: Ditto.
22252 (negdd2, *negdd2_fpr): Merge.
22253 (absdd2, *absdd2_fpr): Merge.
22254 (negtd2, *negtd2_fpr): Merge.
22255 (abstd2, *abstd2_fpr): Merge.
22256 * config/rs6000/e500.h: Delete file.
22257 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22258 TARGET_FPRS by 1 and simplify.
22259 * config/rs6000/rs6000-c.c: Ditto.
22260 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22261 TARGET_DF_SPE by 0.
22262 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22264 * config/rs6000/rs6000.md: Ditto.
22265 (floatdidf2, *floatdidf2_fpr): Merge.
22266 (move_from_CR_gt_bit): Delete.
22267 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22268 (E500_CR_IOR_COMPARE): Delete.
22269 (All patterns that require !TARGET_FPRS): Delete.
22270 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22272 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22274 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22276 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22278 * graphds.c (add_edge): Intitialize edge's attached data.
22279 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22280 pointer parameter. Call pointed function on each edge during
22281 graph traversing. Skip traversing the edge when the function
22283 (graphds_dfs, graphds_scc): Ditto.
22284 (for_each_edge): New parameter. Pass the new parameter to callback
22286 * graphds.h (skip_edge_callback): New function pointer type.
22287 (graphds_dfs, graphds_scc): New function pointer parameter.
22288 (graphds_edge_callback, for_each_edge): New parameter.
22290 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22292 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22293 out code checking if runtime alias check is possible to below ...
22294 Call the new function.
22295 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22296 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22298 2017-06-07 Marek Polacek <polacek@redhat.com>
22301 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22302 TYPE_OVERFLOW_WRAPS check.
22304 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22306 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22307 if versioning is required.
22308 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22309 peeling with the check for versioning.
22311 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22313 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22314 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22315 Set true to new parameter if new ssa variable is defined.
22316 (vect_gen_vector_loop_niters): Refactor. Set range information
22317 for the new vector loop bound variable.
22318 (vect_do_peeling): Ditto.
22320 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22322 * tree-affine.c (ssa.h): Include header file.
22323 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22324 has wrapping overflow behavior.
22326 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22328 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22330 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22332 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22333 (tree_to_aff_combination): ... here.
22335 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22337 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22338 reg_pressure model function.
22339 (ivopts_global_cost_for_size): Delete.
22340 (determine_set_costs, iv_ca_recount_cost): Call new model function
22341 ivopts_estimate_reg_pressure.
22343 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22345 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22346 expensive than udiv. Remove floating point cases from mod.
22348 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22350 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22351 Increase idiv cost.
22353 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22355 * config/aarch64/aarch64.md
22356 (copysignsf3): Fix mask generation.
22358 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22360 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22362 (class dump_manager): Add register_dumps method.
22363 * dumpfile.c: Include langhooks.h.
22364 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22365 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22366 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22367 (dump_manager::dump_register): Start with 512 entries instead of 32.
22368 (dump_manager::register_dumps): New method.
22369 * toplev.c (general_init): Instead of invoking register_dumps
22370 langhook, invoke register_dumps method on the dump manager.
22371 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22374 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22376 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22377 Say that # requires an associated define_split to exist, and that
22378 the define_split must be suitable for use after register allocation.
22380 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22382 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22383 (compute_outgoing_frequencies): Also initialize zero counts.
22384 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22385 around loops; preserve more of profile when nothing changes.
22387 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22389 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22391 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22392 * config/arm/arm-cpu-cdata.h: Regenerate.
22393 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22394 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22395 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22396 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22397 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22399 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22401 * config/arm/t-rmprofile: Likewise.
22402 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22404 2017-06-06 David S. Miller <davem@davemloft.net>
22407 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22408 blockage if function uses alloca.
22410 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22412 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22413 New "uid" fields to hold pretty-print IDs of group and ref.
22414 Memory references are now identified as <group_id>:<ref_id>
22415 instead of using [random] addresses.
22416 (dump_mem_details): Simplify, no functional change.
22417 (dump_mem_ref): Simplify and make output more concise.
22418 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22419 (find_or_create_group): Initialize group uid.
22420 (record_ref): Initialize ref uid. Improve debug output.
22421 (prune_group_by_reuse, should_issue_prefetch_p,)
22422 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22423 (mark_nontemporal_store, determine_loop_nest_reuse):
22424 Improve debug output.
22426 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22428 * dbgcnt.def (prefetch): New debug counter.
22429 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22430 (schedule_prefetches): Stop issueing prefetches if debug counter
22433 2017-06-06 Tom de Vries <tom@codesourcery.com>
22435 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22436 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22438 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22440 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22441 Use aarch64_reg_or_zero predicate for operand 4.
22442 (aarch64_compare_and_swap<mode> define_insn_and_split):
22443 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22444 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22446 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22448 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22449 (arm_compute_save_core_reg_mask): This.
22450 (thumb1_compute_save_reg_mask): Rename into ...
22451 (thumb1_compute_save_core_reg_mask): This.
22452 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22453 (arm_compute_frame_layout): Likewise.
22455 2017-06-06 Richard Biener <rguenther@suse.de>
22457 PR tree-optimization/80974
22458 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22459 keep or clear leaders SSA info.
22461 2017-06-06 Tom de Vries <tom@codesourcery.com>
22463 * config/nvptx/nvptx.c (split_mode_p): New function.
22464 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22466 2017-06-06 Tom de Vries <tom@codesourcery.com>
22468 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22470 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22473 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22476 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22478 * shrink-wrap.c (handle_simple_exit): Update profile.
22479 (try_shrink_wrapping): Upate profile.
22481 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22483 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22484 (tree_guess_outgoing_edge_probabilities): New.
22485 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22486 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22488 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22490 * ipa-split.c (split_function): Initialize return bb profile.
22492 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22494 * profile.c (compute_branch_probabilities): Also initialize
22495 EXIT_BLOCK profile.
22497 2017-06-06 Richard Biener <rguenther@suse.de>
22499 PR tree-optimization/80928
22500 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22501 (vect_analyze_loop_operations): Properly guard analysis for
22503 (vect_transform_loop): Likewise.
22504 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22505 (vect_model_induction_cost): Do not cost for pure SLP.
22506 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22507 of induction in inner loop vectorization.
22508 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22509 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22510 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22512 (vect_analyze_slp_cost_1): Cost induction.
22513 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22514 (vect_get_slp_vect_defs): Likewise.
22515 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22516 (vect_transform_stmt): Handle SLP reductions.
22517 * tree-vectorizer.h (vectorizable_induction): Adjust.
22519 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22521 * config/rs6000/rs6000.c (make_resolver_func): Update
22522 init_lowered_empty_function call.
22524 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22526 * doc/invoke.texi: Document the -fprofile-abs-path option.
22527 * common.opt (fprofile-abs-path): New option.
22528 * gcov-io.h (gcov_write_filename): Declare.
22529 * gcov-io.c (gcov_write_filename): New function.
22530 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22531 * profile.c (output_location): Likewise.
22533 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22535 * shring-wrap.c: Revert accidental commit.
22537 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22539 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22541 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22543 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22545 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22546 profile in callgraph edge.
22547 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22548 (apply_scale): Likewise.
22549 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22550 Also scale profile when inlining function with zero profile.
22551 (initialize_cfun): Update exit block profile even when it is zero.
22552 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22553 when profile is read.
22555 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22557 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22558 (CLONE_*): New constants to define the processors we can generate
22559 code for with the target_clone attribute.
22560 (rs6000_clone_map): New array to identify which clone processors
22561 the current program is running on.
22562 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22563 target_clone attribute.
22564 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22565 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22566 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22567 (cpu_expand_builtin): Add support for target_clone attribute.
22568 (rs6000_valid_attribute_p): Allow "default" attribute.
22569 (get_decl_name): New debug function to simplify printing the
22570 current function name in debugging statements.
22571 (rs6000_clone_priority): New functions to support the target_clone
22572 attribute, and be able to generate code to switch between ISA 2.05
22573 through ISA 3.0 (power6 through power9).
22574 (rs6000_compare_version_priority): Likewise.
22575 (rs6000_get_function_versions_dispatcher): Likewise.
22576 (make_resolver_func): Likewise.
22577 (add_condition_to_bb): Likewise.
22578 (dispatch_function_versions): Likewise.
22579 (rs6000_generate_version_dispatcher_body): Likewise.
22580 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22581 (fusion_gpr_load_p): Fix a spacing issue.
22582 * doc/extend.texi (Common Function Attributes): Document that the
22583 PowerPC supports the target_clone attribute.
22585 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22587 * config/arm/arm.h: explain F symbol found in description of ARM
22588 register allocation in its legend.
22590 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22592 * config/mips/frame-header-opt.c: Include profile-count.h.
22593 * config/riscv/riscv.c: Include profile-count.h
22595 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22597 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22599 (sm_set_flag_if_changed): Add bbs field.
22600 (execute_sm_if_changed_flag_set): Pass BBS.
22601 (execute_sm): Update.
22603 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22605 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22608 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22610 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22611 (peephole2): New peephole2 to emit the above.
22612 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22614 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22616 * config/aarch64/aarch64.c (define_peephole2 above
22617 *sub_<shift>_<mode>): New peephole.
22619 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22621 * config/i386/i386.c (make_resolver_func): Update.
22622 * Makefile.in: Add profile-count.h and profile-count.o
22623 * auto-profile.c (afdo_indirect_call): Update to new API.
22624 (afdo_set_bb_count): Update.
22625 (afdo_propagate_edge): Update.
22626 (afdo_propagate_circuit): Update.
22627 (afdo_calculate_branch_prob): Update.
22628 (afdo_annotate_cfg): Update.
22629 * basic-block.h: Include profile-count.h
22630 (struct edge_def): Turn count to profile_count.
22631 (struct basic_block_def): Likewie.
22632 (REG_BR_PROB_BASE): Move to profile-count.h
22633 (RDIV): Move to profile-count.h
22634 * bb-reorder.c (max_entry_count): Turn to profile_count.
22635 (find_traces): Update.
22636 (rotate_loop):Update.
22637 (connect_traces):Update.
22638 (sanitize_hot_paths):Update.
22639 * bt-load.c (migrate_btr_defs): Update.
22640 * cfg.c (RDIV): Remove.
22641 (init_flow): Use alloc_block.
22642 (alloc_block): Uninitialize count.
22643 (unchecked_make_edge): Uninitialize count.
22644 (check_bb_profile): Update.
22645 (dump_edge_info): Update.
22646 (dump_bb_info): Update.
22647 (update_bb_profile_for_threading): Update.
22648 (scale_bbs_frequencies_int): Update.
22649 (scale_bbs_frequencies_gcov_type): Update.
22650 (scale_bbs_frequencies_profile_count): New.
22651 * cfg.h (update_bb_profile_for_threading): Update.
22652 (scale_bbs_frequencies_profile_count): Declare.
22653 * cfgbuild.c (compute_outgoing_frequencies): Update.
22654 (find_many_sub_basic_blocks): Update.
22655 * cfgcleanup.c (try_forward_edges): Update.
22656 (try_crossjump_to_edge): Update.
22657 * cfgexpand.c (expand_gimple_tailcall): Update.
22658 (construct_exit_block): Update.
22659 * cfghooks.c (verify_flow_info): Update.
22660 (dump_bb_for_graph): Update.
22661 (split_edge): Update.
22662 (make_forwarder_block): Update.
22663 (duplicate_block): Update.
22664 (account_profile_record): Update.
22665 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22666 (get_estimated_loop_iterations): Update.
22667 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22668 (single_likely_exit): Update.
22669 * cfgloopmanip.c (scale_loop_profile): Update.
22671 (set_zero_probability): Update.
22672 (lv_adjust_loop_entry_edge): Update.
22673 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22674 (purge_dead_edges): Update.
22675 (rtl_account_profile_record): Update.
22676 * cgraph.c (cgraph_node::create): Uninitialize count.
22677 (symbol_table::create_edge): Uninitialize count.
22678 (cgraph_update_edges_for_call_stmt_node): Update.
22679 (cgraph_edge::dump_edge_flags): Update.
22680 (cgraph_node::dump): Update.
22681 (cgraph_edge::maybe_hot_p): Update.
22682 * cgraph.h: Include profile-count.h
22683 (create_clone), create_edge, create_indirect_edge): Update.
22684 (cgraph_node): Turn count to profile_count.
22685 (cgraph_edge0: Likewise.
22686 (make_speculative, clone): Update.
22687 (create_edge): Update.
22688 (init_lowered_empty_function): Update.
22689 * cgraphclones.c (cgraph_edge::clone): Update.
22690 (duplicate_thunk_for_node): Update.
22691 (cgraph_node::create_clone): Update.
22692 * cgraphunit.c (cgraph_node::analyze): Update.
22693 (cgraph_node::expand_thunk): Update.
22694 * final.c (dump_basic_block_info): Update.
22695 * gimple-streamer-in.c (input_bb): Update.
22696 * gimple-streamer-out.c (output_bb): Update.
22697 * graphite.c (print_global_statistics): Update.
22698 (print_graphite_scop_statistics): Update.
22699 * hsa-brig.c: Include basic-block.h.
22700 * hsa-dump.c: Include basic-block.h.
22701 * hsa-gen.c (T sum_slice): Update.
22702 (convert_switch_statements):Update.
22703 * hsa-regalloc.c: Include basic-block.h.
22704 * ipa-chkp.c (chkp_produce_thunks): Update.
22705 * ipa-cp.c (struct caller_statistics): Update.
22706 (init_caller_stats): Update.
22707 (gather_caller_stats): Update.
22708 (ipcp_cloning_candidate_p): Update.
22709 (good_cloning_opportunity_p): Update.
22710 (get_info_about_necessary_edges): Update.
22711 (dump_profile_updates): Update.
22712 (update_profiling_info): Update.
22713 (update_specialized_profile): Update.
22714 (perhaps_add_new_callers): Update.
22715 (decide_about_value): Update.
22716 (ipa_cp_c_finalize): Update.
22717 * ipa-devirt.c (struct odr_type_warn_count): Update.
22718 (struct decl_warn_count): Update.
22719 (struct final_warning_record): Update.
22720 (possible_polymorphic_call_targets): Update.
22721 (ipa_devirt): Update.
22722 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22723 * ipa-icf.c (sem_function::merge): Update.
22724 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22725 * ipa-inline.c (compute_uninlined_call_time): Update.
22726 (compute_inlined_call_time): Update.
22727 (want_inline_small_function_p): Update.
22728 (want_inline_self_recursive_call_p): Update.
22729 (edge_badness): Update.
22730 (lookup_recursive_calls): Update.
22731 (recursive_inlining): Update.
22732 (inline_small_functions): Update.
22733 (dump_overall_stats): Update.
22734 (dump_inline_stats): Update.
22735 * ipa-profile.c (ipa_profile_generate_summary): Update.
22736 (ipa_propagate_frequency): Update.
22737 (ipa_profile): Update.
22738 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22739 * ipa-utils.c (ipa_merge_profiles): Update.
22740 * loop-doloop.c (doloop_modify): Update.
22741 * loop-unroll.c (report_unroll): Update.
22742 (unroll_loop_runtime_iterations): Update.
22743 * lto-cgraph.c (lto_output_edge): Update.
22744 (lto_output_node): Update.
22745 (input_node): Update.
22746 (input_edge): Update.
22747 (merge_profile_summaries): Update.
22748 * lto-streamer-in.c (input_cfg): Update.
22749 * lto-streamer-out.c (output_cfg): Update.
22750 * mcf.c (create_fixup_graph): Update.
22751 (adjust_cfg_counts): Update.
22752 (sum_edge_counts): Update.
22753 * modulo-sched.c (sms_schedule): Update.
22754 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22755 * predict.c (maybe_hot_count_p): Update.
22756 (probably_never_executed): Update.
22757 (dump_prediction): Update.
22758 (combine_predictions_for_bb): Update.
22759 (propagate_freq): Update.
22760 (handle_missing_profiles): Update.
22761 (counts_to_freqs): Update.
22762 (rebuild_frequencies): Update.
22763 (force_edge_cold): Update.
22764 * predict.h: Include profile-count.h
22765 (maybe_hot_count_p, counts_to_freqs): UPdate.
22766 * print-rtl-function.c: Do not include cfg.h
22767 * print-rtl.c: Include basic-block.h
22768 * profile-count.c: New file.
22769 * profile-count.h: New file.
22770 * profile.c (is_edge_inconsistent): Update.
22771 (correct_negative_edge_counts): Update.
22772 (is_inconsistent): Update.
22773 (set_bb_counts): Update.
22774 (read_profile_edge_counts): Update.
22775 (compute_frequency_overlap): Update.
22776 (compute_branch_probabilities): Update; Initialize and deinitialize
22778 (branch_prob): Update.
22779 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22780 (edge_gcov_count): New.
22781 (bb_gcov_count): New.
22782 * shrink-wrap.c (try_shrink_wrapping): Update.
22783 * tracer.c (better_p): Update.
22784 * trans-mem.c (expand_transaction): Update.
22785 (ipa_tm_insert_irr_call): Update.
22786 (ipa_tm_insert_gettmclone_call): Update.
22787 * tree-call-cdce.c: Update.
22788 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22789 (gimple_duplicate_sese_tail): Update.
22790 (gimple_account_profile_record): Update.
22791 (execute_fixup_cfg): Update.
22792 * tree-inline.c (copy_bb): Update.
22793 (copy_edges_for_bb): Update.
22794 (initialize_cfun): Update.
22795 (freqs_to_counts): Update.
22796 (copy_cfg_body): Update.
22797 (expand_call_inline): Update.
22798 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22799 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22800 (try_unroll_loop_completely): Update.
22801 (try_peel_loop): Update.
22802 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22803 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22804 * tree-ssa-loop-split.c (connect_loops): Update.
22805 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22806 * tree-ssa-reassoc.c (branch_fixup): Update.
22807 * tree-ssa-tail-merge.c (replace_block_by): Update.
22808 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22809 (compute_path_counts): Update.
22810 (update_profile): Update.
22811 (recompute_probabilities): Update.
22812 (update_joiner_offpath_counts): Update.
22813 (estimated_freqs_path): Update.
22814 (freqs_to_counts_path): Update.
22815 (clear_counts_path): Update.
22816 (ssa_fix_duplicate_block_edges): Update.
22817 (duplicate_thread_path): Update.
22818 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22819 (struct switch_conv_info): Update.
22820 * tree-tailcall.c (decrease_profile): Update.
22821 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22822 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22823 * value-prof.c (check_counter): Update.
22824 (gimple_divmod_fixed_value): Update.
22825 (gimple_mod_pow2): Update.
22826 (gimple_mod_subtract): Update.
22827 (gimple_ic_transform): Update.
22828 (gimple_stringop_fixed_value): Update.
22829 * value-prof.h (gimple_ic): Update.
22831 2017-06-02 Carl Love <cel@us.ibm.com>
22833 * config/rs6000/rs6000-c: Add support for built-in functions
22834 vector double vec_doublee (vector signed int);
22835 vector double vec_doublee (vector unsigned int);
22836 vector double vec_doublee (vector float);
22837 vector double vec_doubleh (vector signed int);
22838 vector double vec_doubleh (vector unsigned int);
22839 vector double vec_doubleh (vector float);
22840 vector double vec_doublel (vector signed int);
22841 vector double vec_doublel (vector unsigned int);
22842 vector double vec_doublel (vector float);
22843 vector double vec_doubleo (vector signed int);
22844 vector double vec_doubleo (vector unsigned int);
22845 vector double vec_doubleo (vector float);.
22846 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22847 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22849 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22850 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22851 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22853 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22854 vec_doublel, vec_doubleh.
22855 * doc/extend.texi: Update the built-in documentation file for the
22856 new built-in functions.
22858 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22861 * ipa-inline-analysis.c (free_growth_caches): Set
22862 edge_removal_hook_holder to NULL after removing it.
22864 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22866 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22867 comparision with zero.
22869 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22870 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22871 for early expansion of vec_min and vec_max builtins.
22872 (builtin_function_type): Add min/max unsigned variants to those
22873 identified as having unsigned arguments.
22875 2017-06-02 Olivier Hainque <hainque@adacore.com>
22877 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22879 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22881 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22882 Use VALL_F16 iterator rather than VALL.
22884 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22886 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22887 Emit CBNZ inside loop when doing a strong exchange and comparing
22888 against zero. Generate the CC flags after the loop.
22890 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22892 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22893 (dl_section_ref): New.
22894 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22895 On AIX, append an expression to subtract the size of the
22896 section length to dl_section_ref.
22898 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22900 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22901 for early expansion of vector absolute builtins.
22903 2017-06-02 Richard Biener <rguenther@suse.de>
22905 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22906 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22908 2017-06-02 Richard Biener <rguenther@suse.de>
22910 PR tree-optimization/80948
22911 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22912 stmt order as well.
22914 2017-06-02 Richard Biener <rguenther@suse.de>
22916 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22918 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22919 for inductions as relevant.
22921 2017-06-02 Richard Biener <rguenther@suse.de>
22923 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22924 (vectorizable_induction): ... this. Remove dead code.
22926 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22928 * builtins. (expand_builtin_alloca): Remove second parameter and
22929 infer its value from the first parameter instead.
22930 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22932 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22934 PR rtl-optimization/80903
22935 * loop-doloop.c (add_test): Unshare sequence.
22937 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22939 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22941 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22943 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22945 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22946 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22947 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22948 (xlogue_layout::get_stub_rtx): Make static.
22949 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22950 (xlogue_layout::compute_stub_managed_regs): Rename to...
22951 (xlogue_layout::count_stub_managed_regs): ...this.
22952 (xlogue_layout::is_stub_managed_reg): New function.
22953 (xlogue_layout::m_stub_names): Rename to...
22954 (xlogue_layout::s_stub_names): ...this, make static.
22955 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22956 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22957 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22958 xlogue_layout::s_stub_names): Instantiate statics.
22959 (stub_managed_regs): Remove.
22960 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22961 (disable_call_ms2sysv_xlogues): Rename to...
22962 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22963 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22965 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22966 change after reload_completed.
22967 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22969 (ix86_expand_prologue): Likewise.
22970 (ix86_expand_epilogue): Likewise.
22971 (ix86_expand_split_stack_prologue): Likewise.
22972 (ix86_compute_frame_layout): Remove frame parameter ...
22973 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22974 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22976 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22977 (ix86_frame): Move from here ...
22978 * config/i386/i386.h (ix86_frame): ... to here.
22979 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22980 complete ix86_frame data structure instead. Remove some_ld_name.
22982 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22984 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22985 symbols that hold a DECL_VALUE_EXPR.
22987 2017-06-01 Martin Jambor <mjambor@suse.cz>
22989 PR tree-optimization/80898
22990 * tree-sra.c (process_subtree_disqualification): Removed.
22991 (disqualify_candidate): Do not acll
22992 process_subtree_disqualification.
22993 (subtree_mark_written_and_enqueue): New function.
22994 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22995 RHS has been disqualified and re-queue LHS if necessary. Apart
22996 from that, ignore disqualified RHS.
22998 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23000 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23001 address fetch for z10 or later.
23003 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23005 * config/arc/arc.md (tst_movb): Add guard when splitting.
23007 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23009 * config/arc/arc.c (arc_can_eliminate): Test against
23010 arc_frame_pointer_needed.
23012 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23014 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23015 to prevent store reordering.
23016 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23017 (type): Add block type.
23018 (stack_tie): Define special instruction to be used in
23021 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23023 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23024 constraint. It is not valid for the pattern.
23025 (noncommutative_binary_comparison): Likewise.
23027 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23029 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23032 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23034 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23035 be used by the reg-alloc.
23037 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23039 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23040 reg-alloc when having mul64 or mul32x16 instructions.
23041 (mulsidi3): Likewise.
23042 (umulsidi3): Likewise.
23043 (mulsi32x16): New pattern.
23044 (mulsi64): Likewise.
23045 (mulsidi64): Likewise.
23046 (umulsidi64): Likewise.
23047 (MUL32x16_REG): Define.
23048 (mul64_600): Use MUL32x16_REG.
23049 (mac64_600): Likewise.
23050 (umul64_600): Likewise.
23051 (umac64_600): Likewise.
23053 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23055 * config/arc/arc.md (mulsi3_700): Make it commutative.
23057 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23059 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23061 (*sign_extendsidi2_insn): Likewise for movstosw.
23063 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23065 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23066 the type of the input discriminant value. Convert the
23067 discriminant value of signedness vary.
23069 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23071 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23072 Add to -Wall section.
23074 2017-06-01 Richard Biener <rguenther@suse.de>
23076 PR middle-end/66313
23077 * fold-const.c (fold_plusminus_mult_expr): If the factored
23078 factor may be zero use a wrapping type for the inner operation.
23079 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23080 and handle moved defs.
23081 (process_assignment): Properly guard the unary op case. Return a
23082 tri-state indicating that moving the stmt before the call may allow
23083 to continue. Pass through to_move.
23084 (find_tail_calls): Handle moving unrelated defs before
23087 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23090 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23091 splitter result in the canonical way.
23093 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23095 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23096 also for 32bit target. Update insn attributes.
23097 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23099 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23101 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23102 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23103 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23104 (_mm_maskz_min_ss): New intrinsics.
23106 2017-05-31 Martin Liska <mliska@suse.cz>
23108 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23109 Change comment style to one we normally use.
23110 (vectorizable_reduction): Likewise.
23111 (vectorizable_induction): Likewise.
23112 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23113 (vectorizable_call): Likewise.
23114 (vectorizable_simd_clone_call): Likewise.
23115 (vectorizable_conversion): Likewise.
23116 (vectorizable_assignment): Likewise.
23117 (vectorizable_shift): Likewise.
23118 (vectorizable_operation): Likewise.
23119 (vectorizable_store): Likewise.
23120 (vectorizable_load): Likewise.
23121 * tree-vectorizer.h: Likewise.
23123 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23125 * passes.c (emergency_dump_function): New.
23126 * tree-pass.h (emergency_dump_function): Declare.
23127 * plugin.c (plugins_internal_error_function): Remove.
23128 * plugin.h (plugins_internal_error_function): Remove declaration.
23129 * toplev.c (internal_error_function): New static function. Use it...
23130 (general_init): ...here.
23132 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23134 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23135 (arc_rtx_costs): Add costs for new patterns.
23136 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23137 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23139 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23141 * tree-ssa-strlen.c (get_next_strinfo): New function.
23142 (get_stridx_plus_constant): Use it.
23143 (zero_length_string): Likewise.
23144 (adjust_related_strinfos): Likewise.
23145 (adjust_last_stmt): Likewise.
23147 2017-05-31 Richard Biener <rguenther@suse.de>
23150 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23151 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23153 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23155 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23156 loop_vinfo argument and use of dependence distance vectors.
23157 Check instead whether the two references differ only in their
23158 initial value and assume that they have the same alignment if the
23159 difference is a multiple of the vector alignment.
23160 (vect_analyze_data_refs_alignment): Update call accordingly.
23162 2017-05-31 Martin Liska <mliska@suse.cz>
23165 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23167 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23169 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23170 (create_intersect_range_checks): Move from ...
23171 * tree-data-ref.c (create_intersect_range_checks_index)
23172 (create_intersect_range_checks): ... to here.
23173 (create_runtime_alias_checks): New function factored from ...
23174 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23175 here. Call above function.
23176 * tree-data-ref.h (create_runtime_alias_checks): New function.
23178 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23180 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23181 segment length for dr_b and compute it in wide_int.
23183 2017-05-31 Richard Biener <rguenther@suse.de>
23185 PR tree-optimization/80906
23186 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23187 and pass through iv_map.
23188 (copy_bb_and_scalar_dependences): Adjust.
23189 (translate_pending_phi_nodes): Likewise.
23190 (copy_loop_close_phi_args): Handle code-generating IVs instead
23193 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23195 * diagnostic-color.c (color_dict): Add "type-diff".
23196 (parse_gcc_colors): Update comment.
23197 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23198 -fdiagnostics-show-template-tree and -fno-elide-type.
23199 (GCC_COLORS): Add type-diff to example.
23201 (-fdiagnostics-show-template-tree): New.
23202 (-fno-elide-type): New.
23203 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23204 the pp_format_decoder callback. Call any m_format_postprocessor's
23206 (pretty_printer::pretty_printer): Initialize
23207 m_format_postprocessor.
23208 (pretty_printer::~pretty_printer): Delete any
23209 m_format_postprocessor.
23210 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23211 (class format_postprocessor): New class.
23212 (struct pretty_printer::format_decoder): Document the new parameters.
23213 (struct pretty_printer::m_format_postprocessor): New field.
23214 * tree-diagnostic.c (default_tree_printer): Update for new
23215 bool and const char ** params.
23216 * tree-diagnostic.h (default_tree_printer): Likewise.
23218 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23220 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23221 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23222 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23223 rs6000_gen_cell_microcode code.
23224 (rs6000_final_prescan_insn): Delete.
23225 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23226 "warn-cell-microcode" entries.
23227 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23228 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23229 throughout. Change cc_reg_not_micro_cr0_operand to
23230 cc_reg_not_cr0_operand throughout.
23231 (*extendhi<mode>2_noload): Delete.
23232 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23233 (mwarn-cell-microcode): Delete.
23234 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23235 -mgen-cell-microcode and -mwarn-cell-microcode.
23237 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23240 * config/i386/constraints.md (Yd): New constraint.
23242 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23243 and (?Yd, r) alternatives. Update insn attributes.
23244 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23245 and (?*Yd, r) alternatives. Update insn attributes.
23246 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23248 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23250 * gimplify.c (gimplify_modify_expr): Don't create a
23251 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23254 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23256 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23258 2017-05-30 Richard Biener <rguenther@suse.de>
23260 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23261 and reduc_def fields.
23262 (STMT_VINFO_REDUC_TYPE): New define.
23263 (STMT_VINFO_REDUC_DEF): Likewise.
23264 (vect_force_simple_reduction): Adjust prototype.
23265 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23266 (vect_is_simple_reduction): Remove check_reduction argument.
23267 (vect_force_simple_reduction): Adjust and set
23268 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23269 (vectorizable_reduction): Do not re-do reduction analysis
23270 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23271 * tree-parloops.c (gather_scalar_reductions): Adjust.
23273 2017-05-30 Richard Biener <rguenther@suse.de>
23275 PR middle-end/80901
23276 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23279 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23281 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23282 Introduce unknown_misalignment parameter and remove vf.
23283 (vect_peeling_hash_get_lowest_cost):
23284 Pass unknown_misalignment parameter.
23285 (vect_enhance_data_refs_alignment):
23286 Fix unsupportable data ref treatment.
23288 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23290 * tree-vect-data-refs.c (vect_get_data_access_cost):
23291 Workaround for SLP handling.
23292 (vect_enhance_data_refs_alignment):
23293 Compute costs for doing no peeling at all, compare to the best
23294 peeling costs so far and avoid peeling if cheaper.
23296 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23298 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23299 Return peeling info and set costs to zero for unlimited cost
23301 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23302 with unknown misalignment. Compute and costs for unknown
23303 misalignment, compare them to the costs for known misalignment
23304 and choose the cheapest for peeling.
23306 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23308 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23309 (vect_get_peeling_costs_all_drs): Create function.
23310 (vect_peeling_hash_get_lowest_cost):
23311 Use vect_get_peeling_costs_all_drs.
23312 (vect_peeling_supportable): Create function.
23313 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23315 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23317 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23318 DR_HAS_NEGATIVE_STEP.
23319 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23320 (vect_enhance_data_refs_alignment): Use.
23321 (vect_duplicate_ssa_name_ptr_info): Use.
23322 * tree-vectorizer.h (dr_misalignment): Use.
23323 (known_alignment_for_access_p): Use.
23325 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23328 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23330 (has_section_name): New function.
23332 2017-05-30 Martin Liska <mliska@suse.cz>
23335 * auto-profile.c (get_function_decl_from_block): Fix
23338 2017-05-30 Richard Biener <rguenther@suse.de>
23340 PR middle-end/80876
23341 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23343 2017-05-30 Martin Liska <mliska@suse.cz>
23345 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23346 * dumpfile.h (struct dump_file_info): Remove ctors.
23348 2017-05-30 Martin Liska <mliska@suse.cz>
23350 * predict.def: Fix GNU coding style.
23352 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23354 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23355 Mark 'to' argument with ATTRIBUTE_UNUSED.
23357 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23359 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23360 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23361 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23364 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23366 * doc/install.texi (Options specification): Restore entry of
23367 --enable-sjlj-exceptions.
23369 2017-05-27 Michael Eager <eager@eagercon.com>
23372 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23374 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23376 * config/microblaze/microblaze.h
23377 (FIXED_REGISTERS): Update in macro.
23378 (CALL_USED_REGISTERS): Update in macro.
23380 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23382 * doc/install.texi: Add links to macOS binary distributions.
23384 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23388 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23390 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23392 2017-05-26 Martin Liska <mliska@suse.cz>
23394 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23396 2017-05-26 Martin Liska <mliska@suse.cz>
23398 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23399 always leading ';; '.
23400 (dump_bb_info): Likewise.
23401 (brief_dump_cfg): Likewise.
23402 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23403 * dumpfile.c: Remove usage of TDF_VERBOSE.
23404 * dumpfile.h (enum dump_kind): Likewise.
23405 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23406 * print-tree.c (debug_verbose): Remove.
23407 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23408 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23409 * tree-diagnostic.c (default_tree_printer): Replace
23410 TDF_DIAGNOSTIC with TDF_SLIM.
23412 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23414 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23415 in parameter loop, rather than loop_vinfo.
23416 (create_intersect_range_checks): Ditto.
23417 (vect_create_cond_for_alias_checks): Update call to above functions.
23419 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23421 PR tree-optimization/80815
23422 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23423 for merging runtime alias checks. Handle negative DR_STEPs.
23425 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23427 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23429 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23430 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23431 out code pruning runtime alias checks.
23432 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23433 factored out from above.
23434 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23436 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23438 (prune_runtime_alias_test_list): New decalaration.
23440 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23442 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23443 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23444 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23445 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23446 (operator==, comp_dr_with_seg_len_pair): Ditto.
23447 (vect_prune_runtime_alias_test_list): Ditto.
23449 2017-05-26 Martin Liska <mliska@suse.cz>
23452 * params.def: Bound partial-inlining-entry-probability param.
23454 2017-05-26 Marek Polacek <polacek@redhat.com>
23457 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23460 2017-05-26 Richard Biener <rguenther@suse.de>
23462 PR tree-optimization/80842
23463 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23466 2017-05-26 Richard Biener <rguenther@suse.de>
23468 PR tree-optimization/80844
23469 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23471 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23473 * doc/md.texi (Machine Constraints): Update x86 family
23474 machine constraints section to match 'config/i386/constraints.md'.
23476 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23478 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23480 2017-05-25 Nathan Sidwell <nathan@acm.org>
23482 * doc/invoke.texi (--enable-languages): Update documentation.
23484 2017-05-25 Martin Liska <mliska@suse.cz>
23486 * dumpfile.c: Add TDF_FOLDING.
23487 * dumpfile.h (enum dump_kind): Likewise.
23488 * genmatch.c (dt_simplify::gen_1): Use it.
23490 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23492 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23494 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23496 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23497 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23499 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23501 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23502 * match.pd (X == C): Rewrite it here.
23503 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23504 with_certain_nonzero_bits2): New predicates.
23505 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23507 2017-05-24 Nathan Sidwell <nathan@acm.org>
23509 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23512 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23515 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23517 * config/powerpcspe: New port. Files are copied from the rs6000
23518 port, with "rs6000" in filenames replaced by "powerpcspe".
23520 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23522 PR rtl-optimization/80754
23523 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23525 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23527 * config/sparc/sparc.md (length): Return the correct value for -mflat
23528 sibcalls to match output_sibcall.
23530 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23534 * config/rs6000/rs6000.c (struct machine_function): Add new field
23536 (rs6000_get_separate_components): Init that field, use it.
23537 (rs6000_components_for_bb): Use the field.
23539 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23541 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23543 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23545 PR middle-end/80823
23546 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23548 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23551 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23552 address against address_operand predicate.
23553 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23555 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23557 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23558 parameters passed indirectly.
23560 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23562 * config/i386/i386.md (*movdi_internal): Remove SSE4
23563 alternative 18 (?r, *v). Update insn attributes.
23564 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23565 Update insn attributes.
23566 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23567 Update insn attributes.
23568 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23569 alternative 1 (r, v). Remove isa attribute.
23570 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23571 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23572 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23574 2017-05-23 Tom de Vries <tom@codesourcery.com>
23576 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23579 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23581 * cgraphunit.c (symbol_table::process_new_functions): Update.
23582 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23583 (inline_generate_summary): Rename to ...
23584 (ipa_fn_summary_generate): ... this one.
23585 (inline_read_summary): Rename to ...
23586 (ipa_fn_summary_read): ... this one.
23587 (inline_write_summary): Rename to ...
23588 (ipa_fn_summary_write): ... this one.
23589 (inline_free_summary): Rename to ...
23590 (ipa_free_fn_summary): ... this one.
23591 (pass_data_local_fn_summary, pass_local_fn_summary,
23592 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23593 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23594 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23595 make_pass_ipa_fn_summary): New.
23596 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23597 inline_write_summary, inline_free_summary): Remove.
23598 (ipa_free_fn_summary) : New.
23599 * ipa-inline.c (ipa_inline): Update.
23600 (pass_ipa_inline): Do not generate summaries.
23601 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23603 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23604 and add pass_ipa_fn_summary.
23605 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23607 (make_pass_inline_parameters): Remove.
23609 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23611 * omp-low.c (struct omp_context): Remove "default_kind" member.
23614 * omp-offload.c (execute_oacc_device_lower): Remove the
23615 parallelism dimensions function attributes for unparallelized
23616 OpenACC kernels constructs.
23618 2017-05-23 Martin Liska <mliska@suse.cz>
23620 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23622 (cgraph_edge::make_speculative): Likewise.
23623 (cgraph_edge::resolve_speculation): Likewise.
23624 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23625 (cgraph_node::dump): Likewise.
23626 * cgraph.h: Likewise.
23627 * cgraphunit.c (analyze_functions): Likewise.
23628 (symbol_table::compile): Likewise.
23629 * ipa-cp.c (print_all_lattices): Likewise.
23630 (determine_versionability): Likewise.
23631 (initialize_node_lattices): Likewise.
23632 (ipcp_verify_propagated_values): Likewise.
23633 (estimate_local_effects): Likewise.
23634 (update_profiling_info): Likewise.
23635 (create_specialized_node): Likewise.
23636 (perhaps_add_new_callers): Likewise.
23637 (decide_about_value): Likewise.
23638 (decide_whether_version_node): Likewise.
23639 (identify_dead_nodes): Likewise.
23640 (ipcp_store_bits_results): Likewise.
23641 * ipa-devirt.c (dump_targets): Likewise.
23642 (ipa_devirt): Likewise.
23643 * ipa-icf.c (sem_item::dump): Likewise.
23644 (sem_function::equals): Likewise.
23645 (sem_variable::equals): Likewise.
23646 (sem_item_optimizer::read_section): Likewise.
23647 (sem_item_optimizer::execute): Likewise.
23648 (congruence_class::dump): Likewise.
23649 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23650 (dump_inline_summary): Likewise.
23651 (estimate_node_size_and_time): Likewise.
23652 (inline_analyze_function): Likewise.
23653 * ipa-inline-transform.c (inline_call): Likewise.
23654 * ipa-inline.c (report_inline_failed_reason): Likewise.
23655 (want_early_inline_function_p): Likewise.
23656 (edge_badness): Likewise.
23657 (update_edge_key): Likewise.
23658 (inline_small_functions): Likewise.
23659 * ipa-profile.c (ipa_profile): Likewise.
23660 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23661 (ipa_make_edge_direct_to_target): Likewise.
23662 (remove_described_reference): Likewise.
23663 (ipa_impossible_devirt_target): Likewise.
23664 (propagate_controlled_uses): Likewise.
23665 (ipa_print_node_params): Likewise.
23666 (ipcp_transform_function): Likewise.
23667 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23668 (propagate_pure_const): Likewise.
23669 * ipa-reference.c (generate_summary): Likewise.
23670 (read_write_all_from_decl): Likewise.
23671 (propagate): Likewise.
23672 (ipa_reference_read_optimization_summary): Likewise.
23673 * ipa-utils.c (ipa_merge_profiles): Likewise.
23674 * ipa.c (walk_polymorphic_call_targets): Likewise.
23675 (symbol_table::remove_unreachable_nodes): Likewise.
23676 (ipa_single_use): Likewise.
23677 * passes.c (execute_todo): Likewise.
23678 * predict.c (drop_profile): Likewise.
23679 * symtab.c (symtab_node::get_dump_name): New function.
23680 (symtab_node::dump_name): Likewise.
23681 (symtab_node::dump_asm_name): Likewise.
23682 (symtab_node::dump_references): Likewise.
23683 (symtab_node::dump_referring): Likewise.
23684 (symtab_node::dump_base): Likewise.
23685 (symtab_node::debug_symtab): Likewise.
23686 * tree-sra.c (convert_callers_for_node): Likewise.
23687 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23688 * value-prof.c (init_node_map): Likewise.
23690 2017-05-23 Martin Liska <mliska@suse.cz>
23692 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23693 and symtab_node::debug_symtab to symbol_table::debug.
23694 * cgraphunit.c (analyze_functions): Use the renamed function.
23695 (symbol_table::compile): Likewise.
23696 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23697 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23698 * passes.c (execute_todo): Likewise.
23699 * symtab.c (symbol_table::dump): New function.
23700 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23702 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23704 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23705 that nonconst implies exec.
23707 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23709 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23710 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23711 (inline_edge_summary_vec): Turn into ...
23712 (ipa_call_summaries): ... this one.
23713 (redirect_to_unreachable, edge_set_predicate,
23714 evaluate_properties_for_edge, inline_summary_alloc,
23715 reset_ipa_call_summary, reset_inline_summary,
23716 inline_summary_t::duplicate): Update.
23717 (inline_edge_duplication_hook): Turn to ...
23718 (ipa_call_summary_t::duplicate): ... this one.
23719 (inline_edge_removal_hook): Turn to ...
23720 (ipa_call_summary_t::remove): ... this one.
23721 (dump_inline_edge_summary): Turn to ...
23722 (dump_ipa_call_summary): ... this one.
23723 (estimate_function_body_sizes): Update.
23724 (inline_update_callee_summaries): Update.
23725 (remap_edge_change_prob): Update.
23726 (remap_edge_summaries): Update.
23727 (inline_merge_summary): Update.
23728 (do_estimate_edge_time): Update.
23729 (inline_generate_summary): Update.
23730 (inline_read_section): Update.
23731 (inline_read_summary): Update.
23732 (inline_free_summary): Update.
23733 * ipa-inline.c (can_inline_edge_p): Update.
23734 (compute_inlined_call_time): Update.
23735 (want_inline_small_function_p): Update.
23736 (edge_badness): Update.
23737 (early_inliner): Update.
23738 * ipa-inline.h (inline_edge_summary): Turn to ...
23739 (ipa_call_summary): ... this one.
23740 (ipa_call_summary_t): New class.
23741 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23742 (ipa_call_summaries): New.
23743 (inline_edge_summary): Remove.
23744 (estimate_edge_growth): Update.
23745 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23746 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23747 * ipa-split.c (execute_split_functions): Update.
23748 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23750 2017-05-23 Tom de Vries <tom@codesourcery.com>
23752 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23753 attributes): Document rdrand effective target.
23755 2017-05-23 Tom de Vries <tom@codesourcery.com>
23757 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23758 attributes): Sort alphabetically.
23760 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23762 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23764 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23767 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23768 V2DF/V2DI splat into two separate patterns, one that handles
23769 registers, and the other that only handles memory. Drop support
23770 for splatting from a GPR on ISA 2.07 and then splitting the
23771 splat into direct move and splat.
23772 (vsx_splat_<mode>_reg): Likewise.
23773 (vsx_splat_<mode>_mem): Likewise.
23775 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23777 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23779 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23781 PR middle-end/80809
23782 * omp-low.c (finish_taskreg_remap): New function.
23783 (finish_taskreg_scan): If unit size of ctx->record_type
23784 is non-constant, unshare the size expression and replace
23785 decls in it with possible outer var refs.
23787 PR middle-end/80809
23788 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23789 GOVD_SHARED rather than GOVD_PRIVATE with it.
23790 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23791 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23793 PR middle-end/80853
23794 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23795 as last argument to build_outer_var_ref for pointer bases of array
23796 section reductions.
23798 2017-05-19 Martin Sebor <msebor@redhat.com>
23800 * print-tree.c (print_node): Print DECL_READ_P flag.
23802 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23804 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23805 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23806 * cgraph.c: Likewise.
23807 * cgraphunit.c: Likewise.
23808 * gengtype.c: Likewise.
23809 * ipa-cp.c: Likewise.
23810 * ipa-devirt.c: Likewise.
23811 * ipa-icf.c: Likewise.
23812 * ipa-predicate.c: Likewise.
23813 * ipa-profile.c: Likewise.
23814 * ipa-prop.c: Likewise.
23815 * ipa-split.c: Likewise.
23817 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23818 edge_predicate_pool, dump_inline_hints,
23819 inline_summary::account_size_time, redirect_to_unreachable,
23820 edge_set_predicate, set_hint_predicate,
23821 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23822 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23823 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23824 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23825 ipa_call_summary_t::remove, initialize_growth_caches,
23826 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23827 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23828 mark_modified, unmodified_parm_1, unmodified_parm,
23829 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23830 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23831 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23832 will_be_nonconstant_predicate, record_modified_bb_info,
23833 get_minimal_bb, record_modified, param_change_prob,
23834 phi_result_unknown_predicate, predicate_for_phi_result,
23835 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23836 estimate_function_body_sizes, compute_inline_parameters,
23837 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23838 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23839 inline_update_callee_summaries, remap_edge_change_prob,
23840 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23841 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23842 inline_analyze_function, inline_summary_t::insert,
23843 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23844 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23845 inline_free_summary): Move to ipa-fnsummary.h
23846 (predicate_t): Remove.
23847 * ipa-fnsummary.c: New file.
23848 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23849 (enum inline_hints_vals, inline_hints, agg_position_info,
23850 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23851 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23852 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23853 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23854 inline_read_summary, inline_write_summary, inline_free_summary,
23855 inline_analyze_function, initialize_inline_failed,
23856 inline_merge_summary, inline_update_overall_summary,
23857 compute_inline_parameters): Move to ipa-fnsummary.h
23858 * ipa-fnsummary.h: New file.
23859 * ipa-inline-transform.h: Include ipa-inline.h.
23860 * ipa-inline.c: LIkewise.
23862 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23864 * ipa-inline.c (edge_badness): Use inlined_time instead of
23865 inline_summaries->get.
23867 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23869 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23871 2017-05-22 Nathan Sidwell <nathan@acm.org>
23873 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23874 (fdump-lang): Document 'raw' option.
23875 * dumpfile.h (TDI_tu): Delete.
23876 * dumpfile.c (dump_files): Remove translation-unit.
23877 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23879 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23881 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23882 command option from $(AWK) call.
23883 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23884 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23885 [FORMAT]: Remove handling of variable.
23886 * config/avr/t-multilib: Regenerate.
23888 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23890 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23892 (dump_inline_summary): Do not print self_time.
23893 (estimate_function_body_sizes): Do not set self_time.
23894 (compute_inline_parameters): Likewise.
23895 (inline_read_section, inline_write_summary): Do not stream self_time.
23896 * ipa-inline.h (inline_summary): Drop self_time.
23898 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23900 * ipa-inline-analysis.c (account_size_time): Rename to ...
23901 (inline_summary::account_size_time): ... this one.
23902 (reset_ipa_call_summary): Turn to ...
23903 (ipa_call_summary::reset): ... this one.
23904 (reset_inline_summary): Turn to ...
23905 (inline_summary::reset): ... this one.
23906 (inline_summary_t::remove): Update.
23907 (inline_summary_t::duplicate): Update.
23908 (ipa_call_summary_t::remove): Update.
23909 (dump_inline_summary): Update.
23910 (estimate_function_body_sizes): Update.
23911 (compute_inline_parameters): Update.
23912 (estimate_node_size_and_time): Update.
23913 (inline_merge_summary): Update.
23914 (inline_update_overall_summary): Update.
23915 (inline_read_section): Update.
23916 (inline_write_summary): Update.
23917 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23918 add account_size_time and reset member functions.
23919 (ipa_call_summary): Add reset function.
23920 * ipa-predicate.h (predicate::operator &): Constify.
23922 2017-05-22 Richard Biener <rguenther@suse.de>
23924 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23926 2017-05-19 Jason Merrill <jason@redhat.com>
23928 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23930 2017-05-19 Marek Polacek <polacek@redhat.com>
23933 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23934 TYPE_OVERFLOW_WRAPS checks.
23936 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23938 * tree-core.h (enum omp_clause_default_kind): Add
23939 "OMP_CLAUSE_DEFAULT_PRESENT".
23940 * tree-pretty-print.c (dump_omp_clause): Handle it.
23941 * gimplify.c (enum gimplify_omp_var_data): Add
23942 "GOVD_MAP_FORCE_PRESENT".
23943 (gimplify_adjust_omp_clauses_1): Map it to
23944 "GOMP_MAP_FORCE_PRESENT".
23945 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23947 * gimplify.c (oacc_default_clause): Clarify.
23949 2017-05-19 Nathan Sidwell <nathan@acm.org>
23951 LANG_HOOK_REGISTER_DUMPS
23952 * toplev.c (general_init): Call register dump lang hook.
23953 * doc/invoke.texi: Document -fdump-lang option family.
23954 * dumpfile.c (dump_files): Remove class dump here.
23955 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23956 * dumpfile.h (tree_dump_index): Remove TDI_class.
23957 * langhooks-def.h (lhd_register_dumps): Declare.
23958 (LANG_HOOKS_REGISTER_DUMPS): Define.
23959 (LANG_HOOKS_INITIALIZER): Add it.
23960 * langhooks.c (lhd_register_dumps): Define.
23961 * langhooks.h (struct lang_hooks): Add register_dumps.
23963 2017-05-19 Nathan Sidwell <nathan@acm.org>
23965 * context.h (context::set_passes): New.
23966 * context.c (context::context): Do not create pass manager.
23967 * toplev.c (general_init): Create pass manager here.
23969 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23971 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23972 use this splitter if two add or or instructions would also work for
23973 the constant we want to generate.
23975 2017-05-19 Richard Biener <rguenther@suse.de>
23978 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23979 predicate evaluation.
23981 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23983 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23985 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23988 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23990 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23991 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23992 (inline_edge_summary_vec): Turn into ...
23993 (ipa_call_summaries): ... this one.
23994 (redirect_to_unreachable, edge_set_predicate,
23995 evaluate_properties_for_edge, inline_summary_alloc,
23996 reset_ipa_call_summary, reset_inline_summary,
23997 inline_summary_t::duplicate): Update.
23998 (inline_edge_duplication_hook): Turn to ...
23999 (ipa_call_summary_t::duplicate): ... this one.
24000 (inline_edge_removal_hook): Turn to ...
24001 (ipa_call_summary_t::remove): ... this one.
24002 (dump_inline_edge_summary): Turn to ...
24003 (dump_ipa_call_summary): ... this one.
24004 (estimate_function_body_sizes): Update.
24005 (inline_update_callee_summaries): Update.
24006 (remap_edge_change_prob): Update.
24007 (remap_edge_summaries): Update.
24008 (inline_merge_summary): Update.
24009 (do_estimate_edge_time): Update.
24010 (inline_generate_summary): Update.
24011 (inline_read_section): Update.
24012 (inline_read_summary): Update.
24013 (inline_free_summary): Update.
24014 * ipa-inline.c (can_inline_edge_p): Update.
24015 (compute_inlined_call_time): Update.
24016 (want_inline_small_function_p): Update.
24017 (edge_badness): Update.
24018 (early_inliner): Update.
24019 * ipa-inline.h (inline_edge_summary): Turn to ...
24020 (ipa_call_summary): ... this one.
24021 (ipa_call_summary_t): New class.
24022 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24023 (ipa_call_summaries): New.
24024 (inline_edge_summary): Remove.
24025 (estimate_edge_growth): Update.
24026 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24027 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24028 * ipa-split.c (execute_split_functions): Update.
24029 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24031 2017-05-19 Richard Biener <rguenther@suse.de>
24033 PR middle-end/80764
24034 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24036 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24038 * config/rs6000/rs6000.c (struct machine_function): Add field
24039 fpr_is_wrapped_separately.
24040 (rs6000_get_separate_components): Use 64 components. Handle the
24041 new FPR components.
24042 (rs6000_components_for_bb): Handle the FPR components.
24043 (rs6000_emit_prologue_components): Handle the FPR components.
24044 (rs6000_emit_epilogue_components): Handle the FPR components.
24045 (rs6000_set_handled_components): Handle the FPR components.
24046 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24047 that are already separately shrink-wrapped.
24048 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24049 that are already separately shrink-wrapped.
24051 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24054 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24057 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24058 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24059 cases where the register allocator uses a move and an offsettable
24060 memory operation to/from a FPR register on ISA 2.06/2.07.
24061 (define_peephole2 for Altivec d-form store): Likewise.
24063 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24066 * config/i386/mmx.md (*mov<mode>_internal): Enable
24067 alternatives 11, 12, 13 and 14 also for 32bit targets.
24068 Remove alternatives 15, 16, 17 and 18.
24069 * config/i386/sse.md (vec_concatv2di): Change
24070 alternative (!x, *y) to (x, ?!*Yn).
24072 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24074 * dumpfile.h (enum dump_kind): Remove stray comma.
24076 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24078 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24079 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24080 predicate::num_conditions
24081 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24082 (CHANGED): turn into predicate::changed.
24083 (agg_position_info): Move to ipa-predicate.h
24084 (add_condition, predicate::add_clause, predicate::operator &=,
24085 predicate::or_with, predicate::evaluate, predicate::probability,
24086 dump_condition, dump_clause, predicate::dump,
24087 predicate::remap_after_duplication, predicate::remap_after_inlining,
24088 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24089 (evaluate_conditions_for_known_args): Update.
24090 (set_cond_stmt_execution_predicate): Update.
24091 * ipa-inline.h: Include ipa-predicate.h
24092 (condition, inline_param_summary, conditions, agg_position_info,
24093 predicate): Move to ipa-predicate.h
24094 * ipa-predicate.c: New file.
24095 * ipa-predicate.h: New file.
24097 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24099 * final.c (leaf_function_p): Check we are not in a sequence.
24101 2017-05-18 Martin Liska <mliska@suse.cz>
24103 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24104 * dumpfile.c (dump_register): Use new enum dump_kind.
24105 (get_dump_file_name): Likewise.
24106 (dump_enable_all): Likewise.
24107 (dump_switch_p_1): Likewise.
24108 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24109 * dumpfile.h (enum dump_kind): New enum type.
24110 (struct dump_file_info): Create constructor and
24111 format fields and comments.
24112 * passes.c (pass_manager::register_one_dump_file):
24114 * statistics.c (statistics_early_init): Likewise.
24115 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24116 TDF_TREE with TDF_SLIM.
24117 (gather_memory_references_ref): Likewise.
24119 2017-05-18 Martin Liska <mliska@suse.cz>
24121 * vec.h (struct vnull): Use it.
24123 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24125 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24126 (true_predicate, false_predicate, true_predicate_p,
24127 false_predicate_p): Remove.
24128 (single_cond_predicate, not_inlined_predicate): Turn to member function
24130 (add_condition): Update.
24131 (add_clause): Turn to...
24132 (predicate::add_clause): ... this one; update; allow passing NULL
24134 (and_predicates): Turn to ...
24135 (predicate::operator &=): ... this one.
24136 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24137 (or_predicates): Turn to ...
24138 (predicate::or_with): ... this one.
24139 (evaluate_predicate): Turn to ...
24140 (predicate::evaluate): ... this one.
24141 (predicate_probability): Turn to ...
24142 (predicate::probability): ... this one.
24143 (dump_condition): Update.
24144 (dump_predicate): Turn to ...
24145 (predicate::dump): ... this one.
24146 (account_size_time): Update.
24147 (edge_set_predicate): Update.
24148 (set_hint_predicate): UPdate.
24149 (evaluate_conditions_for_known_args): Update.
24150 (evaluate_properties_for_edge): Update.
24151 (remap_predicate_after_duplication): Turn to...
24152 (predicate::remap_after_duplication): ... this one.
24153 (remap_hint_predicate_after_duplication): Update.
24154 (inline_summary_t::duplicate): UPdate.
24155 (dump_inline_edge_summary): Update.
24156 (dump_inline_summary): Update.
24157 (set_cond_stmt_execution_predicate): Update.
24158 (set_switch_stmt_execution_predicate): Update.
24159 (compute_bb_predicates): Update.
24160 (will_be_nonconstant_expr_predicate): Update.
24161 (will_be_nonconstant_predicate): Update.
24162 (phi_result_unknown_predicate): Update.
24163 (predicate_for_phi_result): Update.
24164 (array_index_predicate): Update.
24165 (estimate_function_body_sizes): Update.
24166 (estimate_node_size_and_time): Update.
24167 (estimate_ipcp_clone_size_and_time): Update.
24168 (remap_predicate): Rename to ...
24169 (predicate::remap_after_inlining): ... this one.
24170 (remap_hint_predicate): Update.
24171 (inline_merge_summary): Update.
24172 (inline_update_overall_summary): Update.
24173 (estimate_size_after_inlining): Update.
24174 (read_predicate): Rename to ...
24175 (predicate::stream_in): ... this one.
24176 (read_inline_edge_summary): Update.
24177 (write_predicate): Rename to ...
24178 (predicate::stream_out): ... this one.
24179 (write_inline_edge_summary): Update.
24180 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24181 (clause_t): Turn to uint32_t
24182 (predicate): Turn to class; implement constructor and operators
24184 (size_time_entry): Update.
24185 (inline_summary): Update.
24186 (inline_edge_summary): Update.
24188 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24190 * fold-const.c (fold_binary_loc): Move transformation...
24191 * match.pd (C - X CMP X): ... here.
24193 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24195 * config/sparc/sparc.c (sparc_option_override): Set function
24196 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24197 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24199 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24201 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24203 2017-05-18 Marek Polacek <polacek@redhat.com>
24206 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24207 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24208 gimple_assign_load_p.
24210 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24212 PR middle-end/80692
24213 * real.c (do_compare): Give decimal_do_compare preference over
24214 comparing just the signs.
24216 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24218 * doc/md.texi (Canonicalization of Instructions): Describe the
24219 canonical form of instructions that inherently set a condition
24222 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24224 PR middle-end/80775
24225 * tree-cfg.c: Move deletion of unreachable case statements to after
24226 the merging of consecutive case labels.
24228 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24230 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24231 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24232 restoring of callee-saved registers.
24234 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24236 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24237 * config/visium/visium.c (single_set_and_flags): Likewise.
24238 * config/visium/visium.md (Substitutions): Likewise.
24240 2017-05-17 Martin Liska <mliska@suse.cz>
24242 * cfg.c: Introduce dump_flags_t type and
24243 use it instead of int type.
24245 * cfghooks.c: Likewise.
24246 * cfghooks.h (struct cfg_hooks): Likewise.
24247 * cfgrtl.c: Likewise.
24248 * cfgrtl.h: Likewise.
24249 * cgraph.c (cgraph_node::get_body): Likewise.
24250 * coretypes.h: Likewise.
24251 * domwalk.c: Likewise.
24252 * domwalk.h: Likewise.
24253 * dumpfile.c (struct dump_option_value_info): Likewise.
24254 (dump_enable_all): Likewise.
24255 (dump_switch_p_1): Likewise.
24256 (opt_info_switch_p): Likewise.
24257 * dumpfile.h (enum tree_dump_index): Likewise.
24258 (struct dump_file_info): Likewise.
24259 * genemit.c: Likewise.
24260 * generic-match-head.c: Likewise.
24261 * gengtype.c (open_base_files): Likewise.
24262 * gimple-pretty-print.c: Likewise.
24263 * gimple-pretty-print.h: Likewise.
24264 * graph.c (print_graph_cfg): Likewise.
24265 * graphite-scop-detection.c (dot_all_sese): Likewise.
24266 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24267 * loop-unroll.c (report_unroll): Likewise.
24268 * passes.c (pass_manager::register_one_dump_file): Likewise.
24269 * print-tree.c: Likewise.
24270 * statistics.c: Likewise.
24271 * tree-cfg.c: Likewise.
24272 * tree-cfg.h: Likewise.
24273 * tree-dfa.c: Likewise.
24274 * tree-dfa.h: Likewise.
24275 * tree-dump.c (dump_function): Likewise.
24276 * tree-dump.h (struct dump_info): Likewise.
24277 * tree-pretty-print.c: Likewise.
24278 * tree-pretty-print.h: Likewise.
24279 * tree-ssa-live.c: Likewise.
24280 * tree-ssa-live.h: Likewise.
24281 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24282 * tree-vect-loop.c: Likewise.
24283 * tree-vect-slp.c: Likewise.
24285 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24286 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24288 PR tree-optimization/80457
24289 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24290 of all arguments to a statement as scalar_to_vec operations.
24291 (vectorizable_call): Adjust call to vect_model_simple_cost for
24293 (vectorizable_conversion): Likewise.
24294 (vectorizable_assignment): Likewise.
24295 (vectorizable_shift): Likewise.
24296 (vectorizable_operation): Likewise.
24297 (vectorizable_comparison): Likewise.
24298 (vect_is_simple_cond): Record the def types for operands.
24299 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24300 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24301 for statement argument count.
24303 2017-05-16 Carl Love <cel@us.ibm.com>
24305 * config/rs6000/rs6000-c: Add support for built-in functions
24306 vector unsigned long long vec_bperm (vector unsigned long long,
24307 vector unsigned char)
24308 vector signed long long vec_mule (vector signed int,
24310 vector unsigned long long vec_mule (vector unsigned int,
24311 vector unsigned int)
24312 vector signed long long vec_mulo (vector signed int,
24314 vector unsigned long long vec_mulo (vector unsigned int,
24315 vector unsigned int)
24316 vector signed char vec_sldw (vector signed char,
24317 vector signed char,
24319 vector unsigned char vec_sldw (vector unsigned char,
24320 vector unsigned char,
24322 vector signed short vec_sldw (vector signed short,
24323 vector signed short,
24325 vector unsigned short vec_sldw (vector unsigned short,
24326 vector unsigned short,
24328 vector signed int vec_sldw (vector signed int,
24331 vector unsigned int vec_sldw (vector unsigned int,
24332 vector unsigned int,
24334 vector signed long long vec_sldw (vector signed long long,
24335 vector signed long long,
24337 vector unsigned long long vec_sldw (vector unsigned long long,
24338 vector unsigned long long,
24340 * config/rs6000/rs6000-c: Add support for built-in functions
24341 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24342 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24343 * doc/extend.texi: Update the built-in documentation for the
24344 new built-in functions.
24346 2017-05-16 Marek Polacek <polacek@redhat.com>
24350 * tree.c (save_expr): Don't fold the expression.
24352 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24354 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24355 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24356 and (?*y,m). Update insn attributes.
24358 2017-05-16 Martin Liska <mliska@suse.cz>
24360 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24361 flags argument of print_gimple_stmt, print_gimple_expr,
24362 print_generic_stmt and print_generic_expr.
24363 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24364 * coretypes.h: Likewise.
24365 * except.c (dump_eh_tree): Likewise.
24366 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24367 * gimple-pretty-print.h: Likewise.
24368 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24369 (backprop::push_to_worklist): Likewise.
24370 (backprop::pop_from_worklist): Likewise.
24371 (backprop::process_use): Likewise.
24372 (backprop::intersect_uses): Likewise.
24373 (note_replacement): Likewise.
24374 * gimple-ssa-store-merging.c
24375 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24376 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24377 (pass_store_merging::execute): Likewise.
24378 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24379 (ssa_base_cand_dump_callback): Likewise.
24380 (dump_incr_vec): Likewise.
24381 (replace_refs): Likewise.
24382 (replace_mult_candidate): Likewise.
24383 (create_add_on_incoming_edge): Likewise.
24384 (create_phi_basis): Likewise.
24385 (insert_initializers): Likewise.
24386 (all_phi_incrs_profitable): Likewise.
24387 (introduce_cast_before_cand): Likewise.
24388 (replace_one_candidate): Likewise.
24389 * gimplify.c (gimplify_expr): Likewise.
24390 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24391 (set_rename): Likewise.
24392 (rename_uses): Likewise.
24393 (copy_loop_phi_nodes): Likewise.
24394 (add_close_phis_to_merge_points): Likewise.
24395 (copy_loop_close_phi_args): Likewise.
24396 (copy_cond_phi_args): Likewise.
24397 (graphite_copy_stmts_from_block): Likewise.
24398 (translate_pending_phi_nodes): Likewise.
24399 * graphite-poly.c (print_pdr): Likewise.
24400 (dump_gbb_cases): Likewise.
24401 (dump_gbb_conditions): Likewise.
24402 (print_scop_params): Likewise.
24403 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24404 (build_cross_bb_scalars_use): Likewise.
24405 (gather_bbs::before_dom_children): Likewise.
24406 * hsa-dump.c (dump_hsa_immed): Likewise.
24407 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24408 (get_replacement_map): Likewise.
24409 * ipa-inline-analysis.c (dump_condition): Likewise.
24410 (estimate_function_body_sizes): Likewise.
24411 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24412 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24413 * ipa-prop.c (ipa_dump_param): Likewise.
24414 (ipa_print_node_jump_functions_for_edge): Likewise.
24415 (ipa_modify_call_arguments): Likewise.
24416 (ipa_modify_expr): Likewise.
24417 (ipa_dump_param_adjustments): Likewise.
24418 (ipa_dump_agg_replacement_values): Likewise.
24419 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24420 * ipa-pure-const.c (check_stmt): Likewise.
24421 (pass_nothrow::execute): Likewise.
24422 * ipa-split.c (execute_split_functions): Likewise.
24423 * omp-offload.c (dump_oacc_loop_part): Likewise.
24424 (dump_oacc_loop): Likewise.
24425 * trans-mem.c (tm_log_emit): Likewise.
24426 (tm_memopt_accumulate_memops): Likewise.
24427 (dump_tm_memopt_set): Likewise.
24428 (dump_tm_memopt_transform): Likewise.
24429 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24430 (print_loop): Likewise.
24431 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24432 (chkp_gather_checks_info): Likewise.
24433 (chkp_get_check_result): Likewise.
24434 (chkp_remove_check_if_pass): Likewise.
24435 (chkp_use_outer_bounds_if_possible): Likewise.
24436 (chkp_reduce_bounds_lifetime): Likewise.
24437 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24438 (chkp_mark_completed_bounds): Likewise.
24439 (chkp_register_incomplete_bounds): Likewise.
24440 (chkp_mark_invalid_bounds): Likewise.
24441 (chkp_maybe_copy_and_register_bounds): Likewise.
24442 (chkp_build_returned_bound): Likewise.
24443 (chkp_get_bound_for_parm): Likewise.
24444 (chkp_build_bndldx): Likewise.
24445 (chkp_get_bounds_by_definition): Likewise.
24446 (chkp_generate_extern_var_bounds): Likewise.
24447 (chkp_get_bounds_for_decl_addr): Likewise.
24448 * tree-chrec.c (chrec_apply): Likewise.
24449 * tree-data-ref.c (dump_data_reference): Likewise.
24450 (dump_subscript): Likewise.
24451 (dump_data_dependence_relation): Likewise.
24452 (analyze_overlapping_iterations): Likewise.
24453 * tree-inline.c (expand_call_inline): Likewise.
24454 (tree_function_versioning): Likewise.
24455 * tree-into-ssa.c (dump_defs_stack): Likewise.
24456 (dump_currdefs): Likewise.
24457 (dump_names_replaced_by): Likewise.
24458 (dump_update_ssa): Likewise.
24459 (update_ssa): Likewise.
24460 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24461 * tree-parloops.c (build_new_reduction): Likewise.
24462 (try_create_reduction_list): Likewise.
24463 (ref_conflicts_with_region): Likewise.
24464 (oacc_entry_exit_ok_1): Likewise.
24465 (oacc_entry_exit_single_gang): Likewise.
24466 * tree-pretty-print.h: Likewise.
24467 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24468 (get_scalar_evolution): Likewise.
24469 (add_to_evolution): Likewise.
24470 (get_loop_exit_condition): Likewise.
24471 (analyze_evolution_in_loop): Likewise.
24472 (analyze_initial_condition): Likewise.
24473 (analyze_scalar_evolution): Likewise.
24474 (instantiate_scev): Likewise.
24475 (number_of_latch_executions): Likewise.
24476 (gather_chrec_stats): Likewise.
24477 (final_value_replacement_loop): Likewise.
24478 (scev_const_prop): Likewise.
24479 * tree-sra.c (dump_access): Likewise.
24480 (disqualify_candidate): Likewise.
24481 (create_access): Likewise.
24482 (reject): Likewise.
24483 (maybe_add_sra_candidate): Likewise.
24484 (create_access_replacement): Likewise.
24485 (analyze_access_subtree): Likewise.
24486 (analyze_all_variable_accesses): Likewise.
24487 (sra_modify_assign): Likewise.
24488 (initialize_constant_pool_replacements): Likewise.
24489 (find_param_candidates): Likewise.
24490 (decide_one_param_reduction): Likewise.
24491 (replace_removed_params_ssa_names): Likewise.
24492 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24493 * tree-ssa-copy.c (dump_copy_of): Likewise.
24494 (copy_prop_visit_cond_stmt): Likewise.
24495 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24496 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24497 (record_equivalences_from_stmt): Likewise.
24498 * tree-ssa-dse.c (compute_trims): Likewise.
24499 (delete_dead_call): Likewise.
24500 (delete_dead_assignment): Likewise.
24501 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24502 (forward_propagate_into_cond): Likewise.
24503 (pass_forwprop::execute): Likewise.
24504 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24505 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24507 (move_computations_worker): Likewise.
24508 (execute_sm): Likewise.
24509 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24510 (remove_exits_and_undefined_stmts): Likewise.
24511 (remove_redundant_iv_tests): Likewise.
24512 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24513 (adjust_iv_update_pos): Likewise.
24514 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24515 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24516 (value_replacement): Likewise.
24517 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24518 * tree-ssa-pre.c (print_pre_expr): Likewise.
24519 (get_representative_for): Likewise.
24520 (create_expression_by_pieces): Likewise.
24521 (insert_into_preds_of_block): Likewise.
24522 (eliminate_insert): Likewise.
24523 (eliminate_dom_walker::before_dom_children): Likewise.
24524 (eliminate): Likewise.
24525 (remove_dead_inserted_code): Likewise.
24526 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24527 * tree-ssa-reassoc.c (get_rank): Likewise.
24528 (eliminate_duplicate_pair): Likewise.
24529 (eliminate_plus_minus_pair): Likewise.
24530 (eliminate_not_pairs): Likewise.
24531 (undistribute_ops_list): Likewise.
24532 (eliminate_redundant_comparison): Likewise.
24533 (update_range_test): Likewise.
24534 (optimize_range_tests_var_bound): Likewise.
24535 (optimize_vec_cond_expr): Likewise.
24536 (rewrite_expr_tree): Likewise.
24537 (rewrite_expr_tree_parallel): Likewise.
24538 (linearize_expr): Likewise.
24539 (break_up_subtract): Likewise.
24540 (linearize_expr_tree): Likewise.
24541 (attempt_builtin_powi): Likewise.
24542 (attempt_builtin_copysign): Likewise.
24543 (transform_stmt_to_copy): Likewise.
24544 (transform_stmt_to_multiply): Likewise.
24545 (dump_ops_vector): Likewise.
24546 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24547 (print_scc): Likewise.
24548 (set_ssa_val_to): Likewise.
24549 (visit_reference_op_store): Likewise.
24550 (visit_use): Likewise.
24551 (sccvn_dom_walker::before_dom_children): Likewise.
24552 (run_scc_vn): Likewise.
24553 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24555 (expr_hash_elt::print): Likewise.
24556 (const_and_copies::pop_to_marker): Likewise.
24557 (const_and_copies::record_const_or_copy_raw): Likewise.
24558 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24559 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24560 (dump_predicates): Likewise.
24561 (find_uninit_use): Likewise.
24562 (warn_uninitialized_phi): Likewise.
24563 (pass_late_warn_uninitialized::execute): Likewise.
24564 * tree-ssa.c (verify_vssa): Likewise.
24565 (verify_ssa): Likewise.
24566 (maybe_optimize_var): Likewise.
24567 * tree-vrp.c (dump_value_range): Likewise.
24568 (dump_all_value_ranges): Likewise.
24569 (dump_asserts_for): Likewise.
24570 (register_edge_assert_for_2): Likewise.
24571 (vrp_visit_cond_stmt): Likewise.
24572 (vrp_visit_switch_stmt): Likewise.
24573 (vrp_visit_stmt): Likewise.
24574 (vrp_visit_phi_node): Likewise.
24575 (simplify_cond_using_ranges_1): Likewise.
24576 (fold_predicate_in): Likewise.
24577 (evrp_dom_walker::before_dom_children): Likewise.
24578 (evrp_dom_walker::push_value_range): Likewise.
24579 (evrp_dom_walker::pop_value_range): Likewise.
24580 (execute_early_vrp): Likewise.
24582 2017-05-16 Richard Biener <rguenther@suse.de>
24584 * dwarf2out.c (loc_list_from_tree_1): Do not create
24585 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24587 2017-05-16 Richard Biener <rguenther@suse.de>
24589 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24591 (note_variable_value_in_expr): If we resolved the decl ref
24592 do not push to the stack.
24594 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24596 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24597 operations in fast-math mode.
24598 (vaddq_f16): Likewise.
24599 (vmul_f16): Likewise.
24600 (vmulq_f16): Likewise.
24601 (vsub_f16): Likewise.
24602 (vsubq_f16): Likewise.
24603 * config/arm/neon.md (add<mode>3): New.
24605 (fma:<VH:mode>3): New. Also remove outdated comment.
24608 2017-05-16 Martin Liska <mliska@suse.cz>
24612 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24613 (odr_types_equivalent_p): Likewise.
24615 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24617 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24619 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24622 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24623 non-interunit SSE move alternatives with '?'.
24624 (zero-extendsidi peephole2): New peephole to skip intermediate
24625 general register in SSE zero-extend sequence.
24627 2017-05-15 Jeff Law <law@redhat.com>
24629 * reorg.c (relax_delay_slots): Create a new variable to hold
24630 the temporary target rather than clobbering TARGET_LABEL.
24632 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24633 missing argument to extract_bit_field call.
24634 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24636 2017-05-15 Martin Liska <mliska@suse.cz>
24639 * gcc.c (process_command): Do not allow empty argument of -o option.
24641 2017-05-15 Renlin Li <renlin.li@arm.com>
24643 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24644 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24645 * config/aarch64/constraints.md (Usf): Add long call check.
24646 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24647 (call_value): Likewise.
24648 (sibcall): Likewise.
24649 (sibcall_value): Likewise.
24651 (call_value_insn): New.
24652 (sibcall_insn): Update rtx pattern.
24653 (sibcall_value_insn): Likewise.
24654 (call_internal): Remove.
24655 (call_value_internal): Likewise.
24656 (sibcall_internal): Likewise.
24657 (sibcall_value_internal): Likewise.
24658 (call_reg): Likewise.
24659 (call_symbol): Likewise.
24660 (call_value_reg): Likewise.
24661 (call_value_symbol): Likewise.
24663 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24666 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24668 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24670 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24671 compatible with CCGOCmode and with CCZmode.
24673 2017-05-14 Martin Sebor <msebor@redhat.com>
24675 PR middle-end/77671
24676 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24677 (gimple_fold_builtin_snprintf): Same.
24678 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24679 (gimple_fold_builtin_snprintf): Same.
24680 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24681 of character types.
24682 (is_call_safe): New function.
24683 (try_substitute_return_value): Call it.
24684 (try_simplify_call): New function.
24685 (pass_sprintf_length::handle_gimple_call): Call it.
24687 2017-05-14 Martin Sebor <msebor@redhat.com>
24689 PR middle-end/80669
24690 * builtins.c (expand_builtin_stpncpy): Simplify.
24692 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24694 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24695 * config/i386/i386.h
24696 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24697 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24698 (struct machine_function): Add new members call_ms2sysv,
24699 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24700 (struct machine_frame_state): New fields sp_realigned and
24701 sp_realigned_offset.
24702 * config/i386/i386.c
24703 (enum xlogue_stub): New enum.
24704 (enum xlogue_stub_sets): New enum.
24705 (class xlogue_layout): New class.
24706 (struct ix86_frame): New fields stack_realign_allocate_offset,
24707 stack_realign_offset and outlined_save_offset. Modify comments to
24708 detail stack layout when using out-of-line stubs.
24709 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24710 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24711 -mcall-ms2sysv-xlogues.
24712 (stub_managed_regs): New static variable.
24713 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24714 registers managed by out-of-line stub.
24715 (disable_call_ms2sysv_xlogues): New function.
24716 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24717 m->call_ms2sysv when appropriate and compute frame layout for
24719 (sp_valid_at, fp_valid_at): New inline functions.
24720 (choose_basereg): New function.
24721 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24723 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24724 Use align parameter of choose_baseaddr to generated aligned SSE movs
24726 (pro_epilogue_adjust_stack): Modify to track
24727 machine_frame_state::sp_realigned.
24728 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24729 (ix86_nsaved_sseregs): Likewise.
24730 (ix86_emit_save_regs): Likewise.
24731 (ix86_emit_save_regs_using_mov): Likewise.
24732 (ix86_emit_save_sse_regs_using_mov): Likewise.
24733 (get_scratch_register_on_entry): Likewise.
24734 (gen_frame_set): New function.
24735 (gen_frame_load): Likewise.
24736 (gen_frame_store): Likewise.
24737 (emit_outlined_ms2sysv_save): Likewise.
24738 (emit_outlined_ms2sysv_restore): Likewise.
24739 (ix86_expand_prologue): Modify stack re-alignment code and call
24740 emit_outlined_ms2sysv_save when appropriate.
24741 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24742 parameter rtx_insn *insn, which allows the function to be used to only
24743 generate the notes.
24744 (ix86_expand_epilogue): Modify validity checks of frame and stack
24745 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24746 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24747 * config/i386/predicates.md
24748 (save_multiple): New predicate.
24749 (restore_multiple): Likewise.
24750 * config/i386/sse.md
24751 (save_multiple<mode>): New pattern.
24752 (save_multiple_realign<mode>): Likewise.
24753 (restore_multiple<mode>): Likewise.
24754 (restore_multiple_and_return<mode>): Likewise.
24755 (restore_multiple_leave_return<mode>): Likewise.
24756 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24758 2017-05-14 Julia Koval <julia.koval@intel.com>
24760 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24761 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24762 (__builtin_ia32_xsetbv): New builtins.
24763 * config/i386/i386.c (ix86_expand_special_args_builtin):
24765 (ix86_expand_builtin): Special expand for new intrinsics.
24766 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24767 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24768 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24770 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24772 * cfganal.c (inverted_post_order_compute): Change argument type
24774 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24775 * df-core.c (rest_of_handle_df_initialize): Adjust.
24776 (rest_of_handle_df_finish): Likewise.
24777 (df_analyze_1): Likewise.
24778 (df_analyze): Likewise.
24779 (loop_inverted_post_order_compute): Change argument to be a vec *.
24780 (df_analyze_loop): Adjust.
24781 (df_get_n_blocks): Likewise.
24782 (df_get_postorder): Likewise.
24783 * df.h (struct df_d): Change field to be a vec.
24784 * lcm.c (compute_laterin): Adjust.
24785 (compute_available): Likewise.
24786 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24787 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24788 * tree-ssa-pre.c (compute_antic): Likewise.
24790 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24792 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24793 (depth_first_search::depth_first_search): Change structure init
24794 function to this constructor.
24795 (depth_first_search::add_bb): Rename function to this member.
24796 (depth_first_search::execute): Likewise.
24797 (flow_dfs_compute_reverse_finish): Adjust.
24799 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24801 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24802 (longest_simple_path): Likewise.
24803 * shrink-wrap.c (spread_components): Likewise.
24804 (disqualify_problematic_components): Likewise.
24805 (emit_common_heads_for_components): Likewise.
24806 (emit_common_tails_for_components): Likewise.
24807 (insert_prologue_epilogue_for_components): Likewise.
24809 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24811 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24814 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24816 * df-core.c (df_set_blocks): Start using auto_bitmap.
24817 (df_compact_blocks): Likewise.
24818 * df-problems.c (df_rd_confluence_n): Likewise.
24819 * df-scan.c (df_insn_rescan_all): Likewise.
24820 (df_process_deferred_rescans): Likewise.
24821 (df_update_entry_block_defs): Likewise.
24822 (df_update_exit_block_uses): Likewise.
24823 (df_entry_block_bitmap_verify): Likewise.
24824 (df_exit_block_bitmap_verify): Likewise.
24825 (df_scan_verify): Likewise.
24826 * lra-constraints.c (lra_constraints): Likewise.
24827 (undo_optional_reloads): Likewise.
24828 (lra_undo_inheritance): Likewise.
24829 * lra-remat.c (calculate_gen_cands): Likewise.
24830 (do_remat): Likewise.
24831 * lra-spills.c (assign_spill_hard_regs): Likewise.
24832 (spill_pseudos): Likewise.
24833 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24834 (bitmap_set_subtract_values): Likewise.
24836 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24838 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24839 management with auto_bitmap.
24840 (fix_inter_tick): Likewise.
24841 (fix_recovery_deps): Likewise.
24842 * ira.c (add_store_equivs): Likewise.
24843 (find_moveable_pseudos): Likewise.
24844 (split_live_ranges_for_shrink_wrap): Likewise.
24845 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24846 (rtx_reuse_manager::seen_def_p): Likewise.
24847 (rtx_reuse_manager::set_seen_def): Likewise.
24848 * print-rtl.h (class rtx_reuse_manager): Likewise.
24850 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24852 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24854 (migrate_btr_def): Likewise.
24855 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24856 * df-core.c (loop_post_order_compute): Likewise.
24857 (loop_inverted_post_order_compute): Likewise.
24858 * hsa-common.h: Likewise.
24859 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24860 * init-regs.c (initialize_uninitialized_regs): Likewise.
24861 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24862 (inline_small_functions): Likewise.
24863 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24864 * ira.c (combine_and_move_insns): Likewise.
24865 (build_insn_chain): Likewise.
24866 * loop-invariant.c (find_invariants): Likewise.
24867 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24868 * predict.c (tree_predict_by_opcode): Likewise.
24869 (predict_paths_leading_to): Likewise.
24870 (predict_paths_leading_to_edge): Likewise.
24871 (estimate_loops_at_level): Likewise.
24872 (estimate_loops): Likewise.
24873 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24874 (spread_components): Likewise.
24875 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24876 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24877 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24878 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24879 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24880 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24881 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24882 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24883 (mark_threaded_blocks): Likewise.
24884 (thread_through_all_blocks): Likewise.
24885 * tree-ssa.c (verify_ssa): Likewise.
24886 (execute_update_addresses_taken): Likewise.
24887 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24889 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24891 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24893 (post_order_compute): Likewise.
24894 (inverted_post_order_compute): Likewise.
24895 (pre_and_rev_post_order_compute_fn): Likewise.
24897 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24899 * genrecog.c (int_set::int_set): Explicitly construct our
24900 auto_vec base class.
24901 * vec.h (auto_vec::auto_vec): New constructor.
24903 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24905 * bitmap.h (class auto_bitmap): New constructor taking
24906 bitmap_obstack * argument.
24908 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24910 * bitmap.h (class auto_bitmap): Change type of m_bits to
24911 bitmap_head, and adjust ctor / dtor and member operators.
24913 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24915 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24916 when returned register mode doesn't match original mode.
24918 2017-05-12 Jeff Law <law@redhat.com>
24919 Jakub Jelinek <jakub@redhat.com>
24921 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24922 we look for cc setter after the compare-elim changes.
24923 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24924 within the vector to match what compare-elim now expects.
24925 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24926 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24928 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24929 after the compare-elim changes.
24930 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24931 the vector to match what compare-elim now expects.
24932 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24933 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24934 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24935 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24936 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24938 * config/visium/visium.c (single_set_and_flags): Fix where
24939 we look for cc setter after the compare-elim changes.
24940 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24941 with the vector to match what compare-elim now expects.
24942 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24943 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24944 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24945 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24946 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24947 (neg<mode>2_insn_set_overflow): Likewise.
24949 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24951 PR middle-end/79794
24952 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24953 maybe_expand_insn call, set ops[0].target. If still set after call,
24954 set alt_rtl. Add extra arg to recursive calls.
24955 (extract_bit_field): Add alt_rtl argument. Pass to
24957 * expmed.h (extract_bit_field): Fix prototype.
24958 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24959 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24960 to extract_bit_field_calls.
24961 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24962 Pass alt_rtl to extract_bit_field calls.
24963 * calls.c (store_unaligned_arguments_into_psuedos)
24964 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24965 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24967 * optabs.h (struct expand_operand): Add target bitfield.
24969 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24971 * compare-elim.c (try_eliminate_compare): Canonicalize
24972 operation with embedded compare to
24973 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24974 (set (reg) (operation)].
24976 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24978 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24981 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24982 cost of adding a carry flag for ADC instruction.
24983 [case MINUS]: Ignore the cost of subtracting a carry flag
24984 for SBB instruction.
24986 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24988 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24990 * config/rs6000/bmiintrin.h: New file.
24991 * config/rs6000/bmi2intrin.h: New file.
24992 * config/rs6000/x86intrin.h: New file.
24994 2017-05-12 Jeff Law <law@redhat.com>
24996 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24999 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25001 PR middle-end/80707
25002 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25004 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25006 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25007 early expansion of vector divide builtins.
25008 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25009 builtins identified as having unsigned arguments.
25011 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25013 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25014 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25015 expansion of vector logical operations (and, andc, or, xor,
25018 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25020 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25021 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25023 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25025 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25026 early GIMPLE expansion of vector multiplies.
25028 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25030 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25031 TARGET_HAVE_MOVT conditional.
25032 (movt splitter): Likewise.
25034 2017-05-12 Richard Biener <rguenther@suse.de>
25036 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25037 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25038 Fold all stmts not inplace.
25040 2017-05-12 Richard Biener <rguenther@suse.de>
25042 PR tree-optimization/80713
25043 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25044 inserted_exprs bit for not removed stmts.
25046 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25048 PR middle-end/69921
25049 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25050 parallelized" attribute for parallelized OpenACC kernels.
25051 * omp-offload.c (execute_oacc_device_lower): Use it.
25053 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25054 Set "oacc kernels" attribute.
25055 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25056 parameter. Adjust all users.
25057 (oacc_fn_attrib_kernels_p): Remove function.
25058 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25059 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25060 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25061 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25062 assert "oacc kernels" attribute is set.
25064 2017-05-11 Carl Love <cel@us.ibm.com>
25066 * config/rs6000/rs6000-c: Add support for built-in functions
25067 vector unsigned char vec_popcnt (vector signed char)
25068 vector unsigned char vec_popcnt (vector unsigned char)
25069 vector unsigned short vec_popcnt (vector signed short)
25070 vector unsigned short vec_popcnt (vector unsigned short)
25071 vector unsigned int vec_popcnt (vector signed int)
25072 vector unsigned int vec_popcnt (vector unsigned int)
25073 vector unsigned long long vec_popcnt (vector signed long long)
25074 vector unsigned long long vec_popcnt (vector unsigned long long)
25075 vector signed long long vec_slo (vector signed long long,
25076 vector signed char)
25077 vector signed long long vec_slo (vector signed long long,
25078 vector unsigned char)
25079 vector unsigned long long vec_slo (vector unsigned long long,
25080 vector signed char)
25081 vector unsigned long long vec_slo (vector unsigned long long,
25082 vector unsigned char)
25083 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25084 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25085 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25086 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25087 * doc/extend.texi: Update the built-in documentation file for the
25088 new built-in functions.
25090 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25092 * attribs.h (sorted_attr_string): Move machine independent
25093 functions for target clone support from the i386 port to common
25094 code. Rename ix86_function_versions to common_function_versions.
25095 Rename make_name to make_unique_name.
25096 (common_function_versions): Likewise.
25097 (make_unique_name): Likewise.
25098 (make_dispatcher_decl): Likewise.
25099 (is_function_default_version): Likewise.
25100 * attribs.c (attr_strcmp): Likewise.
25101 (sorted_attr_string): Likewise.
25102 (common_function_versions): Likewise.
25103 (make_unique_name): Likewise.
25104 (make_dispatcher_decl): Likewise.
25105 (is_function_default_version): Likewise.
25106 * config/i386/i386.c (attr_strcmp): Likewise.
25107 (sorted_attr_string): Likewise.
25108 (ix86_function_versions): Likewise.
25109 (make_name): Likewise.
25110 (make_dispatcher_decl): Likewise.
25111 (is_function_default_version): Likewise.
25112 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25114 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25117 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25118 Account for direct move costs for vec_construct of integer
25121 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25124 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25125 (UNSPEC_STX_ATOMIC): Ditto.
25126 (loaddi_via_sse): New insn.
25127 (storedi_via_sse): Ditto.
25128 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25129 Update corresponding peephole2 patterns.
25130 (atomic_storedi_fpu): Ditto.
25132 2017-05-11 Julia Koval <julia.koval@intel.com>
25134 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25135 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25137 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25138 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25139 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25141 2017-05-11 Nathan Sidwell <nathan@acm.org>
25143 * graphite-poly.c: Include dumpfile.h.
25145 * dumpfle.h (dump_function): Declare here ...
25146 * tree-dump.h (dump_function): ... not here.
25147 * dumpfile.c: #include tree-cfg.h.
25148 (dump_function): Move here from ...
25149 * tree-dump.c (dump_function): ... here.
25150 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25151 * graphite-poly.c: Don't include tree-dump.h.
25152 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25153 * print-tree.c: Likewise.
25154 * stor-layout.c: Likewise.
25155 * tree-nested.c: Likewise.
25157 * dumpfile.c (dump_start): Use TDF_FLAGS.
25158 (dump_enable_all): Fix TDF_KIND check thinko.
25160 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25162 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25163 array entries to represent two legal parameterizations of the
25164 overloaded __builtin_cmpb function, as represented by the
25165 P6_OV_BUILTIN_CMPB constant.
25166 (altivec_resolve_overloaded_builtin): Add special case handling
25167 for the __builtin_cmpb function, as represented by the
25168 P6_OV_BUILTIN_CMPB constant.
25169 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25170 (BU_P6_64BIT_2): New macro.
25171 (BU_P6_OVERLOAD_2): New macro
25172 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25173 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25174 (CMPB): Add overload support to represent both 32-bit and 64-bit
25175 compare-bytes function.
25176 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25177 support for TARGET_CMPB.
25178 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25179 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25180 documentation of the __builtin_cmpb overloaded built-in function.
25182 2017-05-11 Richard Biener <rguenther@suse.de>
25184 PR tree-optimization/80705
25185 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25186 bases are not vectorizable.
25188 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25190 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25191 when counting register pressure.
25193 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25195 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25196 (iv_ca_more_deps): Renamed to ...
25197 (iv_ca_compare_deps): ... this.
25198 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25200 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25202 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25204 (determine_group_iv_costs): ... here.
25205 (find_inv_vars_cb): Record inv var if it's not recorded before.
25207 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25209 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25210 (get_shiftadd_cost): Ditto.
25212 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25214 * tree-ssa-address.c: Include header file.
25215 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25217 (add_to_parts): Refactor.
25218 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25219 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25222 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25224 PR tree-optimization/53090
25225 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25227 (extract_cond_operands): Detect condition with IV on both sides
25228 and return COMP_IV_EXPR_2.
25229 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25230 (rewrite_use_compare): Simplify by removing call to function
25231 extract_cond_operands.
25233 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25235 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25236 (extract_cond_operands): Detect condition comparing against non-
25237 invariant bound and return appropriate enum value.
25238 (find_interesting_uses_cond): Update use of extract_cond_operands.
25239 Handle its return value accordingly.
25240 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25242 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25244 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25245 nonlinear iv_use computation in loop invariant sensitive way.
25247 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25249 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25250 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25254 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25255 (dump_cand): Support iv_cand.inv_exprs.
25256 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25258 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25261 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25263 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25265 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25266 as local function. Include necessary header files.
25267 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25269 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25271 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25273 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25275 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25276 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25277 RSHIFT_EXPR and BIT_NOT_EXPR.
25279 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25281 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25282 (adjust_setup_cost): New parameter supporting round up adjustment.
25283 (struct address_cost_data): Delete.
25284 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25285 (split_address_cost, ptr_difference_cost): Delete.
25286 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25287 (struct ainc_cost_data): New struct.
25288 (get_address_cost_ainc): New function.
25289 (get_address_cost, get_computation_cost): Reimplement.
25290 (determine_group_iv_cost_address): Record inv_expr for all uses of
25292 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25293 (iv_ca_has_deps): Reimplemented to ...
25294 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25296 (iv_ca_extend): Call iv_ca_more_deps.
25298 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25300 * tree-ssa-address.c (struct mem_address): Move to header file.
25301 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25302 * tree-ssa-address.h (struct mem_address): Move from C file.
25303 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25305 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25307 * tree-affine.h (aff_combination_type): New interface.
25308 (aff_combination_zero_p): Remove static.
25309 (aff_combination_const_p): New interface.
25310 (aff_combination_singleton_var_p): New interfaces.
25312 2017-05-11 Richard Biener <rguenther@suse.de>
25314 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25315 Skip unreachable blocks and destinations.
25316 (eliminate): Move stmt removal and fixup ...
25317 (fini_eliminate): ... here. Skip inserted exprs.
25318 (pass_pre::execute): Move fini_pre after fini_eliminate.
25319 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25320 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25321 PRE to get rid of dead code that has invalid SSA form and
25322 split critical edges again.
25324 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25326 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25328 2017-05-11 Richard Biener <rguenther@suse.de>
25330 * passes.c (execute_function_todo): Verify loops if they are
25331 said to be up-to-date.
25332 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25333 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25335 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25338 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25339 handle calling assemble_external ourself.
25342 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25343 modes with zero size. Enhance comment.
25345 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25347 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25348 built-ins for vec_xl and vec_xst with short and char pointer
25351 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25353 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25354 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25355 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25356 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25357 (_mm_maskz_min_round_ss): New intrinsics.
25358 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25359 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25360 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25361 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25362 (__builtin_ia32_minss_mask_round): New builtins.
25363 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25364 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25365 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25367 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25368 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25370 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25373 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25375 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25376 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25377 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25378 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25379 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25380 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25381 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25382 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25383 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25384 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25385 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25386 (__builtin_ia32_mulss_mask_round): New builtins.
25387 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25388 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25389 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25391 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25392 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25394 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25397 2017-05-10 Julia Koval <julia.koval@intel.com>
25399 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25400 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25401 (_mm256_setr_m128i): New intrinsics.
25403 2017-05-10 Julia Koval <julia.koval@intel.com>
25405 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25406 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25407 (_mm_maskz_rcp14_ss): New intrinsics.
25408 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25409 (__builtin_ia32_rcp14ss_mask): New builtins.
25410 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25412 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25414 PR tree-optimization/51513
25415 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25416 (assert_unreachable_fallthru_edge_p): Use it.
25417 (group_case_labels_stmt): Likewise.
25418 * tree-cfg.h: Prototype it.
25419 * stmt.c: Include cfghooks.h and tree-cfg.h.
25420 (emit_case_dispatch_table) <gap_label>: New local variable.
25421 Use it to fill dispatch table gaps.
25422 Test for default_label before updating probabilities.
25423 (expand_case) <default_label>: Remove unneeded initialization.
25424 Test for unreachable default case statement and remove its edge.
25425 Set default_label accordingly.
25426 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25428 2017-05-10 Carl Love <cel@us.ibm.com>
25430 * config/rs6000/rs6000-c: Add support for built-in functions
25431 vector signed char vec_neg (vector signed char)
25432 vector signed short int vec_neg (vector short int)
25433 vector signed int vec_neg (vector signed int)
25434 vector signed long long vec_neg (vector signed long long)
25435 vector float vec_neg (vector float)
25436 vector double vec_neg (vector double)
25437 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25439 * config/rs6000/altivec.h: Add define for vec_neg
25440 * doc/extend.texi: Update the built-in documentation for the
25441 new built-in functions.
25443 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25445 PR tree-optimization/77644
25446 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25448 2017-05-10 Nathan Sidwell <nathan@acm.org>
25450 * dumpfile.h (TDI_lang_all): New.
25451 (TDF_KIND): New. Renumber others
25452 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25454 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25456 (get_dump_file_name): Adjust suffix generation.
25457 (dump_enable_all): Use TDF_KIND.
25458 * doc/invoke.texi (-fdump-lang-all): Document.
25460 * dumpfile.h: Tabify.
25462 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25465 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25466 Move member access before delete.
25468 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25470 * tree-inline.c (expand_call_inline): Split block at stmt
25473 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25476 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25477 are now unused after splitting mov{sf,sd}_hardfloat.
25478 (f32_lr2): Likewise.
25479 (f32_lm): Likewise.
25480 (f32_lm2): Likewise.
25481 (f32_li): Likewise.
25482 (f32_li2): Likewise.
25483 (f32_lv): Likewise.
25484 (f32_sr): Likewise.
25485 (f32_sr2): Likewise.
25486 (f32_sm): Likewise.
25487 (f32_sm2): Likewise.
25488 (f32_si): Likewise.
25489 (f32_si2): Likewise.
25490 (f32_sv): Likewise.
25491 (f32_dm): Likewise.
25492 (f32_vsx): Likewise.
25493 (f32_av): Likewise.
25494 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25495 For movsf, order stores so the VSX stores occur before the GPR
25496 store which encourages the register allocator to use a traditional
25497 FPR instead of a GPR. For movsd, order the stores so that the GPR
25498 store comes before the VSX stores to allow the power6 to work.
25499 This is due to the power6 not having a 32-bit integer store
25500 instruction from a FPR.
25501 (movsf_hardfloat): Likewise.
25502 (movsd_hardfloat): Likewise.
25504 2017-05-09 Martin Sebor <msebor@redhat.com>
25506 PR translation/80280
25507 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25510 PR translation/80280
25511 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25512 data member added in r247778.
25513 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25515 2017-05-09 Nathan Sidwell <nathan@acm.org>
25517 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25519 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25522 2017-05-09 Marek Polacek <polacek@redhat.com>
25524 * doc/invoke.texi: Fix typo.
25526 2017-05-09 Richard Biener <rguenther@suse.de>
25528 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25529 (vrp_val_is_min): Likewise.
25530 (set_value_range_to_value): Likewise.
25531 (set_value_range_to_nonnegative): Likewise.
25532 (gimple_assign_nonzero_p): Likewise.
25533 (gimple_stmt_nonzero_p): Likewise.
25534 (vrp_int_const_binop): Likewise. Remove unreachable case.
25535 (adjust_range_with_scev): Adjust comments.
25536 (compare_range_with_value): Likewise.
25537 (extract_range_from_phi_node): Likewise.
25538 (test_for_singularity): Likewise.
25540 2017-05-09 Richard Biener <rguenther@suse.de>
25542 * tree-vrp.c (get_single_symbol): Add assert that we don't
25543 get overflowed constants as invariant part.
25544 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25545 checks. Use wi::cmp instead of recursing for integer constants.
25546 (compare_values): Just ignore whether we assumed undefined
25547 overflow instead of failing the compare.
25548 (extract_range_for_var_from_comparison_expr): Add comment before the
25549 TREE_NO_WARNING sets.
25550 (test_for_singularity): Likewise.
25551 (extract_range_from_comparison): Do not disable optimization
25552 when we assumed undefined overflow.
25553 (extract_range_basic): Remove init of unused var.
25555 2017-05-09 Richard Biener <rguenther@suse.de>
25557 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25558 (extract_range_from_multiplicative_op_1): Adjust.
25559 (extract_range_from_binary_expr_1): Use int_const_binop.
25561 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25564 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25565 rs6000_store_data_bypass_p in seven define_bypass directives and
25566 in several comments.
25567 * config/rs6000/rs6000-protos.h: Add prototype for
25568 rs6000_store_data_bypass_p function.
25569 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25570 function implements slightly different (rs6000-specific) semantics
25571 than store_data_bypass_p, returning false rather than aborting
25572 with assertion error when arguments do not satisfy the
25573 requirements of store data bypass.
25574 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25575 rs6000_store_data_bypass_p.
25577 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25579 * config/xtensa/xtensa-protos.h
25580 (xtensa_initial_elimination_offset): New declaration.
25581 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25582 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25583 macro definition, add case for FRAME_POINTER_REGNUM when
25584 FRAME_GROWS_DOWNWARD.
25585 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25586 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25587 xtensa_initial_elimination_offset.
25589 2017-05-08 Nathan Sidwell <nathan@acm.org>
25591 * doc/invoke.texi: Alphabetize -fdump options.
25593 2017-05-08 Martin Sebor <msebor@redhat.com>
25595 PR translation/80280
25596 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25598 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25600 * target.def (compute_frame_layout): New optional target hook.
25601 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25602 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25603 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25605 * reload1.c (verify_initial_elim_offsets): Likewise.
25606 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25607 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25608 (arm_get_frame_offsets): Split up into this ...
25609 (arm_compute_frame_layout): ... and this function.
25611 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25613 * config/aarch64/constraints.md (Usa): New constraint.
25614 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25616 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25618 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25619 with_multilib_list after it has been checked.
25621 2017-05-08 Richard Biener <rguenther@suse.de>
25623 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25624 (bitmap_set_subtract_values): Likewise.
25626 2017-05-08 Richard Biener <rguenther@suse.de>
25628 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25629 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25631 (gimple_stmt_nonzero_warnv_p): Rename to ...
25632 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25634 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25635 (extract_range_basic): Adjust, do not disable propagation on
25636 strict overflow sensitive simplification.
25637 (vrp_visit_cond_stmt): Likewise.
25639 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25641 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25642 body size unconditionally.
25644 2017-05-07 Jeff Law <law@redhat.com>
25647 2017-05-06 Jeff Law <law@redhat.com>
25648 PR tree-optimization/78496
25649 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25652 PR tree-optimization/78496
25653 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25654 (simplify_stmt_using_ranges): Call it.
25655 (vrp_dom_walker::before_dom_children): Extract equivalences
25656 from an ASSERT_EXPR with an equality comparison against a
25659 2017-05-06 Jeff Law <law@redhat.com>
25661 PR tree-optimization/78496
25662 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25665 PR tree-optimization/78496
25666 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25667 (simplify_stmt_using_ranges): Call it.
25668 (vrp_dom_walker::before_dom_children): Extract equivalences
25669 from an ASSERT_EXPR with an equality comparison against a
25672 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25674 * lra-constraints.c (lra_copy_reg_equiv): New function.
25675 (split_reg): Use it to copy equivalence information from the
25676 original register to the spill register.
25678 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25680 PR rtl-optimization/75964
25681 * simplify-rtx.c (simplify_const_relational_operation): Remove
25682 invalid handling of comparisons of integer ABS.
25684 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25686 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25687 initialize to zero.
25688 (init_regs): Remove declaration.
25689 (function_arg_advance_32): Initialize error_p as boolean variable.
25691 2017-05-05 Nathan Sidwell <nathan@acm.org>
25693 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25694 lines. Use for (;;).
25696 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25698 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25699 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25700 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25701 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25702 VF=2 that require versioning.
25704 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25706 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25709 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25711 * diagnostic.h (diagnostic_override_option_index): Convert from
25712 macro to inline function.
25714 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25716 * diagnostic.c (last_module_changed_p): New function.
25717 (set_last_module): New function.
25718 (diagnostic_report_current_module): Convert macro usage to
25719 the above functions.
25720 * diagnostic.h (diagnostic_context::last_module): Strengthen
25721 from const line_map * to const line_map_ordinary *.
25722 (diagnostic_last_module_changed): Delete macro.
25723 (diagnostic_set_last_module): Delete macro.
25725 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25727 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25728 with diagnostic_report_diagnostic.
25729 (diagnostic_n_impl_richloc): Likewise.
25730 * diagnostic.h (report_diagnostic): Delete macro.
25731 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25732 with diagnostic_report_diagnostic.
25733 * substring-locations.c (format_warning_va): Likewise.
25735 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25737 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25738 save/restor of format_spec. Move option-printing code to...
25739 (print_option_information): ...this new function, and
25740 reimplement by simply printing to the pretty_printer,
25741 rather than appending to the format string.
25743 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25745 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25746 handling logic into...
25747 (update_effective_level_from_pragmas): ...this new function.
25749 2017-05-04 Andrew Waterman <andrew@sifive.com>
25751 * config/riscv/riscv.opt (mstrict-align): New option.
25752 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25753 (SLOW_UNALIGNED_ACCESS): Define.
25754 (riscv_slow_unaligned_access): Declare.
25755 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25757 (riscv_slow_unaligned_access): New variable.
25758 (rocket_tune_info): Set slow_unaligned_access to true.
25759 (optimize_size_tune_info): Set slow_unaligned_access to false.
25760 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25761 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25762 (riscv_option_override): Set riscv_slow_unaligned_access.
25763 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25765 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25767 * config/riscv/riscv.md: Unify indentation.
25769 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25774 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25776 (extendsi<mode>2): Add support for doing sign extension via
25777 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25778 don't have ISA 3.0 instructions.
25779 (extendsi<mode>2 splitter): Likewise.
25780 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25781 generate the normal insns since SImode can now go in vector
25782 registers. Disallow the special UNSPECs needed for previous
25783 machines to hide SImode being used. Add new insns
25784 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25785 (fix_trunc<mode>si2_stfiwx): Likewise.
25786 (fix_trunc<mode>si2_internal): Likewise.
25787 (fixuns_trunc<mode>si2): Likewise.
25788 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25789 (fctiw<u>z_<mode>_smallint): Likewise.
25790 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25791 of floating point to 32-bit integer from doing a direct move to
25792 the GPR registers to do a store.
25793 (fctiwz_<mode>): Break long line.
25795 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25797 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25798 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25799 (addr_list, addr_offset_valid_p): New.
25800 (split_address_groups): Check offset validity with above function.
25801 (gt-tree-ssa-loop-ivopts.h): Include header file.
25803 2017-05-05 Nathan Sidwell <nathan@acm.org>
25805 * config.gcc (arm*-*-*): Add missing 'fi'.
25807 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25809 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25810 included in -fopt-info does not matter.
25811 * doc/optinfo.texi (-fopt-info): Fix description of default
25812 behavour. Explicitly say order of options included in -fopt-info
25815 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25817 * config.gcc: Allow combinations of aprofile and rmprofile values for
25818 --with-multilib-list.
25819 * config/arm/t-multilib: New file.
25820 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25821 variables. Remove setting of ISA and floating-point ABI in
25822 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25823 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25824 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25825 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25827 * config/arm/t-rmprofile: Likewise except for the matches changes.
25828 * doc/install.texi (--with-multilib-list): Document the combination of
25829 aprofile and rmprofile values and warn about pitfalls in doing that.
25831 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25833 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25834 (movdi_aarch64): Likewise.
25836 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25838 PR tree-optimization/80632
25839 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25841 (build_arrays): Initialize it for virtual phis.
25842 (fix_phi_nodes): Use it for virtual phis.
25844 PR tree-optimization/80558
25845 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25846 [x, y] op z into [x op, y op z] for op & or | if conditions
25849 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25850 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25853 * config/arm/arm.md (use_literal_pool): Remove.
25854 (64-bit immediate split): No longer takes cost into consideration
25855 if arm_disable_literal_pool is enabled.
25856 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25857 used when arm_disable_literal_pool is enabled.
25858 (arm_max_const_double_inline_cost): Remove use of
25859 arm_disable_literal_pool.
25860 (push_minipool_fix): Add assert.
25861 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25862 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25863 (no_literal_pool_sf_immediate): New.
25865 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25867 PR tree-optimization/80613
25868 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25869 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25871 2017-05-05 Richard Biener <rguenther@suse.de>
25873 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25875 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25877 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25878 of this flag from insn conditions due to removal from r247495.
25880 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25882 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25884 (arm_early_store_addr_dep_ptr): Likewise.
25885 * config/arm/aarch-common-protos.h
25886 (arm_early_load_addr_dep_ptr): Add prototype.
25887 (arm_early_store_addr_dep_ptr): Likewise.
25888 * config/arm/cortex-a53.md: Add new bypasses.
25890 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25892 * tree.c (next_type_uid): Change type to unsigned.
25893 (type_hash_canon): Decrement back next_type_uid if
25894 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25895 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25898 2017-05-04 Martin Sebor <msebor@redhat.com>
25900 * builtins.c: Fix a trivial typo in a comment.
25902 PR middle-end/79234
25903 * builtins.c (check_sizes): Adjust to handle reading past the end.
25904 Avoid printing excessive upper bound of ranges. Use %E to print
25905 tree nodes instead of converting them to %wu.
25906 (expand_builtin_memchr): New function.
25907 (compute_dest_size): Rename...
25908 (compute_objsize): ...to this.
25909 (expand_builtin_memcpy): Adjust.
25910 (expand_builtin_mempcpy): Adjust.
25911 (expand_builtin_strcat): Adjust.
25912 (expand_builtin_strcpy): Adjust.
25913 (check_strncat_sizes): Adjust.
25914 (expand_builtin_strncat): Adjust.
25915 (expand_builtin_strncpy): Adjust and simplify.
25916 (expand_builtin_memset): Adjust.
25917 (expand_builtin_bzero): Adjust.
25918 (expand_builtin_memcmp): Adjust.
25919 (expand_builtin): Handle memcmp.
25920 (maybe_emit_chk_warning): Check strncat just once.
25922 2017-05-04 Martin Sebor <msebor@redhat.com>
25924 PR preprocessor/79214
25925 PR middle-end/79222
25926 PR middle-end/79223
25927 * builtins.c (check_sizes): Add inlining context and issue
25928 warnings even when -Wno-system-headers is set.
25929 (check_strncat_sizes): Same.
25930 (expand_builtin_strncat): Same.
25931 (expand_builtin_memmove): New function.
25932 (expand_builtin_stpncpy): Same.
25933 (expand_builtin): Handle memmove and stpncpy.
25935 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25937 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25938 which is not used any more.
25940 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25942 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25944 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25946 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25947 (cortexa53_tunings): Likewise.
25948 (cortexa57_tunings): Likewise.
25949 (cortexa72_tunings): Likewise.
25950 (cortexa73_tunings): Likewise.
25952 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25954 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25955 Set loop alignment to 8.
25957 2017-05-04 Martin Sebor <msebor@redhat.com>
25959 PR translation/80280
25960 * builtins.c (expand_builtin_object_size): Add missing quoting to
25961 %D and like directives.
25962 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25963 (hsa_type_for_tree_type): Same.
25964 (verify_function_arguments): Same.
25965 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25966 * varasm.c (get_section): Same.
25969 2017-05-04 Martin Sebor <msebor@redhat.com>
25971 PR translation/80280
25972 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25974 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25976 * config/aarch64/aarch64.c (generic_addrcost_table):
25977 Change HI/TI mode setting.
25979 2017-05-04 Martin Jambor <mjambor@suse.cz>
25981 PR tree-optimization/80622
25982 * tree-sra.c (comes_initialized_p): New function.
25983 (build_accesses_from_assign): Only set write lazily when
25984 comes_initialized_p is false.
25985 (analyze_access_subtree): Use comes_initialized_p.
25986 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25987 instead of testing for PARM_DECL.
25989 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25991 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25992 constraint on operand 0 to allow more general addressing modes.
25993 Adjust output template.
25994 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25996 * config/aarch64/aarch64-protos.h
25997 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25998 * config/aarch64/constraints.md (Dp): New address constraint.
25999 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26002 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26004 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26005 update use of estimate_ipcp_clone_size_and_time.
26006 (estimate_local_effects): Update use of
26007 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26008 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26009 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26010 Return nonspecialized time.
26012 2017-05-04 Richard Biener <rguenther@suse.de>
26014 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26015 for the last VUSE which def dominates the PHI. Directly call
26017 (get_continuation_for_phi_1): Remove.
26019 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26021 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26022 to explain the use of truncating division. Cap the number of
26023 iterations to the maximum given by nb_iterations_upper_bound,
26026 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26028 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26029 * configure: Regenerate.
26030 * config.in: Regenerate.
26031 * config/i386/driver-mingw32.c: new file.
26032 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26033 * config.host: Link driver-mingw32.o on MinGW host.
26034 * doc/install.texi: Document new --enable-mingw-wildcard configure
26037 2017-05-04 Marek Polacek <polacek@redhat.com>
26039 PR tree-optimization/80612
26040 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26042 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26043 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26045 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26046 (movt splitter): Likewise.
26047 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26048 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26049 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26050 block for Thumb-1 with MOVT.
26051 (thumb2_legitimate_address_p): Move code block ...
26052 (can_avoid_literal_pool_for_label_p): ... into this new function.
26053 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26055 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26056 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26057 "M-profile targets with the MOVT instruction".
26059 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26061 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26062 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26063 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26065 2017-05-04 Martin Liska <mliska@suse.cz>
26067 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26068 variable cond_code.
26070 2017-05-04 Richard Biener <rguenther@suse.de>
26072 * tree.c (array_at_struct_end_p): Handle arrays at struct
26073 end with flexarrays more conservatively. Refactor and treat
26074 arrays of arrays or aggregates more strict. Fix
26075 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26076 * tree.c (array_at_struct_end_p): Adjust prototype.
26077 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26078 * gimple-fold.c (get_range_strlen): Likewise.
26079 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26081 2017-05-04 Richard Biener <rguenther@suse.de>
26083 PR tree-optimization/31130
26084 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26086 (supports_overflow_infinity): Likewise.
26087 (is_negative_overflow_infinity): Likewise.
26088 (is_positive_overflow_infinity): Likewise.
26089 (is_overflow_infinity): Likewise.
26090 (stmt_overflow_infinity): Likewise.
26091 (overflow_infinity_range_p): Likewise.
26092 (usable_range_p): Remove as always returning true.
26093 (make_overflow_infinity): Remove.
26094 (negative_overflow_infinity): Likewise.
26095 (positive_overflow_infinity): Likewise.
26096 (avoid_overflow_infinity): Likewise.
26097 (set_value_range): Adjust accordingly.
26098 (set_value_range_to_nonnegative): Likewise, remove now unused
26099 overflow_infinity arg.
26100 (vrp_operand_equal_p): Adjust.
26101 (update_value_range): Likewise.
26102 (range_int_cst_singleton_p): Likewise.
26103 (operand_less_p): Likewise.
26104 (compare_values_warnv): Likewise.
26105 (extract_range_for_var_from_comparison_expr): Likewise.
26106 (vrp_int_const_binop): Likewise.
26107 (zero_nonzero_bits_from_vr): Likewise.
26108 (extract_range_from_multiplicative_op_1): Likewise.
26109 (extract_range_from_binary_expr_1): Likewise.
26110 (extract_range_from_unary_expr): Likewise.
26111 (extract_range_from_comparison): Likewise.
26112 (extract_range_basic): Likewise.
26113 (adjust_range_with_scev): Likewise.
26114 (compare_ranges): Likewise.
26115 (compare_range_with_value): Likewise.
26116 (dump_value_range): Likewise.
26117 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26119 (simplify_cond_using_ranges): Adjust.
26121 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26123 * brig-builtins.def: Added a builtin for class_f64.
26124 * builtin-types.def: Added a builtin type needed by class_f64.
26126 2017-05-03 Jason Merrill <jason@redhat.com>
26128 * timevar.def: Add TV_CONSTEXPR.
26130 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26132 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26134 2017-05-03 Martin Jambor <mjambor@suse.cz>
26136 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26137 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26138 * ipa-cp.c (ipcp_propagate_stage): Do not call
26139 ipa_update_after_lto_read.
26140 * ipa-inline.c (ipa_inline): Likewise.
26142 2017-05-03 Martin Jambor <mjambor@suse.cz>
26144 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26145 tag. Added a default constructor and a destructor.
26146 (ipa_edge_args_sum_t): New class;
26147 (ipa_edge_args_sum): Declare.
26148 (ipa_edge_args_vector): Remove declaration.
26149 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26150 (ipa_free_edge_args_substructures): Remove declaration.
26151 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26152 (ipa_edge_args_info_available_for_edge_p): Likewise.
26153 * ipa-prop.c (ipa_edge_args_vector): Removed.
26154 (edge_removal_hook_holder): Likewise.
26155 (edge_duplication_hook_holder): Likewise.
26156 (ipa_edge_args_sum): New variable.
26157 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26158 ipa_edge_args_vector.
26159 (ipa_free_edge_args_substructures): Likewise.
26160 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26161 ipa_edge_args_vector.
26162 (ipa_edge_removal_hook): Turned into method
26163 ipa_edge_args_sum_t::remove.
26164 (ipa_edge_duplication_hook): Turned into method
26165 ipa_edge_args_sum_t::duplicate.
26166 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26167 registering edge hooks.
26168 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26169 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26170 ipa_edge_args_sum instead of ipa_edge_args_vector.
26171 * ipa-profile.c (ipa_profile): Likewise.
26173 2017-05-03 Martin Jambor <mjambor@suse.cz>
26175 * symbol-summary.h (function_summary): New method exists.
26176 (function_summary::symtab_removal): Deallocate through release.
26177 (call_summary): New class.
26178 (gt_ggc_mx): New overload.
26179 (gt_pch_nx): Likewise.
26180 (gt_pch_nx): Likewise.
26182 2017-05-03 Jeff Law <law@redhat.com>
26184 PR tree-optimization/78496
26185 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26186 from simplify_cond_using_ranges. Split off code to walk
26187 backwards through casts into ...
26188 (simplify_cond_using_ranges_2): New function.
26189 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26190 (execute_vrp): After identifying jump threads, call
26191 simplify_cond_using_ranges_2.
26193 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26196 * ipa-inline.h (inline_summary): Add ctor.
26197 (create_ggc): Do not use ggc_cleared_alloc.
26199 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26200 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26202 * gcc.c (handle_braces): Support escaping in switch matching
26204 * doc/invoke.texi (Spec Files): Document it.
26205 Remove superfluous @code markup in items.
26207 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26209 * diagnostic-show-locus.c (struct column_range): New struct.
26210 (get_affected_columns): New function.
26211 (get_printed_columns): New function.
26212 (struct correction): New struct.
26213 (correction::ensure_capacity): New function.
26214 (correction::ensure_terminated): New function.
26215 (struct line_corrections): New struct.
26216 (line_corrections::~line_corrections): New dtor.
26217 (line_corrections::add_hint): New function.
26218 (layout::print_trailing_fixits): Reimplement in terms of the new
26220 (selftest::test_overlapped_fixit_printing): New function.
26221 (selftest::diagnostic_show_locus_c_tests): Call it.
26223 2017-05-03 Nathan Sidwell <nathan@acm.org>
26225 Canonicalize canonical type hashing
26226 * tree.h (type_hash_canon_hash): Declare.
26227 * tree.c (type_hash_list, attribute_hash_list): Move into
26228 type_hash_canon_hash.
26229 (build_type_attribute_qual_variant): Break out hash code calc into
26230 type_hash_canon_hash.
26231 (type_hash_canon_hash): New. Generic type hash computation.
26232 (build_range_type_1, build_array_type_1, build_function_type,
26233 build_method_type_directly, build_offset_type, build_complex_type,
26234 make_vector_type): Call it.
26236 2017-05-03 Richard Biener <rguenther@suse.de>
26238 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26239 When all DRs have unknown misaligned do not always peel
26240 when there is a store but apply the same costing model as if
26241 there were only loads.
26243 2017-05-03 Richard Biener <rguenther@suse.de>
26246 PR tree-optimization/80492
26247 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26248 compare_base_decls returning dont-know properly.
26250 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26252 * config/arm/iterators.md (CCSI): New mode iterator.
26253 (arch): New mode attribute.
26254 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26255 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26256 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26257 code iterator for success result mode.
26258 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26259 the corresponding new insn generators.
26261 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26264 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26265 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26267 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26269 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26270 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26271 (DDR_A): Wrap DDR argument in brackets.
26272 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26273 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26274 (DDR_REVERSED_P): Likewise.
26276 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26278 PR tree-optimization/79472
26279 * tree-switch-conversion.c (struct switch_conv_info): Add
26280 contiguous_range and default_case_nonstandard fields.
26281 (collect_switch_conv_info): Compute contiguous_range and
26282 default_case_nonstandard fields, don't clear final_bb if
26283 contiguous_range and only the default case doesn't have the required
26285 (check_all_empty_except_final): Set default_case_nonstandard instead
26286 of failing if contiguous_range and the default case doesn't have empty
26288 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26289 and only the default case doesn't have the required constants. Skip
26291 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26292 if default_case_nonstandard.
26293 (build_constructors): Build constant 1 just once. Assert that default
26294 values aren't inserted in between cases if contiguous_range. Skip
26296 (build_arrays): Skip virtual phis.
26297 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26298 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26299 Handle virtual phis.
26300 (gen_inbound_check): Handle default_case_nonstandard case.
26301 (process_switch): Adjust check_final_bb caller. Call
26302 gather_default_values with the first non-default case instead of
26303 default case if default_case_nonstandard.
26305 2017-05-02 Nathan Sidwell <nathan@acm.org>
26307 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26308 check. Fix formatting.
26310 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26312 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26313 errors when comparing specialized and unspecialized times.
26315 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26317 * diagnostic-show-locus.c
26318 (layout::should_print_annotation_line_p): Make private.
26319 (layout::print_annotation_line): Make private.
26320 (layout::annotation_line_showed_range_p): Make private.
26321 (layout::show_ruler): Make private.
26322 (layout::print_source_line): Make private. Pass in line and
26323 line_width, rather than calling location_get_source_line. Drop
26325 (layout::print_leading_fixits): New method.
26326 (layout::print_any_fixits): Rename to...
26327 (layout::print_trailing_fixits): ...this, and make private.
26328 Don't print newline fixits.
26329 (diagnostic_show_locus): Move logic for printing one row into...
26330 (layout::print_line): ...this new function. Move the
26331 location_get_source_line call and error-handling from
26332 print_source_line to here. Call print_leading_fixits, and rename
26333 print_any_fixits to print_trailing_fixits.
26334 (selftest::test_fixit_insert_containing_newline): Update now that
26335 newlines are partially supported.
26336 (selftest::test_fixit_insert_containing_newline_2): New test.
26337 (selftest::test_fixit_replace_containing_newline): Update comments.
26338 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26339 * edit-context.c (class added_line): New class.
26340 (class edited_line): Describe newline handling in comment.
26341 (edited_line::actually_edited_p): New method.
26342 (edited_line::print_content): Delete redundant decl.
26343 (edited_line::m_predecessors): New field.
26344 (edited_file::print_content): Call edited_line::print_content.
26345 (edited_file::print_diff): Update to support newlines.
26346 (edited_file::print_diff_hunk): Likewise.
26347 (edited_file::print_run_of_changed_lines): New function.
26348 (edited_file::print_diff_line): Convert to...
26349 (print_diff_line): ...this.
26350 (edited_file::get_effective_line_count): New function.
26351 (edited_line::edited_line): Initialize new field m_predecessors.
26352 (edited_line::~edited_line): Clean up m_predecessors.
26353 (edited_line::apply_fixit): Handle newlines.
26354 (edited_line::get_effective_line_count): New function.
26355 (edited_line::print_content): New function.
26356 (edited_line::print_diff_lines): New function.
26357 (selftest::test_applying_fixits_insert_containing_newline): New
26359 (selftest::test_applying_fixits_replace_containing_newline): New
26361 (selftest::insert_line): New function.
26362 (selftest::test_applying_fixits_multiple_lines): Add example of
26364 (selftest::edit_context_c_tests): Call the new tests.
26366 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26368 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26369 parameter cand. Update dump information.
26370 (get_computation_cost): Update uses.
26372 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26374 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26375 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26376 (get_computation_at, rewrite_use_address): Update use of
26377 get_computation_aff.
26379 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26381 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26382 (get_computation): Delete.
26383 (get_computation_cost): Implement like get_computation_cost_at.
26384 Use get_computation_at.
26385 (get_computation_cost_at): Delete.
26386 (rewrite_use_nonlinear_expr): Use get_computation_at.
26387 (rewrite_use_compare, remove_unused_ivs): Ditto.
26389 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26391 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26393 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26395 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26396 (ivopts_global_cost_for_size): Rename parameter and update uses.
26397 (iv_ca_recount_cost): Update uses.
26398 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26399 candidates seperately in n_invs and n_cands.
26400 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26402 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26404 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26405 (find_inv_vars_cb): New.
26406 (find_depends): Renamed to ...
26407 (find_inv_vars): ... this.
26408 (add_candidate_1, force_var_cost): Call find_inv_vars.
26409 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26411 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26413 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26414 inv_vars. Add inv_exprs.
26415 (struct iv_cand): Rename depends_on to inv_vars.
26416 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26417 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26418 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26419 (dump_cand): Dump inv_vars.
26420 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26421 (record_invariant, find_depends, add_candidate_1): Ditto.
26422 (set_group_iv_cost, force_var_cost): Ditto.
26423 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26424 (get_computation_cost_at, get_computation_cost): Ditto.
26425 (determine_group_iv_cost_generic): Ditto.
26426 (determine_group_iv_cost_address): Ditto.
26427 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26428 (determine_group_iv_costs): Ditto.
26429 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26430 (iv_ca_set_remove_invariants): Renamed to ...
26431 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26432 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26433 (iv_ca_set_add_invariants): Renamed to ...
26434 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26435 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26436 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26437 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26438 (create_new_ivs): Remove useless dump.
26440 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26442 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26444 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26445 (iv_ca_set_no_cp, create_new_iv): Ditto.
26447 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26449 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26451 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26453 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26454 function tree_check2.
26456 2017-05-02 Martin Liska <mliska@suse.cz>
26458 * doc/gcov.texi: Add missing preposition.
26459 * gcov.c (function_info::function_info): Properly fill up
26460 all member variables.
26462 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26464 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26466 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26468 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26470 2017-05-02 Martin Liska <mliska@suse.cz>
26473 * lto-streamer-in.c (lto_read_tree_1): Remove
26474 LTO_STREAMER_DEBUG.
26475 * lto-streamer.c (struct tree_hash_entry): Likewise.
26476 (struct tree_entry_hasher): Likewise.
26477 (tree_entry_hasher::hash): Likewise.
26478 (tree_entry_hasher::equal): Likewise.
26479 (lto_streamer_init): Likewise.
26480 (lto_orig_address_map): Likewise.
26481 (lto_orig_address_get): Likewise.
26482 (lto_orig_address_remove): Likewise.
26483 * lto-streamer.h: Likewise.
26484 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26485 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26487 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26489 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26490 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26491 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26492 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26493 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26494 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26495 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26496 (mm_maskz_sub_ss): New intrinsics.
26497 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26498 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26499 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26500 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26501 (__builtin_ia32_subss_mask_round): New builtins.
26502 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26503 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26504 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26506 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26507 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26509 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26512 2017-05-02 Martin Jambor <mjambor@suse.cz>
26514 PR tree-optimization/78687
26515 * tree-sra.c (access): New field parent.
26516 (process_subtree_disqualification): New function.
26517 (disqualify_candidate): Call it.
26518 (build_accesses_from_assign): Reset write flag if creating an
26520 (build_access_subtree): Fill in parent field and also prpagate
26521 down grp_write flag.
26522 (create_artificial_child_access): New parameter set_grp_write, set
26523 grp_write to its value.
26524 (propagate_subaccesses_across_link): Also propagate grp_write flag
26526 (propagate_all_subaccesses): Push the closest parent back to work
26527 queue if add_access_to_work_queue returned true.
26529 2017-05-02 Richard Biener <rguenther@suse.de>
26531 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26532 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26533 -fstrict-overflow documentation.
26534 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26535 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26536 flag_strict_overflow.
26537 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26538 * lto-opts.c (lto_write_options): Do not stream it.
26539 * lto-wrapper.c (merge_and_complain): Do not handle it.
26540 * opts.c (default_options_table): Do not set -fstrict-overflow.
26541 (finish_options): Likewise do not clear it when sanitizing.
26542 * simplify-rtx.c (simplify_const_relational_operation): Do not
26543 test flag_strict_overflow.
26545 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26547 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26548 using enabled attribute.
26549 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26550 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26551 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26552 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26553 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26554 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26555 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26556 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26557 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26558 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26560 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26562 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26564 2017-05-02 Richard Biener <rguenther@suse.de>
26566 PR tree-optimization/80591
26568 2017-04-10 Richard Biener <rguenther@suse.de>
26570 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26573 2017-05-02 Richard Biener <rguenther@suse.de>
26575 PR tree-optimization/80549
26576 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26577 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26578 headers if they do not have a preheader.
26580 2017-05-02 Martin Liska <mliska@suse.cz>
26583 * common.opt: Fix typo.
26584 * doc/invoke.texi: Likewise.
26586 2017-05-01 Jan Beulich <jbeulich@suse.com>
26588 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26589 swapping, add (x,x,m,x,n) alternative.
26591 2017-05-01 Nathan Sidwell <nathan@acm.org>
26593 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26594 unnecessary unadjusted_alignment check.
26596 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26599 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26601 * gimplify.c (gimplify_cilk_detach): New function.
26602 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26603 * tree-core.h: Document EXPR_CILK_SPAWN.
26604 * tree.h (EXPR_CILK_SPAWN): Define.
26606 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26608 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26609 to use new fixit_hint representation, using the "replace" logic.
26610 (get_line_span_for_fixit_hint): Likewise.
26611 (layout::print_any_fixits): Likewise.
26612 (selftest::test_one_liner_many_fixits): Rename to...
26613 (selftest::test_one_liner_many_fixits_1): ...this, and update
26614 comment and expected output to reflect that the multiple fix-it
26615 hints are now consolidated into one insertion.
26616 (selftest::test_one_liner_many_fixits_2): New test.
26617 (selftest::test_diagnostic_show_locus_one_liner): Update for
26619 (selftest::test_fixit_consolidation): Update for fix-it API
26621 * diagnostic.c (print_parseable_fixits): Likewise.
26622 * edit-context.c (edited_line::m_line_events): Convert from
26623 auto_vec <line_event *> to auto_vec <line_event>.
26624 (class line_event): Convert from abstract base class to a concrete
26625 class, taking over the role of replace_event.
26626 (class insert_event): Delete.
26627 (class replace_event): Rename to class line_event. Convert to
26629 (edit_context::add_fixits): Reimplement.
26630 (edit_context::apply_insert): Delete.
26631 (edit_context::apply_replace): Rename to...
26632 (edit_context::apply_fixit): ...this. Convert to half-open range.
26633 (edited_file::apply_insert): Delete.
26634 (edited_file::apply_replace): Rename to...
26635 (edited_file::apply_fixit): ...this.
26636 (edited_line::~edited_line): Drop deletion of events.
26637 (edited_line::apply_insert): Delete.
26638 (edited_line::apply_replace): Rename to...
26639 (edited_line::apply_fixit): ...this. Convert to half-open range.
26640 Update for change to type of m_line_events.
26641 * edit-context.h (edit_context::apply_insert): Delete.
26642 (edit_context::apply_replace): Rename to...
26643 (edit_context::apply_fixit): ...this.
26645 2017-05-01 Martin Sebor <msebor@redhat.com>
26647 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26650 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26653 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26654 __get_cpuid_max returns 0.
26655 (__get_cpuid_count): Ditto.
26657 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26659 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26660 replacement expression is another instance of one of its arguments.
26662 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26665 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26666 check for stack push/pop autoinc.
26667 * config/i386/i386.c (ix86_agi_dependent): Return false
26668 if the only reason why modified_in_p returned true is that
26669 addr is SP based and set_insn is a push or pop.
26671 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26673 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26676 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26679 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26680 (account_size_time): Use two predicates - exec_pred and
26682 (evaluate_conditions_for_known_args): Compute both clause and
26684 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26685 (inline_summary_t::duplicate): Update.
26686 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26688 (compute_inline_parameters): Likewise.
26689 (estimate_edge_size_and_time): Update caluclation of time.
26690 (estimate_node_size_and_time): Compute both time and nonspecialized
26692 (estimate_ipcp_clone_size_and_time): Update.
26693 (inline_merge_summary): Update.
26694 (do_estimate_edge_time): Update.
26695 (do_estimate_edge_size): Update.
26696 (do_estimate_edge_hints): Update.
26697 (inline_read_section, inline_write_summary): Stream both new predicates.
26698 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26700 (compute_inlined_call_time): Cleanup.
26701 (big_speedup_p): Update.
26702 (edge_badness): Update.
26703 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26704 (size_time_entry): Replace predicate by exec_predicate and
26705 nonconst_predicate.
26706 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26707 (estimate_edge_time): Return also nonspec_time.
26708 (reset_edge_growth_cache): Update.
26710 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26712 PR rtl-optimization/80491
26713 * ifcvt.c (noce_process_if_block): When looking for x setter
26714 with missing else_bb, don't check only the insn right before
26715 cond_earliest, but look for the last insn that x is modified in
26716 within the same bb.
26718 PR rtl-optimization/80491
26719 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26721 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26723 PR tree-optimization/80487
26724 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26726 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26728 PR tree-optimization/79697
26729 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26730 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26731 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26733 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26734 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26736 2017-04-28 Martin Sebor <msebor@redhat.com>
26738 PR tree-optimization/80523
26739 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26740 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26742 (maybe_warn, format_directive, parse_directive): Use new functions.
26743 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26745 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26747 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26749 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26751 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26752 target_header_dir): Set correctly.
26753 * configure: Regenerated.
26754 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26755 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26756 instead of SYSTEM_HEADER_DIR.
26758 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26760 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26761 (estimate_local_effects): Likewise.
26762 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26763 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26764 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26765 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26766 do_estimate_edge_time, estimate_edge_time): Likewise.
26767 * ipa-inline-analysis.c (estimate_node_size_and_time,
26768 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26769 (estimate_time_after_inlining): Remove.
26771 2017-04-28 Martin Liska <mliska@suse.cz>
26773 * doc/gcov.texi: Enhance documentation of gcov.
26775 2017-04-28 Martin Liska <mliska@suse.cz>
26777 * doc/gcov.texi: Sort options in alphabetic order.
26778 * doc/gcov-dump.texi: Likewise.
26779 * doc/gcov-tool.texi: Likewise.
26780 * gcov.c (print_usage): Likewise.
26781 * gcov-dump.c (print_usage): Likewise.
26782 * gcov-tool.c (print_merge_usage_message): Likewise.
26783 (print_rewrite_usage_message): Likewise.
26784 (print_overlap_usage_message): Likewise.
26786 2017-04-28 Martin Liska <mliska@suse.cz>
26788 PR gcov-profile/53915
26789 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26791 2017-04-28 Martin Liska <mliska@suse.cz>
26793 PR gcov-profile/79891
26794 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26795 is marked by compiler as living on a line.
26796 (get_cycles_count): Remove usage of the union.
26797 (output_intermediate_file): Likewise.
26798 (find_source): Fix GNU coding style.
26799 (accumulate_line_counts): Remove old non-all block mode.
26800 (output_lines): Remove usage of the union.
26801 * profile.c (output_location): Include all BBs, even if
26802 belonging to a same line (and file) as a previous BB.
26804 2017-04-28 Martin Liska <mliska@suse.cz>
26806 * gcov.c (process_args): Handle new argument 'w'.
26807 (read_graph_file): Assign ID to BBs.
26808 (output_branch_count): Display BB # if verbose flag is set.
26809 (output_lines): Likewise for arcs.
26810 (print_usage): Add '--verbose' option help.
26811 * doc/gcov.texi: Document --verbose (-w) option.
26813 2017-04-28 Martin Liska <mliska@suse.cz>
26815 * gcov.c (struct block_location_info): New struct.
26816 (process_file): Fill up the new structure.
26817 (read_graph_file): Replace usage of encoding by the newly added
26819 (add_line_counts): Likewise.
26820 (accumulate_line_counts): Remove usage of the union.
26821 (function_info::function_info): New function.
26822 (function_info::~function_info): Likewise.
26823 (process_file): Call delete instead of release_function.
26824 (release_function): Release the function.
26825 (release_structures): Call delete instead of release_function.
26826 (solve_flow_graph): Replace usage of num_blocks.
26827 (find_exception_blocks): Likewise.
26828 (output_lines): Fix GNU coding style.
26830 2017-04-28 Martin Liska <mliska@suse.cz>
26833 * coverage.c (coverage_remove_note_file): New function.
26834 * coverage.h: Declare the function.
26835 * toplev.c (finalize): Clean if an error has been seen.
26837 2017-04-28 Martin Liska <mliska@suse.cz>
26839 PR gcov-profile/80031
26840 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26841 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26842 * gcov.c (read_graph_file): Read just number of blocks.
26843 * profile.c (branch_prob): Do not stream 0 flags per a basic
26846 2017-04-28 Martin Liska <mliska@suse.cz>
26848 * gcov-dump.c (tag_*): Add new argument to declarations.
26849 (dump_gcov_file): Likewise.
26850 (tag_blocks): Add and use new argument depth.
26851 (tag_arcs): Likewise.
26852 (tag_lines): Likewise.
26853 (tag_counters): Likewise.
26854 (tag_summary): Likewise.
26855 (dump_working_sets): Use depth to do a proper indentation.
26857 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26860 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26861 * symtab.c (symtab_node::debug_symtab): Move definition here.
26863 2017-04-28 Richard Biener <rguenther@suse.de>
26865 * lto-streamer.h (LTO_major_version): Bump to 7.
26867 2017-04-28 Richard Biener <rguenther@suse.de>
26869 * tree-vrp.c (assert_info): New struct.
26870 (add_assert_info): New helper.
26871 (register_edge_assert_for_2): Refactor to add asserts to a vector
26873 (register_edge_assert_for_1): Likewise.
26874 (register_edge_assert_for): Likewise.
26875 (finish_register_edge_assert_for): New helper actually registering
26876 asserts where live on edge.
26877 (find_conditional_asserts): Adjust.
26878 (find_switch_asserts): Likewise.
26879 (evrp_dom_walker::try_find_new_range): Generalize.
26880 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26882 2017-04-27 Marek Polacek <polacek@redhat.com>
26885 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26886 arg10 and arg11 to itype.
26888 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26890 * doc/extend.texi (Object Size Checking): Improve grammar.
26892 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26895 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26896 that the logic for permitting reciprocal estimates matches that
26899 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26902 * tree.c (type_cache_hasher::equal): Only compare
26903 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26904 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26905 non-aggregate element types.
26906 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26907 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26909 2017-04-27 Richard Biener <rguenther@suse.de>
26911 PR middle-end/80533
26912 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26913 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26914 keeping a reference to a trailing array.
26916 2017-04-27 Richard Biener <rguenther@suse.de>
26918 PR middle-end/80539
26919 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26920 being in loop-closed SSA form conservatively.
26921 (chrec_fold_multiply_poly_poly): Likewise.
26923 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26925 PR middle-end/79665
26926 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26927 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26929 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26932 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26933 (aarch64_function_arg_alignment): Return unsigned int again, but still
26934 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26935 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26936 Don't emit -Wpsabi note.
26937 (aarch64_function_arg_boundary): Likewise.
26938 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26941 2017-04-26 Nathan Sidwell <nathan@acm.org>
26943 * tree.h (crc32_unsigned_n): Declare.
26944 (crc32_unsigned, crc32_unsigned): Make inline.
26945 * tree.c (crc32_unsigned_bits): Replace with ...
26946 (crc32_unsigned_n): ... this.
26947 (crc32_unsigned, crc32_byte): Remove.
26948 (crc32_string): Remove unnecessary braces.
26950 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26952 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26953 * ipa-inline-analysis.c (MAX_TIME): Remove.
26954 (account_size_time): Use sreal for time.
26955 (dump_inline_summary): Update.
26956 (estimate_function_body_sizes): Update.
26957 (estimate_edge_size_and_time): Update.
26958 (estimate_calls_size_and_time): Update.
26959 (estimate_node_size_and_time): Update.
26960 (inline_merge_summary): Update.
26961 (inline_update_overall_summary): Update.
26962 (estimate_time_after_inlining): Update.
26963 (inline_read_section): Update.
26964 (inline_write_summary): Update.
26965 * ipa-inline.c (compute_uninlined_call_time): Update.
26966 (compute_inlined_call_time): Update.
26967 (recursive_inlining): Update.
26968 (inline_small_functions): Update.
26969 (dump_overall_stats): Update.
26970 * ipa-inline.h: Include sreal.h.
26971 (size_time_entry): Turn time to sreal.
26972 (inline_summary): Turn self_time nad time to sreal.
26974 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26976 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26978 (sreal::stream_out, sreal::stream_in): New.
26979 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26981 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26983 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26986 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26989 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26991 (dimode_scalar_chain::compute_convert_gain): Ditto.
26992 (dimode_scalar_chain::make_vector_copies): Ditto.
26993 (dimode_scalar_chain::convert_reg): Ditto.
26994 (dimode_scalar_chain::convert_insn): Ditto.
26995 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26996 (VI248_AVX512BW_1): New mode iterator.
26997 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26998 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27001 2017-04-25 Martin Sebor <msebor@redhat.com>
27003 PR tree-optimization/80497
27004 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27005 constants are representable in HOST_WIDE_INT.
27006 (parse_directive): Ditto.
27008 2017-04-25 Martin Sebor <msebor@redhat.com>
27011 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27012 (new_zero_array): Adjust signature.
27013 (dom_info::dom_init): Used unsigned rather that size_t.
27014 (dom_info::dom_info): Same.
27016 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27017 Jakub Jelinek <jakub@redhat.com>
27020 * config/arm/arm.c: Include gimple.h.
27021 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27022 returns negative, increment ncrn only if it returned positive.
27023 (arm_needs_doubleword_align): Return int instead of bool,
27024 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27025 members, but if there is any such non-FIELD_DECL
27026 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27027 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27028 returns negative, increment nregs only if it returned positive.
27029 (arm_setup_incoming_varargs): Likewise.
27030 (arm_function_arg_boundary): Emit -Wpsabi note if
27031 arm_needs_doubleword_align returns negative, return
27032 DOUBLEWORD_ALIGNMENT only if it returned positive.
27034 2017-04-25 Marek Polacek <polacek@redhat.com>
27037 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27038 first argument to type.
27040 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27043 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27044 type checks to test for compatibility instead of equality.
27046 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27047 Jakub Jelinek <jakub@redhat.com>
27050 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27052 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27053 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27054 the alignment computation, but return their maximum in warn_alignment.
27055 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27056 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27058 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27059 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27062 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27064 * config/arc/simdext.md (dmpyh): Fix typo.
27066 2017-04-25 Richard Biener <rguenther@suse.de>
27068 PR tree-optimization/80492
27069 * alias.c (compare_base_decls): Handle registers with asm
27070 specification conservatively.
27071 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27072 compare_base_decls returning dont-know properly.
27074 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27076 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27077 (legitimate_offset_address_p): New function.
27078 (arc_legitimate_address_p): Use above function.
27080 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27082 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27084 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27086 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27087 ACCH registers whenever they are available.
27089 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27091 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27092 double regs fix when not used.
27094 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27096 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27098 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27099 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27101 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27103 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27104 when emitting short ADD instructions.
27106 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27108 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27110 (cmpsi_cc_c_insn): Likewise.
27111 (cbranchsi4_scratch): Compute proper instruction length using
27112 compact_hreg_operand.
27113 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27115 2017-04-25 Richard Biener <rguenther@suse.de>
27117 PR middle-end/80509
27118 * passes.c (pass_manager::pass_manager): Initialize
27119 m_name_to_pass_map.
27121 2017-04-25 Richard Biener <rguenther@suse.de>
27123 PR tree-optimization/79201
27124 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27126 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27129 * config/s390/vector.md: Split MEM->GPR vector moves for
27130 non-s_operand addresses.
27132 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27135 * config/s390/predicates.md (reload_const_wide_int_operand): New
27137 * config/s390/s390.md ("movti"): Remove d/P alternative.
27138 ("movti_bigconst"): New pattern definition.
27140 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27143 * s390-protos.h (s390_expand_cs_hqi): Removed.
27144 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27145 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27146 modes as well as CCZ1mode and CCZmode.
27147 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27148 signature of s390_emit_compare_and_swap.
27149 (s390_expand_cs_hqi): Likewise, make static.
27150 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27151 compare-and-swap, in some cases.
27152 (s390_expand_cs): Wrapper function.
27153 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27155 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27156 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27157 patterns for small and large integers. Forbid symref memory operands.
27158 Move expander to s390.c. Require cc register.
27159 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27160 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27161 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27162 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27163 symref memory operands. Remove CC mode and call s390_match_ccmode
27165 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27167 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27169 * config/s390/s390.md (define_peephole2): New peephole to help
27170 combining the load-and-test pattern with volatile memory.
27172 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27174 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27175 with CCZmode for TARGET_Z196.
27177 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27179 PR rtl-optimization/80501
27180 * combine.c (make_compound_operation_int): Set subreg_code to SET
27181 even for AND with mask of the sign bit of mode.
27183 PR rtl-optimization/80500
27184 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27185 sum's initial value.
27187 2017-04-25 Julian Brown <julian@codesourcery.com>
27188 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27190 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27192 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27194 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27196 2017-04-25 Julian Brown <julian@codesourcery.com>
27197 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27199 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27200 (thunderx2t99_sha): New Reservation.
27202 2017-04-25 Julian Brown <julian@codesourcery.com>
27203 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27205 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27206 type for 1-element load.
27208 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27210 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27212 2017-04-24 Martin Jambor <mjambor@suse.cz>
27214 PR tree-optimization/80293
27215 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27216 char arrays not totally scalarizable if it is false.
27217 (analyze_all_variable_accesses): Pass correct value in the new
27218 parameter. Add a statistics counter.
27220 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27222 PR middle-end/79931
27223 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27225 2017-04-24 Richard Biener <rguenther@suse.de>
27227 PR tree-optimization/80494
27228 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27229 out for complex types.
27231 2017-04-24 Richard Biener <rguenther@suse.de>
27233 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27234 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27235 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27236 (DFS): Adjust and never fail.
27237 (sccvn_dom_walker::fail): Remove.
27238 (sccvn_dom_walker::before_dom_children): Adjust.
27239 (run_scc_vn): Likewise and never fail.
27240 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27241 (pass_fre::execute): Likewise.
27243 2017-04-24 Richard Biener <rguenther@suse.de>
27245 PR tree-optimization/79725
27246 * tree-ssa-sink.c (statement_sink_location): Return whether
27247 failure reason was zero uses. Move that check later.
27248 (sink_code_in_bb): Deal with zero uses by removing the stmt
27251 2017-04-24 Richard Biener <rguenther@suse.de>
27254 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27255 pointer-based references.
27257 2017-04-24 Richard Biener <rguenther@suse.de>
27260 * pass_manager.h (pass_manager::operator new): Remove.
27261 (pass_manager::operator delete): Likewise.
27262 * passes.c (pass_manager::operator new): Remove.
27263 (pass_manager::operator delete): Likewise.
27264 (pass_manager::pass_manager): Zero individual pass members.
27266 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27269 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27270 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27271 Check "XEXP (src, 1)" operand here.
27272 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27273 Check "XEXP (src, 1)" operand here.
27274 (dimode_scalar_chain::make_vector_copies): Detect count register
27275 of a shift instruction. Zero extend count register from QImode
27276 to DImode to satisfy vector shift pattern count operand predicate.
27277 Substitute vector shift count operand with a DImode copy.
27278 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27281 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27283 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27284 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27285 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27286 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27287 (UNSPEC_NOREX_MEM): Remove definition.
27289 2017-04-21 Richard Biener <rguenther@suse.de>
27291 PR tree-optimization/79547
27292 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27293 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27294 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27295 without any constraints.
27297 2017-04-21 Richard Biener <rguenther@suse.de>
27299 PR tree-optimization/78847
27300 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27302 2017-04-21 Richard Biener <rguenther@suse.de>
27304 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27305 (build_distinct_type_copy): Likewise.
27306 (build_variant_type_copy): Likewise.
27307 * tree.c (build_qualified_type): Pass down mem-stat info.
27308 (build_distinct_type_copy): Likewise.
27309 (build_variant_type_copy): Likewise.
27311 2017-04-21 Richard Biener <rguenther@suse.de>
27313 PR tree-optimization/80237
27314 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27316 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27317 for a simplified result.
27319 2016-04-21 Richard Biener <rguenther@suse.de>
27321 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27322 sth as strict as a simple_iv but a chrec without symbols and an
27323 operand defined in the loop we are peeling (and not some subloop).
27324 (propagate_constants_for_unrolling): Propagate all constants.
27326 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27329 * config/i386/i386.c (print_reg): Remove assert for disalowed
27330 regno values, call output_operand_lossage instead.
27332 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27335 * config/i386/constraints.md (Yc): New register constraint.
27336 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27337 Use Yc constraint for alternative 2 of operand 0. Remove
27338 preferred_for_speed attribute.
27340 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27342 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27343 lastprivate clauses in SIMT case.
27345 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27347 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27349 2017-04-20 Richard Biener <rguenther@suse.de>
27351 PR tree-optimization/57796
27352 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27353 as N scalar stores.
27354 (vect_model_load_cost): Cost gathers as N scalar loads.
27356 2017-04-20 Richard Biener <rguenther@suse.de>
27358 * ggc-page.c (ggc_allocated_p): Rename to ...
27359 (safe_lookup_page_table_entry): ... this and return the lookup
27361 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27363 2017-04-20 Richard Biener <rguenther@suse.de>
27365 PR tree-optimization/80453
27366 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27367 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27368 from the conditions.
27369 (vn_phi_eq): Pass them down.
27370 (vn_phi_lookup): Record them.
27371 (vn_phi_insert): Likewise.
27373 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27375 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27376 uninitialized variable warning to avoid buffer overrun.
27378 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27381 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27382 is suppressed for '{ 0 }' in C.
27384 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27386 * BASE-VER: Set to 8.0.0.
27388 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27390 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27391 priority .init_array and .fini_array section with SECTION_NOTYPE
27394 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27396 PR middle-end/80423
27397 * tree.h (build_array_type): Add typeless_storage default argument.
27398 * tree.c (type_cache_hasher::equal): Also compare
27399 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27400 (build_array_type): Add typeless_storage argument, set
27401 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27403 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27404 (build_array_type): Likewise. Add typeless_storage argument.
27406 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27407 Jakub Jelinek <jakub@redhat.com>
27409 PR tree-optimization/80426
27410 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27411 operation on symbolic operands, also compute the overflow for the
27412 invariant part when the operation degenerates into a negation.
27414 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27417 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27418 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27421 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27423 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27426 * config/avr/avr.c (tree.h): Include it.
27427 (cgraph.h): Include it.
27428 (avr_encode_section_info): Don't warn for uninitialized progmem
27429 variable if it's just an alias.
27431 2017-04-19 Richard Biener <rguenther@suse.de>
27434 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27435 when needed by AutoPGO.
27437 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27440 * doc/lto.texi: Remove an extra 'that'.
27442 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27444 PR rtl-optimization/80429
27445 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27446 are only used in debug insns.
27448 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27449 Vladimir Makarov <vmakarov@redhat.com>
27451 * config/sparc/predicates.md (input_operand): Add comment. Return
27452 true for any memory operand when LRA is in progress.
27453 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27455 2017-04-18 Jeff Law <law@redhat.com>
27458 * mips.md ({return,simple_return}_internal): Do not overwrite
27461 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27463 PR tree-optimization/80443
27464 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27465 instead of adding 1, subtract -1 and similarly instead of subtracting
27468 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27470 PR rtl-optimization/80357
27471 * haifa-sched.c (tmp_bitmap): New variable.
27472 (model_recompute): Handle duplicate use records.
27473 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27474 (free_global_sched_pressure_data): Free it.
27476 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27479 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27480 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27481 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27482 instead of SYSTEM_HEADER_DIR.
27484 2017-04-18 Jeff Law <law@redhat.com>
27486 PR middle-end/80422
27487 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27488 predecessors after walking up the insn chain.
27490 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27493 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27494 sizetype type into debug info.
27496 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27499 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27500 unneeded test for TARGET_UPPER_REGS_SF.
27501 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27503 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27506 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27507 instead of gsi_after_labels.
27509 2017-04-18 Jeff Law <law@redhat.com>
27511 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27515 2017-04-13 Jeff Law <law@redhat.com>
27516 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27517 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27519 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27522 * config/avr/avr.c (intl.h): Include it.
27523 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27525 2017-04-18 Martin Liska <mliska@suse.cz>
27527 PR gcov-profile/78783
27528 * gcov-tool.c (gcov_output_files): Validate that destination
27529 file is either removed by the tool or by a user.
27531 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27532 Guy Benyei <guybe@mellanox.com>
27534 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27535 block, and do not negate it, the stored id is already negative.
27537 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27539 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27541 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27544 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27545 masks of options that should be turned off if the VSX vector
27546 options are turned off.
27547 (OTHER_P8_VECTOR_MASKS): Likewise.
27548 (OTHER_VSX_VECTOR_MASKS): Likewise.
27549 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27550 rs6000_disable_incompatible_switches to validate no type switches
27552 (rs6000_incompatible_switch): New function to disallow turning on
27553 other vector options if -mno-vsx, -mno-power8-vector, or
27554 -mno-power9-vector are specified.
27556 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27558 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27560 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27562 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27563 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27564 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27565 (ARG_POINTER_CFA_OFFSET): Likewise.
27567 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27569 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27570 conditions to take advantage of various optimizations.
27572 2017-04-13 Jeff Law <law@redhat.com>
27574 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27575 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27576 (zero_extendsidi2_dext): Likewise.
27578 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27581 * fold-const.c (fold_ternary_loc): Revert
27582 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27585 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27587 PR rtl-optimization/80343
27588 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27589 new scratch pseudo.
27591 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27594 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27595 to ubsan_encode_value.
27597 2017-04-13 Jeff Law <law@redhat.com>
27599 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27600 appearing in DEBUG_INSNs.
27602 2017-04-13 Martin Liska <mliska@suse.cz>
27604 PR gcov-profile/80413
27605 * gcov-io.c (gcov_write_string): Copy to buffer just when
27606 allocated size is greater than zero.
27608 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27611 * dwarf2out.c (decls_for_scope): Ignore declarations of
27612 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27614 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27617 * ipa-visibility.c (non_local_p): Fix typos.
27618 (localize_node): When localizing symbol in same comdat group,
27619 dissolve the group only when we know external symbols are going
27621 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27623 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27625 PR tree-optimization/79390
27626 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27627 order does not result in usable sequence, retry with reversed operand
27633 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27634 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27635 op0 instead of fold_convert_loc (loc, type, arg0).
27637 2017-04-12 Jeff Law <law@redhat.com>
27639 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27640 has a delay slot in the generated code.
27642 * config/cris/cris.md (cris_preferred_reload_class): Return
27643 GENNONACR_REGS rather than GENERAL_REGS.
27645 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27648 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27649 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27650 signedness of the result type.
27652 2017-04-12 Richard Biener <rguenther@suse.de>
27653 Jeff Law <law@redhat.com>
27655 PR tree-optimization/80359
27656 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27657 trim stores to TARGET_MEM_REFs.
27659 2017-04-12 Richard Biener <rguenther@suse.de>
27661 PR tree-optimization/79390
27662 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27663 threading case even more.
27665 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27668 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27669 for quad_address_p for TImode, instead of just not indexed_address.
27671 2017-04-12 Richard Biener <rguenther@suse.de>
27672 Bernd Edlinger <bernd.edlinger@hotmail.de>
27674 PR middle-end/79671
27675 * alias.c (component_uses_parent_alias_set_from): Handle
27676 TYPE_TYPELESS_STORAGE.
27677 (get_alias_set): Likewise.
27678 * tree-core.h (tree_type_common): Add typeless_storage flag.
27679 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27680 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27681 for types containing members with TYPE_TYPELESS_STORAGE.
27682 (place_field): Likewise.
27683 (layout_type): Likewise for ARRAY_TYPE.
27684 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27685 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27686 TYPE_TYPELESS_STORAGE.
27687 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27689 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27692 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27693 first argument to type.
27695 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27699 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27700 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27701 (rs6000_expand_binop_builtin): Likewise.
27702 (rs6000_expand_ternop_builtin): Likewise; also add missing
27703 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27704 vshasigma built-ins.
27705 * doc/extend.texi: Document that vec_xxpermdi's third argument
27706 must be a constant.
27708 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27710 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27711 Use shift_const cost parameter when calculating gain of STV shifts.
27713 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27715 PR rtl-optimization/70478
27716 * lra-constraints.c (process_alt_operands): Check memory for
27717 disfavoring memory insn operand.
27719 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27721 PR middle-end/80100
27722 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27723 left shift in unsigned HOST_WIDE_INT type.
27725 PR rtl-optimization/80385
27726 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27727 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27730 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27731 if they have any depend clauses.
27733 2017-04-11 Martin Liska <mliska@suse.cz>
27736 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27737 * ipa-split.c (split_function): Create a local comdat symbol
27738 if caller is in a comdat group.
27740 2017-04-11 Martin Liska <mliska@suse.cz>
27743 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27746 2017-04-11 Martin Sebor <msebor@redhat.com>
27748 PR middle-end/80364
27749 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27750 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27752 (directive::set_width, directive::set_precision, format_character):
27754 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27757 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27760 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27761 conflict, set target->arch_name instead of target->cpu_name.
27763 2017-04-11 Richard Biener <rguenther@suse.de>
27765 PR tree-optimization/80374
27766 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27767 build_zero_cst, remove fold_convertible_p check again.
27769 2017-04-11 Martin Liska <mliska@suse.cz>
27772 * ubsan.c (instrument_object_size): Do not instrument register
27775 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27778 * config/i386/i386-builtin-types.def
27779 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27780 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27781 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27782 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27783 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27784 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27785 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27786 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27787 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27788 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27789 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27790 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27791 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27792 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27793 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27794 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27795 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27796 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27797 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27798 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27799 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27800 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27801 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27802 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27803 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27804 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27805 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27806 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27807 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27808 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27809 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27810 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27811 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27812 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27813 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27814 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27815 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27816 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27817 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27818 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27819 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27820 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27821 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27822 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27823 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27824 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27826 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27827 flag to second_arg_count, handle 4 argument function type _COUNT
27828 aliases, handle second_arg_count on second argument rather than last.
27830 2017-04-10 Jeff Law <law@redhat.com>
27832 PR tree-optimization/80374
27833 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27834 record anything if we can not convert integer_zero_node to the
27837 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27840 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27841 Enhance special handling given to the TARGET_P9_MINMAX option in
27842 relation to certain other options.
27844 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27846 PR tree-optimization/80153
27847 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27848 remove POINTER_PLUS_EXPR's base part directly, rather than through
27851 2017-04-10 Richard Biener <rguenther@suse.de>
27852 Bin Cheng <bin.cheng@arm.com>
27854 PR tree-optimization/80153
27855 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27856 the first element of pointer type aff_tree. Build result expr in
27858 (add_elt_to_tree): Convert to type unconditionally. Remove other
27859 fold_convert calls.
27860 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27861 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27863 2017-04-10 Richard Biener <rguenther@suse.de>
27865 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27868 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27870 PR rtl-optimization/70478
27871 * lra-constraints.c (curr_small_class_check): New.
27872 (update_and_check_small_class_inputs): New.
27873 (process_alt_operands): Update curr_small_class_check. Disfavor
27874 alternative insn memory operands. Check available regs for small
27877 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27880 * config/mips/mips.opt (-mvirt): Update description.
27881 * doc/invoke.texi (-mvirt): Likewise.
27883 2017-04-10 Richard Biener <rguenther@suse.de>
27885 PR middle-end/80362
27886 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27887 looking for NEGATE_EXPR in -A / -B to A / B folding.
27889 2017-04-10 Martin Liska <mliska@suse.cz>
27891 PR gcov-profile/80224
27892 * gcov.c (print_usage): Fix usage string.
27893 (get_gcov_intermediate_filename): Remove.
27894 (output_gcov_file): Use both for normal and intermediate format.
27895 (generate_results): Do not initialize special file for
27896 intermediate format.
27898 2017-04-10 Richard Biener <rguenther@suse.de>
27900 PR tree-optimization/80304
27901 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27904 2017-04-10 Nathan Sidwell <nathan@acm.org>
27907 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27908 (rs6000_init_builtins): Use it.
27910 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27912 * config/arm/arm.md (<mrc>): Add mode to SET source.
27913 (<mrrc>): Likewise.
27915 2017-04-10 Richard Biener <rguenther@suse.de>
27917 PR middle-end/80344
27918 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27920 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27923 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27924 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27925 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27926 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27927 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27928 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27929 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27930 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27931 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27932 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27933 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27934 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27935 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27936 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27937 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27938 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27939 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27940 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27941 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27942 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27943 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27944 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27945 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27947 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27949 PR rtl-optimization/70478
27950 * lra-constraints.c: Reverse the last patch.
27952 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27954 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27955 Add comment for WCHAR_T.
27957 2017-04-08 Martin Liska <mliska@suse.cz>
27960 2017-04-07 Martin Liska <mliska@suse.cz>
27963 * ipa-split.c (split_function): Add function part to a same comdat
27966 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27969 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27971 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27973 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27974 (vec_pairing): ...to this.
27975 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27976 (rs6000_sched_init): Adjust for name change.
27977 (struct rs6000_sched_context): Likewise.
27978 (rs6000_init_sched_context): Likewise.
27979 (rs6000_set_sched_context): Likewise.
27981 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27987 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27989 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27990 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27991 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27993 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27995 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27997 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27999 PR rtl-optimization/70703
28000 * ira-color.c (update_conflict_hard_regno_costs): Use
28001 int64_t instead of HOST_WIDE_INT.
28003 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28005 PR rtl-optimization/70478
28006 * lra-constraints.c (process_alt_operands): Disfavor alternative
28007 insn memory operands.
28009 2017-04-07 Jeff Law <law@redhat.com>
28011 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28012 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28014 2017-04-07 Martin Liska <mliska@suse.cz>
28017 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28018 Show error message instead of an ICE.
28020 2017-04-07 Martin Liska <mliska@suse.cz>
28023 * ipa-split.c (split_function): Add function part to a same comdat
28026 2017-04-07 Richard Biener <rguenther@suse.de>
28028 PR middle-end/80341
28029 * tree.c (get_unwidened): Also handle ! for_type case for
28031 * convert.c (do_narrow): Split out from ...
28032 (convert_to_integer_1): ... here. Do not pass final truncation
28033 type to get_unwidened for TRUNC_DIV_EXPR.
28035 2017-04-07 Richard Biener <rguenther@suse.de>
28037 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28039 (aff_combination_const): Adjust.
28040 (aff_combination_scale): Likewise.
28041 (aff_combination_add_elt): Likewise.
28042 (aff_combination_add_cst): Likewise.
28043 (aff_combination_convert): Likewise.
28044 (add_elt_to_tree): Likewise. Remove unused argument.
28045 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28047 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28049 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28051 * config/arm/arm.c (arm_default_short_enums): Use
28052 ARM_DEFAULT_SHORT_ENUMS.
28053 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28055 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28058 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28059 members with redundant out-of-class redeclaration.
28061 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28064 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28065 * config/i386/i386.md (*zero_extendsidi2):
28066 Add (?*x,*x) and (?*v,*v) alternatives.
28068 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28071 * config/i386/i386.c (ix86_expand_builtin)
28072 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28073 mode from insn data. Convert operands to insn operand mode.
28074 Copy operands that don't satisfy insn predicate to a register.
28076 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28078 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28081 2017-04-06 Richard Biener <rguenther@suse.de>
28083 PR tree-optimization/80334
28084 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28085 preserve alignment of accesses.
28087 2017-04-06 Richard Biener <rguenther@suse.de>
28089 PR tree-optimization/80262
28090 * tree-sra.c (build_ref_for_offset): Preserve address-space
28092 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28093 Drop useless address-space information on MEM_REF offsets.
28095 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28097 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28099 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28101 PR rtl-optimization/70703
28102 * ira-color.c (update_conflict_hard_regno_costs): Use
28103 HOST_WIDE_INT instead of long.
28105 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28108 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28109 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28111 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28112 for x86_64 target. Handle -m3dnowa option.
28114 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28116 PR rtl-optimization/70703
28117 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28118 (update_conflict_hard_regno_costs): Use long instead of unsigned
28119 arithmetic for cost calculation.
28121 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28122 Bernd Edlinger <bernd.edlinger@hotmail.de>
28125 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28128 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28131 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28132 ptr_mode with Pmode throughout.
28133 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28134 into probe_stack_range and use DImode.
28136 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28139 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28140 call_eh_return is true.
28142 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28144 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28145 Initialize last_match_fntype_index.
28147 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28150 * tree-nvr.c: Include internal-fn.h.
28151 (pass_return_slot::execute): Ignore internal calls without
28154 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28155 Richard Biener <rguenther@suse.de>
28158 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28159 captures used multiple times, except for the last use.
28160 * generic-match-head.c: Include gimplify.h.
28162 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28164 PR tree-optimization/79390
28165 * target.h (struct noce_if_info): Declare.
28166 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28167 * target.def (noce_conversion_profitable_p): New target hook.
28168 * ifcvt.h (struct noce_if_info): New type, moved from ...
28169 * ifcvt.c (struct noce_if_info): ... here.
28170 (noce_conversion_profitable_p): Renamed to ...
28171 (default_noce_conversion_profitable_p): ... this. No longer
28173 (noce_try_store_flag_constants, noce_try_addcc,
28174 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28175 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28176 instead of noce_conversion_profitable_p.
28177 * config/i386/i386.c: Include ifcvt.h.
28178 (ix86_option_override_internal): Don't override
28179 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28180 (ix86_noce_conversion_profitable_p): New function.
28181 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28182 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28183 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28184 * doc/tm.texi: Regenerated.
28186 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28188 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28191 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28194 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28195 instructions for small multiply cores.
28197 2017-04-04 Jeff Law <law@redhat.com>
28199 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28201 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28202 that are not set by the loop over the elements.
28204 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28207 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28208 int mode, convert_modes it to mode as unsigned, otherwise use
28209 lowpart_subreg to mode rather than SImode.
28210 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28211 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28212 Use DImode instead of SImode for the shift count operand.
28213 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28216 2017-04-04 Richard Biener <rguenther@suse.de>
28218 PR middle-end/80281
28219 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28220 arithmetic done for the negate or the plus. Simplify.
28221 (A - (-B) -> A + B): Likewise.
28222 * fold-const.c (split_tree): Make sure to not negate pointers.
28224 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28226 PR rtl-optimization/60818
28227 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28228 a compare of comparisons with the thing compared if this results
28229 in a different machine mode.
28231 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28233 * alias.c (base_alias_check): Fix typo in comment.
28234 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28235 * cgraphunit.c (symbol_table::compile): Likewise.
28236 * collect2.c (maybe_run_lto_and_relink): Likewise.
28237 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28238 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28239 * config/avr/avr.c (avr_map_op_t): Likewise.
28240 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28241 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28242 * config/epiphany/epiphany.md (movcc): Likewise.
28243 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28244 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28246 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28247 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28248 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28249 * config/sh/sh.c (sh_rtx_costs): Likewise.
28250 * fold-const.c (fold_truth_andor): Likewise.
28251 * genautomata.c (collapse_flag): Likewise.
28252 * gengtype.h (struct type::u::s): Likewise.
28253 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28254 * input.c (FORMAT_AMOUNT): Likewise.
28255 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28256 (known_aggs_to_agg_replacement_list): Likewise.
28257 * ipa-inline-analysis.c: Likewise.
28258 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28259 * ipa-polymorphic-call.c
28260 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28261 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28262 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28264 * modulo-sched.c (apply_reg_moves): Likewise.
28265 * omp-expand.c (build_omp_regions_1): Likewise.
28266 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28267 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28268 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28269 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28270 * value-prof.c: Likewise.
28271 * var-tracking.c (val_reset): Likewise.
28273 2017-04-03 Richard Biener <rguenther@suse.de>
28275 PR tree-optimization/80275
28276 * fold-const.c (split_address_to_core_and_offset): Handle
28279 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28281 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28282 descriptors is at least equal to that of functions.
28284 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28286 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28288 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28291 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28292 (mov<IMOD4:mode>): New expander.
28293 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28295 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28297 PR rtl-optimization/79405
28298 * fwprop.c (propagations_left): New variable.
28299 (forward_propagate_into): Decrement it.
28300 (fwprop_init): Initialize it.
28301 (fw_prop): If the variable has reached zero, stop propagating.
28302 (fwprop_addr): Ditto.
28304 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28307 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28308 a FUNCTION_DECL, pass it as decl instead of origin to
28311 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28313 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28316 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28319 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28320 TARGET_VSX_SMALL_INTEGER.
28322 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28324 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28325 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28327 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28329 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28330 extraction from odd-numbered MSA register.
28332 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28334 PR middle-end/80173
28335 * expmed.c (store_bit_field_1): Don't attempt to create
28336 a word subreg out of hard registers wider than word if they
28337 have HARD_REGNO_NREGS of 1 for their mode.
28339 PR middle-end/80163
28340 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28341 conversions to integer types wider than word and pointer.
28344 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28345 (rtx_equal_for_cselib_p): Pass 0 to it.
28346 * cselib.c (cselib_hasher::equal): Likewise.
28347 (rtx_equal_for_cselib_1): Add depth argument. If depth
28348 is 128, don't look up VALUE locs and punt. Increment
28349 depth in recursive calls when walking VALUE locs.
28351 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28353 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28354 (make_gcov_file_name): Use the canonical path name for generating
28356 (read_line): Fix handling of files with ascii null bytes.
28358 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28360 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28361 to initialise a vector register instead
28362 of using a const_int.
28364 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28366 PR translation/80189
28367 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28368 diagnostic messages.
28370 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28373 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28374 (dfp_diex_<mode>): Update mode of operand 1.
28375 * doc/extend.texi (dxex, dxexq): Document change to return type.
28376 (diex, diexq): Document change to argument type.
28378 2017-03-30 Martin Jambor <mjambor@suse.cz>
28381 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28382 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28383 it reflects the signature changes performed at the callee side.
28384 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28385 to cgraph_build_function_type_skip_args.
28386 (build_function_decl_skip_args): Adjust call to the above function.
28388 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28391 * config/i386/sse.md
28392 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28393 register as dest whenever it is a MEM not rtx_equal_p to the
28394 corresponding dup operand, and when forcing into reg move the
28395 reg into the memory afterwards.
28396 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28397 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28398 for the force_reg mode.
28399 (avx512vl_vextractf128<mode>): Use register as dest either
28400 always when a MEM, or when it is a MEM not rtx_equal_p to the
28401 corresponding dup operand, or even not when it is a CONST_VECTOR
28402 depending on the mode and lo vs. hi.
28403 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28405 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28406 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28407 Likewise. Require that operands[2] is even.
28408 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28409 Remove extraneous parens. Require that operands[2] is a multiple
28411 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28412 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28413 disallow memory then.
28415 2017-03-30 Richard Biener <rguenther@suse.de>
28417 PR tree-optimization/77498
28418 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28419 to non-constants over backedges.
28421 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28423 PR rtl-optimization/80233
28424 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28425 as last_combined_insn. Do not test for BARRIER_P separately.
28427 2017-03-29 Andreas Schwab <schwab@suse.de>
28430 * calls.c (prepare_call_address): Convert funexp to Pmode before
28431 copying to temp reg.
28433 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28435 PR tree-optimization/80158
28436 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28437 Handle possible future case of more than one alternate
28439 (replace_rhs_if_not_dup): Likewise.
28440 (replace_one_candidate): Likewise.
28442 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28444 PR rtl-optimization/80193
28445 * ira.c (ira): Do not check allocation for LRA.
28447 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28449 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28450 (nvptx_output_simt_exit): Declare.
28451 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28452 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28453 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28454 (nvptx_declare_function_name): ...here. Emit declaration of local
28455 memory space buffer for omp_simt_enter insn.
28456 (nvptx_output_unisimt_switch): New.
28457 (nvptx_output_softstack_switch): New.
28458 (nvptx_output_simt_enter): New.
28459 (nvptx_output_simt_exit): New.
28460 * config/nvptx/nvptx.h (struct machine_function): New fields
28461 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28462 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28463 (UNSPECV_SIMT_EXIT): Ditto.
28464 (omp_simt_enter_insn): New insn.
28465 (omp_simt_enter): New expansion.
28466 (omp_simt_exit): New insn.
28467 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28469 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28470 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28471 (expand_GOMP_SIMT_EXIT): New.
28472 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28473 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28474 (GOMP_SIMT_EXIT): Ditto.
28475 * target-insns.def (omp_simt_enter): New insn.
28476 (omp_simt_exit): Ditto.
28477 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28479 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28480 (lower_rec_input_clauses): Likewise.
28481 (lower_lastprivate_clauses): Handle SIMT privatization.
28483 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28484 (ompdevlow_adjust_simt_enter): New.
28485 (find_simtpriv_var_op): New.
28486 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28487 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28489 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28490 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28491 (copy_decl_for_dup_finish): Ditto.
28493 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28495 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28498 * config/i386/i386.c (ix86_option_override_internal): Always
28499 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28501 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28503 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28505 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28507 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28508 mark new edge's irreducible flag accordign to it.
28509 (vect_do_peeling): Check loop preheader edge's irreducible flag
28510 and pass it to function slpeel_add_loop_guard.
28512 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28514 PR tree-optimization/80218
28515 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28516 Update block frequencies and counts.
28518 2017-03-28 Richard Biener <rguenther@suse.de>
28520 PR tree-optimization/78644
28521 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28522 of a simplification result we may not use it at all.
28524 2017-03-28 Richard Biener <rguenther@suse.de>
28527 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28528 without arguments, generate default definition of a SSA name.
28530 2017-03-28 Richard Biener <rguenther@suse.de>
28532 PR middle-end/80222
28533 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28534 TYPE_REF_CAN_ALIAS_ALL references.
28535 * fold-const.c (fold_indirect_ref_1): Likewise.
28537 2017-03-28 Martin Liska <mliska@suse.cz>
28540 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28541 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28543 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28544 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28546 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28547 (EXTRA_SPECS): Define.
28548 (SUBTARGET_EXTRA_SPECS): Likewise.
28549 (SUBTARGET_CPP_SPEC): Likewise.
28550 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28551 SUBTARGET_EXTRA_SPECS.
28552 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28554 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28556 * config/arc/simdext.md (vst64_insn): Update pattern.
28557 (vld32wh_insn): Likewise.
28558 (vld32wl_insn): Likewise.
28559 (vld64_insn): Likewise.
28560 (vld32_insn): Likewise.
28562 2017-03-28 Marek Polacek <polacek@redhat.com>
28565 * fold-const.c (fold_comparison): Use protected_set_expr_location
28566 instead of SET_EXPR_LOCATION.
28568 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28570 * tree.c (add_expr): Avoid name lookup warning.
28572 2017-03-27 Jeff Law <law@redhat.com>
28574 PR tree-optimization/80216
28575 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28576 function name. Limit recursion depth.
28577 (record_temporary_equivalences): Corresponding changes.
28579 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28581 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28584 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28587 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28589 * cfgcleanup.c (reg_note_cfa_p): New array.
28590 (insns_have_identical_cfa_notes): New function.
28591 (old_insns_match_p): Don't cross-jump in between /f
28592 and non-/f instructions. If both i1 and i2 are frame related,
28593 verify all CFA notes, their order and content.
28595 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28598 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28599 HImode and SImode with zero extend to DImode to one insn.
28600 (bswap<mode>2_extenddi): Likewise.
28601 (bswapsi2_extenddi): Likewise.
28602 (bswaphi2_extendsi): Likewise.
28603 (bswaphi2): Combine bswap HImode and SImode into one insn.
28604 Separate memory insns from swapping register.
28605 (bswapsi2): Likewise.
28606 (bswap<mode>2): Likewise.
28607 (bswaphi2_internal): Delete, no longer used.
28608 (bswapsi2_internal): Likewise.
28609 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28610 store, and gpr<-gpr swap insns.
28611 (bswap<mode>2_store): Likewise.
28612 (bswaphi2_reg): Register only splitter, combine with the splitter.
28613 (bswaphi2 splitter): Likewise.
28614 (bswapsi2_reg): Likewise.
28615 (bswapsi2 splitter): Likewise.
28616 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28617 the insns into load, store, and register/register insns.
28618 (bswapdi2_ldbrx): Likewise.
28619 (bswapdi2_load): Likewise.
28620 (bswapdi2_store): Likewise.
28621 (bswapdi2_reg): Likewise.
28623 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28625 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28626 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28628 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28631 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28633 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28634 special handling for target option conflicts between dform
28635 options (-mpower9-dform, -mpower9-dform-vector,
28636 -mpower9-dform-scalar) and -mno-direct-move.
28638 2017-03-27 Richard Biener <rguenther@suse.de>
28640 PR tree-optimization/80181
28641 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28643 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28645 * config/arc/predicates.md (move_double_src_operand): Replace the
28646 call to move_double_src_operand with a call to address_operand.
28648 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28650 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28651 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28652 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28654 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28656 * config/arc/predicates.md (long_immediate_loadstore_operand):
28657 Consider scaled addresses cases.
28659 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28661 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28662 restored when in interrupt.
28663 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28664 doesn't have delay slot.
28666 2017-03-27 Richard Biener <rguenther@suse.de>
28669 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28670 inlined thunk clones.
28672 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28675 * asan.c (instrument_derefs): Copy over last operand from
28676 original COMPONENT_REF to the new COMPONENT_REF with
28677 DECL_BIT_FIELD_REPRESENTATIVE.
28678 * ubsan.c (instrument_object_size): Likewise.
28680 2017-03-27 Richard Biener <rguenther@suse.de>
28682 PR tree-optimization/80170
28683 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28684 sure DR/SCEV didnt fold in constants we do not see when looking
28685 at the reference base alignment.
28687 2017-03-27 Richard Biener <rguenther@suse.de>
28689 PR middle-end/80171
28690 * gimple-fold.c (fold_ctor_reference): Properly guard against
28691 NULL return value from canonicalize_constructor_val.
28693 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28696 * config/i386/i386.c (ix86_expand_builtin)
28697 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28698 flags reg setting and flags reg using instructions.
28699 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28700 clobbering instructions to zero extend op2.
28702 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28704 * doc/install.texi (Configuration) <--with-aix-soname>:
28705 Update link to AIX ld.
28707 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28709 PR rtl-optimization/80160
28710 PR rtl-optimization/80159
28711 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28712 reg_alternate_class into account.
28714 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28717 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28718 to consider in curr_insn_transform.
28720 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28722 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28723 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28724 and emit_mode_inner.
28726 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28728 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28729 argument to the overloaded builtin variants. Use the new flag to
28730 deprecate certain builtin variants.
28731 * config/s390/s390-builtin-types.def: Add new builtin types.
28732 * config/s390/s390-builtins.h: Support new flags field for
28733 overloaded builtins.
28734 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28735 (s390_macro_to_expand): Enable vector float data type.
28736 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28737 builtins by incrementing the __VEC__ version number.
28738 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28740 (s390_resolve_overloaded_builtin): Emit error messages depending
28741 on the builtin flags.
28742 * config/s390/s390.c (s390_expand_builtin): Support additional
28743 flags argument. Change error message to match the messages
28744 emitted in s390-c.c.
28745 * config/s390/s390.md: New UNSPEC_* constants.
28746 (op_type): Add new instruction types.
28747 * config/s390/vecintrin.h: Add new builtins and test data class
28749 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28750 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28751 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28752 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28753 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28754 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28756 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28757 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28758 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28759 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28761 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28762 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28763 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28764 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28766 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28767 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28768 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28769 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28770 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28771 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28772 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28774 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28775 ("vec_scatter_element<V_HW_4:mode>_DI")
28776 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28777 ("vec_fpint<mode>", "vflls")
28778 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28779 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28780 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28781 ("*vec_cmphe<mode>_cc"): ... these.
28783 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28784 mode constant instead of magic value.
28786 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28788 * config/s390/s390.c (s390_expand_vec_compare): Support other
28789 vector floating point modes than just V2DF.
28790 (s390_expand_vcond): Likewise.
28791 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28792 (s390_cannot_change_mode_class): Prevent mode changes between TF
28793 and V1TF in vector registers.
28794 * config/s390/s390.md (DF, SF): New mode attributes.
28795 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28796 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28797 SFmode support for VRs.
28798 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28800 (VFT, VF_HW): New mode iterators.
28801 (vw, sdx): New mode attributes.
28802 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28803 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28804 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28805 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28806 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28807 also the new vector floating point modes. Renaming to ...
28809 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28810 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28811 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28812 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28813 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28814 ("vec_unordered<mode>"): ... these.
28816 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28817 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28818 ("*vec_extendv2df"): New insn definitions.
28820 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28822 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28823 ("mulditi3_2", "*muldi3_sign"): New patterns.
28824 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28825 rename the pattern definition.
28827 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28829 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28831 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28833 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28835 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28836 instruction if possible.
28837 * config/s390/vector.md (vec_halfnumelts): New mode
28839 ("*vec_vllezlf<mode>"): New pattern.
28841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28843 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28844 ("popcountv4si2", "popcountv2di2"): Rename to ...
28845 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28846 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28848 ("popcount<mode>2_vxe"): New pattern.
28850 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28852 * common/config/s390/s390-common.c (processor_flags_table): Add
28854 * config.gcc: Add arch12.
28855 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28856 Default to arch12 for unknown CPU model numbers.
28857 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28858 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28859 PROCESSOR_max sanity check.
28860 * config/s390/s390-opts.h (enum processor_type): Add
28862 * config/s390/s390.c (processor_table): Add arch12.
28863 (s390_expand_builtin): Add check for B_VXE flag.
28864 (s390_issue_rate): Add PROCESSOR_ARCH12.
28865 (s390_get_sched_attrmask): Likewise.
28866 (s390_get_unit_mask): Likewise.
28867 (s390_sched_score): Enable z13 scheduling for arch12.
28868 (s390_sched_reorder): Likewise.
28869 (s390_sched_variable_issue): Likewise.
28870 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28872 (s390_tune_attr): Use z13 scheduling also for arch12.
28873 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28874 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28875 (TARGET_VXE_P): New macros.
28876 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28877 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28878 * config/s390/s390.opt: Add arch12 as processor_type.
28880 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28882 * config/s390/s390.md
28883 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28884 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28885 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28887 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28888 Rename expanders to ...
28890 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28891 ("fixuns_truncdddi2_emu"): ... these.
28893 ("fixuns_trunc<mode>si2_emu"): New expander.
28895 ("*fixuns_truncdfdi2_z13"): Rename to ...
28896 ("*fixuns_truncdfdi2_vx"): ... this.
28898 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28900 * config/s390/2964.md: Remove the single element vector compare
28901 instructions which are no longer used.
28902 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28904 (s390_canonicalize_comparison): Remove handling of DFmode
28906 (s390_expand_vec_compare_scalar): Remove function.
28907 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28908 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28910 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28912 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28914 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28916 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28917 will anyway by matched by mov<mode>_64dfp.
28919 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28921 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28922 vlef/vstef. Add missing operand to vleif.
28924 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28926 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28927 pair for all vector types with 64 bit elements.
28928 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28929 * config/s390/vector.md (V_HW_64): ... here.
28930 (V_128_NOSINGLE): New mode iterator.
28931 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28932 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28933 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28934 ("*vec_load_pairv2di"): Change to ...
28935 ("*vec_load_pair<mode>"): ... this one.
28937 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28939 * config/s390/constraints.md: Add comments.
28940 (jKK): Reject element sizes > 8 bytes.
28941 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28943 * config/s390/s390.md: Add the s_operand checks formerly in
28944 s390_split_ok_p to various splitters where they are still
28946 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28947 for 128 bit vectors. Plus two splitters.
28949 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28951 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28954 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28957 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28958 error if the boundary argument is not constant.
28960 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28962 PR rtl-optimization/80112
28963 * loop-doloop.c (doloop_condition_get): Don't check condition
28964 if cmp isn't SET with IF_THEN_ELSE src.
28966 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28968 PR tree-optimization/80158
28969 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28970 replacing a candidate statement, also replace it for the
28971 candidate's alternate interpretation.
28972 (replace_rhs_if_not_dup): Likewise.
28973 (replace_one_candidate): Likewise.
28975 2017-03-24 Richard Biener <rguenther@suse.de>
28977 PR tree-optimization/80167
28978 * graphite-isl-ast-to-gimple.c
28979 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28981 (translate_isl_ast_to_gimple::get_rename): Likewise.
28983 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28985 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28986 handling of certain combinations of target options, including the
28987 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28988 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28990 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28993 * config/arm/arm.md (*load_multiple): Add reload_completed to
28994 matching condition.
28996 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28997 Richard Biener <rguenth@suse.de>
28999 PR tree-optimization/79908
29000 PR tree-optimization/80136
29001 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29002 been cast away, gimplify_and_add suffices.
29004 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29006 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29008 2017-03-23 Richard Biener <rguenther@suse.de>
29010 PR tree-optimization/80032
29011 * gimplify.c (gimple_push_cleanup): Forced unconditional
29012 cleanups still have to go to the conditional_cleanups
29015 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29017 PR tree-optimization/80072
29018 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29020 (next_operand_entry_id): Change type to unsigned int.
29021 (sort_by_operand_rank): Make sure to return the right return value
29022 even if unsigned fields are bigger than INT_MAX.
29023 (struct oecount): Change cnt and id type to unsigned int.
29024 (oecount_hasher::equal): Formatting fix.
29025 (oecount_cmp): Make sure to return the right return value
29026 even if unsigned fields are bigger than INT_MAX.
29027 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29030 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29031 TREE_READONLY on result if writing it more than once.
29034 * doc/invoke.texi (-fsanitize=thread): Document that with
29035 -fnon-call-exceptions atomics are not able to throw
29039 * tsan.c: Include tree-eh.h.
29040 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29041 maybe_clean_or_replace_eh_stmt where needed.
29042 (instrument_memory_accesses): Add cfg_changed argument.
29043 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29044 if it returned true.
29045 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29047 PR rtl-optimization/63191
29048 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29049 wrapper function, moved the whole old content into ...
29050 (ix86_delegitimize_address_1): ... this. New inline function.
29051 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29052 true as last argument instead of ix86_delegitimize_address.
29054 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29056 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29057 cortexa57_branch_cost.
29059 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29061 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29063 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29066 * doc/md.texi (Constraints): Document wA constraint.
29067 * config/rs6000/constraints.md (wA): New.
29068 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29069 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29070 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29071 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29073 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29076 * gimplify.c (is_oacc_declared): New function.
29077 (oacc_default_clause): Use it to set default flags for acc declared
29078 variables inside parallel regions.
29079 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29080 declared variables.
29081 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29082 declare attribute to any decl as necessary.
29084 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29087 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29088 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29089 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29090 (arm_arch_lpae): This.
29091 * config/arm/arm.c (arm_arch7ve): Rename into ...
29092 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29093 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29096 2017-03-22 Martin Liska <mliska@suse.cz>
29099 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29100 error message instead of an ICE.
29102 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29104 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29106 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29108 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29113 2017-03-21 Martin Sebor <msebor@redhat.com>
29115 * doc/extend.texi: Use "cannot" instead of "can't."
29116 * doc/hostconfig.texi: Same.
29117 * doc/install.texi: Same.
29118 * doc/invoke.texi: Same.
29119 * doc/loop.texi: Same.
29120 * doc/md.texi: Same.
29121 * doc/objc.texi: Same.
29122 * doc/rtl.texi: Same.
29123 * doc/tm.texi: Same.
29124 * doc/tm.texi.in: Same.
29125 * doc/trouble.texi: Same.
29127 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29130 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29131 (collect_checksum_attributes): Set it.
29132 (die_checksum_ordered): Use it.
29134 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29136 PR tree-optimization/79908
29137 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29138 change: For a VA_ARG whose LHS has been cast away, use
29139 force_gimple_operand to construct the side effects.
29141 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29143 PR translation/80001
29144 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29145 more amenable to translation.
29146 (oacc_loop_auto_partitions): Likewise.
29148 2017-03-21 Marek Polacek <polacek@redhat.com>
29149 Martin Sebor <msebor@redhat.com>
29151 PR tree-optimization/80109
29152 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29153 on INTEGRAL_TYPE_P.
29155 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29156 Segher Boessenkool <segher@kernel.crashing.org>
29159 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29160 check reg_used_between_p between insn and one of succ or succ2
29161 depending on if succ is artificial insn not inserted into insn
29164 2017-03-21 Martin Liska <mliska@suse.cz>
29166 PR gcov-profile/80081
29167 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29168 * doc/gcc.texi: Include gcov-dump stuff.
29169 * doc/gcov-dump.texi: New file.
29171 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29173 PR rtl-optimization/79150
29174 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29175 conditional jump, if the jump is the last insn of the loop.
29177 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29178 Richard Biener <rguenth@suse.de>
29180 PR tree-optimization/79908
29181 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29182 been cast away, use force_gimple_operand to construct the side
29185 2017-03-21 Martin Liska <mliska@suse.cz>
29187 PR libfortran/79956
29188 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29191 2017-03-21 Brad Spengler <spender@grsecurity.net>
29194 * plugin.c (htab_hash_plugin): New function.
29195 (add_new_plugin): Use it and adjust.
29196 (parse_plugin_arg_opt): Adjust.
29197 (init_one_plugin): Likewise.
29199 2017-03-21 Richard Biener <rguenther@suse.de>
29201 PR tree-optimization/80032
29202 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29203 if set force the cleanup to happen unconditionally.
29204 (gimplify_target_expr): Push inserted clobbers with force_uncond
29205 to avoid them being removed by control-dependent DCE.
29207 2017-03-21 Richard Biener <rguenther@suse.de>
29209 PR tree-optimization/80122
29210 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29211 va_arg_pack_len when the inlined call stmt requires pack
29213 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29215 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29218 * tsan.c (instrument_builtin_call): If the memory model argument
29219 is not a constant, assume it is valid.
29222 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29225 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29227 PR rtl-optimization/79910
29228 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29229 if its dest is used by an insn before I2 (other than the combined
29230 insns themselves, which are properly handled already).
29232 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29235 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29237 * combine.c (record_used_regs): New static function.
29238 (try_combine): Handle situations where there is an additional
29239 instruction between I2 and I3 which needs to have a LOG_LINK
29243 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29245 * combine.c (try_combine): Delete redundant i1 test. Call
29246 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29248 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29251 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29252 alternatives 13/14.
29254 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29256 PR tree-optimization/80054
29257 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29258 the optimization if a PHI or any of its arguments is not dominated
29259 by the candidate's basis. Use gphi* rather than gimple* as
29261 (replace_profitable_candidates): Clean up a gimple* variable that
29262 should be a gphi* variable.
29264 2017-03-20 Martin Sebor <msebor@redhat.com>
29267 * doc/extend.texi (attribute constructor): Document present limitation.
29269 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29272 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29273 __POWER9_VECTOR__ #ifdef control, change template definition to
29274 use Power9-specific built-in function.
29275 (vec_any_eq): Likewise.
29276 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29277 to control outcomes from this test.
29278 (vector_ae_<mode>p): For VEC_F modes, likewise.
29280 2017-03-20 Ian Lance Taylor <iant@google.com>
29282 * config/i386/i386.c (ix86_function_regparm): Save an extra
29283 register for -fsplit-stack with DECL_STATIC_CHAIN.
29285 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29288 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29289 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29291 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29293 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29295 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29298 2017-03-20 Marek Polacek <polacek@redhat.com>
29301 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29303 2017-03-20 Richard Biener <rguenther@suse.de>
29305 PR tree-optimization/80113
29306 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29307 allocate extra SSA name for PHI def.
29308 (add_close_phis_to_outer_loops): Likewise.
29309 (add_close_phis_to_merge_points): Likewise.
29310 (copy_loop_close_phi_args): Likewise.
29311 (copy_cond_phi_nodes): Likewise.
29313 2017-03-20 Martin Liska <mliska@suse.cz>
29315 PR middle-end/79753
29316 * tree-chkp.c (chkp_build_returned_bound): Do not build
29317 returned bounds for a LHS that's not a BOUNDED_P type.
29319 2017-03-20 Martin Liska <mliska@suse.cz>
29323 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29324 COMPLEX_CST and VECTOR_CST.
29326 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29329 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29330 target operand. A new splitter adds the clobber statement in case
29331 the target operand is dead anyway.
29333 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29335 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29336 to age-old versions of binutils and glibc.
29338 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29340 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29342 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29344 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29346 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29348 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29349 requirement for binutils 2.13.
29351 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29353 * combine.c (try_combine): Delete redundant i1 test. Call
29354 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29356 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29358 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29359 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29361 <riscv64-*-elf>: Re-arrange section
29362 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29363 <riscv32-*-linux>: Likewise.
29364 <riscv64-*-elf>: Likewise
29365 <riscv64-*-linux>: Likewise.
29367 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29370 * aarch64.opt(verbose-cost-dump): Fix typo.
29372 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29375 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29376 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29378 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29380 * reload.c (find_reloads): When reloading a nonoffsettable address,
29381 use RELOAD_OTHER for it and its address reloads.
29383 PR rtl-optimization/79910
29384 * combine.c (record_used_regs): New static function.
29385 (try_combine): Handle situations where there is an additional
29386 instruction between I2 and I3 which needs to have a LOG_LINK
29389 2017-03-17 Jeff Law <law@redhat.com>
29391 PR tree-optimization/71437
29392 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29393 conditional in the hash table first.
29394 (vrp_dom_walker::before_dom_children): Extract condition from
29395 ASSERT_EXPR. Record condition, its inverion and any implied
29396 conditions as well.
29398 2017-03-17 Marek Polacek <polacek@redhat.com>
29399 Markus Trippelsdorf <markus@trippelsdorf.de>
29401 PR tree-optimization/80079
29402 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29405 2017-03-17 Richard Biener <rguenther@suse.de>
29407 PR middle-end/80075
29408 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29409 Properly verify the LHS before the RHS possibly claims to be
29411 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29414 2017-03-17 Martin Jambor <mjambor@suse.cz>
29416 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29417 (List of -O2 options): Likewise.
29418 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29421 2017-03-17 Tom de Vries <tom@codesourcery.com>
29423 * gcov-dump.c (print_usage): Print bug_report_url.
29425 2017-03-17 Richard Biener <rguenther@suse.de>
29427 PR middle-end/80050
29428 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29429 (parser::peek): Likewise.
29431 2017-03-17 Richard Biener <rguenther@suse.de>
29433 PR tree-optimization/80048
29434 * sese.c (free_sese_info): Properly release rename_map and
29435 copied_bb_map elements.
29437 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29439 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29440 Add linked-list forward and backlinks. Insert on
29441 construction, remove on destruction.
29442 (class pass_store_merging): Add m_stores_head field.
29443 (pass_store_merging::terminate_and_process_all_chains):
29444 Iterate over m_stores_head list.
29445 (pass_store_merging::terminate_all_aliasing_chains):
29447 (pass_store_merging::execute): Check for debug stmts first.
29448 Push new chains onto the m_stores_head stack.
29450 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29453 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29454 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29455 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29457 2017-03-16 Jeff Law <law@redhat.com>
29459 PR tree-optimization/71437
29460 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29461 member function. Implementation moved into after_dom_children
29462 member function and into the threader's thread_outgoing_edges
29464 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29465 some code into new thread_outgoing_edges.
29466 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29467 definition. Simplify marker handling (do it here). Assume we always
29468 have the available expression and the const/copies tables.
29469 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29471 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29472 * tree-vrp.c (equiv_stack): No longer file scoped.
29473 (vrp_dom_walker): New class.
29474 (vrp_dom_walker::before_dom_children): New member function.
29475 (vrp_dom_walker::after_dom_children): Likewise.
29476 (identify_jump_threads): Setup domwalker. Use it rather than
29477 walking edges in a random order by hand. Simplify setup/finalization.
29478 (finalize_jump_threads): Remove.
29479 (vrp_finalize): Do not call identify_jump_threads here.
29480 (execute_vrp): Do it here instead and call thread_through_all_blocks
29483 PR tree-optimization/71437
29484 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29486 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29488 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29489 (dom_opt_dom_walker::thread_across_edge): Remove
29490 handle_dominating_asserts argument. All callers changed.
29491 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29492 changes. Remove calls to lhs_of_dominating_assert. Other
29493 uses of handle_dominating_asserts turn into unconditional code
29494 (simplify_control_stmt_condition_1): Likewise.
29495 (simplify_control_stmt_condition): Likewise.
29496 (thread_through_normal_block, thread_across_edge): Likewise.
29497 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29498 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29499 object if it is not an SSA_NAME.
29500 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29501 before calling into the VRP specific simplifiers.
29502 (identify_jump_threads): Remove handle_dominating_asserts
29505 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29508 * tree-diagnostic.c (default_tree_printer): No longer static.
29509 * tree-diagnostic.h (default_tree_printer): New prototype.
29511 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29513 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29514 Change ins into fmov.
29516 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29518 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29519 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29520 Use h_con constraint for operand 1.
29521 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29522 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29524 2017-03-15 Jeff Law <law@redhat.com>
29526 PR tree-optimization/71437
29527 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29528 (record_temporary_equivalences): Use it.
29530 PR tree-optimization/71437
29531 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29532 tree-ssa-scopedtables.
29533 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29534 (record_conditions, record_cond, vuse_eq): Likewise.
29535 (record_edge_info): Adjust to API tweak of record_conditions.
29536 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29537 (record_temporary_equivalences, optimize_stmt): Likewise.
29538 (eliminate_redundant_computations): Likewise.
29539 (record_equivalences_from_stmt): Likewise.
29540 * tree-ssa-scopedtables.c: Include options.h and params.h.
29541 (vuse_eq): New function, moved from tree-ssa-dom.c
29542 (build_and_record_new_cond): Likewise.
29543 (record_conditions): Likewise. Accept vector of conditions rather
29544 than edge_equivalence structure for first argument.
29545 for the first argument.
29546 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29547 from tree-ssa-dom.c.
29548 (avail_exprs_stack::record_cond): Likewise.
29549 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29550 from tree-ssa-dom.c.
29551 (avail_exprs_stack): Add new member functions lookup_avail_expr
29553 (record_conditions): Declare.
29555 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29558 * lra-constraints.c (process_alt_operands): Increase reject for
29559 reloading an input/output operand.
29561 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29564 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29565 insns to convert from signed/unsigned char/short to IEEE 128-bit
29567 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29569 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29572 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29573 subreg of inner mode for values already in registers.
29575 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29577 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29578 iteration reg is used after the loop.
29580 2017-03-14 Martin Sebor <msebor@redhat.com>
29582 PR tree-optimization/79800
29583 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29584 precision in negative-positive range.
29585 (format_floating): Call non-const overload with adjusted precision.
29587 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29590 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29593 2017-03-14 Martin Sebor <msebor@redhat.com>
29595 PR middle-end/80020
29596 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29597 * builtins.def (aligned_alloc): Use it.
29600 * Makefile.in (GTFILES): Add calls.c.
29601 * calls.c: Include "gt-calls.h".
29603 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29605 PR rtl-optimization/79728
29606 * regs.h (struct target_regs): New field
29607 x_contains_allocatable_regs_of_mode.
29608 (contains_allocatable_regs_of_mode): New macro.
29609 * reginfo.c (init_reg_sets_1): Initialize it, and change
29610 contains_reg_of_mode so it includes global regs as well.
29611 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29612 rather than contains_regs_of_mode.
29614 2017-03-14 Martin Liska <mliska@suse.cz>
29616 * doc/invoke.texi: Document options that can't be combined with
29617 -fcheck-pointer-bounds.
29619 2017-03-14 Martin Liska <mliska@suse.cz>
29621 PR middle-end/79831
29622 * doc/invoke.texi (-Wchkp): Document the option.
29624 2017-03-14 Martin Liska <mliska@suse.cz>
29626 * Makefile.in: Install gcov-dump.
29628 2017-03-14 Martin Liska <mliska@suse.cz>
29630 * multiple_target.c (expand_target_clones): Bail out for
29631 an invalid attribute.
29633 2017-03-14 Richard Biener <rguenther@suse.de>
29635 * alias.c (struct alias_set_entry): Pack properly.
29636 * cfgloop.h (struct loop): Likewise.
29637 * cse.c (struct set): Likewise.
29638 * ipa-utils.c (struct searchc_env): Likewise.
29639 * loop-invariant.c (struct invariant): Likewise.
29640 * lra-remat.c (struct cand): Likewise.
29641 * recog.c (struct change_t): Likewise.
29642 * rtl.h (struct address_info): Likewise.
29643 * symbol-summary.h (function_summary): Likewise.
29644 * tree-loop-distribution.c (struct partition): Likewise.
29645 * tree-object-size.c (struct object_size_info): Likewise.
29646 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29647 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29648 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29649 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29650 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29651 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29652 (struct _stmt_vec_info): Likewise.
29654 2017-03-14 Martin Liska <mliska@suse.cz>
29657 * multiple_target.c (create_dispatcher_calls): Check that
29658 a target can create a function dispatcher.
29660 2017-03-14 Martin Liska <mliska@suse.cz>
29663 * multiple_target.c (expand_target_clones): Drop local.local
29664 flag for default implementation.
29666 2017-03-14 Richard Biener <rguenther@suse.de>
29668 PR tree-optimization/80030
29669 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29671 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29673 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29674 gcc_fallthrough() instead of __attribute__((fallthrough));
29676 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29678 * doc/gcc.texi: Remove "up" link to (DIR).
29679 * doc/gccint.texi: Ditto.
29681 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29683 * doc/install.texi (Specific) <avr>: Remove reference to
29686 2017-03-13 Jeff Law <law@redhat.com>
29688 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29689 attribute rather than comments.
29691 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29692 match_scratch operand is highest.
29694 2017-03-13 Martin Liska <mliska@suse.cz>
29696 PR middle-end/78339
29697 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29698 is a CHKP clone, use original declaration.
29700 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29702 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29703 (arc_conditional_register_usage): Use a different allocation order
29704 when optimizing for size.
29705 * common/config/arc/arc-common.c (arc_option_optimization_table):
29706 Section anchors default on when optimizing for size.
29708 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29710 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29712 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29714 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29715 * config/arc/arc.md (cpu_facility): Add cd variant.
29716 (*movqi_insn): Add code density variant.
29717 (*movhi_insn): Likewise.
29718 (*movqi_insn): Likewise.
29719 (*addsi3_mixed): Likewise.
29720 (subsi3_insn): Likewise.
29722 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29724 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29726 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29728 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29729 expressions with MINUS and UNARY ops.
29731 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29734 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29736 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29737 between vec_select and vector argument.
29738 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29739 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29740 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29741 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29742 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29743 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29745 2017-03-13 Richard Biener <rguenther@suse.de>
29748 * params.def (vect-max-peeling-for-alignment): Fix typo.
29750 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29752 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29753 issue that only occurred with binutils below 2.18.
29755 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29757 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29758 refer to binutils 2.11/2.12 minimum.
29760 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29762 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29763 ftp.kernel.org and simplify binutils requirement.
29765 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29767 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29769 (Optimize Options): Ditto. Also remove redundancy.
29771 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29773 PR translation/79848
29774 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29776 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29777 to G_ to avoid double translation.
29779 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29781 PR translation/79923
29782 * auto-profile.c (get_combined_location): Convert leading
29783 character of diagnostics to lower case and remove trailing period.
29784 (read_profile): Likewise for various diagnostics.
29785 * config/arm/arm.c (arm_option_override): Remove trailing period
29786 from various diagnostics.
29787 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29788 (msp430_expand_delay_cycles): Likewise.
29790 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29793 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29794 full command-line argument, rather than just "str".
29795 (aarch64_validate_march): Likewise.
29796 (aarch64_validate_mtune): Likewise.
29798 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29800 PR rtl-optimization/78911
29801 * lra-assigns.c (must_not_spill_p): New function.
29802 (spill_for): Use it.
29804 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29806 PR tree-optimization/79981
29807 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29808 ATOMIC_COMPARE_EXCHANGE ifn result.
29809 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29810 IFN_ATOMIC_COMPARE_EXCHANGE.
29812 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29815 * opts.c (parse_sanitizer_options): Add missing question mark to
29816 "did you mean" message.
29818 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29820 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29822 (VMULEUH_UNS): Likewise.
29823 (VMULOUB_UNS): Likewise.
29824 (VMULOUH_UNS): Likewise.
29825 * config/rs6000/rs6000.c (builtin_function_type): Remove
29826 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29828 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29831 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29832 x with result of extra_parsing_for_operand_code_0.
29833 (function_reader::extra_parsing_for_operand_code_0): Convert
29834 return type from void to rtx, returning x. When reading
29835 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29836 larger size containing struct block_symbol.
29838 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29841 -mfloat128-hardware without -m64.
29843 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29846 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29847 entries to the case statement that marks unsigned arguments to
29848 overloaded functions.
29850 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29852 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29853 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29855 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29858 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29859 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29861 2017-03-10 Martin Liska <mliska@suse.cz>
29865 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29866 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29869 2017-03-10 Olivier Hainque <hainque@adacore.com>
29871 * tree-switch-conversion (array_value_type): Start by resetting
29872 candidate type to it's main variant.
29874 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29876 PR rtl-optimization/79909
29877 * combine.c (try_combine): Use simplify_replace_rtx on individual
29878 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29879 of the whole CALL_INSN_FUNCTION_USAGE.
29881 PR tree-optimization/79972
29882 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29883 get_range_info on SSA_NAMEs. Formatting fixes.
29885 2017-03-10 Richard Biener <rguenther@suse.de>
29886 Jakub Jelinek <jakub@redhat.com>
29888 PR tree-optimization/77975
29889 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29890 edge to be constant.
29891 (get_val_for): For constant x return it. Formatting fix.
29892 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29893 would use the same bases as the current one.
29895 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29897 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29898 instead of vec_select for V1TImode.
29899 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29901 (VSX_LE_128): Add V1TI to this mode iterator.
29902 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29903 (*vsx_le_perm_store_<mode>): Likewise.
29904 (pre-reload splitter for VSX stores): Likewise.
29905 (post-reload splitter for VSX stores): Likewise.
29906 (*vsx_xxpermdi2_le_<mode>): Likewise.
29907 (*vsx_lxvd2x2_le_<mode>): Likewise.
29908 (*vsx_stxvd2x2_le_<mode>): Likewise.
29910 2017-03-09 Michael Eager <eager@eagercon.com>
29912 Correct failures with --enable-checking=yes,rtl.
29914 * config/microblaze/microblaze.c (microblaze_expand_shift):
29915 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29916 test for const0_rtx.
29917 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29918 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29920 2017-03-09 Richard Biener <rguenther@suse.de>
29922 PR tree-optimization/79977
29923 * graphite-scop-detection.c (scop_detection::merge_sese):
29924 Handle the case of extra exits to blocks dominating the entry.
29926 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29928 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29931 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29933 PR rtl-optimization/79949
29934 * lra-constraints.c (process_alt_operands): Check memory when
29935 trying to predict a cycle. Print about the overall increase.
29937 2017-03-09 Richard Biener <rguenther@suse.de>
29939 PR middle-end/79971
29940 * gimple-expr.c (useless_type_conversion_p): Preserve
29941 TYPE_SATURATING for fixed-point types.
29943 2017-03-09 Richard Biener <rguenther@suse.de>
29946 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29947 alignment of BLKmode params.
29949 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29952 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29953 (VALL_NO_V2Q): Likewise.
29955 * config/aarch64/aarch64-simd.md
29956 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29958 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29959 VALL_NO_V2Q mode iterator.
29960 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29962 2017-03-09 Martin Liska <mliska@suse.cz>
29964 PR tree-optimization/79631
29965 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29966 tree_int_cst_sign_bit just for INTEGER constants.
29968 2017-03-09 Martin Liska <mliska@suse.cz>
29972 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29975 2017-03-09 Marek Polacek <polacek@redhat.com>
29978 * tree.c (inchash::add_expr): Handle TREE_VEC.
29980 2017-03-09 Martin Liska <mliska@suse.cz>
29983 (chkp_narrow_size_and_offset): New function.
29984 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29985 (void chkp_parse_bit_field_ref): New function.
29986 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29987 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29989 2017-03-09 Martin Liska <mliska@suse.cz>
29992 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29993 (chkp_find_bounds_1): Remove gcc_unreachable.
29995 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29998 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29999 BUILT_IN_SYNC*, determine the access type from the size suffix and
30000 always build a MEM_REF with that type. Handle forgotten
30001 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30004 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30005 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30006 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30007 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30008 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30009 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30010 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30011 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30012 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30013 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30014 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30015 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30016 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30017 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30018 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30019 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30020 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30021 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30022 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30023 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30024 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30025 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30026 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30027 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30028 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30029 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30030 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30031 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30032 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30033 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30034 definitions outside of __OPTIMIZE__ guarded section.
30037 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30038 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30039 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30040 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30043 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30045 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30046 ("vfenez<mode>"): Add missing constraints.
30048 2017-03-08 Martin Sebor <msebor@redhat.com>
30051 * config/nds32/nds32.c (nds32_option_override):
30052 Fix misspelled diagnostic.
30054 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30057 * gimplify.c (gimplify_omp_for): Replace index var in outer
30058 taskloop statement with an artificial variable and add
30059 OMP_CLAUSE_PRIVATE clause for it.
30061 2017-03-08 Richard Biener <rguenther@suse.de>
30063 PR tree-optimization/79955
30064 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30065 for accesses that are completely outside of the variable.
30067 2017-03-08 Andrew Haley <aph@redhat.com>
30069 PR tree-optimization/79943
30070 * tree-ssa-loop-split.c (compute_new_first_bound): When
30071 calculating the new upper bound, (END-BEG) should be added, not
30074 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30076 * config/avr/avr.md (setmemhi): Make sure match_dup
30077 operand number comes before match_scratch.
30079 2017-03-08 Richard Biener <rguenther@suse.de>
30081 PR tree-optimization/79920
30082 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30083 with ncopies == 1 to ...
30084 (vect_transform_slp_perm_load): ... here. Properly compute
30085 all element loads by iterating VF times over the group. Do
30086 not handle ncopies (computed in a broken way) in
30087 vect_create_mask_and_perm.
30089 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30092 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30093 is a uniform vector, use uniform_vector_p return value instead of
30094 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30096 2017-03-07 Marek Polacek <polacek@redhat.com>
30098 PR middle-end/79809
30099 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30100 (alloca_call_type): Likewise.
30102 2017-03-07 Martin Liska <mliska@suse.cz>
30104 * gcov.c (process_args): Put comment to correct location.
30106 2017-03-07 Martin Liska <mliska@suse.cz>
30108 PR middle-end/68270
30109 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30110 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30111 (chkp_narrow_bounds_for_field): Likewise.
30112 (chkp_parse_array_and_component_ref): Pass one more argument to
30115 2017-03-07 Richard Biener <rguenther@suse.de>
30117 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30120 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30122 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30123 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30125 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30128 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30129 to end of description.
30130 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30132 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30134 PR rtl-optimization/79901
30135 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30137 (*avx512f_<code><mode>3<mask_name>): ... this.
30138 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30139 iterator instead of VI8_AVX2_AVX512BW.
30141 PR rtl-optimization/79901
30142 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30143 min/max expander, expand it using expand_vec_cond_expr.
30146 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30149 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30152 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30153 to void * for PCH reasons.
30154 * dwarf2out.c (output_loc_operands, output_die): Cast
30155 v.val_vec.array to unsigned char *.
30157 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30160 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30163 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30165 PR rtl-optimization/79571
30166 * lra-constraints.c (process_alt_operands): Calculate static
30167 reject and subtract it from overall when only addresses will be
30170 2017-03-06 Julia Koval <julia.koval@intel.com>
30173 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30174 incoming stack boundary to 128 for 64-bit targets.
30176 2017-03-06 Richard Biener <rguenther@suse.de>
30178 PR tree-optimization/79894
30179 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30180 to NULL after folding it.
30182 2017-03-06 Richard Biener <rguenther@suse.de>
30184 PR tree-optimization/79824
30185 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30186 check disabling peeling for gaps.
30188 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30190 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30191 attributes): Document gettimeofday.
30193 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30195 * config/s390/s390.c (s390_option_override_internal): Set
30196 PARAM_MIN_VECT_LOOP_BOUND
30198 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30200 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30201 * config/s390/s390.md: Likewise.
30203 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30206 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30207 (<avx2_avx512>_perm<mode>): Rename to ...
30208 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30210 (<avx512>_perm<mode>_mask): Rename to ...
30211 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30213 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30214 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30215 instead of VI8F_256_512.
30216 (avx512f_perm<mode>): New define_expand.
30217 (avx512f_perm<mode>_mask): Likewise.
30218 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30219 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30221 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30223 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30224 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30226 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30228 2017-03-06 Martin Liska <mliska@suse.cz>
30231 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30232 when having a SSA NAME w/o VAR_DECL assigned to it.
30234 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30236 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30237 msa_dpsub_<su>_d): Fix MODE for vec_select.
30239 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30241 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30243 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30245 2017-03-06 Richard Biener <rguenther@suse.de>
30247 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30248 * plugin.c (register_plugin_info): Likewise.
30249 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30251 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30253 * config/i386/sse.md (sse_storehps, sse_storelps,
30254 avx_<castmode><avxsizesuffix>_<castmode>,
30255 avx512f_<castmode><avxsizesuffix>_<castmode>,
30256 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30257 in condition that at least one operand is not a MEM.
30259 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30261 PR middle-end/79805
30262 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30263 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30265 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30266 gimple_call_nothrow_p flag based on whether original builtin can throw.
30267 If it can, emit following stmts on the fallthrough edge.
30268 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30269 don't create new bb if inserting just debug stmts on the edge, try to
30270 insert them on the fallthru bb or just reset debug stmts.
30272 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30275 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30276 restore recog_data (including the operand rtxes inside it) around
30277 the call to get_insn_template.
30279 2017-03-03 Martin Sebor <msebor@redhat.com>
30281 PR tree-optimization/79699
30282 * context.c (context::~context): Free MPFR caches to avoid
30283 a memory leak on program exit.
30285 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30287 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30288 Use wide_int::ulow () instead of .elt (0).
30290 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30292 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30293 (*pushxf): Limit oF constraint to 32bit targets and add oC
30294 constraint for 64bit targets.
30295 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30296 (*pushdf): Change rmF constraint to rmC.
30298 2017-03-03 Martin Liska <mliska@suse.cz>
30300 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30301 Remove unused variable.
30303 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30306 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30307 is a memory operand, increase num_memory.
30308 (ix86_expand_args_builtin): Likewise.
30310 2017-03-03 Jan Hubicka <jh@suse.cz>
30313 * ipa-devirt.c (maybe_record_node): Properly handle
30314 __cxa_pure_virtual visibility.
30316 2017-03-03 Martin Liska <mliska@suse.cz>
30318 PR tree-optimization/79803
30319 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30321 (pass_loop_prefetch::execute): Disabled optimization if an
30322 assumption about L1 cache size is not met.
30324 2017-03-03 Martin Liska <mliska@suse.cz>
30326 PR rtl-optimization/79574
30327 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30328 (hash_scan_set): Likewise.
30329 (dump_hash_table): Likewise.
30330 (hoist_code): Likewise.
30332 2017-03-03 Richard Biener <rguenther@suse.de>
30334 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30335 in place of uhigh ().
30336 (fixed_convert_from_real): Likewise.
30338 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30341 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30343 2017-03-03 Richard Biener <rguenther@suse.de>
30345 PR middle-end/79818
30346 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30347 TYPE_OVERFLOW_UNDEFINED check.
30349 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30351 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30353 (vector_ae_<mode>_p): Likewise.
30354 (vector_nez_<mode>_p): Likewise.
30355 (vector_ne_v2di_p): Likewise.
30356 (vector_ae_v2di_p): Likewise.
30357 (vector_ne_<mode>_p): Likewise.
30358 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30360 (vsx_tsqrt<mode>2_fe): Likewise.
30362 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30365 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30367 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30369 PR rtl-optimization/79780
30370 * cprop.c (one_cprop_pass): When second and further conditional trap
30371 in a single basic block is turned into an unconditional trap, turn it
30372 into a deleted note to avoid RTL verification failures.
30374 2017-03-02 Richard Biener <rguenther@suse.de>
30376 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30378 2017-03-02 Richard Biener <rguenther@suse.de>
30380 PR tree-optimization/79345
30382 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30383 param and abort the walk, returning -1 if it is hit.
30384 (walk_aliased_vdefs): Take a limit param and pass it on.
30385 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30386 defaulting to 0 and return a signed int.
30387 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30388 (check_defs): New helper.
30389 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30390 about uninitialized memory.
30391 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30392 bogus uninitialized warning.
30393 (fixed_convert_from_real): Likewise.
30395 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30397 PR tree-optimization/66768
30398 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30399 iv_use if base object can't be determined.
30401 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30403 PR tree-optimization/79345
30404 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30405 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30406 (get_pattern_stats): Initialize it.
30407 * genemit.c (gen_expand): Verify match_scratch numbers come after
30408 match_operand/match_dup numbers.
30409 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30410 match_scratch numbers.
30411 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30413 * config/s390/s390.md (trunctdsd2): Likewise.
30415 2017-03-02 Richard Biener <rguenther@suse.de>
30417 * wide-int.h (wide_int_storage::operator=): Implement in terms
30420 2017-03-02 Richard Biener <rguenther@suse.de>
30422 PR tree-optimization/79777
30423 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30424 the to insert expression to sth existing.
30426 2017-03-01 Martin Sebor <msebor@redhat.com>
30428 PR middle-end/79692
30429 * gimple-ssa-sprintf.c
30430 (directive::known_width_and_precision): New function.
30431 (format_integer): Use it.
30432 (get_mpfr_format_length): Consider the full range of precision
30433 when computing %g output with the # flag. Set the likely byte
30434 count to 3 rather than 1 when precision is indeterminate.
30435 (format_floating): Correct the lower bound of precision.
30437 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30439 * doc/invoke.texi: Document default code model for 64-bit Linux.
30441 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30444 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30445 udiv rather than div since input pattern is unsigned.
30447 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30449 * config/i386/i386.c (print_reg): Warn for values of
30450 unsupported size in integer register.
30452 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30455 * config/rs6000/predicates.md (current_file_function_operand): Do
30456 not allow self calls to be local if the function is replaceable.
30458 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30461 * config/rs6000/altivec.h (vec_ctz and others): Change the
30462 preprocessor macro that controls conditional compilation from
30463 _ARCH_PWR9 to __POWER9_VECTOR__.
30464 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30465 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30466 control (instead of _ARCH_PWR9 control) so that template
30467 definition uses power9-specific function.
30468 (vec_any_eq): Likewise.
30469 (vec_all_ne): Change macro definition to use a power9-specific
30470 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30471 _ARCH_PWR9 control).
30472 (vec_any_eq) Likewise.
30473 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30474 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30475 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30476 support for xvcmpnedp instruction.
30477 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30478 macro expansion so that Power9 implementation of vec_all_ne does
30479 not use the AltiVec predicate framework.
30480 (VCMPNEH_P): Likewise.
30481 (VCMPNEW_P): Likewise.
30482 (VCMPNED_P): Likewise.
30483 (VCMPNEFP_P): Likewise.
30484 (VCMPNEDP_P): Likewise.
30485 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30486 implementation of vec_any_eq to not use AltiVec predicate
30488 (VCMPAEH_P): Likewise.
30489 (VCMPAEW_P): Likewise.
30490 (VCMPAED_P): Likewise.
30491 (VCMPAEFP_P): Likewise.
30492 (VCMPAEDP_P): Likewise.
30493 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30494 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30495 not use the AltiVec predicate framework.
30496 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30497 of vec_any_eq to not use AltiVec predicate framework.
30498 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30499 support for predefined __POWER9_VECTOR__ macro to indicate that
30500 Power9 instruction selection is enabled.
30501 (altivec_overloaded_builtins): Remove extraneous
30502 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30503 function argument types RS6000_BTI_bool_V16QI and
30504 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30505 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30506 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30507 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30508 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30509 Power9 for implementations of vec_cmpne. Change the signature for
30510 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30511 (representing vec_all_ne) to remove the previously described first
30512 argument of type RS6000_BTI_INTSI, as this was an artifact of
30513 reliance on the AltiVec predicate framework, which is no longer
30514 used in the implementation of these functions. Add
30515 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30516 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30517 since, unlike the AltiVec predicate framework implementation, we
30518 do not share function descriptors between vec_alle and vec_anyeq.
30519 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30520 set of modes that receive special treatment even when
30521 TARGET_P9_VECTOR is true. The special treatment emits code that
30522 does not depend on Power9 instructions.
30523 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30524 define_expand to not rely on AltiVec predicate framework.
30525 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30527 (vector_ne_v2di_p): Change this define_expand to not rely on
30528 AltiVec predicate framework.
30529 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30531 (vector_ne_<mode>_p): Change this define_expand to not rely on
30532 AltiVec predicate framework.
30533 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30535 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30536 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30537 define_insn pattern.
30538 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30539 define_insn pattern because the xvcmpne<VSs>. instruction is not
30541 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30542 instruction is not supported.
30544 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30546 * config/nvptx/nvptx.c: Include intl.h.
30548 2017-03-01 Martin Jambor <mjambor@suse.cz>
30551 * ipa-prop.h (ipa_bits): Removed field known.
30552 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30553 to pointers. Adjusted their comments to warn about their sharing.
30554 (ipcp_transformation_summary): Change bits to a vector of pointers.
30555 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30556 (ipa_get_ipa_bits_for_value): Declare.
30557 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30558 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30559 (ipa_bits_hash_table): Likewise.
30560 (ipa_vr_ggc_hash_traits): Likewise.
30561 (ipa_vr_hash_table): Likewise.
30562 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30563 being pointers and vr_known being removed.
30564 (ipa_set_jf_unknown): Likewise.
30565 (ipa_get_ipa_bits_for_value): New function.
30566 (ipa_set_jfunc_bits): Likewise.
30567 (ipa_get_value_range): New overloaded functions.
30568 (ipa_set_jfunc_vr): Likewise.
30569 (ipa_compute_jump_functions_for_edge): Use the above functions to
30570 construct bits and vr parts of jump functions.
30571 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30572 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30574 (ipcp_grow_transformations_if_necessary): Also allocate
30575 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30577 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30578 them. Fix too long lines.
30579 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30580 vr_known being removed.
30581 (ipa_read_jump_function): Use new setter functions to construct bits
30582 and vr parts of jump functions or set them to NULL.
30583 (write_ipcp_transformation_info): Adjust for bits being pointers.
30584 (read_ipcp_transformation_info): Likewise.
30585 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30587 Include gt-ipa-prop.h.
30588 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30590 (ipcp_store_bits_results): Likewise.
30591 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30592 Do not write to existing jump functions but use a temporary instead.
30594 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30597 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30598 attempt to use its first operand as BIT_FIELD_REF base.
30600 2017-03-01 Richard Biener <rguenther@suse.de>
30602 PR middle-end/79721
30603 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30604 interpolating formula in wrapping arithmetic.
30605 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30607 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30609 PR tree-optimization/79734
30610 * tree-vect-generic.c (expand_vector_condition): Optimize
30611 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30612 Handle VEC_COND_EXPR where comparison has different inner width from
30613 type's inner width.
30615 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30617 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30618 markup, and similar issues. Remove @opindex entries for things
30619 that aren't options. Add missing -mmpy-option entries.
30621 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30623 PR tree-optimization/79737
30624 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30625 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30626 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30627 instead of byte_size. Formatting fix.
30628 (shift_bytes_in_array_right): Formatting fix.
30630 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30633 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30634 condition on optimize for the leaf function test.
30636 2017-02-28 Martin Liska <mliska@suse.cz>
30639 * read-rtl-function.c (function_reader::handle_unknown_directive):
30640 Bail out when one uses -flto.
30642 2017-02-28 Martin Liska <mliska@suse.cz>
30644 * common.opt: Replace space with tabular for options of <number>
30646 * config/i386/i386.opt: Show <number> value for
30647 -mlarge-data-threshold.
30648 * opts.c (print_filtered_help): Do not display number in hexadecimal
30651 2017-02-28 Martin Liska <mliska@suse.cz>
30653 * common.opt: Fix --help=option -Q for options which are of
30656 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30658 * config/i386/i386.c (print_reg): Error out for values
30659 of 8-bit size in invalid integer register.
30661 2017-02-28 Martin Sebor <msebor@redhat.com>
30663 PR tree-optimization/79691
30664 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30666 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30669 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30670 gcc_unreachable with output_operand_lossage.
30672 2017-02-28 Richard Biener <rguenther@suse.de>
30674 PR tree-optimization/79740
30675 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30677 (visit_nary_op): Insert the nary into the hashtable if we
30678 pattern-matched sth.
30679 * tree-ssa-pre.c (eliminate_insert): Robustify.
30681 2017-02-28 Richard Biener <rguenther@suse.de>
30683 PR middle-end/79731
30684 * fold-const.c (decode_field_reference): Reject out-of-bound
30687 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30689 * config/i386/i386.c: Include intl.h.
30690 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30691 instead of just cond ? "..." : "...".
30692 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30693 * coverage.c (read_counts_file): Likewise.
30694 * omp-offload.c: Include intl.h.
30695 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30696 of just cond ? "..." : "...".
30697 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30698 of just cond ? "..." : "...".
30700 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30703 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30705 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30706 'tune for' CPU name.
30707 * config/arm/arm-cpu-data.h: Regenerated.
30709 2017-02-28 Richard Biener <rguenther@suse.de>
30711 PR tree-optimization/79732
30712 * tree-inline.c (expand_call_inline): Do not shadow var.
30714 2017-02-28 Richard Biener <rguenther@suse.de>
30716 PR tree-optimization/79723
30717 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30718 address-space properly.
30720 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30722 * doc/optinfo.texi (Optimization groups): Fix option used for
30724 * doc/invoke.texi (-fopt-info): Document "omp".
30725 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30726 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30727 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30728 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30729 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30731 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30733 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30734 OPTGROUP_OMP with "omp".
30736 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30739 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30740 for arithmetic shift of unsigned V2DI.
30742 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30744 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30745 arc/linux.h headers.
30746 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30747 (LINK_SPEC): Likewise.
30748 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30749 (EXTRA_SPECS): Likewise.
30750 (STARTFILE_SPEC): Likewise.
30751 (ENDFILE_SPEC): Likewise.
30752 (LIB_SPEC): Likewise.
30753 (TARGET_SDATA_DEFAULT): Likewise.
30754 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30755 (MULTILIB_DEFAULTS): Likewise.
30756 (DWARF2_UNWIND_INFO): Likewise.
30757 * config/arc/big.h: New file.
30758 * config/arc/elf.h: Likewise.
30759 * config/arc/linux.h: Likewise.
30760 * config/arc/t-uClibc: Remove.
30762 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30764 PR tree-optimization/77536
30765 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30766 (tree_transform_and_unroll_loop): Use above function to compute the
30767 estimated niter of unrolled loop and use it when scaling profile.
30768 Also use count info rather than frequency if it's non-zero.
30769 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30770 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30771 (vect_transform_loop): Call above function.
30773 2017-02-27 Richard Biener <rguenther@suse.de>
30775 PR tree-optimization/45397
30776 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30777 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30778 (visit_nary_op): Add pattern matching for CSEing sign-changed
30779 or truncated operations with wider ones.
30781 2017-02-27 Richard Biener <rguenther@suse.de>
30783 PR tree-optimization/79690
30784 * tree-vect-stmts.c (vectorizable_store): Use vector type
30785 built from the DR with address-space.
30787 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30789 * doc/invoke.texi (Optimize Options): Refine the description
30790 of asan-use-after-return.
30792 2017-02-25 Alan Modra <amodra@gmail.com>
30794 PR rtl-optimization/79584
30795 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30796 base, not ad->base_term, the reg within base. Remove assertion
30797 that ad->base == ad->base_term. Replace gen_int_mode using
30798 bogus mode with const0_rtx.
30800 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30802 PR middle-end/79396
30803 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30804 FMA_EXPR like tcc_binary or tcc_unary.
30806 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30809 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30811 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30812 (output_loc_operands): Handle DW_OP_call_ref and
30813 DW_OP_GNU_variable_value.
30814 (struct variable_value_struct): New type.
30815 (struct variable_value_hasher): Likewise.
30816 (variable_value_hash): New variable.
30817 (string_types): Remove.
30818 (copy_loc_descr): New function.
30819 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30820 (prepend_loc_descr_to_each): New function.
30821 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30822 instead of add_loc_descr_to_each if the first argument is single
30823 location list and the second has multiple.
30824 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30825 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30826 when looking for variable value which doesn't have other location info.
30827 (loc_list_from_tree): Formatting fix.
30828 (gen_array_type_die): Simplify DW_AT_string_length handling.
30829 (adjust_string_types): Remove.
30830 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30831 string_types. Call resolve_variable_values.
30832 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30833 (resolve_addr_in_expr): Likewise. Add A argument.
30834 (copy_deref_exprloc): Remove deref argument. Adjust for the
30835 original expression being DW_OP_GNU_variable_value with optionally
30836 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30837 optionally after it.
30838 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30839 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30840 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30841 (variable_value_hasher::hash, variable_value_hasher::equal): New
30843 (resolve_variable_value_in_expr, resolve_variable_value,
30844 resolve_variable_values, note_variable_value_in_expr,
30845 note_variable_value): New functions.
30846 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30849 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30852 * opts.h (handle_generated_option): Add GENERATED_P argument.
30853 * opts-common.c (handle_option): Adjust function comment.
30854 (handle_generated_option): Add GENERATED_P argument, pass it to
30856 (control_warning_option): Pass false to handle_generated_option
30858 * opts.c (maybe_default_option): Pass true to handle_generated_option
30860 * optc-gen.awk: Likewise.
30862 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30864 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30865 a REG, look at the REG it is a SUBREG of.
30866 (splitter for cmpeqsi_t): Ditto.
30868 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30870 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30871 the special USEs with the pattern of the insn, not the insn itself.
30873 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30876 * doc/invoke.texi: Document -mload-store-pairs.
30878 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30879 Sandra Loosemore <sandra@codesourcery.com>
30881 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30882 argument isn't a CONST_INT.
30883 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30884 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30885 (nios2_validate_compare): Bypass alternate compare logic if *op2
30886 is not a CONST_INT.
30887 (ldstwm_operation_p): Return false if first_base is not a REG or
30888 if first_offset is not a CONST_INT.
30890 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30892 * config/cris/cris.md: Use correct operand in a define_peephole2.
30894 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30896 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30898 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30900 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30901 this_insn if it is an INSN or JUMP_INSN.
30902 (force_offsettable): Look at base, not at addr.
30903 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30904 on things that aren't necessarily CONST_INTs.
30906 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30908 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30909 -mfpmath=sse is the default also for x86-32 targets with SSE2
30910 instruction set when @option{-ffast-math} is enabled
30912 2017-02-24 Jeff Law <law@redhat.com>
30914 PR rtl-optimizatoin/79286
30915 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30918 2017-02-24 Richard Biener <rguenther@suse.de>
30920 PR tree-optimization/79389
30921 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30924 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30926 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30927 function comment to reflect reality.
30928 (loop_exits_before_overflow): Fix typo in function description.
30930 2017-02-24 Richard Biener <rguenther@suse.de>
30932 PR tree-optimization/79389
30933 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30934 properly that a threading opportunity exists. Detect conditional
30935 copy/constant propagation opportunities.
30937 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30939 * config/visium/visium.md (type): Add trap.
30940 (b): New mode attribute.
30941 (*btst): Rename into...
30942 (*btst<mode>): ...this and adjust.
30943 (*cbranchsi4_btst_insn): Rename into...
30944 (*cbranch<mode>4_btst_insn): ...this and adjust.
30945 (trap): New define_insn.
30947 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30949 PR tree-optimization/79389
30950 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30951 (noce_reversed_cond_code): New function.
30952 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30953 reversed_comparison_code. Formatting fix.
30954 (noce_try_store_flag): Test rev_cond != NULL in addition to
30955 reversed_comparison_code.
30956 (noce_try_store_flag_constants): Likewise.
30957 (noce_try_store_flag_mask): Likewise.
30958 (noce_try_addcc): Use rev_cond if non-NULL instead of
30959 reversed_comparison_code.
30960 (noce_try_cmove_arith): Likewise. Formatting fixes.
30961 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30962 (noce_find_if_block): Initialize rev_cond.
30963 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30964 instead of false as last argument never attempt to reverse it
30967 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30969 PR tree-optimization/79663
30970 * tree-predcom.c (combine_chains): Process refs in reverse order
30971 only for ZERO length chains, and add explaining comment.
30973 2017-02-23 Jeff Law <law@redhat.com>
30975 PR tree-optimization/79578
30976 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30977 in call to operand_equal_p.
30979 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30982 * config/i386/cpuid.h: Fix another undefined behavior.
30984 2017-02-23 Richard Biener <rguenther@suse.de>
30986 PR tree-optimization/79683
30987 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30988 vector types for data-refs.
30990 2017-02-23 Martin Liska <mliska@suse.cz>
30992 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30994 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30996 PR middle-end/79665
30997 * internal-fn.c (get_range_pos_neg): Moved to ...
30998 * tree.c (get_range_pos_neg): ... here. No longer static.
30999 * tree.h (get_range_pos_neg): New prototype.
31000 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31001 are known to be in between 0 and signed maximum inclusive, try to
31002 expand both unsigned and signed divmod and use the cheaper one from
31005 2017-02-22 Jeff Law <law@redhat.com>
31007 PR tree-optimization/79578
31008 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31009 to compare base operands.
31011 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31014 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31015 gpc_reg_operand instead of fpr_reg_operand.
31017 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31019 * config/mips/mips.c (mips_return_in_memory): Force FP
31020 vector types to be returned in memory for o32 ABI.
31022 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31024 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31025 instead of DW_TAG_member for static data member declarations and don't
31026 set no_linkage_name for static inline data members.
31027 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31030 2017-02-22 Martin Liska <mliska@suse.cz>
31032 * doc/invoke.texi: Replace inequality signs with square brackets
31035 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31038 * lra-constraints.c (simplify_operand_subreg): Handle
31039 WORD_REGISTER_OPERATIONS targets.
31041 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31044 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31045 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31046 elimination by swapping fld*.
31048 2017-02-22 Richard Biener <rguenther@suse.de>
31050 PR tree-optimization/79673
31051 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31052 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31053 irrelevant address-space qualifiers and avoiding a
31054 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31056 2017-02-22 Richard Biener <rguenther@suse.de>
31058 PR tree-optimization/79666
31059 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31060 to not symbolically negate if that may introduce undefined
31063 2017-02-22 Martin Liska <mliska@suse.cz>
31066 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31067 * data-streamer-out.c (streamer_write_gcov_count_stream):
31069 * value-prof.c (stream_out_histogram_value): Make assert more
31070 precise based on type of counter.
31072 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31075 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31076 Use nonimmediate_operand instead of memory_operand for operand 1.
31077 (float-extend standard_x87sse_constant_load splitter): Ditto.
31079 2017-02-21 Jeff Law <law@redhat.com>
31081 PR tree-optimization/79621
31082 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31083 blocks with edges to themselves.
31085 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31088 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31089 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31090 Use gimple_call_builtin_p.
31093 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31094 on temporarily removed DEBUG_INSNs.
31096 PR tree-optimization/79649
31097 * tree-loop-distribution.c (classify_partition): Give up on
31098 non-generic address space loads/stores.
31100 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31102 * doc/loop.texi (Loop manipulation): Remove nonexistent
31103 tree_ssa_loop_version from the documentation.
31104 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31106 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31109 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31110 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31111 * config/rs6000/rs6000.c: Include except.h.
31112 (rs6000_expand_split_stack_prologue): Call
31113 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31115 2017-02-21 Martin Jambor <mjambor@suse.cz>
31118 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31119 have been analyzed.
31121 2017-02-21 Martin Jambor <mjambor@suse.cz>
31123 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31124 for backward compatibility only.
31125 * doc/invoke.texi (Option Summary): Remove all references to
31126 -fipa-cp-alignment.
31128 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31132 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31134 * lra-constraints.c (curr_insn_transform): Handle
31135 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31137 2017-02-21 Martin Liska <mliska@suse.cz>
31139 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31141 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31143 PR translation/79638
31144 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31146 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31149 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31150 (arm_function_ok_for_sibcall): Return false for an indirect call by
31151 descriptor if all the argument registers are used.
31152 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31153 alignment of the function.
31155 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31157 PR tree-optimization/61441
31158 * simplify-rtx.c (simplify_const_unary_operation): For
31159 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31160 the sNaN unmodified.
31162 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31164 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31165 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31166 instead of SYSTEM_HEADER_DIR.
31168 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31169 Martin Liška <mliska@suse.cz>
31171 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31172 Fix typos and grammar, use active voice, and clarify.
31174 2017-02-20 Marek Polacek <polacek@redhat.com>
31176 PR middle-end/79537
31177 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31180 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31182 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31185 * config/i386/i386.c (ix86_expand_builtin): Handle
31186 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31187 ix86_builtins_isa[fcode].isa as a requirement of those
31188 flags and any other flag in the bitmask.
31189 (ix86_init_mmx_sse_builtins): Use 0 instead of
31190 ~OPTION_MASK_ISA_64BIT as mask.
31191 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31192 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31193 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31194 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31196 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31199 * lra-constraints.c (split_reg): Check requested split mode
31200 is supported by the register.
31202 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31204 * lra-constraints.c (simplify_operand_subreg): Remove early
31207 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31210 * lra-constraints.c (curr_insn_transform): Tighten condition
31211 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31213 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31216 * lra-constraints.c (curr_insn_transform): Handle
31217 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31219 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31222 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31224 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31226 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31229 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31232 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31234 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31237 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31239 * final.c (last_columnnum, override_columnnum): New variables.
31240 (final_start_function): Set last_columnnum, pass it to begin_prologue
31241 hook and pass 0 to dwarf2out_begin_prologue.
31242 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31243 to source_line debug hook.
31244 (notice_source_line): Compute last_columnnum and for debug_column_info
31245 return true on column changes.
31246 * debug.h (struct gcc_debug_hooks): Add column argument to
31247 source_line and begin_prologue hooks.
31248 (debug_nothing_int_charstar_int_bool): Remove prototype.
31249 (debug_nothing_int_int_charstar,
31250 debug_nothing_int_int_charstar_int_bool): New prototypes.
31251 (dwarf2out_begin_prologue): Add column argument.
31252 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31253 begin_prologue hooks.
31254 (debug_nothing_int_charstar_int_bool): Remove.
31255 (debug_nothing_int_int_charstar,
31256 debug_nothing_int_int_charstar_int_bool): New functions.
31257 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31258 through to dwarf2out_source_line.
31259 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31260 (dwarf2out_source_line): Add column argument, emit it if requested.
31261 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31263 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31264 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31265 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31266 through to dwarf2out_begin_prologue.
31267 (vmsdbgout_source_line): Add column argument, pass it through to
31268 dwarf2out_source_line.
31269 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31270 dbxout_source_line caller.
31271 (dbxout_source_line): Add column argument.
31273 * common.opt (gno-column-info, gcolumn-info): New options.
31274 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31275 (check_die): Also test for multiple DW_AT_decl_column attributes.
31276 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31277 DW_AT_decl_column if requested.
31278 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31280 (gen_variable_die): Likewise.
31281 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31282 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31285 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31286 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31287 (ix86_handle_option): Handle OPT_m3dnowa.
31288 * doc/invoke.texi (-m3dnowa): Document.
31289 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31290 -m3dnowa instead of -m3dnow -march=athlon.
31293 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31294 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31296 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31299 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31300 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31301 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31302 generator for vsx_xxpermdi_<mode>_be.
31303 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31304 force big-endian semantics.
31305 (vsx_xxpermdi_<mode>_be): New define_expand with same
31306 implementation as previous version of vsx_xxpermdi_<mode>.
31308 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31310 PR tree-optimization/79327
31311 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31312 variable, its initialization and use.
31314 2017-02-17 Julia Koval <julia.koval@intel.com>
31316 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31317 (OPTION_MASK_ISA_PKU_UNSET): New.
31318 (ix86_handle_option): Handle -mrdpid.
31319 * config/i386/cpuid.h (bit_RDPID): New.
31320 * config/i386/driver-i386.c (host_detect_local_cpu):
31321 Detect RDPID feature.
31322 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31323 * config/i386/i386-c.c (ix86_target_macros_internal):
31325 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31326 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31327 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31328 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31329 * config/i386/i386.md (define_insn "rdpid"): New.
31330 * config/i386/i386.opt Add -mrdpid.
31331 * config/i386/immintrin.h (_rdpid_u32): New.
31333 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31335 PR rtl-optimization/79541
31336 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31337 instead of transforming it into USE.
31339 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31341 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31342 If HONOR_SNANS (SFmode) force the input to a register.
31343 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31344 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31345 an frsp or similar insn.
31347 2017-02-17 Martin Liska <mliska@suse.cz>
31349 PR rtl-optimization/79577
31350 * params.def (selsched-max-sched-times): Increase minimum to 1.
31352 2017-02-17 Martin Liska <mliska@suse.cz>
31354 PR rtl-optimization/79574
31355 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31357 2017-02-17 Martin Liska <mliska@suse.cz>
31359 PR tree-optimization/79529
31360 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31361 ssa_defined_default_def_p to handle cases which are implicitly
31363 * tree-ssa.c (ssa_defined_default_def_p): New function.
31364 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31365 which are implicitly defined.
31366 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31368 2017-02-17 Richard Biener <rguenther@suse.de>
31370 PR middle-end/79576
31371 * params.def (max-ssa-name-query-depth): Limit to 10.
31373 2017-02-17 Richard Biener <rguenther@suse.de>
31375 PR tree-optimization/79552
31376 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31379 2017-02-17 Richard Biener <rguenther@suse.de>
31382 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31384 2017-02-17 Marek Polacek <polacek@redhat.com>
31386 PR middle-end/79536
31387 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31388 (fold_negate_expr): New wrapper.
31390 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31392 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31393 Correct terminology and de-emphasize pre-standard behavior.
31395 2017-02-16 Alan Modra <amodra@gmail.com>
31397 PR rtl-optimization/79286
31398 * ira.c (def_dominates_uses): New function.
31399 (update_equiv_regs): Don't create an equivalence for insns that
31400 may trap where the register def does not dominate the use.
31402 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31404 PR rtl-optimization/78127
31405 * lra.c (lra): Call lra_eliminate before finish the loop after
31408 2017-02-16 Richard Biener <rguenther@suse.de>
31410 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31412 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31413 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31414 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31415 (isl_val_int_from_wi): New function.
31416 (extract_affine_gmp): Rename to ...
31417 (extract_affine_wi): ... this, take a widest_int.
31418 (extract_affine_int): Just wrap extract_affine_wi.
31419 (add_param_constraints): Use isl_val_int_from_wi.
31420 (add_loop_constraints): Likewise, and extract_affine_wi.
31422 2017-02-15 Jeff Law <law@redhat.com>
31424 PR middle-end/79521
31425 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31426 ira_init_register_move_cost_if_necessary.
31428 2017-02-15 Martin Sebor <msebor@redhat.com>
31430 PR middle-end/32003
31431 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31432 removed in a prior commit.
31434 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31436 PR tree-optimization/79347
31437 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31438 counters during peeling.
31440 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31442 * Makefile.in (site.exp): Remove "set ISLVER".
31444 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31447 * real.c (real_from_integer): Call real_convert even for decimal.
31449 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31452 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31454 2017-02-14 Andrew Pinski <apinski@cavium.com>
31456 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31457 cores and change the partno/implementer to be correct.
31458 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31459 the 'B" as the implementer.
31460 * config/aarch64/aarch64-tune.md: Regenerate.
31462 2017-02-14 Carl Love <cel@us.ibm.com>
31464 * config/rs6000/rs6000.c: Add case statement entry to make the
31465 xvcvuxdsp built-in argument unsigned.
31466 * config/rs6000/vsx.md: Fix the source and return operand types so they
31467 match the instruction definitions from the ISA document. Fix typo
31468 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31471 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31474 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31475 member early_clobber_alts.
31476 * lra-lives.c (reg_early_clobber_p): New.
31477 (process_bb_lives): Use it.
31478 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31479 (debug_operand_data): Initialize early_clobber_alts.
31480 (setup_operand_alternative): Set up early_clobber_alts.
31481 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31482 alternatives to new_insn_reg.
31483 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31485 (lra_update_insn_regno_info): Pass the new arg.
31487 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31489 PR middle-end/79505
31490 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31491 (new_oacc_loop_raw): Don't clear already cleared fields.
31494 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31495 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31496 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31498 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31501 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31503 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31506 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31507 the extra instruction to the right place to store 128-bit constant
31510 2017-02-14 Martin Sebor <msebor@redhat.com>
31512 PR middle-end/79448
31513 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31514 warning for strings of unknown length.
31516 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31518 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31520 2017-02-14 Jeff Law <law@redhat.com>
31523 * ira-costs.c (scan_one_insn): Initialize register move costs
31524 for pseudos seen in USE/CLOBBER insns.
31526 PR tree-optimization/79095
31527 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31528 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31529 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31530 if the operands are known to be not equal, then the resulting range
31532 (intersect_ranges): If the new range is ~[0,0] and the old range is
31533 wide, then prefer ~[0,0].
31534 * tree-vrp.c (overflow_comparison_p_1): New function.
31535 (overflow_comparison_p): New function.
31536 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31537 if NAME is used in an overflow test.
31538 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31539 overflow check that can be expressed as an equality test, then adjust
31540 ops to be that equality test.
31542 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31544 * config/s390/s390-builtin-types.def: Remove flags argument.
31545 * config/s390/s390.c (s390_init_builtins): Likewise.
31547 2017-02-14 Martin Liska <mliska@suse.cz>
31549 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31550 vector. Fix trailing white spaces.
31552 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31554 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31557 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31559 PR rtl-optimization/68664
31560 * config/arm/arm.c (arm_sched_can_speculate_insn):
31561 New function. Declare prototype.
31562 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31564 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31566 PR rtl-optimization/68664
31567 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31569 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31571 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31573 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31574 max skip bytes for function, loop and jump.
31576 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31578 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31579 ABS_EXPR for gimple dump.
31581 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31584 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31586 PR tree-optimization/79408
31587 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31588 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31589 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31590 also if rhs1 is INTEGER_CST.
31592 2017-02-14 Richard Biener <rguenther@suse.de>
31594 PR middle-end/79432
31595 * tree-into-ssa.c (insert_phi_nodes): When the function can
31596 have abnormal edges rewrite SSA names with broken use-def
31597 dominance out of SSA and register them for PHI insertion.
31599 2017-02-13 Martin Sebor <msebor@redhat.com>
31601 PR middle-end/79496
31602 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31603 clearing info.nowrite flag when snprintf size argument is a range.
31605 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31607 * cprop.c (cprop_jump): Add missing space in string literal.
31608 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31609 (get_constraint_for_component_ref): Likewise.
31610 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31611 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31612 * lra-constraints.c (process_alt_operands): Likewise.
31613 * ipa-inline.c (inline_small_functions): Likewise.
31614 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31615 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31616 * trans-mem.c (diagnose_tm_1_op): Likewise.
31617 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31618 (grid_parallel_clauses_gridifiable): Likewise.
31620 * config/nvptx/mkoffload.c (process): Add space in between
31623 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31624 "MOD4_SSE_REGS" and "ALL_REGS".
31626 * spellcheck.c (test_data): Add , in between "foo" and "food".
31628 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31631 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31632 boundary crossing check and subsequent code generation agree.
31634 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31636 * config/aarch64/aarch64.c (has_memory_op): Delete.
31637 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31640 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31642 PR rtl-optimization/79388
31643 PR rtl-optimization/79450
31644 * combine.c (distribute_notes): When removing TEM_INSN for which
31645 corresponding dest has last value recorded, invalidate that last
31648 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31650 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31651 of explicit '@'. Add missing assembly comment marker on branch costs
31654 2017-02-13 Nathan Sidwell <nathan@acm.org>
31656 * gengtype-lex.l (<in_struct>): Add '/'.
31658 2017-02-13 Martin Liska <mliska@suse.cz>
31661 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31663 2017-02-13 Richard Biener <rguenther@suse.de>
31665 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31667 * configure: Re-generate.
31668 * config.in: Likewise.
31669 * graphite-dependences.c: Simplify as if
31670 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31671 * graphite-isl-ast-to-gimple.c: Likewise.
31672 * graphite-optimize-isl.c: Likewise.
31673 * graphite-poly.c: Likewise.
31674 * graphite-sese-to-poly.c: Likewise.
31675 * graphite.h: Likewise.
31676 * toplev.c: Include isl/version.h and use isl_version () for
31677 printing the ISL version.
31678 * doc/install.texi: Update ISL requirement.
31680 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31682 * doc/standards.texi (Standards): Update reference to
31685 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31687 * doc/extend.texi (Named Address Spaces): sourceware.org now
31689 * doc/install.texi (Binaries): Ditto.
31692 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31694 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31695 standard terminology "stringize"/"stringizing" throughout.
31696 * doc/cppinternals.texi: Likewise.
31698 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31700 * doc/extend.texi: Fix some spelling mistakes and typos.
31701 * doc/invoke.texi: Likewise.
31703 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31706 * params.def (inline-min-speedup) Change from 10 to 8.
31708 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31710 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31713 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31716 * ipa-inline-analysis.c (get_minimal_bb): New function.
31717 (record_modified): Use it.
31718 (remap_edge_change_prob): Handle also ancestor functions.
31720 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31722 * doc/contrib.texi (Contributors): Remove broken link into
31723 the Mauve CVS repository.
31725 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31727 PR middle-end/79454
31728 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31729 result computation whenever lhs doesn't have vector mode, not
31730 just when it has BLKmode.
31732 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31734 * doc/makefile.texi (profiledbootstrap): Refer to the
31735 installation instructions only in textual form.
31737 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31740 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31742 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31744 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31745 (Specific): Update mingw-w64 reference.
31747 (Specific): Remove broken link to Renesas RX processor.
31749 2017-02-10 Richard Biener <rguenther@suse.de>
31751 * toplev.c (process_options): Do not mention obsolete graphite
31752 options when printing sorry message about missing graphite support.
31753 Mention -floop-nest-optimize.
31755 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31757 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31758 (vtst_p16): Likewise.
31759 (vtstq_p8): Likewise.
31760 (vtstq_p16): Likewise.
31762 (vtstq_p64): Likewise.
31763 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31764 (vset_lane_p64): New.
31765 (vsetq_lane_p64): New.
31767 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31769 PR tree-optimization/79411
31770 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31771 stmt operands are SSA_NAMEs used in abnormal phis.
31772 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31775 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31778 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31781 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31783 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31785 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31787 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31791 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31792 not arbitrary TREE_CONSTANT.
31795 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31796 "omp declare target link" attribute unless is_global_var.
31797 * omp-offload.c (find_link_var_op): Likewise.
31799 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31800 Chung-Lin Tang <cltang@codesourcery.com>
31802 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31804 (gimplify_adjust_omp_clauses): Don't delete TILE.
31805 (gimplify_omp_for): Deal with TILE.
31806 * internal-fn.c (expand_GOACC_TILE): New function.
31807 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31809 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31810 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31812 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31813 avoid DIV for outermost collapse var.
31814 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31815 Remove out of date comments, fix whitespace.
31816 * omp-general.c (omp_extract_for_data): Deal with tiling.
31817 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31818 adjust OLF_DIM_BASE value.
31819 (struct omp_for_data): Add tiling field.
31820 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31821 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31822 for auto loops. Remove default auto determining, moved to
31823 oacc_loop_fixed_partitions.
31824 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31825 stmts, add e_mask field.
31826 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31827 (oacc_thread_numbers): Use oacc_dim_call.
31828 (oacc_xform_tile): New.
31829 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31830 (finish_oacc_loop): Adjust for ifns vector.
31831 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31832 add case for GOACC_TILE fns.
31833 (oacc_loop_xform_loop): Delete.
31834 (oacc_loop_process): Iterate over call list directly, and add
31835 handling for GOACC_TILE fns.
31836 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31838 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31839 vector partitioning to outer loops. Assign 2 partitions to loops
31840 when available. Add TILE handling.
31841 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31842 (execite_oacc_device_lower): Process GOACC_TILE fns,
31843 ignore unknown specs.
31844 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31845 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31846 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31848 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31850 * configure.ac (ACX_BUGURL): Update.
31851 * configure: Regenerate.
31853 2017-02-09 Richard Biener <rguenther@suse.de>
31855 PR tree-optimization/69823
31856 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31857 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31859 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31861 * config/arc/arc-c.def: Add __NPS400__ definition.
31862 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31863 (TARGET_NPS400): Define.
31865 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31867 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31869 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31870 pointer, arch_info.
31871 (arc_cpu_types): Fill the arch_info field with a pointer into the
31872 arc_arch_types table.
31873 (arc_selected_cpu): Declare.
31874 * config/arc/arc.c (arc_selected_cpu): Make global.
31875 (arc_selected_arch): Delete.
31876 (arc_base_cpu): Delete.
31877 (arc_override_options): Remove references to deleted variables,
31878 update access to arch information.
31879 (ARC_OPT): Update access to arch information.
31880 (ARC_OPTX): Likewise.
31881 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31882 (TARGET_ARC600): Update access to arch information.
31883 (TARGET_ARC601): Likewise.
31884 (TARGET_ARC700): Likewise.
31885 (TARGET_EM): Likewise.
31886 (TARGET_HS): Likewise.
31887 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31890 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31893 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31894 condition/operands for integer GE/LE/GEU/LEU operations.
31896 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31898 PR translation/79397
31899 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31902 2017-02-08 Martin Jambor <mjambor@suse.cz>
31905 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31906 whether allocation happened.
31907 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31908 nothing was allocated.
31910 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31912 PR tree-optimization/79408
31913 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31914 constant, but SSA_NAME with a known integer range, use the minimum
31915 of that range instead of op1 to determine if modulo can be replaced
31916 with its first operand.
31918 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31920 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31922 2017-02-08 Richard Biener <rguenther@suse.de>
31924 PR tree-optimization/71824
31925 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31926 Check all loops contained in the merged region.
31928 2017-02-07 Andrew Pinski <apinski@cavium.com>
31930 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31932 2017-02-07 Andrew Pinski <apinski@cavium.com>
31934 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31935 (thunderxt88): Likewise.
31936 (thunderxt81): Disable LSE and change v8.1 to v8.
31937 (thunderxt83): Likewise.
31939 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31940 Richard Biener <rguenther@suse.de>
31942 PR middle-end/79399
31943 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31944 type from int to size_t.
31945 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31947 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31949 PR rtl-optimization/79386
31950 * cprop.c (bypass_conditional_jumps): Initialize
31951 bypass_last_basic_block already before splitting bbs after
31952 unconditional traps...
31953 (bypass_conditional_jumps): ... rather than here.
31956 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31957 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31958 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31959 fix -masm=intel patterns.
31961 2017-02-07 Richard Biener <rguenther@suse.de>
31963 PR tree-optimization/79256
31964 PR middle-end/79278
31965 * builtins.c (get_object_alignment_2): Use min_align_of_type
31966 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31967 and ADJUST_FIELD_ALIGN.
31969 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31971 * doc/tm.texi: Regenerate.
31972 * stor-layout.c (layout_decl): Adjust.
31973 (update_alignment_for_field): Likewise.
31974 (place_field): Likewise.
31975 (min_align_of_type): Likewise.
31976 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31977 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31978 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31979 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31980 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31981 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31982 * config/i386/i386.c (x86_field_alignment): Likewise.
31983 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31984 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31985 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31986 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31987 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31988 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31992 2017-01-30 Richard Biener <rguenther@suse.de>
31994 PR tree-optimization/79256
31995 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31996 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31999 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32001 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32002 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32003 builtins to SImode and emit a zero-extend, if necessary.
32005 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32007 * docs/invoke.texi (RISC-V Options): Alphabetize.
32009 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32011 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32014 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32016 * config/riscv/riscv.c: New file.
32017 * common/config/riscv/riscv-common.c: Likewise.
32018 * config.gcc: Likewise.
32019 * config/riscv/constraints.md: Likewise.
32020 * config/riscv/elf.h: Likewise.
32021 * config/riscv/generic.md: Likewise.
32022 * config/riscv/linux.h: Likewise.
32023 * config/riscv/multilib-generator: Likewise.
32024 * config/riscv/peephole.md: Likewise.
32025 * config/riscv/pic.md: Likewise.
32026 * config/riscv/predicates.md: Likewise.
32027 * config/riscv/riscv-builtins.c: Likewise.
32028 * config/riscv/riscv-c.c: Likewise.
32029 * config/riscv/riscv-ftypes.def: Likewise.
32030 * config/riscv/riscv-modes.def: Likewise.
32031 * config/riscv/riscv-opts.h: Likewise.
32032 * config/riscv/riscv-protos.h: Likewise.
32033 * config/riscv/riscv.h: Likewise.
32034 * config/riscv/riscv.md: Likewise.
32035 * config/riscv/riscv.opt: Likewise.
32036 * config/riscv/sync.md: Likewise.
32037 * config/riscv/t-elf-multilib: Likewise.
32038 * config/riscv/t-linux: Likewise.
32039 * config/riscv/t-linux-multilib: Likewise.
32040 * config/riscv/t-riscv: Likewise.
32041 * configure.ac: Likewise.
32042 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32043 Waterman as RISC-V maintainers.
32044 * doc/install.texi: Add RISC-V entries.
32045 * doc/invoke.texi: Add RISC-V options section.
32046 * doc/md.texi: Add RISC-V constraints section.
32047 * configure: Regenerated.
32049 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32052 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32053 false values to be constant vectors with all 0 or all 1 bits set.
32054 (vcondu<mode><mode>): Likewise.
32055 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32057 (fpmask_comparison_operator): Update comment.
32058 (vecint_comparison_operator): New predicate.
32059 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32060 vector conditionals when the true and false values are constant
32061 vectors with all 0 bits or all 1 bits set.
32063 2017-02-06 Martin Sebor <msebor@redhat.com>
32065 PR tree-optimization/79376
32066 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32068 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32070 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32071 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32072 to simplify split condition.
32074 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32076 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32077 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32080 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32082 PR rtl-optimization/68664
32083 * target.def (can_speculate_insn): New hook.
32084 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32085 * doc/tm.texi: Regenerate.
32086 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32087 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32088 (rs6000_sched_can_speculate_insn): New function.
32090 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32092 PR tree-optimization/79284
32093 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32094 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32095 vectorizable_mask_load_store, vectorizable_operation,
32096 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32097 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32098 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32099 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32100 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32101 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32102 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32103 is_gimple_assign (stmt). Replace another such test with
32104 is_gimple_assign (stmt).
32106 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32109 * config/avr/avr.c (rtl-iter.h): Include it.
32110 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32111 (avr_legitimate_combined_insn): ...and implementation.
32113 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32115 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32116 * config/s390/s390.c (s390_const_operand_ok)
32117 (s390_canonicalize_comparison, s390_extract_part)
32118 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32119 (s390_contiguous_bitmask_p, s390_rtx_costs)
32120 (legitimize_pic_address): Likewise.
32121 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32122 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32123 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32124 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32125 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32127 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32129 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32130 REGNO($0) == REGNO($1).
32132 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32134 * config/s390/linux.h(SIZE_TYPE): Add comment.
32136 2017-02-06 Julian Brown <julian@codesourcery.com>
32137 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32138 Virendra Pathak <virendra.pathak@broadcom.com>
32140 * config/aarch64/aarch64-cores.def: Change the scheduler
32142 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32143 * config/aarch64/thunderx2t99.md: New file.
32145 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32147 * doc/standards.texi (Go Language): Update link to language
32150 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32152 * tree-eh.c (lower_resx): Sanitize profile.
32153 (cleanup_empty_eh_move_lp): Likewise.
32155 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32158 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32160 * cfgloopmanip.h (loop_version): Update prototype.
32161 * modulo-sched.c (sms_schedule): Update call of loop_version.
32162 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32163 * tree-parloops.c (gen_parallel_loop): Likewise.
32164 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32165 * tree-ssa-loop-split.c (split_loop): Likewise.
32166 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32167 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32169 2017-02-05 Martin Liska <mliska@suse.cz>
32172 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32174 (print_operand_address): Initialize a struct to zero.
32176 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32178 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32179 garbage collector only in textual form.
32181 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32183 * doc/extend.texi (x86 specific memory model extensions for
32184 transactional memory): Simplify a phrase.
32186 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32189 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32190 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32191 (atomic_storedi_1): Likewise.
32193 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32195 PR tree-optimization/79338
32196 * tree-parloops.c (gather_scalar_reductions): Don't call
32197 vect_analyze_loop_form for loop->inner before destroying loop's
32200 2017-02-03 Martin Sebor <msebor@redhat.com>
32202 PR tree-optimization/79327
32203 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32204 when precision has resulted in leading zeros.
32205 (format_integer): Adjust the likely counter to assume an unknown
32206 argument that may be zero is non-zero.
32208 2017-02-03 Jason Merrill <jason@redhat.com>
32211 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32212 avoid copying non-taken branch.
32214 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32216 PR tree-optimization/79340
32217 * tree-vect-loop.c (vectorizable_reduction): Release
32218 vec_defs elements after safe_splicing them into other vectors.
32221 PR tree-optimization/79327
32222 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32223 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32225 (format_integer): Use wide_int_to_tree instead of build_int_cst
32226 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32227 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32228 of shortest and longest sequence.
32230 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32232 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32233 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32235 2017-02-03 Walter Lee <walt@tilera.com>
32238 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32239 after initial stackframe link reg save.
32240 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32242 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32245 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32246 wu for stxssp alternative.
32248 2017-02-03 Martin Sebor <msebor@redhat.com>
32250 PR tree-optimization/79352
32251 * gimple-fold.c (get_range_strlen): Add argument.
32252 (get_range_strlen): Change return type to bool.
32253 (get_maxval_strlen): Pass in a dummy argument.
32254 * gimple-fold.h (get_range_strlen): Change return type to bool.
32255 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32256 * tree.h (array_at_struct_end_p): Add argument.
32257 * tree.c (array_at_struct_end_p): Handle it.
32259 2017-02-03 Martin Liska <mliska@suse.cz>
32262 * multiple_target.c (create_dispatcher_calls): Redirect edge
32263 from a caller of a dispatcher.
32264 (expand_target_clones): Make the clones local.
32265 (ipa_target_clone): Do both target clones and resolvers.
32266 (ipa_dispatcher_calls): Remove the pass.
32267 (pass_dispatcher_calls::gate): Likewise.
32268 (make_pass_dispatcher_calls): Likewise.
32269 * passes.def (pass_target_clone): Put as very first IPA early
32272 2017-02-03 Martin Liska <mliska@suse.cz>
32274 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32275 in case of a function with ifunc attribute.
32277 2017-02-03 Martin Liska <mliska@suse.cz>
32279 * cgraph.c (cgraph_node::dump): Dump function version info.
32280 * symtab.c (symtab_node::dump_base): Add missing new line.
32282 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32284 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32285 (ifcombine_ifandif): Use it.
32287 2017-02-03 Martin Liska <mliska@suse.cz>
32289 * doc/invoke.texi: Document default value for
32290 use-after-scope-direct-emission-threshold.
32292 2017-02-03 Martin Liska <mliska@suse.cz>
32294 PR tree-optimization/79339
32295 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32296 (format_floating): Likewise.
32298 2017-02-03 Martin Liska <mliska@suse.cz>
32301 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32303 (ipa_node_params_t::remove): Likewise.
32304 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32305 initialization from removed ipa_node_params_t::insert.
32306 (ipa_node_params::~ipa_node_params): Move from removed
32307 ipa_node_params_t::release.
32308 * symbol-summary.h (symbol_summary::m_released): New member.
32309 Do not release a summary twice. Do not allow to call finalizer
32310 for types of a summary that live in GGC memory.
32312 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32314 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32317 2017-02-02 Martin Sebor <msebor@redhat.com>
32319 PR middle-end/79275
32320 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32321 (format_string): Tighten up the range of output for non-constant
32322 strings and correct the expected range for wide non-constant strings.
32324 2017-02-02 Martin Sebor <msebor@redhat.com>
32326 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32328 PR middle-end/32003
32329 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32331 (-fdump-tree-@var): Add to index and document how to come up
32332 with pass-specific option and dump file names.
32333 (-fdump-passes): Clarify where to look for output.
32335 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32337 PR middle-end/77445
32338 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32339 statistics of the analyzed path; allow threading for speed when
32340 any of BBs along the path are optimized for speed.
32342 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32344 PR middle-end/78468
32345 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32346 settings of the virtual registers.
32349 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32351 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32352 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32355 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32357 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32358 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32360 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32362 * config/s390/s390.md: Add missing comments with the expanded
32364 * config/s390/vector.md: Likewise.
32365 * config/s390/vx-builtins.md: Likewise.
32367 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32370 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32371 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32372 conditions on a single line.
32374 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32376 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32377 __S390_VX__ to __VX__.
32379 2017-02-01 Andrew Pinski <apinski@cavium.com>
32381 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32382 stmt_info to record_stmt_cost.
32383 (vect_get_known_peeling_cost): Pass stmt_info if known to
32385 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32386 cpu_vector_cost field into
32387 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32388 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32389 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32390 splitting of scalar_stmt_cost and vec_stmt_cost.
32391 (thunderx_vector_cost): Likewise.
32392 (cortexa57_vector_cost): LIkewise.
32393 (exynosm1_vector_cost): Likewise.
32394 (xgene1_vector_cost): Likewise.
32395 (thunderx2t99_vector_cost): Improve after the splitting of the two
32397 (aarch64_builtin_vectorization_cost): Update for the splitting of
32398 scalar_stmt_cost and vec_stmt_cost.
32400 2017-02-01 Torvald Riegel <triegel@redhat.com>
32401 Richard Henderson <rth@redhat.com>
32403 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32404 conditional on existance of a fast atomic load.
32405 * optabs-query.c (can_atomic_load_p): New function.
32406 * optabs-query.h (can_atomic_load_p): Declare it.
32407 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32408 no fast atomic load is available for the particular size of access.
32409 (expand_atomic_compare_and_swap): Likewise.
32410 (expand_atomic_load): Likewise.
32411 (expand_atomic_store): Likewise.
32412 (expand_atomic_fetch_op): Likewise.
32413 * testsuite/lib/target-supports.exp
32414 (check_effective_target_sync_int_128): Remove x86 because it provides
32415 no fast atomic load.
32416 (check_effective_target_sync_int_128_runtime): Likewise.
32418 2017-02-01 Richard Biener <rguenther@suse.de>
32420 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32421 (graphite_transform_loops): Provide opt-info for optimized nests.
32422 * tree-parloop.c (parallelize_loops): Provide opt-info for
32423 parallelized loops.
32425 2017-02-01 Richard Biener <rguenther@suse.de>
32427 PR middle-end/79315
32428 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32429 was not set before.
32431 2017-02-01 Richard Biener <rguenther@suse.de>
32433 PR tree-optimization/71824
32434 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32435 Verify the loops are valid in the merged SESE region.
32436 (scop_detection::can_represent_loop_1): Check analyzing the
32437 evolution of the number of iterations in the region succeeds.
32439 2017-01-31 Ian Lance Taylor <iant@golang.org>
32441 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32442 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32444 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32446 PR preprocessor/79210
32447 * input.c (get_substring_ranges_for_loc): Replace line_width
32448 assertion with error-handling.
32450 2017-01-31 Richard Biener <rguenther@suse.de>
32452 PR tree-optimization/77318
32453 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32454 (create_pw_aff_from_tree): Take loop parameter.
32455 (add_condition_to_pbb): Pass loop of the condition to
32456 create_pw_aff_from_tree.
32458 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32460 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32461 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32463 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32467 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32469 (convert_int_to_float128): Likewise.
32470 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32471 (convert_int_to_float128): Likewise.
32472 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32473 (UNSPEC_IEEE128_CONVERT): Likewise.
32474 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32475 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32476 Use local variables for IBM extended format.
32477 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32478 (fix_trunc<mode>si2_fprs): Likewise.
32479 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32480 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32481 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32482 to know that we can now have integers of all sizes in vector
32484 (fix<uns>_<mode>di2_hw): Likewise.
32485 (float<uns>_<mode>si2_hw): Likewise.
32486 (fix_<mode>si2_hw): Likewise.
32487 (fixuns_<mode>si2_hw): Likewise.
32488 (float<uns>_<mode>di2_hw): Likewise.
32489 (float_<mode>di2_hw): Likewise.
32490 (float_<mode>si2_hw): Likewise.
32491 (floatuns_<mode>di2_hw): Likewise.
32492 (floatuns_<mode>si2_hw): Likewise.
32493 (xscvqp<su>wz_<mode>): Delete, no longer used.
32494 (xscvqp<su>dz_<mode>): Likewise.
32495 (xscv<su>dqp_<mode>): Likewise.
32496 (ieee128_mfvsrd_64bit): Likewise.
32497 (ieee128_mfvsrd_32bit): Likewise.
32498 (ieee128_mfvsrwz): Likewise.
32499 (ieee128_mtvsrw): Likewise.
32500 (ieee128_mtvsrd_64bit): Likewise.
32501 (ieee128_mtvsrd_32bit): Likewise.
32503 2017-01-31 Martin Liska <mliska@suse.cz>
32506 * ipa-prop.c (ipa_free_all_node_params): Call release method
32507 instead of ~sumbol_summary to not to trigger double times
32510 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32512 PR tree-optimization/71691
32513 * bitmap.h (class auto_bitmap): New.
32514 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32515 is_maybe_undefined instead of ssa_undefined_value_p.
32517 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32519 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32520 __S390_ARCH_LEVEL__ to __ARCH__.
32522 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32524 PR tree-optimization/79267
32525 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32526 if should_remove_lhs_p is true.
32528 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32531 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32532 (add_alignment_attribute): New.
32533 (base_type_die): Add alignment attribute.
32534 (subrange_type_die): Likewise.
32535 (modified_type_die): Likewise.
32536 (gen_array_type_die): Likewise.
32537 (gen_descr_array_type_die: Likewise.
32538 (gen_enumeration_type_die): Likewise.
32539 (gen_subprogram_die): Likewise.
32540 (gen_variable_die): Likewise.
32541 (gen_field_die): Likewise.
32542 (gen_ptr_to_mbr_type_die): Likewise.
32543 (gen_struct_or_union_type_die): Likewise.
32544 (gen_subroutine_type_die): Likewise.
32545 (gen_typedef_die): Likewise.
32546 (base_type_cmp): Compare alignment attribute.
32548 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32551 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32552 (setb_unsigned) New pattern for setb with CCUNS.
32553 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32554 subfc./subfe sequence to avoid overflow problems. Generate a
32555 shorter sequence with cmpld/setb for power9.
32556 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32557 for generating subfc. instruction.
32558 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32559 now uses this instruction.
32561 2017-01-30 Ian Lance Taylor <iant@google.com>
32564 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32565 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32567 2017-01-30 Martin Sebor <msebor@redhat.com>
32569 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32570 Move constant to the right of a relational operator.
32571 (get_mpfr_format_length, format_character, format_string): Ditto.
32572 (should_warn_p, maybe_warn): Same.
32574 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32576 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32579 * asan.c (get_translation_unit_decl): Remove function.
32580 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32582 2017-01-30 Martin Liska <mliska@suse.cz>
32584 PR gcov-profile/79259
32585 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32586 -fprofile-generate.
32588 2017-01-30 Martin Liska <mliska@suse.cz>
32591 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32592 Initialize variables with NULL value.
32594 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32597 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32599 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32601 2017-01-30 Richard Biener <rguenther@suse.de>
32603 PR tree-optimization/79276
32604 * tree-vrp.c (process_assert_insertions): Properly adjust common
32605 when removing a duplicate.
32607 2017-01-30 Richard Biener <rguenther@suse.de>
32609 PR tree-optimization/79256
32610 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32611 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32613 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32615 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32618 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32619 ("*r<noxa>sbg_<mode>_sll_bitmask")
32620 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32621 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32622 Use contiguous_bitmask_nowrap_operand.
32624 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32627 * config/rs6000/altivec.h (vec_xl): Revise #define.
32628 (vec_xst): Likewise.
32630 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32632 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32634 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32636 PR rtl-optimization/79194
32637 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32638 traps before call to bypass_conditional_jumps.
32640 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32642 PR tree-optimization/71374
32643 * lra-constraints.c (check_conflict_input_operands): New.
32644 (match_reload): Use it.
32646 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32649 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32650 account to calculate conflict_set.
32652 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32654 PR rtl-optimization/78559
32655 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32656 other_insn in combine.
32658 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32660 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32661 uint16_type_node for BT_UINT16.
32663 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32665 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32666 "RTL Tests" to menu.
32667 (GIMPLE Tests): New node.
32668 (RTL Tests): New node.
32670 2017-01-27 Richard Biener <rguenther@suse.de>
32672 PR tree-optimization/79245
32673 * tree-loop-distribution.c (distribute_loop): Apply cost
32674 modeling also to detected patterns.
32676 2017-01-27 Richard Biener <rguenther@suse.de>
32678 PR tree-optimization/71433
32679 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32680 (compare_assert_loc): New function.
32681 (process_assert_insertions): Sort and optimize assert locations
32682 to remove duplicates and push down identical assertions on
32683 edges to their destination block.
32685 2017-01-27 Richard Biener <rguenther@suse.de>
32687 PR tree-optimization/79244
32688 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32689 out SSA names even if abnormal.
32691 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32693 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32694 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32695 instead of MPFR_RNDN.
32697 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32700 * arm.c (arm_option_override): Don't call build_target_option_node
32701 until after doing all option overrides.
32702 (arm_valid_target_attribute_tree): Likewise.
32704 2017-01-27 Martin Liska <mliska@suse.cz>
32706 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32707 for {cd}tors and C++ {cd}tors.
32709 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32711 * config/s390/s390.md ("*setmem_long_and")
32712 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32714 2017-01-26 Martin Sebor <msebor@redhat.com>
32716 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32719 2017-01-26 Martin Sebor <msebor@redhat.com>
32721 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32722 HAVE_DFmode before using XFmode or DFmode.
32723 (parse_directive): Avoid using the z length modifier to avoid
32724 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32726 PR middle-end/78703
32727 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32728 to accept adjustment as an array.
32729 (get_int_range): New function.
32730 (struct directive): Make width and prec arrays.
32731 (directive::set_width, directive::set_precision): Call get_int_range.
32732 (format_integer, format_floating): Handle width and precision ranges.
32733 (format_string, parse_directive): Same.
32735 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32738 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32739 comdat_type_p set, just clone them, but keep the children in the
32743 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32744 which have direct callers with -fvar-tracking-assignments enabled
32746 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32747 inside of type units.
32749 2017-01-26 Martin Sebor <msebor@redhat.com>
32751 PR middle-end/78703
32752 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32754 (struct format_result): Replace number_chars, number_chars_min,
32755 and number_chars_max with a single member of struct result_range.
32757 (format_result::operator+=): Adjust.
32758 (struct fmtresult): Remove bounded. Handle likely and unlikely
32760 (fmtresult::adjust_for_width_or_precision): New function.
32761 (fmtresult:type_max_digits): New function.
32762 (bytes_remaining): Handle likely and unlikely counters.
32763 (min_bytes_remaining): Remove.
32764 (format_percent): Simplify.
32765 (format_integer, format_floating): Set likely and unlikely counters.
32766 (get_string_length, format_character, format_string): Same.
32767 (format_plain, should_warn_p): New function.
32768 (maybe_warn): Call should_warn_p. Update diagnostic messages
32769 and handle those for all directives, including plain strings.
32770 (format_directive): Handle likely and unlikely counters.
32771 Remove unnecessary quoting from diagnostics. Add an informational
32773 (add_bytes): Remove.
32774 (pass_sprintf_length::compute_format_length): Simplify.
32775 (try_substitute_return_value): Handle likely and unlikely counters.
32777 2017-01-26 Carl Love <cel@us.ibm.com>
32779 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32780 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32782 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32785 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32786 endianess for subregs into account.
32787 * lra-constraints.c (lra_constraints): Do risky transformations
32788 always on the first iteration.
32789 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32790 last_call_used_reg_set.
32791 (process_bb_lives): Define and use last_call_used_reg_set.
32792 * lra.c (lra): Always continue after lra_constraints on the first
32795 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32797 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32799 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32801 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32803 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32804 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32805 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32806 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32807 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32808 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32809 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32810 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32811 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32813 2017-01-26 Marek Polacek <polacek@redhat.com>
32816 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32817 for the third operand.
32819 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32821 PR middle-end/79236
32822 * omp-low.c (struct omp_context): Add simt_stmt field.
32823 (scan_omp_for): Return omp_context *.
32824 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32825 context to the _simt_ SIMD stmt.
32826 (lower_omp_for): For combined SIMD with sibling _simt_
32827 SIMD, make sure to use the same decls in _looptemp_
32828 clauses as in the sibling.
32830 2017-01-26 David Sherwood <david.sherwood@arm.com>
32832 PR middle-end/79212
32833 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32836 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32839 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32840 emit fld b; fld a; if possible.
32842 * brig-builtins.def: Update copyright years.
32843 * config/arm/arm_acle_builtins.def: Update copyright years.
32845 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32848 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32849 constraint instead of o for the stxsd instruction.
32851 2017-01-25 Carl Love <cel@us.ibm.com>
32853 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32854 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32856 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32858 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32860 2017-01-25 Richard Biener <rguenther@suse.de>
32862 PR tree-optimization/69264
32863 * target.def (vector_alignment_reachable): Improve documentation.
32864 * doc/tm.texi: Regenerate.
32865 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32867 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32868 earlier changes with respect to TYPE_USER_ALIGN.
32869 (vector_alignment_reachable_p): Likewise. Improve dumping.
32871 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32874 * config/arm/arm.md (xordi3): Force constant operand into a register
32877 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32879 * doc/invoke.texi (-fstore-merging): Correct default optimization
32880 levels at which it is enabled.
32881 (-O): Move -fstore-merging from list to...
32884 2017-01-25 Richard Biener <rguenther@suse.de>
32887 * omp-expand.c: Include debug.h.
32888 (expand_omp_taskreg): Make sure to generate early debug before
32889 outlining anything from a function.
32890 (expand_omp_target): Likewise.
32891 (grid_expand_target_grid_body): Likewise.
32893 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32896 * asan.c (get_translation_unit_decl): New function.
32897 (asan_add_global): Extract modules file name from globals
32898 TRANSLATION_UNIT_DECL name.
32900 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32903 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32904 for long calls with APCS frame and VFP.
32906 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32908 * cfg.c (original_copy_tables_initialized_p): New function.
32909 * cfg.h (original_copy_tables_initialized_p): New decl.
32910 * cfgrtl.c (relink_block_chain): Guard the call to
32911 free_original_copy_tables with a call to
32912 original_copy_tables_initialized_p.
32913 * cgraph.h (symtab_node::native_rtl_p): New decl.
32914 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32915 (symtab_node::needed_p): Don't assert for early assembly output
32916 for __RTL functions.
32917 (cgraph_node::finalize_function): Set "force_output" for __RTL
32919 (cgraph_node::analyze): Bail out early for __RTL functions.
32920 (analyze_functions): Update assertion to support __RTL functions.
32921 (cgraph_node::expand): Bail out early for __RTL functions.
32922 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32924 * function.h (struct function): Update comment for field
32926 * gimple-expr.c: Include "tree-pass.h".
32927 (gimple_has_body_p): Return false for __RTL functions.
32928 * Makefile.in (OBJS): Add run-rtl-passes.o.
32929 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32931 (gcc::pass_manager::get_clean_slate): New accessor.
32932 * passes.c: Include "insn-addr.h".
32933 (should_skip_pass_p): Add logging. Update logic for running
32934 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32935 property-provider override so it is only done for gimple passes.
32937 (skip_pass): New function.
32938 (execute_one_pass): Call skip_pass when skipping passes.
32939 * read-md.c (md_reader::read_char): Support filtering
32940 the input to a subset of line numbers.
32941 (md_reader::md_reader): Initialize fields
32942 m_first_line and m_last_line.
32943 (md_reader::read_file_fragment): New function.
32944 * read-md.h (md_reader::read_file_fragment): New decl.
32945 (md_reader::m_first_line): New field.
32946 (md_reader::m_last_line): New field.
32947 * read-rtl-function.c (function_reader::create_function): Only
32948 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32949 curr_properties. Set DECL_INITIAL to a dummy block.
32950 (read_rtl_function_body_from_file_range): New function.
32951 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32953 * run-rtl-passes.c: New file.
32954 * run-rtl-passes.h: New file.
32956 2017-01-24 Jeff Law <law@redhat.com>
32958 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32961 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32963 PR tree-optimization/79159
32964 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32965 (record_nonwrapping_iv): Improve boundary using above function if no
32966 value range information.
32968 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32969 Martin Jambor <mjambor@suse.cz>
32971 * brig-builtins.def: New file.
32972 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32973 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32974 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32975 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32976 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32977 * builtin-types.def (BT_INT8): New.
32978 (BT_INT16): Likewise.
32979 (BT_UINT8): Likewise.
32980 (BT_UINT16): Likewise.
32981 (BT_FN_ULONG): Likewise.
32982 (BT_FN_UINT_INT): Likewise.
32983 (BT_FN_UINT_ULONG): Likewise.
32984 (BT_FN_UINT_LONG): Likewise.
32985 (BT_FN_UINT_PTR): Likewise.
32986 (BT_FN_ULONG_PTR): Likewise.
32987 (BT_FN_INT8_FLOAT): Likewise.
32988 (BT_FN_INT16_FLOAT): Likewise.
32989 (BT_FN_UINT32_FLOAT): Likewise.
32990 (BT_FN_UINT16_FLOAT): Likewise.
32991 (BT_FN_UINT8_FLOAT): Likewise.
32992 (BT_FN_UINT64_FLOAT): Likewise.
32993 (BT_FN_UINT16_UINT32): Likewise.
32994 (BT_FN_UINT32_UINT16): Likewise.
32995 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32996 (BT_FN_INT_PTR_INT): Likewise.
32997 (BT_FN_UINT_PTR_UINT): Likewise.
32998 (BT_FN_LONG_PTR_LONG): Likewise.
32999 (BT_FN_ULONG_PTR_ULONG): Likewise.
33000 (BT_FN_VOID_UINT64_UINT64): Likewise.
33001 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33002 (BT_FN_INT8_INT8_INT8): Likewise.
33003 (BT_FN_INT16_INT16_INT16): Likewise.
33004 (BT_FN_INT_INT_INT): Likewise.
33005 (BT_FN_UINT_FLOAT_UINT): Likewise.
33006 (BT_FN_FLOAT_UINT_UINT): Likewise.
33007 (BT_FN_ULONG_UINT_UINT): Likewise.
33008 (BT_FN_ULONG_UINT_PTR): Likewise.
33009 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33010 (BT_FN_UINT_UINT_UINT): Likewise.
33011 (BT_FN_VOID_UINT_PTR): Likewise.
33012 (BT_FN_UINT_UINT_PTR: Likewise.
33013 (BT_FN_UINT32_UINT64_PTR): Likewise.
33014 (BT_FN_INT_INT_UINT_UINT): Likewise.
33015 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33016 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33017 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33018 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33019 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33020 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33021 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33022 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33023 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33024 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33025 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33026 * doc/frontends.texi: List BRIG FE.
33027 * doc/install.texi (Testing): Add BRIG tesring requirements.
33028 * doc/invoke.texi (Overall Options): Mention BRIG.
33029 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33031 2017-01-24 Richard Biener <rguenther@suse.de>
33033 PR translation/79208
33034 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33036 2017-01-24 Martin Jambor <mjambor@suse.cz>
33039 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33040 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33041 and known_contexts.
33043 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33045 PR middle-end/79123
33046 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33047 casts from signed to unsigned really don't have a range.
33049 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33051 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33052 GMP_RNDx for compatiblity.
33054 2017-01-24 Martin Liska <mliska@suse.cz>
33057 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33058 that would prevent us to call alloca with -1 as argument.
33060 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33062 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33063 Avoid -Wformat-security warning.
33065 2017-01-23 Andrew Pinski <apinski@cavium.com>
33067 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33070 2017-01-23 Martin Sebor <msebor@redhat.com>
33072 PR middle-end/78703
33073 * gimple-ssa-sprintf.c (warn_level): New global.
33074 (format_integer): Use it here and throughout the rest of the file.
33075 Use the same switch to compute sign as base.
33076 (maybe_warn): New function.
33077 (format_directive): Factor out warnings into maybe_warn.
33078 Add debugging output. Use warn_level.
33079 (add_bytes): Use warn_level.
33080 (pass_sprintf_length::compute_format_length): Add debugging output.
33081 (try_substitute_return_value): Same.
33082 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33084 PR middle-end/78703
33085 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33086 (struct fmtresult, format_integer, format_floating): Adjust.
33087 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33088 (get_string_length, format_string,format_directive): Same.
33089 (pass_sprintf_length::compute_format_length): Same.
33090 (try_substitute_return_value): Simplify slightly.
33092 PR middle-end/78703
33093 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33094 (fmtresult::operator+=): Outlined.
33095 (struct fmtresult): Add ctors.
33096 (struct conversion_spec): Rename...
33097 (struct directive): ...to this. Add and remove data members.
33098 (directive::set_width, directive::set_precision): New functions.
33099 (format_percent): Use fmtresult ctor.
33100 (get_width_and_precision): Remove.
33101 (format_integer): Make naming changes. Avoid computing width and
33103 (format_floating): Same. Adjust indentation.
33104 (format_character, format_none): New functions.
33105 (format_string): Moved character handling to format_character.
33106 (format_directive): Remove arguments, change return type.
33107 (parse_directive): New function.
33108 (pass_sprintf_length::compute_format_length): Move directive
33109 parsing to parse_directive.
33111 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33113 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33114 (assign_assembler_name_if_needed): ... this.
33115 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33116 (assign_assembler_name_if_needed): ... this.
33117 (free_lang_data_in_cgraph): Adjust callers.
33118 * cgraphunit.c (cgraph_node::analyze): Likewise.
33119 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33122 2017-01-23 Richard Biener <rguenther@suse.de>
33124 PR tree-optimization/79088
33125 PR tree-optimization/79188
33126 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33127 resetting loop bounds after last path deletion. Reset loop
33128 bounds of the target loop, make code match the comments.
33129 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33130 Make sure loops need no fixups.
33132 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33134 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33135 exponent support with double type for first argument.
33136 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33137 type returned by __builtin_vec_extract_sig,
33138 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33139 functions from "vector int" to "vector unsigned int" or from
33140 "vector long long int" to "vector unsigned long long int".
33141 Changed type returned by __builtin_vec_extract_exp,
33142 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33143 functions from "vector int" to "vector unsigned int" or from
33144 "vector long long int" to "vector unsigned long long int".
33145 Changed return type of __builtin_vec_test_data_class,
33146 __builtin_vec_test_data_class_sp, and
33147 __builtin_vec_test_data_class_dp from "vector int" to
33148 "vector bool int" or from "vector long long int" to "vector bool
33149 long long int" and changed second argument type from "unsigned
33150 int" to "int". Added new overloaded function forms "vector float
33151 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33152 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33153 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33154 double, vector unsigned long long int)" and "vector double
33155 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33156 long int)". Changed return type of
33157 __builtin_scalar_test_data_class and
33158 __builtin_scalar_test_data_class_sp and
33159 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33160 int" and changed second argument from "unsigned int" to "int".
33161 Changed type returned by __builtin_scalar_test_neg,
33162 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33163 from "int" to "bool int". Added new overloaded function form
33164 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33165 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33166 exponent double-precision with floating point first argument.
33167 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33168 documentation of scalar_test_data_class, scalar_test_neg,
33169 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33170 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33171 vec_test_data_class built-in functions to reflect refinements in
33172 their type signatures.
33174 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33176 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33178 (aarch64_elf_asm_destructor): Likewise.
33180 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33182 PR rtl-optimization/78634
33183 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33184 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33185 * ifcvt.c (noce_try_cmove): Add missing cost check.
33187 PR rtl-optimization/71724
33188 * combine.c (if_then_else_cond): Look for situations where it is
33189 beneficial to undo the work of one of the recursive calls.
33191 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33193 PR tree-optimization/70754
33194 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33195 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33196 combined stmt before it if not NULL.
33197 (combine_chains): Process refs reversely and compute dominance point
33200 2017-01-23 Martin Liska <mliska@suse.cz>
33202 PR tree-optimization/79196
33203 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33204 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33206 (strlen_optimize_stmt): Call the renamed function.
33208 2017-01-23 Michael Matz <matz@suse.de>
33210 PR tree-optimization/78384
33211 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33213 2017-01-23 Richard Biener <rguenther@suse.de>
33215 PR tree-optimization/79186
33216 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33217 both incoming edges before moving an assert.
33219 2017-01-23 Martin Jambor <mjambor@suse.cz>
33221 * ipa-prop.c (load_from_param_1): Removed.
33222 (load_from_unmodified_param): Bits from load_from_param_1 put back
33224 (load_from_param): Removed.
33225 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33226 with stmt. Reverted back to use of load_from_unmodified_param.
33228 2017-01-23 Martin Jambor <mjambor@suse.cz>
33231 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33232 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33233 field a pointer to garbage collected vector, mark lattices and
33234 ipcp_orig_node with GTY((skip)).
33235 (ipa_get_param_count): Adjust to descriptors being a pointer.
33236 (ipa_get_param): Likewise.
33237 (ipa_get_type): Likewise.
33238 (ipa_get_param_move_cost): Likewise.
33239 (ipa_set_param_used): Likewise.
33240 (ipa_get_controlled_uses): Likewise.
33241 (ipa_set_controlled_uses): Likewise.
33242 (ipa_is_param_used): Likewise.
33243 (ipa_node_params_t): Move into garbage collector. New methods insert
33245 (ipa_node_params_sum): Annotate wth GTY(()).
33246 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33248 (ipa_load_from_parm_agg): Adjust declaration.
33249 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33250 * ipa-profile.c (ipa_profile): Likewise.
33251 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33252 (ipa_populate_param_decls): Make descriptors parameter garbage
33254 (ipa_dump_param): Adjust to descriptors being a pointer.
33255 (ipa_alloc_node_params): Likewise.
33256 (ipa_initialize_node_params): Likewise.
33257 (load_from_param_1): Make descriptors parameter garbage collected.
33258 (load_from_unmodified_param): Likewise.
33259 (load_from_param): Likewise.
33260 (ipa_load_from_parm_agg): Likewise.
33261 (ipa_node_params::~ipa_node_params): Removed.
33262 (ipa_free_all_node_params): Remove call to delete operator.
33263 (ipa_node_params_t::insert): New.
33264 (ipa_node_params_t::remove): Likewise.
33265 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33266 copy known_csts and known_contexts vectors.
33267 (ipa_read_node_info): Adjust to descriptors being a pointer.
33268 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33270 (ipcp_transform_function): Make descriptors variable garbage
33273 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33275 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33276 * config/i386/avx512dqintrin.h: Ditto.
33277 * config/i386/avx512fintrin.h: Ditto.
33278 * config/i386/i386.c: Handle new builtins.
33279 * config/i386/i386-builtin.def: Add new builtins.
33280 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33281 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33283 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33284 Martin Liska <mliska@suse.cz>
33286 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33287 * asan.c (asan_expand_poison_ifn): Support stores and use
33288 appropriate ASAN report function.
33289 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33290 * internal-fn.def (ASAN_POISON_USE): Declare.
33291 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33292 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33293 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33294 ASAN_POISON calls w/o LHS.
33295 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33296 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33297 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33298 * gimplify.c (asan_poison_variables): Add attribute
33299 use_after_scope_memory to variables that really needs to live
33301 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33302 having the attribute.
33304 2017-01-23 Martin Liska <mliska@suse.cz>
33306 * asan.c (create_asan_shadow_var): New function.
33307 (asan_expand_poison_ifn): Likewise.
33308 * asan.h (asan_expand_poison_ifn): New declaration.
33309 * internal-fn.c (expand_ASAN_POISON): Likewise.
33310 * internal-fn.def (ASAN_POISON): New builtin.
33311 * sanopt.c (pass_sanopt::execute): Expand
33312 asan_expand_poison_ifn.
33313 * tree-inline.c (copy_decl_for_dup_finish): Make function
33315 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33316 * tree-ssa.c (is_asan_mark_p): New function.
33317 (execute_update_addresses_taken): Rewrite local variables
33318 (identified just by use-after-scope as addressable) into SSA.
33320 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33322 * doc/install.texi (Specific): opensource.apple.com uses https
33323 now. Remove trailing slash.
33325 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33327 * README.Portability: Remove note on an Irix compatibility issue.
33329 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33331 * gcov.c (INCLUDE_ALGORITHM): Define.
33332 (INCLUDE_VECTOR): Define.
33333 No longer include <vector> and <algorithm> directly.
33335 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33337 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33339 * doc/invoke.texi (Code Gen Options): Ditto.
33341 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33344 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33346 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33348 rtl-optimization/79125
33349 * cprop.c (local_cprop_pass): Handle cases where we make an
33350 unconditional trap.
33352 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33356 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33357 read from, for big endian.
33359 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33361 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33362 register pauth builtins for LP64 only.
33364 2017-01-20 Marek Polacek <polacek@redhat.com>
33367 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33370 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33372 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33374 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33375 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33376 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33378 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33381 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33382 in big-endian mode when they are not a single duplicated value.
33384 2017-01-20 Richard Biener <rguenther@suse.de>
33386 * BASE-VER: Bump to 7.0.1.
33388 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33390 * omp-low.c (omplow_simd_context): New struct. Use it...
33391 (lower_rec_simd_input_clauses): ...here and...
33392 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33393 references to idx, lane, max_vf, is_simt.
33395 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33397 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33400 2017-01-20 Martin Jambor <mjambor@suse.cz>
33402 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33403 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33405 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33406 (GTFILES): Rename hsa.c to hsa-common.c.
33407 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33408 * hsa-dump.c: Likewise.
33409 * hsa-gen.c: Likewise.
33410 * hsa-regalloc.c: Likewise.
33411 * ipa-hsa.c: Likewise.
33412 * omp-expand.c: Likewise.
33413 * omp-low.c: Likewise.
33414 * toplev.c: Likewise.
33416 2017-01-20 Marek Polacek <polacek@redhat.com>
33419 * doc/invoke.texi: Document -Wduplicated-branches.
33420 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33421 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33422 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33423 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33424 return 0 only when not OEP_LEXICOGRAPHIC.
33425 (fold_build_cleanup_point_expr): Use the expression
33426 location when building CLEANUP_POINT_EXPR.
33427 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33428 * tree.c (add_expr): Handle error_mark_node.
33430 2017-01-20 Martin Liska <mliska@suse.cz>
33433 * tree-profile.c (init_ic_make_global_vars): Do not call
33435 (gimple_init_gcov_profiler): Likewise.
33437 2017-01-20 Martin Liska <mliska@suse.cz>
33440 * cgraph.h (maybe_create_reference): Remove argument and
33442 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33444 * ipa-cp.c (create_specialized_node): Likewise.
33445 * symtab.c (symtab_node::maybe_create_reference): Handle
33446 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33448 2017-01-20 Martin Liska <mliska@suse.cz>
33450 * read-rtl-function.c (function_reader::create_function): Use
33451 build_decl instread of build_decl_stat.
33453 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33455 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33456 * config/i386/avx512dqintrin.h: Ditto.
33457 * config/i386/avx512fintrin.h: Ditto.
33458 * config/i386/i386-builtin-types.def: Add new types.
33459 * config/i386/i386.c: Handle new types.
33460 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33461 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33462 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33463 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33464 (__builtin_ia32_kshiftridi): New.
33465 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33467 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33471 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33472 define to rs6000_init_stack_protect_guard.
33473 (rs6000_init_stack_protect_guard): New function.
33475 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33476 Yunqiang Su <yunqiang.su@imgtec.com>
33478 * config.gcc (supported_defaults): Add madd4.
33479 (with_madd4): Add validation.
33480 (all_defaults): Add madd4.
33481 * config/mips/mips.opt (mmadd4): New option.
33482 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33484 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33486 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33487 (ISA_HAS_FUSED_MADD4): Likewise.
33488 * doc/invoke.texi (-mmadd4): Document the new option.
33489 * doc/install.texi (--with-madd4): Document the new option.
33491 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33493 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33494 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33495 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33496 (aarch64_init_pauth_hint_builtins): New.
33497 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33498 (aarch64_expand_builtin): Expand new builtins.
33500 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33502 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33503 * combine-stack-adj.c (no_unhandled_cfa): Handle
33504 REG_CFA_TOGGLE_RA_MANGLE.
33505 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33506 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33507 info for return address signing.
33508 (aarch64_expand_epilogue): Likewise.
33510 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33512 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33513 * config/aarch64/aarch64-protos.h
33514 (aarch64_return_address_signing_enabled): New declaration.
33515 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33517 (aarch64_expand_prologue): Sign return address before it's pushed onto
33519 (aarch64_expand_epilogue): Authenticate return address fetched from
33521 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33522 (aarch64_attributes): New function attributes for "sign-return-address".
33523 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33524 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33525 ("*do_return"): Generate combined instructions according to key index.
33526 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33527 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33529 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33530 * config/aarch64/aarch64.opt (msign-return-address=): New.
33531 * doc/extend.texi (AArch64 Function Attributes): Documents
33532 "sign-return-address=".
33533 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33535 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33537 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33538 overall option summary.
33540 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33542 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33543 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33544 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33545 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33547 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33549 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33550 -mpower9-minmax by default for -mcpu=power9.
33551 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33552 128-bit floating point.
33554 2017-01-20 Alan Modra <amodra@gmail.com>
33556 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33557 optimizing for size.
33559 2017-01-20 Alan Modra <amodra@gmail.com>
33562 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33563 for strcmp and strncmp from corresponding builtin decl.
33565 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33567 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33568 instead of i386/rtems-64.h.
33569 * config/i386/rtems-64.h: Remove.
33571 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33575 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33577 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33579 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33581 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33582 Change int to HOST_WIDE_INT.
33583 * config/aarch64/aarch64-protos.h
33584 (aarch64_simd_gen_const_vector_dup): Likewise.
33585 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33587 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33589 * langhooks-def.h (lhd_type_for_size): New decl.
33590 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33591 * langhooks.c (lhd_type_for_size): New function, taken from
33594 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33596 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33597 define_bypass for CR latency.
33598 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33599 (power9-alu2): Add define_bypass for CR latency.
33601 (power9-mul): Update insn latency.
33602 (power9-mul-compare): Update insn latency, bypass latency and remove
33605 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33607 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33609 * config/aarch64/aarch64.md
33610 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33611 aarch64_nopcrelative_literal_loads.
33612 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33614 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33616 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33617 TARGET_LOONGSON_3A.
33618 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33620 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33623 * config.gcc (supported_defaults): Add lxc1-sxc1.
33624 (with_lxc1_sxc1): Add validation.
33625 (all_defaults): Add lxc1-sxc1.
33626 * config/mips/mips.opt (mlxc1-sxc1): New option.
33627 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33629 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33630 __mips_no_lxc1_sxc1.
33631 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33632 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33633 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33635 2017-01-19 Richard Biener <rguenther@suse.de>
33637 PR tree-optimization/72488
33638 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33639 sure to restore SSA info.
33640 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33642 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33644 PR rtl-optimization/79121
33645 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33646 of the inner type when shifting an extended value.
33648 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33651 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33652 interposable aliases.
33654 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33657 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33658 Use the evmergelohi instruction.
33659 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33660 (mov_sitf_e500_subreg8_2_be): Likewise.
33661 (mov_sitf_e500_subreg12_2_le): Likewise.
33662 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33663 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33664 (mov_sitf_e500_subreg8_2_le): Likewise.
33665 (mov_sitf_e500_subreg12_2_be): Likewise.
33667 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33669 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33670 attribute from vecsimple to vecperm.
33671 (altivec_vbpermq2): Likewise.
33673 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33676 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33678 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33679 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33680 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33681 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33682 case where N arg is SIZE_MAX.
33683 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33684 (cmpstrsi): Add pattern.
33686 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33689 __builtin_vec_revb builtins.
33690 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33691 built-in functions to support generation of the ISA 3.0 XXBR<x>
33692 vector byte reverse instructions.
33693 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33694 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33695 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33696 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33697 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33698 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33699 (P9V_BUILTIN_VEC_REVB): Likewise.
33700 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33701 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33702 (p9_xxbrq_v16qi): Likewise.
33703 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33704 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33705 (p9_xxbrh_v8hi): Likewise.
33706 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33707 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33708 vec_revb built-in functions.
33710 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33712 PR rtl-optimization/78952
33713 * config/i386/i386.md (any_extract): New code iterator.
33714 (*insvqi_2): Use any_extract for source operand.
33715 (*insvqi_3): Use any_shiftrt for source operand.
33717 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33719 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33721 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33723 2017-01-18 Matthias Klose <doko@ubuntu.com>
33725 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33727 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33729 * config/rs6000/altivec.h (vec_bperm): Change #define.
33730 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33731 (altivec_vbpermq2): New define_insn.
33732 (altivec_vbpermd): Likewise.
33733 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33734 function interface.
33735 (VBPERMD): Likewise.
33736 (VBPERM): New polymorphic function interface.
33737 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33738 Add entries for P9V_BUILTIN_VEC_VBPERM.
33739 * doc/extend.texi: Add interfaces for vec_bperm.
33741 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33743 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33744 first letter of error messages.
33745 (s390_resolve_overloaded_builtin): Likewise.
33746 * config/s390/s390.c (s390_expand_builtin): Likewise.
33747 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33748 (s390_valid_target_attribute_inner_p): Likewise.
33749 * config/s390/s390.md ("tabort"): Likewise.
33751 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33753 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33754 (ISA_AVOID_DIV_HILO): New macro.
33755 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33756 (ISA_HAS_DDIV): Likewise.
33758 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33760 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33762 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33764 * doc/invoke.texi (fabi-version): Spelling fix.
33766 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33769 * doc/invoke.texi (fabi-version): Mention mangling fix for
33772 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33775 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33777 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33780 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33781 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33783 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33785 (stack_protect_test): Ditto.
33786 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33787 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33789 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33790 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33791 -mstack-protector-guard-offset=.
33792 (RS/6000 and PowerPC Options): Ditto.
33794 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33796 * config/i386/i386.h (MASK_CLASS_P): New define.
33797 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33798 there are no registers from different register sets also when
33799 mask registers are used. Update function comment.
33800 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33801 to (*k/*r) and (*k/*km) alternatives.
33803 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33805 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33806 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33807 (EH_RETURN_HANDLER_RTX): New define.
33808 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33809 Force frame pointer in EH return functions.
33810 (aarch64_expand_epilogue): Add barrier for eh_return.
33811 (aarch64_final_eh_return_addr): Remove.
33812 (aarch64_eh_return_handler_rtx): New function.
33813 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33815 (aarch64_eh_return_handler_rtx): New prototype.
33817 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33819 * config/rs6000/altivec.h (vec_rlmi): New #define.
33820 (vec_vrlnm): Likewise.
33821 (vec_rlnm): Likewise.
33822 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33823 (UNSPEC_VRLNM): Likewise.
33824 (VIlong): New mode iterator.
33825 (altivec_vrl<VI_char>mi): New define_insn.
33826 (altivec_vrl<VI_char>nm): Likewise.
33827 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33829 (VRLDNM): Likewise.
33830 (RLNM): New polymorphic function entry.
33831 (VRLWMI): New monomorphic function entry.
33832 (VRLDMI): Likewise.
33833 (RLMI): New polymorphic function entry.
33834 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33835 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33836 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33839 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33842 * dwarf2out.c (field_byte_offset): Restore the
33843 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33844 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33845 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33848 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33851 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33853 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33856 * dwarf2out.c (add_data_member_location_attribute): For constant
33857 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33858 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33859 DW_AT_byte_size attributes.
33861 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33863 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33864 after forcing to constant memory when the code model is medium.
33866 2017-01-17 Julia Koval <julia.koval@intel.com>
33869 * config/i386/avx512fintrin.h
33870 (_mm512_i32gather_ps): Change __addr type to void const*.
33871 (_mm512_mask_i32gather_ps): Ditto.
33872 (_mm512_i32gather_pd): Ditto.
33873 (_mm512_mask_i32gather_pd): Ditto.
33874 (_mm512_i64gather_ps): Ditto.
33875 (_mm512_mask_i64gather_ps): Ditto.
33876 (_mm512_i64gather_pd): Ditto.
33877 (_mm512_mask_i64gather_pd): Ditto.
33878 (_mm512_i32gather_epi32): Ditto.
33879 (_mm512_mask_i32gather_epi32): Ditto.
33880 (_mm512_i32gather_epi64): Ditto.
33881 (_mm512_mask_i32gather_epi64): Ditto.
33882 (_mm512_i64gather_epi32): Ditto.
33883 (_mm512_mask_i64gather_epi32): Ditto.
33884 (_mm512_i64gather_epi64): Ditto.
33885 (_mm512_mask_i64gather_epi64): Ditto.
33886 (_mm512_i32scatter_ps): Change __addr type to void*.
33887 (_mm512_mask_i32scatter_ps): Ditto.
33888 (_mm512_i32scatter_pd): Ditto.
33889 (_mm512_mask_i32scatter_pd): Ditto.
33890 (_mm512_i64scatter_ps): Ditto.
33891 (_mm512_mask_i64scatter_ps): Ditto.
33892 (_mm512_i64scatter_pd): Ditto.
33893 (_mm512_mask_i64scatter_pd): Ditto.
33894 (_mm512_i32scatter_epi32): Ditto.
33895 (_mm512_mask_i32scatter_epi32): Ditto.
33896 (_mm512_i32scatter_epi64): Ditto.
33897 (_mm512_mask_i32scatter_epi64): Ditto.
33898 (_mm512_i64scatter_epi32): Ditto.
33899 (_mm512_mask_i64scatter_epi32): Ditto.
33900 (_mm512_i64scatter_epi64): Ditto.
33901 (_mm512_mask_i64scatter_epi64): Ditto.
33902 * config/i386/avx512pfintrin.h
33903 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33904 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33905 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33906 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33907 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33908 (_mm512_prefetch_i32scatter_ps): Ditto.
33909 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33910 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33911 (_mm512_prefetch_i64scatter_pd): Ditto.
33912 (_mm512_prefetch_i64scatter_ps): Ditto.
33913 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33914 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33915 * config/i386/avx512vlintrin.h
33916 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33917 (_mm_mmask_i32gather_ps): Ditto.
33918 (_mm256_mmask_i32gather_pd): Ditto.
33919 (_mm_mmask_i32gather_pd): Ditto.
33920 (_mm256_mmask_i64gather_ps): Ditto.
33921 (_mm_mmask_i64gather_ps): Ditto.
33922 (_mm256_mmask_i64gather_pd): Ditto.
33923 (_mm_mmask_i64gather_pd): Ditto.
33924 (_mm256_mmask_i32gather_epi32): Ditto.
33925 (_mm_mmask_i32gather_epi32): Ditto.
33926 (_mm256_mmask_i32gather_epi64): Ditto.
33927 (_mm_mmask_i32gather_epi64): Ditto.
33928 (_mm256_mmask_i64gather_epi32): Ditto.
33929 (_mm_mmask_i64gather_epi32): Ditto.
33930 (_mm256_mmask_i64gather_epi64): Ditto.
33931 (_mm_mmask_i64gather_epi64): Ditto.
33932 (_mm256_i32scatter_ps): Change __addr type to void*.
33933 (_mm256_mask_i32scatter_ps): Ditto.
33934 (_mm_i32scatter_ps): Ditto.
33935 (_mm_mask_i32scatter_ps): Ditto.
33936 (_mm256_i32scatter_pd): Ditto.
33937 (_mm256_mask_i32scatter_pd): Ditto.
33938 (_mm_i32scatter_pd): Ditto.
33939 (_mm_mask_i32scatter_pd): Ditto.
33940 (_mm256_i64scatter_ps): Ditto.
33941 (_mm256_mask_i64scatter_ps): Ditto.
33942 (_mm_i64scatter_ps): Ditto.
33943 (_mm_mask_i64scatter_ps): Ditto.
33944 (_mm256_i64scatter_pd): Ditto.
33945 (_mm256_mask_i64scatter_pd): Ditto.
33946 (_mm_i64scatter_pd): Ditto.
33947 (_mm_mask_i64scatter_pd): Ditto.
33948 (_mm256_i32scatter_epi32): Ditto.
33949 (_mm256_mask_i32scatter_epi32): Ditto.
33950 (_mm_i32scatter_epi32): Ditto.
33951 (_mm_mask_i32scatter_epi32): Ditto.
33952 (_mm256_i32scatter_epi64): Ditto.
33953 (_mm256_mask_i32scatter_epi64): Ditto.
33954 (_mm_i32scatter_epi64): Ditto.
33955 (_mm_mask_i32scatter_epi64): Ditto.
33956 (_mm256_i64scatter_epi32): Ditto.
33957 (_mm256_mask_i64scatter_epi32): Ditto.
33958 (_mm_i64scatter_epi32): Ditto.
33959 (_mm_mask_i64scatter_epi32): Ditto.
33960 (_mm256_i64scatter_epi64): Ditto.
33961 (_mm256_mask_i64scatter_epi64): Ditto.
33962 (_mm_i64scatter_epi64): Ditto.
33963 (_mm_mask_i64scatter_epi64): Ditto.
33964 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33965 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33966 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33967 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33968 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33969 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33970 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33971 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33972 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33973 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33974 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33975 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33976 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33977 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33978 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33979 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33980 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33981 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33982 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33983 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33984 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33985 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33986 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33987 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33988 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33989 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33990 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33991 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33992 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33993 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33994 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33995 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33996 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33997 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33998 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33999 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34000 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34001 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34002 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34003 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34004 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34005 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34006 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34007 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34008 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34009 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34010 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34011 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34012 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34013 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34014 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34015 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34016 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34017 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34018 definitions accordingly.
34020 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34021 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34024 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34027 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34030 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34031 hard regs for inner regno.
34033 2017-01-17 Martin Liska <mliska@suse.cz>
34036 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34037 assumption and add comment.
34039 2017-01-17 Nathan Sidwell <nathan@acm.org>
34041 * ipa-visibility.c (localize_node): New function, broken out of ...
34042 (function_and_variable_visibility): ... here. Call it.
34044 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34046 PR middle-end/77445
34047 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34048 correctly set frequency of oudgoing edge.
34049 (duplicate_thread_path): Fix profile updating.
34051 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34054 * configure.ac: Add GCC_BASE_VER.
34055 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34056 version from BASE-VER file.
34057 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34058 (gcc.o): Depend on $(BASEVER).
34059 * common.opt (dumpfullversion): New option.
34060 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34061 * doc/invoke.texi: Document -dumpfullversion.
34062 * doc/install.texi: Document --with-gcc-major-version-only.
34063 * configure: Regenerated.
34065 2017-01-17 Richard Biener <rguenther@suse.de>
34067 PR tree-optimization/71433
34068 * tree-vrp.c (register_new_assert_for): Merge same asserts
34069 on all incoming edges.
34070 (process_assert_insertions_for): Handle insertions at the
34073 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34075 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34076 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34078 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34081 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34084 2017-01-17 Alan Modra <amodra@gmail.com>
34087 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34088 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34089 symbolic stack limit when pic.
34091 2017-01-16 Martin Sebor <msebor@redhat.com>
34093 PR tree-optimization/78608
34094 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34096 2017-01-16 Jeff Law <law@redhat.com>
34099 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34100 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34101 for several include directories that may be relative to sysroot.
34102 * config/i386/x-mingw32 (gplus_includedir): Define.
34103 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34104 (native_system_includedir): Likewise.
34105 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34106 override if TARGET_SYSTEM_ROOT is defined.
34107 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34109 PR tree-optimization/79090
34110 PR tree-optimization/33562
34111 PR tree-optimization/61912
34112 PR tree-optimization/77485
34113 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34114 and computed trims into the dump file.
34116 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34118 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34120 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34123 * gimplify.c (gimplify_init_constructor): If want_value and
34124 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34128 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34129 sequence. Formatting fixes.
34130 (doloop_optimize): Formatting fixes.
34133 * gcc.c (debug_level_greater_than_spec_func): New function.
34134 (static_spec_functions): Add debug-level-gt spec function.
34135 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34137 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34138 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34139 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34140 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34141 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34142 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34144 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34146 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34147 QImode fixups to general and mask registers only.
34149 2017-01-16 Carl Love <cel@us.ibm.com>
34151 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34152 for built-in functions
34153 vector signed char vec_nabs (vector signed char)
34154 vector signed short vec_nabs (vector signed short)
34155 vector signed int vec_nabs (vector signed int)
34156 vector signed long long vec_nabs (vector signed long long)
34157 vector float vec_nabs (vector float)
34158 vector double vec_nabs (vector double)
34159 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34161 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34162 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34163 * doc/extend.texi: Update the documentation file for the new built-in
34166 2017-01-16 Martin Sebor <msebor@redhat.com>
34168 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34171 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34173 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34174 UNSPEC_VSX__XXSPLTD to require special splat handling.
34176 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34179 * system.h: Poison strndup.
34181 2017-01-16 Alan Modra <amodra@gmail.com>
34184 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34187 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34189 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34191 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34193 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34194 call recog here. Assert that INSN_CODE (insn) is non-negative.
34196 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34199 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34201 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34202 in the currently scheduled RTL fragment.
34204 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34206 PR rtl-optimization/78751
34207 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34210 2017-01-14 Jeff Law <law@redhat.com>
34212 PR tree-optimization/79090
34213 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34214 variable length stores.
34215 (compute_trims): Delete dead assignment to *trim_tail.
34216 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34219 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34221 PR rtl-optimization/78626
34222 PR rtl-optimization/78727
34223 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34224 of a block, and split such blocks after everything else is finished.
34226 2017-01-14 Alan Modra <amodra@gmail.com>
34229 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34230 target legitimate_combined_insn.
34231 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34232 (rs6000_legitimate_combined_insn): New function.
34233 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34235 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34236 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34237 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34239 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34241 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34243 2017-01-13 Jeff Law <law@redhat.com>
34245 PR tree-optimization/33562
34246 PR tree-optimization/61912
34247 PR tree-optimization/77485
34248 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34250 (delete_dead_assignment): Likewise.
34251 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34252 statement to delete_dead_call and delete_dead_assignment.
34254 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34257 * substring-locations.c (format_warning_va): Strengthen case 1 so
34258 that both endpoints of the substring must be within the format
34259 range for just the substring to be printed.
34261 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34263 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34264 * config/i386/i386.c (ix86_target_string): Add missing options
34265 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34266 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34267 flags_other and ix86_target_other to flags2_other. Display unknown
34269 (ix86_valid_target_attribute_inner_p): Add missing options and
34270 reorder options by implied ISAs, as in ix86_target_string.
34272 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34274 * hash-table.h (hash_table::too_empty_p): New function.
34275 (hash_table::expand): Use it.
34276 (hash_table::traverse): Likewise.
34277 (hash_table::empty_slot): Use sizeof (value_type) instead of
34278 sizeof (PTR) to convert bytes to elements. Shrink the table
34279 if the current size is excessive for the current number of
34282 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34284 * ira-costs.c (record_reg_classes): Break from the inner loop
34285 early once alt_fail is known to be true. Update outer loop
34286 handling accordingly.
34288 2017-01-13 Jeff Law <law@redhat.com>
34290 * tree-ssa-dse.c (decrement_count): New function.
34291 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34292 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34293 when we know the partially dead statement is a mem* function.
34295 PR tree-optimization/61912
34296 PR tree-optimization/77485
34297 * tree-ssa-dse.c: Include expr.h.
34298 (maybe_trim_constructor_store): New function.
34299 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34301 PR tree-optimization/33562
34302 PR tree-optimization/61912
34303 PR tree-optimization/77485
34304 * doc/invoke.texi: Document new dse-max-object-size param.
34305 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34306 * tree-ssa-dse.c: Include params.h.
34307 (dse_store_status): New enum.
34308 (initialize_ao_ref_for_dse): New, partially extracted from
34310 (valid_ao_ref_for_dse, normalize_ref): New.
34311 (setup_live_bytes_from_ref, compute_trims): Likewise.
34312 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34313 (maybe_trim_partially_dead_store): Likewise.
34314 (maybe_trim_complex_store): Likewise.
34315 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34316 Track what bytes live from the original store. Return tri-state
34317 for dead, partially dead or live.
34318 (dse_dom_walker): Add constructor, destructor and new private members.
34319 (delete_dead_call, delete_dead_assignment): New extracted from
34321 (dse_optimize_stmt): Make a member of dse_dom_walker.
34322 Use initialize_ao_ref_for_dse.
34324 PR tree-optimization/33562
34325 PR tree-optimization/61912
34326 PR tree-optimization/77485
34327 * sbitmap.h (bitmap_count_bits): Prototype.
34328 (bitmap_clear_range, bitmap_set_range): Likewise.
34329 * sbitmap.c (bitmap_clear_range): New function.
34330 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34332 2017-01-13 Martin Liska <mliska@suse.cz>
34335 * function.c (set_cfun): Add new argument force.
34336 * function.h (set_cfun): Likewise.
34337 * ipa-inline-transform.c (inline_call): Use the function when
34338 strict alising from is dropped for function we inline to.
34340 2017-01-13 Richard Biener <rguenther@suse.de>
34342 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34343 for dumping GIMPLE INTEGER_CSTs.
34345 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34347 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34348 to 201112L since C++17.
34350 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34353 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34354 if -fsanitize=kernel-address is present.
34356 2017-01-13 Richard Biener <rguenther@suse.de>
34358 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34359 as _Literal ( type ) number in case usual suffixes do not
34360 preserve all information.
34362 2017-01-13 Richard Biener <rguenther@suse.de>
34364 PR tree-optimization/77283
34365 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34366 and ssa-iterators.h.
34367 (is_feasible_trace): Implement a cost model based on joiner
34370 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34373 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34374 char or short to __float128/_Float128 directly.
34376 2017-01-12 Martin Sebor <msebor@redhat.com>
34378 to -Wformat-overflow.
34379 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34380 (min_bytes_remaining): Same.
34381 (get_string_length): Same.
34382 (format_string): Same.
34383 (format_directive): Same.
34385 (pass_sprintf_length::handle_gimple_call): Same.
34387 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34389 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34390 info.nowrite calls with no lhs that can't throw. Return bool
34391 whether gsi_remove has been called or not.
34392 (pass_sprintf_length::handle_gimple_call): Return bool whether
34393 try_substitute_return_value called gsi_remove. Formatting fix.
34394 (pass_sprintf_length::execute): Don't use gsi_remove if
34395 handle_gimple_call returned true.
34398 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34399 be removed due to side-effects, don't remove following barrier nor
34400 turn the successor edge into fallthru edge.
34402 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34405 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34406 element-reversing loads and stores as not swappable.
34408 2017-01-12 Nathan Sidwell <nathan@acm.org>
34409 Nicolai Stange <nicstange@gmail.com>
34411 * combine.c (try_combine): Don't ignore result of overlap checking
34412 loop. Combine overlap & asm check into single loop.
34414 2017-01-12 Richard Biener <rguenther@suse.de>
34416 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34417 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34419 2017-01-12 Richard Biener <rguenther@suse.de>
34421 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34422 and TS_TARGET_OPTION directly derive from TS_BASE.
34423 * tree-core.h (tree_optimization_option): Derive from tree_base.
34424 (tree_target_option): Likewise.
34426 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34428 * config/i386/i386.c (memory_address_length): Increase len
34429 only when rip_relative_addr_p returns false.
34431 2017-01-11 Julia Koval <julia.koval@intel.com>
34433 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34434 (OPTION_MASK_ISA_SGX_SET): New.
34435 (ix86_handle_option): Handle OPT_msgx.
34436 * config.gcc: Added sgxintrin.h.
34437 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34438 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34439 * config/i386/i386.c (ix86_target_string): Add -msgx.
34441 (ix86_option_override_internal): Handle new options.
34442 (ix86_valid_target_attribute_inner_p): Add sgx.
34443 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34444 * config/i386/i386.opt: Add msgx.
34445 * config/i386/sgxintrin.h: New file.
34446 * config/i386/x86intrin.h: Add sgxintrin.h.
34448 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34451 * fold-const.c (maybe_nonzero_address): Return 1 for function
34453 (tree_single_nonzero_warnv_p): Don't handle function local objects
34457 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34460 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34463 * opts.c: Include "spellcheck.h"
34464 (struct string_fragment): New struct.
34465 (struct edit_distance_traits<const string_fragment &>): New
34467 (get_closest_sanitizer_option): New function.
34468 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34470 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34472 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34474 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34475 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34476 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34477 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34478 for initial die_offset if dwarf_split_debug_info.
34479 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34480 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34482 (output_skeleton_debug_sections): Formatting fix. Use
34483 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34484 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34486 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34488 * config/arm/cortex-a53.md: Add bypasses for
34489 cortex_a53_r2f_cvt.
34490 (cortex_a53_r2f): Only use for transfers.
34491 (cortex_a53_f2r): Likewise.
34492 (cortex_a53_r2f_cvt): Add reservation for conversions.
34493 (cortex_a53_f2r_cvt): Likewise.
34495 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34497 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34498 to all inlined functions, change static to extern.
34500 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34503 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34505 (arm_assemble_integer): Likewise.
34507 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34509 * config.gcc: Use new awk script to check CPU, FPU and architecture
34510 parameters for --with-... options.
34511 * config/arm/parsecpu.awk: New file
34512 * config/arm/arm-cpus.in: New file.
34513 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34515 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34517 * config/arm/t-arm: Update dependency rules.
34518 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34519 of processing .def files.
34520 * config/arm/genopt.sh: Deleted.
34521 * config/arm/gentune.sh: Deleted.
34522 * config/arm/arm-cores.def: Deleted.
34523 * config/arm/arm-arches.def: Deleted.
34524 * config/arm/arm-fpus.def: Deleted.
34525 * config/arm/arm-tune.md: Regenerated.
34526 * config/arm/arm-tables.opt: Regenerated.
34527 * config/arm/arm-cpu.h: New generated file.
34528 * config/arm/arm-cpu-data.h: New generated file.
34529 * config/arm/arm-cpu-cdata.h: New generated file.
34531 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34534 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34536 (input_varpool_node): Unpack dynamically_initialized bit.
34538 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34540 PR rtl-optimization/79032
34541 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34542 the alignment of the adjusted memory reference against that of MODE,
34543 instead of the alignment of the original memory reference.
34545 2017-01-11 Martin Jambor <mjambor@suse.cz>
34547 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34549 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34550 decorated functions.
34552 2017-01-11 Richard Biener <rguenther@suse.de>
34554 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34555 set range/nonnull info for PHI results. Do not set it on
34556 stmts marked for removal.
34558 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34560 * expr.c (store_field): In the bitfield case, fetch the return value
34561 from the registers before applying a single big-endian adjustment.
34562 Always do a final load for a BLKmode value not larger than a word.
34564 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34567 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34568 that we correctly handle column numbers greater than
34569 LINE_MAP_MAX_COLUMN_NUMBER.
34571 2017-01-10 Martin Sebor <msebor@redhat.com>
34573 PR middle-end/78245
34574 * gimple-ssa-sprintf.c (get_destination_size): Call
34575 {init,fini}object_sizes.
34576 * tree-object-size.c (addr_object_size): Adjust.
34577 (pass_through_call): Adjust.
34578 (pass_object_sizes::execute): Adjust.
34579 * tree-object-size.h (fini_object_sizes): Declare.
34581 2017-01-10 Martin Sebor <msebor@redhat.com>
34583 PR tree-optimization/78775
34584 * builtins.c (get_size_range): Move...
34585 * calls.c: ...to here.
34586 (alloc_max_size): Accept zero argument.
34587 (operand_signed_p): Remove.
34588 (maybe_warn_alloc_args_overflow): Call get_size_range.
34589 * calls.h (get_size_range): Declare.
34591 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34593 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34594 from TI's devices.csv file as of September 2016.
34595 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34597 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34599 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34600 * doc/invoke.texi: Likewise.
34601 * doc/md.texi: Likewise.
34602 * doc/objc.texi: Likewise.
34604 2017-01-10 Joshua Conner <joshconner@google.com>
34606 * config/arm/fuchsia-elf.h: New file.
34607 * config/fuchsia.h: New file.
34608 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34609 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34611 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34613 2016-01-10 Richard Biener <rguenther@suse.de>
34615 PR tree-optimization/79034
34616 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34617 Propagate out degenerate PHIs in the joiner.
34619 2017-01-10 Martin Liska <mliska@suse.cz>
34621 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34622 (sort_congruence_classes_by_decl_uid): Likewise.
34623 (sort_congruence_class_groups_by_decl_uid): Likewise.
34624 (sem_item_optimizer::merge_classes): Sort class, groups in these
34625 classes and members in the groups by DECL_UID of declarations.
34626 This would make merge operations stable.
34628 2017-01-10 Martin Liska <mliska@suse.cz>
34630 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34631 usage of m_classes_vec.
34632 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34633 (sem_item_optimizer::get_group_by_hash): Likewise.
34634 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34635 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34636 (sem_item_optimizer::verify_classes): Likewise.
34637 (sem_item_optimizer::process_cong_reduction): Likewise.
34638 (sem_item_optimizer::dump_cong_classes): Likewise.
34639 (sem_item_optimizer::merge_classes): Likewise.
34640 * ipa-icf.h (congruence_class_hash): Rename from
34641 congruence_class_group_hash. Remove declaration of m_classes_vec.
34643 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34645 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34646 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34647 * config.gcc: Add avx512vpopcntdqintrin.h.
34648 * config/i386/avx512vpopcntdqintrin.h: New.
34649 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34650 * config/i386/i386-builtin-types.def: Add new types.
34651 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34652 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34653 __builtin_ia32_vpopcountq_v8di_mask): New.
34654 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34655 __AVX512VPOPCNTDQ__.
34656 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34657 (PTA_AVX512VPOPCNTDQ): Define.
34658 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34659 TARGET_AVX512VPOPCNTDQ_P): Define.
34660 * config/i386/i386.opt: Add mavx512vpopcntdq.
34661 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34662 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34664 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34666 PR middle-end/77484
34667 * predict.def (PRED_CALL): Set to 67.
34669 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34671 * expr.c (store_field): In the bitfield case, if the value comes from
34672 a function call and is of an aggregate type returned in registers, do
34673 not modify the field mode; extract the value in all cases if the mode
34674 is BLKmode and the size is not larger than a word.
34676 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34679 * config/i386/cpuid.h: Fix undefined behavior.
34681 2017-01-04 Jeff Law <law@redhat.com>
34683 PR tree-optimization/79007
34684 PR tree-optimization/67955
34685 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34686 conservative for pt.null when flag_non_call_exceptions is on.
34688 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34690 PR translation/79019
34691 PR translation/79020
34692 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34693 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34694 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34696 * config/avr/avr.opt (maccumulate-args): Likewise.
34697 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34698 * common.opt (freport-bug): Likewise.
34699 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34700 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34701 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34702 translatable string.
34703 * config/i386/i386.c (function_value_32): Likewise.
34704 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34705 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34707 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34708 * common/config/msp430/msp430-common.c (msp430_handle_option):
34710 * symtab.c (symtab_node::verify_base): Likewise.
34711 * opts.c (set_debug_level): Likewise.
34712 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34713 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34714 missing whitespace to translatable strings.
34715 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34716 * config/sh/superh.h: Likewise.
34717 * config/i386/xopintrin.h: Likewise.
34718 * config/i386/znver1.md: Likewise.
34719 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34720 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34721 * double-int.h (struct double_int): Likewise.
34722 * double-int.c (div_and_round_double): Likewise.
34723 * wide-int.cc: Likewise.
34724 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34725 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34726 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34727 (crossjumps_occurred): ... this.
34728 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34731 PR tree-optimization/78899
34732 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34733 returning bool return struct loop *, NULL for failure and the new
34735 (versionable_outer_loop_p): Don't version outer loop if it has
34736 dont_vectorized bit set.
34737 (tree_if_conversion): When versioning outer loop, ensure
34738 tree_if_conversion is performed also on the inner loop of the
34739 non-vectorizable outer loop copy.
34740 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34741 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34742 prevent vectorization of it.
34743 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34744 the outer loop vectorization of the non-scalar version is attempted
34745 before vectorization of the inner loop in scalar version. If
34746 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34747 vectorization of its inner loop.
34748 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34749 has 2 inner loops, rename also on edges from bb whose single pred
34750 is outer_loop->header. Fix typo in function comment.
34752 2017-01-09 Martin Sebor <msebor@redhat.com>
34755 * asan.c (asan_emit_stack_protection): Increase local buffer size
34756 to avoid snprintf truncation warning.
34758 2017-01-09 Andrew Pinski <apinski@cavium.com>
34760 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34761 to reference thunderx2t99 for the tuning structure
34762 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34764 (thunderx2t99_extra_costs): This.
34765 * config/aarch64/aarch64-tune.md: Regenerate.
34766 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34767 (vulcan_addrcost_table): This.
34768 (vulcan_regmove_cost): Rename to ...
34769 (thunderx2t99_regmove_cost): This.
34770 (vulcan_vector_cost): Rename to ...
34771 (thunderx2t99_vector_cost): this.
34772 (vulcan_branch_cost): Rename to ...
34773 (thunderx2t99_branch_cost): This.
34774 (vulcan_tunings): Rename to ...
34775 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34776 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34778 2017-01-09 Martin Jambor <mjambor@suse.cz>
34782 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34783 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34784 (propagate_vr_accross_jump_function): Use the above function for all
34785 value range computations for pass-through jump functions and type
34786 converasion from explicit value range values.
34787 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34788 parameters from TYPE_ARG_TYPES.
34789 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34790 (ipa_write_node_info): Stream type of the actual argument.
34791 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34793 2017-01-09 Martin Liska <mliska@suse.cz>
34796 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34797 (lookup_compiler): Do not show error message with have_E.
34799 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34801 PR tree-optimization/78938
34802 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34803 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34804 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34805 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34808 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34810 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34813 2017-01-09 Richard Biener <rguenther@suse.de>
34815 PR tree-optimization/78997
34816 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34817 name condition properly.
34819 2017-01-09 Richard Biener <rguenther@suse.de>
34822 * dwarf2out.c (is_cxx): New overload with context.
34823 (is_naming_typedef_decl): Use it.
34825 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34827 * invoke.texi (Option Summary): Correct spacing in option lists
34828 and add line breaks to fix over-long lines.
34830 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34832 PR middle-end/17660
34834 * extend.texi (Common Variable Attributes): Add xref to GCC
34835 Internals manual to explain mode attribute keywords.
34837 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34840 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34841 and Preprocessor Options.
34842 (Options for Linking): Document -pthread here....
34843 (RS/6000 and PowerPC Options): ...not here.
34844 (Solaris 2 Options): ...or here.
34845 * doc/cppopts.texi: Document -pthread.
34847 2017-01-08 Martin Sebor <msebor@redhat.com>
34849 PR middle-end/77708
34850 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34851 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34852 New member functions.
34853 (format_directive): Used them.
34855 (pass_sprintf_length::handle_gimple_call): Same.
34856 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34857 to avoid truncation for any argument.
34858 (extract_affine_mul): Same.
34859 * tree.c (get_file_function_name): Same.
34861 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34863 PR middle-end/77484
34864 * predict.def (PRED_INDIR_CALL): Set to 86.
34866 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34868 PR preprocessor/54124
34869 * doc/cppopts.texi: Reformat -d subtable to list the full name
34870 of the options. Add cross-reference to the docs for the general
34871 compiler -d options.
34872 * doc/invoke.texi (Developer Options): Add cross-reference to the
34873 preprocessor-specific -d option documentation.
34875 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34877 PR preprocessor/13498
34878 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34879 redudant material, and reflect new command-line options.
34880 (System Headers): Likewise.
34882 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34884 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34885 -isystem, and -idirafter. Copy-edit.
34886 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34887 default for -ftrack-macro-expansion. Delete obsolete and
34888 badly-formatted implementation details about -fdebug-cpp output.
34889 * doc/cppwarnopts.texi: Copy-edit.
34891 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34894 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34895 that the transition from a max line width >= 1<<10 to narrower
34896 lines works correctly.
34898 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34900 * doc/options.texi (PerFunction): New.
34901 * opt-functions.awk (switch_flags): Map both Optimization and
34902 PerFunction to CL_OPTIMIZATION.
34903 * opth-gen.awk: Test for PerFunction flag along with
34905 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34906 it only when the latter is present. Skip those that don't in
34907 the hash function generator.
34908 * common.opt (fvar-tracking): Mark as PerFunction instead of
34910 (fvar-tracking-assignments): Likewise.
34911 (fvar-tracking-assignments-toggle): Likewise.
34912 (fvar-tracking-uninit): Likewise.
34914 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34916 PR translation/79018
34917 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34920 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34923 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34924 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34925 TARGET_LONG_JUMP_TABLE_OFFSETS.
34926 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34927 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34928 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34929 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34930 * config/m68k/m68k.md (tablejump expander): Likewise.
34931 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34932 TARGET_LONG_JUMP_TABLE_OFFSETS.
34933 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34934 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34936 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34937 David Holsgrove <david.holsgrove@xilinx.com>
34939 * common/config/microblaze/microblaze-common.c
34940 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34941 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34943 * config/microblaze/microblaze.c (microblaze_must_save_register)
34944 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34946 (microblaze_eh_return): New function.
34947 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34948 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34949 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34950 * config/microblaze/microblaze.md (eh_return): New pattern.
34952 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34954 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34955 GCC_DIAGNOSTIC_STRINGIFY): Define.
34957 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34959 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34961 * config/arm/arm.md (<mcrr>): New.
34963 * config/arm/arm.c (arm_arch5te): New.
34964 (arm_option_override): Set arm_arch5te.
34965 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34967 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34968 (arm_mcrr_qualifiers): ... this. New.
34969 (MRRC_QUALIFIERS): Define to...
34970 (arm_mrrc_qualifiers): ... this. New.
34971 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34973 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34974 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34975 (MRRCI, mrrc, MRRC): New.
34976 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34977 VUNSPEC_MRRC2): New.
34979 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34981 * config/arm/arm.md (<mcr>): New.
34983 * config/arm/arm.c (arm_coproc_builtin_available): Add
34984 support for mcr, mrc, mcr2 and mrc2.
34985 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34986 (arm_mcr_qualifiers): ... this. New.
34987 (MRC_QUALIFIERS): Define to ...
34988 (arm_mrc_qualifiers): ... this. New.
34989 (MCR_QUALIFIERS): Define to ...
34990 (arm_mcr_qualifiers): ... this. New.
34991 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34993 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34994 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34995 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34996 VUNSPEC_MRC2): New.
34998 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35000 * config/arm/arm.md (*ldc): New.
35004 * config/arm/arm.c (arm_coproc_builtin_available): Add
35005 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35006 (arm_coproc_ldc_stc_legitimate_address): New.
35007 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35008 'qualifier_const_pointer'.
35009 (LDC_QUALIFIERS): Define to...
35010 (arm_ldc_qualifiers): ... this. New.
35011 (STC_QUALIFIERS): Define to...
35012 (arm_stc_qualifiers): ... this. New.
35013 * config/arm/arm-protos.h
35014 (arm_coproc_ldc_stc_legitimate_address): New.
35015 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35016 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35017 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35018 stc2, stcl, stc2l): New.
35019 * config/arm/constraints.md (Uz): New.
35020 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35021 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35022 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35023 VUNSPEC_STC2L): New.
35025 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35027 * config/arm/arm.md (<cdp>): New.
35028 * config/arm/arm.c (neon_const_bounds): Rename this ...
35029 (arm_const_bounds): ... this.
35030 (arm_coproc_builtin_available): New.
35031 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35032 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35033 (CDP_QUALIFIERS): Define to...
35034 (arm_cdp_qualifiers): ... this. New.
35036 (arm_expand_builtin_args): Add case for 6 arguments.
35037 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35038 (arm_const_bounds): ... this.
35039 (arm_coproc_builtin_available): New.
35040 * config/arm/arm_acle.h (__arm_cdp): New.
35042 * config/arm/arm_acle_builtins.def (cdp): New.
35044 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35045 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35046 'arm_const_bounds'.
35047 * config/arm/types.md (coproc): New.
35048 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35049 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35050 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35051 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35053 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35055 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35056 (UBINOP_QUALIFIERS): New.
35058 (acle_builtin_data): New. Change comment.
35059 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35060 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35061 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35062 arm_acle_builtins.def.
35063 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35064 (arm_init_acle_builtins): New.
35065 (CRC32_BUILTIN): Remove.
35066 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35067 crc32cb, crc32ch and crc32cw.
35068 (arm_init_crc32_builtins): Remove.
35069 (arm_init_builtins): Use arm_init_acle_builtins rather
35070 than arm_init_crc32_builtins.
35071 (arm_expand_acle_builtin): New.
35072 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35073 * config/arm/arm_acle_builtins.def: New.
35075 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35077 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35078 (arm_builtin_datum): ... this.
35079 (arm_init_neon_builtin): Rename to ...
35080 (arm_init_builtin): ... this. Add a new parameters PREFIX
35081 and USE_SIG_IN_NAME.
35082 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35083 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35084 'arm_builtin_datum'.
35085 (arm_init_vfp_builtins): Likewise.
35086 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35087 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35088 (arm_expand_neon_args): Rename to ...
35089 (arm_expand_builtin_args): ... this. Rename builtin_arg
35090 enum values and differentiate between ARG_BUILTIN_MEMORY
35091 and ARG_BUILTIN_NEON_MEMORY.
35092 (arm_expand_neon_builtin_1): Rename to ...
35093 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35094 values, arm_expand_builtin_args and add bool parameter NEON.
35095 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35096 (arm_expand_vfp_builtin): Likewise.
35097 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35099 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35101 PR middle-end/77484
35102 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35103 * predict.c (tree_estimate_probability_bb): Reverse direction of
35104 polymorphic call predictor.
35106 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35108 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35109 (determine_pass_name_match): ...this new function and...
35110 (should_skip_pass_p): ...this new function.
35112 2017-01-06 Nathan Sidwell <nathan@acm.org>
35114 * ipa-visibility.c (function_and_variable_visibility): Reformat
35115 comments and long lines. Remove extrneous if.
35116 * symtab.c (symtab_node::make_decl_local): Fix code format.
35117 (symtab_node::set_section_for_node): Fix comment typo.
35119 2017-01-06 Martin Liska <mliska@suse.cz>
35122 * lra-constraints.c: Rename invariant to lra_invariant.
35123 * predict.c (set_even_probabilities): Initialize e to NULL.
35125 2017-01-05 Martin Sebor <msebor@redhat.com>
35127 PR tree-optimization/78910
35128 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35129 (format_integer): Correct off-by-one error in the handling
35130 of precision with negative numbers in signed conversions..
35132 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35134 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35136 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35138 PR tree-optimization/71016
35139 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35140 factor_out_conditional_conversion. Formatting fix.
35141 (factor_out_conditional_conversion): Add cond_stmt argument.
35142 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35143 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35146 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35148 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35149 read-rtl-function.o, and selftest-rtl.o.
35150 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35151 (selftest::aarch64_test_loading_full_dump): New function.
35152 (selftest::aarch64_run_selftests): New function.
35153 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35154 selftest::aarch64_run_selftests.
35155 * config/i386/i386.c
35156 (selftest::ix86_test_loading_dump_fragment_1): New function.
35157 (selftest::ix86_test_loading_call_insn): New function.
35158 (selftest::ix86_test_loading_full_dump): New function.
35159 (selftest::ix86_test_loading_unspec): New function.
35160 (selftest::ix86_run_selftests): Call the new functions.
35161 * emit-rtl.c (maybe_set_max_label_num): New function.
35162 * emit-rtl.h (maybe_set_max_label_num): New decl.
35163 * function.c (instantiate_decls): Guard call to
35164 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35165 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35167 * gensupport.c (gen_reader::gen_reader): Pass "false"
35168 for new "compact" param of rtx_reader.
35169 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35170 rather than an empty string for NULL strings.
35171 * read-md.c: Potentially include config.h rather than bconfig.h.
35172 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35173 (have_error): New global, copied from errors.c.
35174 (md_reader::read_name): Rename to...
35175 (md_reader::read_name_1): ...this, adding "out_loc" param,
35176 and converting "missing name or number" to returning false, rather
35178 (md_reader::read_name): Reimplement in terms of read_name_1.
35179 (md_reader::read_name_or_nil): New function.
35180 (md_reader::read_string): Handle "(nil)" by returning NULL.
35181 (md_reader::md_reader): Add new param "compact".
35182 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35183 (md_reader::read_file): New method.
35184 * read-md.h (md_reader::md_reader): Add new param "compact".
35185 (md_reader::read_file): New method.
35186 (md_reader::is_compact): New accessor.
35187 (md_reader::read_name): Convert return type from void to file_location.
35188 (md_reader::read_name_or_nil): New decl.
35189 (md_reader::read_name_1): New decl.
35190 (md_reader::m_compact): New field.
35191 (noop_reader::noop_reader): Pass "false" for new "compact" param
35193 (rtx_reader::rtx_reader): Add new "compact" param.
35194 (rtx_reader::read_rtx_operand): Make virtual and convert return
35195 type from void to rtx.
35196 (rtx_reader::read_until): New decl.
35197 (rtx_reader::handle_any_trailing_information): New virtual function.
35198 (rtx_reader::postprocess): New virtual function.
35199 (rtx_reader::finalize_string): New virtual function.
35200 (rtx_reader::m_in_call_function_usage): New field.
35201 (rtx_reader::m_reuse_rtx_by_id): New field.
35202 * read-rtl-function.c: New file.
35203 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35204 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35205 (selftest::verify_three_block_rtl_cfg): New decl.
35206 * read-rtl-function.h: New file.
35207 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35208 For host, include function.h, memmodel.h, and emit-rtl.h.
35209 (one_time_initialization): New function.
35210 (struct compact_insn_name): New struct.
35211 (compact_insn_names): New array.
35212 (find_code): Handle insn codes in compact dumps.
35213 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35214 (bind_subst_iter_and_attr): Likewise.
35215 (add_condition_to_string): Likewise.
35216 (add_condition_to_rtx): Likewise.
35217 (apply_attribute_uses): Likewise.
35218 (add_current_iterators): Likewise.
35219 (apply_iterators): Likewise.
35220 (initialize_iterators): Guard usage of apply_subst_iterator with
35221 #ifdef GENERATOR_FILE.
35222 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35223 (md_reader::read_mapping): Likewise.
35224 (add_define_attr_for_define_subst): Likewise.
35225 (add_define_subst_attr): Likewise.
35226 (read_subst_mapping): Likewise.
35227 (check_code_iterator): Likewise.
35228 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35230 (one_time_initialization): New function.
35231 (rtx_reader::read_until): New method.
35232 (read_flags): New function.
35233 (parse_reg_note_name): New function.
35234 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35235 Handle reuse_rtx ids.
35236 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35237 Add parsing support for RTL dumps, mirroring the special-cases in
35238 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35239 values, and calling handle_any_trailing_information.
35240 (rtx_reader::read_rtx_operand): Convert return type from void
35241 to rtx, returning return_rtx. Handle case 'e'. Call
35242 finalize_string on XSTR and XTMPL fields.
35243 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35244 "(nil)" values were omitted. Call the postprocess vfunc on the
35246 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35247 class ctor. Initialize m_in_call_function_usage. Call
35248 one_time_initialization.
35249 * rtl-tests.c (selftest::test_uncond_jump): Call
35250 set_new_first_and_last_insn.
35251 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35252 * selftest-rtl.c: New file.
35253 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35254 (selftest::get_insn_by_uid): New decl.
35255 * selftest-run-tests.c (selftest::run_tests): Call
35256 read_rtl_function_c_tests.
35257 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35258 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35261 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35263 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35264 operands in a special way. Assert that pos+len <= mode precision.
35266 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35268 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35269 3 argument Alias with unlimited for the negative form.
35270 (fno-vect-cost-model): Removed.
35272 2017-01-05 Martin Liska <mliska@suse.cz>
35274 * hsa-gen.c (gen_hsa_divmod): New function.
35275 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35277 2017-01-05 Martin Liska <mliska@suse.cz>
35280 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35283 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35285 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35286 small constant length operands.
35288 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35290 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35291 between loop iterations.
35293 2017-01-05 Martin Liska <mliska@suse.cz>
35296 * gimplify.c (gimplify_decl_expr): Compare to
35297 asan_poisoned_variables instread of checking flags.
35298 (gimplify_target_expr): Likewise.
35299 (gimplify_expr): Likewise.
35300 (gimplify_function_tree): Conditionally initialize
35301 asan_poisoned_variables.
35303 2017-01-04 Jeff Law <law@redhat.com>
35305 PR tree-optimizatin/78812
35306 * rtl.h (contains_mem_rtx_p): Prototype.
35307 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35308 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35309 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35310 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35311 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35313 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35315 * input.c (assert_char_at_range): Default-initialize actual_range.
35317 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35319 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35320 to match the caller.
35322 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35324 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35325 insns after final jump in test to emit dummy move.
35327 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35329 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35330 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35332 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35334 * multiple_target.c (create_dispatcher_calls): Init e_next.
35335 * tree-ssa-loop-split.c (split_loop): Init border.
35336 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35339 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35344 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35345 (altivec_register_operand): Do not return true if the operand
35346 contains a SUBREG mixing SImode and SFmode.
35347 (vsx_register_operand): Likewise.
35348 (vsx_reg_sfsubreg_ok): New predicate.
35349 (vfloat_operand): Do not return true if the operand contains a
35350 SUBREG mixing SImode and SFmode.
35351 (vint_operand): Likewise.
35352 (vlogical_operand): Likewise.
35353 (gpc_reg_operand): Likewise.
35354 (int_reg_operand): Likewise.
35355 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35356 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35357 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35359 (rs6000_emit_move_si_sf_subreg): New helper function.
35360 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35361 fixup SUBREGs involving SImode and SFmode.
35362 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35363 numbers for the new peephole2 optimization.
35364 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35365 the GLIBC math library that do AND/IOR/XOR operations on single
35366 precision floating point.
35367 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35368 target macros to say whether we need to avoid SUBREGs mixing
35370 (TARGET_ALLOW_SF_SUBREG): Likewise.
35371 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35372 (UNSPEC_SI_FROM_SF): Likewise.
35373 (iorxor): Change spacing.
35374 (and_ior_xor): New iterator for AND, IOR, and XOR.
35375 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35376 (movdi_from_sf_zero_ext): Likewise.
35377 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35378 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35379 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35380 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35381 (fms<mode>4): Likewise.
35382 (fnma<mode>4): Likewise.
35383 (fnms<mode>4): Likewise.
35384 (nfma<mode>4): Likewise.
35385 (nfms<mode>4): Likewise.
35387 2017-01-04 Marek Polacek <polacek@redhat.com>
35390 * doc/invoke.texi: Document -Wpointer-compare.
35392 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35394 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35397 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35398 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35401 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35404 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35405 documentation of the powerpc_popcntb_ok attribute.
35406 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35407 code to issue warning messages if a requested CPU configuration is
35408 not supported by the binary (assembler and loader) toolchain.
35409 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35410 made to define a built-in function that has been disabled.
35411 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35412 made to define a built-in function that has been disabled.
35413 (altivec_init_builtins): Add comment explaining why definition
35414 of the DST built-in functions is not preceded by an assertion
35415 check. Add assertions to prevent ICE if attempts are made to
35416 define an altivec predicate or an abs* built-in function that has
35418 (htm_init_builtins): Add comment explaining why definition of the
35419 htm built-in functions is not preceded by an assertion check.
35421 2017-01-04 Jeff Law <law@redhat.com>
35423 PR tree-optimizatin/67955
35424 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35425 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35426 the points-to solution does not include pt_null. Use DECL_PT_UID
35429 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35431 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35432 Use gen_int_mode instead of gen_lopwart for const_int operands.
35434 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35436 PR tree-optimization/71563
35437 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35438 out of range value - has low bits known to be zero.
35440 2017-01-04 Alan Modra <amodra@gmail.com>
35442 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35443 * configure: Regenerate.
35444 * config.in: Regenerate.
35446 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35449 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35450 a substring of the message, but strcmp with the whole message. Ifdef
35451 ENABLE_NLS, translate the message first using dgettext.
35453 2017-01-03 Jeff Law <law@redhat.com>
35455 PR tree-optimizatin/78856
35456 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35457 (mark_threaded_blocks): Remove code to truncate thread paths that
35458 cross multiple loop headers. Instead invalidate the cached loop
35459 iteration information and handle case of a thread path walking
35460 into an irreducible region.
35462 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35465 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35466 assertions. Add support for doing the signbit if the IEEE 128-bit
35467 floating point value is in a GPR.
35468 * config/rs6000/rs6000.md (Fsignbit): Delete.
35469 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35470 Update the length attribute if the value is in a GPR.
35471 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35472 the sign or zero extension instruction, since the value is always 0/1.
35473 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35476 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35477 extracting SImode to a GPR register so that we can generate a
35478 store, limit the vector to be in a traditional Altivec register
35479 for the vextuwrx instruction.
35481 2017-01-03 Ian Lance Taylor <iant@google.com>
35483 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35485 2017-01-03 Martin Sebor <msebor@redhat.com>
35487 PR tree-optimization/78696
35488 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35489 precision. Use MPFR for %f for greater fidelity. Correct handling
35491 (pass_sprintf_length::compute_format_length): Set width and precision
35492 specified by asrerisk to void_node for vararg functions.
35493 (try_substitute_return_value): Adjust dump output.
35495 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35497 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35499 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35501 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35502 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35503 -mlra/-mno-lra was passed to the compiler.
35505 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35507 PR rtl-optimization/65618
35508 * emit-rtl.c (try_split): Move initialization of "before" and
35509 "after" to just before the call to emit_insn_after_setloc.
35511 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35513 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35515 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35517 * dwarf2out.c (gen_enumeration_type_die): When
35518 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35520 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35522 PR tree-optimization/78965
35523 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35524 Change first argument from const call_info & to call_info &. For %n
35525 set info.nowrite to false.
35527 PR middle-end/78901
35528 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35529 possibly throwing calls.
35531 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35532 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35533 and fns handling, rather than in a separate case SSA_NAME.
35535 2017-01-02 Jeff Law <law@redhat.com>
35537 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35538 fixes for first_period and second_period variables.
35540 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35543 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35544 (*insvqi_1): New insn pattern.
35545 (*insvqi_1_mem_rex64): Ditto.
35546 (*insvqi_2): Ditto.
35547 (*insvqi_3): Rename from *insvqi.
35549 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35551 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35553 * doc/cfg.texi (Edges): Remove reference to Java.
35554 (Maintaining the CFG): Ditto.
35556 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35558 PR middle-end/77674
35559 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35560 transparent aliases.
35562 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35564 PR middle-end/77484
35565 * predict.def (PRED_CALL): Update hitrate.
35566 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35567 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35568 into direct/indirect/polymorphic variants.
35570 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35572 Update copyright years.
35574 * gcc.c (process_command): Update copyright notice dates.
35575 * gcov-dump.c (print_version): Ditto.
35576 * gcov.c (print_version): Ditto.
35577 * gcov-tool.c (print_version): Ditto.
35578 * gengtype.c (create_file): Ditto.
35579 * doc/cpp.texi: Bump @copying's copyright year.
35580 * doc/cppinternals.texi: Ditto.
35581 * doc/gcc.texi: Ditto.
35582 * doc/gccint.texi: Ditto.
35583 * doc/gcov.texi: Ditto.
35584 * doc/install.texi: Ditto.
35585 * doc/invoke.texi: Ditto.
35587 Copyright (C) 2017 Free Software Foundation, Inc.
35589 Copying and distribution of this file, with or without modification,
35590 are permitted in any medium without royalty provided the copyright
35591 notice and this notice are preserved.