1 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
4 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6 2017-11-17 Richard Biener <rguenther@suse.de>
8 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
11 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
14 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
15 of replacing call to strncmp with corresponding call to strcmp when
18 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
20 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
21 option prefer-avx256 for skylake-avx512 configuration.
22 * config/i386/i386.c (ix86_option_override_internal): Ditto.
23 (get_builtin_code_for_version): Ditto.
25 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
26 Monk Chiang <sh.chiang04@gmail.com>
28 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
29 (FIXED_REGISTERS): Reserve more register numbers.
30 (CALL_USED_REGISTERS): Likewise.
31 (REG_ALLOC_ORDER): Likewise.
32 (REG_CLASS_CONTENTS): Likewise.
33 (REGISTER_NAMES): Likewise.
35 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
36 Kito Cheng <kito.cheng@gmail.com>
38 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
40 * config/nds32/iterators.md: Add vector mode iterators and attributes.
42 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
44 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
45 parameter list for vec_splats.
47 2017-11-16 Joseph Myers <joseph@codesourcery.com>
49 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
51 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
54 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
57 * doc/install.texi: Document incompatibility of Perl >=5.6.26
58 with the required version of automake 1.11.6.
60 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
62 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
64 (DU_C2_3_power9): Correct reservation combinations.
65 (FP_DIV_power9, VEC_DIV_power9): New.
66 (power9-alu): Split out rotate/shift...
67 (power9-rot): ...to here, correct dispatch resource.
68 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
70 (power9-fp): Correct latency.
71 (power9-sdiv): Add div/sqrt resource.
72 (power9-ddiv): Correct latency, add div/sqrt resource.
73 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
74 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
76 (power9-qpdiv, power9-qpmul): Adjust resource usage.
78 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
80 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
81 switch statement mapping KF built-ins to TF built-ins if we don't
82 have the proper ISA 3.0 assembler support.
84 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
86 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
87 (lower_emutls_function_body): Do not compute it.
89 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
91 * ipa-split.c (split_bb_info): Turn time to sreal.
92 (split_point): Likewise.
93 (dump_split_point): Likewise.
94 (fine_split_points): Likewise.
95 (execute_split_functions): Only zero split_bbs; turn time to sreals.
97 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
99 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
102 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
104 * predict.c (combine_predictions_for_bb): Preserve zero predicted
106 (expensive_function_p): Remove useless assert.
107 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
109 2017-11-16 Martin Sebor <msebor@redhat.com>
111 PR tree-optimization/82588
112 PR tree-optimization/82583
113 * tree-vrp.c (check_array_ref): Handle flexible array members,
114 string literals, and inner indices.
115 (search_for_addr_array): Add detail to diagnostics.
117 2017-11-16 Nathan Sidwell <nathan@acm.org>
121 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
122 (SET_DECL_ASSEMBLER_NAME): Forward to
123 overwrite_decl_assembler_name.
124 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
125 (overwrite_decl_assembler_name): Declare.
126 * tree.c (overwrite_decl_assembler_name): New.
127 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
128 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
129 (LANG_HOOKS_INITIALIZER): Add it.
130 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
131 * langhooks.c (lhd_set_decl_assembler_name): Use
132 SET_DECL_ASSEMBLER_NAME.
133 (lhd_overwrite_decl_assembler_name): Default implementation.
135 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
136 Jackson Woodruff <jackson.woodruff@arm.com>
138 PR tree-optimization/71026
139 * match.pd: Canonicalize constant multiplies in division.
141 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
143 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
144 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
145 Remove OPT_fomit_frame_pointer entry.
146 * common/config/arc/arc-common.c: Likewise.
147 * common/config/arm/arm-common.c: Likewise.
148 * common/config/avr/avr-common.c: Likewise.
149 * common/config/c6x/c6x-common.c: Likewise.
150 * common/config/cr16/cr16-common.c: Likewise.
151 * common/config/cris/cris-common.c: Likewise.
152 * common/config/epiphany/epiphany-common.c: Likewise.
153 * common/config/fr30/fr30-common.c: Likewise.
154 * common/config/frv/frv-common.c: Likewise.
155 * common/config/ia64/ia64-common.c: Likewise.
156 * common/config/iq2000/iq2000-common.c: Likewise.
157 * common/config/lm32/lm32-common.c: Likewise.
158 * common/config/m32r/m32r-common.c: Likewise.
159 * common/config/mcore/mcore-common.c: Likewise.
160 * common/config/microblaze/microblaze-common.c: Likewise.
161 * common/config/mips/mips-common.c: Likewise.
162 * common/config/mmix/mmix-common.c: Likewise.
163 * common/config/mn10300/mn10300-common.c: Likewise.
164 * common/config/nios2/nios2-common.c: Likewise.
165 * common/config/pa/pa-common.c: Likewise.
166 * common/config/pdp11/pdp11-common.c: Likewise.
167 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
168 * common/config/riscv/riscv-common.c: Likewise.
169 * common/config/rs6000/rs6000-common.c: Likewise.
170 * common/config/rx/rx-common.c: Likewise.
171 * common/config/s390/s390-common.c: Likewise.
172 * common/config/sh/sh-common.c: Likewise.
173 * common/config/sparc/sparc-common.c: Likewise.
174 * common/config/tilegx/tilegx-common.c: Likewise.
175 * common/config/tilepro/tilepro-common.c: Likewise.
176 * common/config/v850/v850-common.c: Likewise.
177 * common/config/visium/visium-common.c: Likewise.
178 * common/config/xstormy16/xstormy16-common.c: Likewise.
179 * common/config/xtensa/xtensa-common.c: Likewise.
180 * invoke.texi (-fomit-frame-pointer): Update documentation.
182 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
184 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
186 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
188 * cfg.c (scale_bbs_frequencies_int,
189 cale_bbs_frequencies_gcov_type): Remove.
190 * cfg.h (scale_bbs_frequencies_int,
191 cale_bbs_frequencies_gcov_type): Remove.
193 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
195 * tree-ssa-loop-manip.c
196 (scale_dominated_blocks_in_loop): Update to profile counts.
197 (tree_transform_and_unroll_loop): Likewise.
199 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
201 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
202 scale_bbs_frequencies_int.
204 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
206 * final.c (compute_alignments): Use counts rather than frequencies.
208 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
210 * cfgloopanal.c: Include sreal.h
211 (average_num_loop_insns): Use counts and sreal for accounting.
213 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
215 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
218 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
219 Kito Cheng <kito.cheng@gmail.com>
221 * config/nds32/constraints.md: Provide more constraints.
222 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
223 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
224 support constraints usage.
226 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
228 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
230 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
231 Kito Cheng <kito.cheng@gmail.com>
233 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
234 * config/nds32/nds32.opt: Refine the layout.
235 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
236 TARGET_EXT_STRING): Support new options.
237 * config/nds32/nds32.h: Likewise.
238 * config/nds32/nds32.md: Likewise.
239 * config/nds32/nds32-predicates.c: Likewise.
240 * config/nds32/constraints.md: Likewise.
241 * common/config/nds32/nds32-common.c: Likewise.
243 2017-11-16 Julia Koval <julia.koval@intel.com>
246 * config/i386/gfniintrin.h: Add sse check.
247 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
249 2017-11-16 Julia Koval <julia.koval@intel.com>
251 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
252 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
253 (ix86_handle_option): Handle -mavx512vbmi2.
254 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
255 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
256 * config/i386/i386-c.c (__AVX512VBMI2__): New.
257 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
258 (ix86_valid_target_attribute_inner_p): Ditto.
259 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
260 * config/i386/i386.opt (mavx512vbmi2): New option.
261 * doc/invoke.texi: Add new option.
263 [2017-11-16 Julia Koval <julia.koval@intel.com>
265 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
266 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
267 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
268 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
269 _mm512_gf2p8mul_epi8): New intrinsics.
270 * config/i386/i386-builtin-types.def
271 (V64QI_FTYPE_V64QI_V64QI): New type.
272 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
273 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
274 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
275 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
276 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
277 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
279 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
281 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
282 explicitly as a stream of bytes.
284 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
286 * config/rs6000/altivec.h (vec_xst_be): New #define.
287 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
288 and externalize from *altivec_vperm_<mode>_internal.
289 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
291 (XL_BE_V8HI): Likewise.
292 (XL_BE_V4SI): Likewise.
293 (XL_BE_V4SI): Likewise.
294 (XL_BE_V2DI): Likewise.
295 (XL_BE_V4SF): Likewise.
296 (XL_BE_V2DF): Likewise.
297 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
298 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
299 all array entries with these keys: VSX_BUILTIN_VEC_XL,
300 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
301 VSX_BUILTIN_VEC_XST_BE.
302 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
303 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
305 (altivec_init_builtins): Replace conditional calls to def_builtin
306 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
307 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
308 with unconditional calls. Remove calls to def_builtin for
309 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
310 __builtin_vec_xst_be.
311 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
312 to define_expand, and add alternate RTL generation for P8.
313 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
315 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
316 add alternate RTL generation for P8.
317 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
318 vsx_ld_elemrev_v16qi.
319 (vsx_st_elemrev_v8hi): Convert define_insn
320 to define_expand, and add alternate RTL generation for P8.
321 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
323 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
324 add alternate RTL generation for P8.
325 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
326 vsx_st_elemrev_v16qi.
328 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
331 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
332 TARGET_AVX512ER check.
333 (ix86_option_override_internal): Set MASK_VZEROUPPER if
334 neither -mzeroupper nor -mno-zeroupper is used and
335 TARGET_EMIT_VZEROUPPER is set.
336 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
337 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
339 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
341 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
342 folding of vector compares.
343 (fold_build_vec_cmp): New helper function.
344 (fold_compare_helper): New helper function.
345 (builtin_function_type): Add compare builtins to the list of functions
346 having unsigned arguments. Cosmetic updates to comment indentation.
347 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
348 the not+eq combination.
350 2017-11-15 Bin Cheng <bin.cheng@arm.com>
352 PR tree-optimization/82726
353 PR tree-optimization/70754
354 * tree-predcom.c (order_drefs_by_pos): New function.
355 (combine_chains): Move code setting has_max_use_after to...
356 (try_combine_chains): ...here. New parameter. Sort combined chains
357 according to position information.
358 (tree_predictive_commoning_loop): Update call to above function.
359 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
361 2017-11-15 Bin Cheng <bin.cheng@arm.com>
363 PR tree-optimization/82726
365 2017-01-23 Bin Cheng <bin.cheng@arm.com>
367 PR tree-optimization/70754
368 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
369 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
370 combined stmt before it if not NULL.
371 (combine_chains): Process refs reversely and compute dominance point
375 2017-02-23 Bin Cheng <bin.cheng@arm.com>
377 PR tree-optimization/79663
378 * tree-predcom.c (combine_chains): Process refs in reverse order
379 only for ZERO length chains, and add explaining comment.
381 2017-11-15 Tamar Christina <tamar.christina@arm.com>
383 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
384 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
385 * doc/invoke.texi (ARM Options): Add armv8.3-a.
387 2017-11-15 Tamar Christina <tamar.christina@arm.com>
389 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
391 2017-11-15 Martin Liska <mliska@suse.cz>
393 * tree-cfg.c (pass_warn_function_return::execute):
394 Compare warn_return_type for greater than zero.
396 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
400 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
401 to return true on Xeon and not on Xeon Phi.
402 (ix86_check_avx256_register): Changed to ...
403 (ix86_check_avx_upper_register): ... this. Add extra check for
404 VALID_AVX512F_REG_OR_XI_MODE.
405 (ix86_avx_u128_mode_needed): Changed
406 ix86_check_avx256_register to ix86_check_avx_upper_register.
407 (ix86_check_avx256_stores): Changed to ...
408 (ix86_check_avx_upper_stores): ... this. Changed
409 ix86_check_avx256_register to ix86_check_avx_upper_register.
410 (ix86_avx_u128_mode_after): Changed
411 avx_reg256_found to avx_upper_reg_found. Changed
412 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
413 (ix86_avx_u128_mode_entry): Changed
414 ix86_check_avx256_register to ix86_check_avx_upper_register.
415 (ix86_avx_u128_mode_exit): Ditto.
416 * config/i386/i386.h: (host_detect_local_cpu): New define.
418 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
420 * config/arm/xgene1.md (xgene1): Split into automatons
421 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
422 (xgene1_f_load): Adjust reservations and/or types.
423 (xgene1_f_store): Likewise.
424 (xgene1_load_pair): Likewise.
425 (xgene1_store_pair): Likewise.
426 (xgene1_fp_load1): Likewise.
427 (xgene1_load1): Likewise.
428 (xgene1_store1): Likewise.
429 (xgene1_move): Likewise.
430 (xgene1_alu): Likewise.
431 (xgene1_simd): Likewise.
432 (xgene1_bfm): Likewise.
433 (xgene1_neon_load1): Likewise.
434 (xgene1_neon_store1): Likewise.
435 (xgene1_neon_logic): Likewise.
436 (xgene1_neon_st1): Likewise.
437 (xgene1_neon_ld1r): Likewise.
438 (xgene1_alu_cond): Added.
439 (xgene1_shift_reg): Likwise.
440 (xgene1_bfx): Likewise.
441 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
443 2017-11-15 Jakub Jelinek <jakub@redhat.com>
446 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
448 (can_widen_mult_without_libcall): New function.
449 (expand_mul_overflow): If only checking unsigned mul overflow,
450 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
451 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
452 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
453 (expand_DIVMOD): Formatting fix.
454 * expmed.h (expand_mult): Add NO_LIBCALL argument.
455 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
456 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
458 PR tree-optimization/82977
459 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
460 constructed temporary to strlen_to_stridx.put.
462 2017-11-15 Martin Liska <mliska@suse.cz>
464 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
465 * configure: Regenerate.
467 2017-11-15 Martin Liska <mliska@suse.cz>
470 * config/sh/sh-mem.cc: Use proper probability for
473 2017-11-14 Jeff Law <law@redhat.com>
475 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
476 the red zone for stack_clash_protection_final_dynamic_probe targets
477 when the total dynamic stack size is zero bytes.
479 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
480 blocks is post order.
482 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
484 * dumpfile.h (TDF_COMPARE_DEBUG): New.
485 * final.c (rest_of_clean_state): Set it for the
486 -fcompare-debug dump.
487 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
488 class when TDF_COMPARE_DEBUG is set.
490 * dwarf2out.c (gen_producer_string): Discard
493 2017-11-15 Joseph Myers <joseph@codesourcery.com>
496 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
497 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
498 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
499 (__TGMATH_REAL_2_3): Remove macros.
500 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
501 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
503 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
505 (remquo): Define using __TGMATH_REAL_3.
507 2017-11-14 Jeff Law <law@redhat.com>
509 * vr-values.c: New file with contents extracted from tree-vrp.c.
510 * Makefile.in (OBJS): Add vr-values.o
511 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
512 (set_value_range, set_and_canonicalize_value_range): Likewise.
513 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
514 (value_range_constant_singleton, symbolic_range_p): Likewise.
515 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
516 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
517 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
518 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
519 (find_case_label_range, find_case_label_index): Likewise.
520 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
521 (range_int_cst_singleton_p, value_inside_range): Likewise.
522 (get_single_symbol): Likewise.
523 (switch_update): Move structure definition here.
524 (to_remove_edges, to_update_switch_stmts): Provide externs.
525 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
526 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
527 (vrp_val_is_min, set_value_range): Likewise.
528 (set_and_canonicalize_value_range, copy_value_range): Likewise.
529 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
530 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
531 (range_is_nonnull, range_int_cst_p): Likewwise.
532 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
533 (get_single_symbol, operand_less_p): Likewise
534 (compare_values_warnv, compare_values): Likewise.
535 (value_inside_range, value_range_constant_singleton): Likewise.
536 (zero_nonzero_bitgs_from_vr): Likewise.
537 (extract_range_from_binary_expr_1): Likewise.
538 (overflow_comparison_p): Likewise.
539 (to_remove_edges, to_update_switch_stmts): Likewise.
540 (find_case_label-index, find_case_label_range): Likewise.
541 (switch_update, set_value_range_to_nonnegative): Remove.
542 (set_value_range_to_truthvalue): Likewise.
543 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
544 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
545 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
546 (find_case_label_ranges, test_for_singularity): Likewise.
547 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
548 (x_vr_values): Move to its remaining use site.
550 2017-11-10 Jeff Law <law@redhat.com>
552 * vr-values.h (VR_INITIALIZER): Move #define here.
553 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
554 * Makefile.in (OBJS): Add tree-evrp.o
555 * tree-vrp.h (assert_info): Move structure definition here.
556 (set_value_range_to_varying): Prototype.
557 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
558 (infer_value_range, register_edge_assert_for): Likewise.
559 (stmt_interesting_for_vrp): Likewise.
560 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
561 (set_value_range_to_varying): No longer static.
562 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
563 (infer_value_range, register_edge_assert_for): Likewise.
565 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
567 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
568 generate the XXBRD instruction.
570 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
571 (rs6000_builtin_type_compatible): Treat _Float128 and long double
572 as being compatible if -mabi=ieeelongdouble.
573 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
574 to setup float128 built-ins with hardware support.
575 (BU_FLOAT128_HW_2): Likewise.
576 (BU_FLOAT128_HW_3): Likewise.
577 (BU_FLOAT128_HW_VSX_1): Likewise.
578 (BU_FLOAT128_HW_VSX_2): Likewise.
579 (scalar_extract_expq): Change float128 built-in functions to
580 accommodate having both KFmode and TFmode functions. Use the
581 KFmode variant as the default.
582 (scalar_extract_sigq): Likewise.
583 (scalar_test_neg_qp): Likewise.
584 (scalar_insert_exp_q): Likewise.
585 (scalar_insert_exp_qp): Likewise.
586 (scalar_test_data_class_qp): Likewise.
587 (sqrtf128_round_to_odd): Delete processing the round to odd
588 built-in functions as special built-in functions, and define them
589 as float128 built-ins. Use the KFmode variant as the default.
590 (truncf128_round_to_odd): Likewise.
591 (addf128_round_to_odd): Likewise.
592 (subf128_round_to_odd): Likewise.
593 (mulf128_round_to_odd): Likewise.
594 (divf128_round_to_odd): Likewise.
595 (fmaf128_round_to_odd): Likewise.
596 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
597 support for KFmode and TFmode xststdcqp calls.
598 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
599 point, switch the built-in handlers for the get/set float128
600 exponent, get float128 mantissa, float128 test built-ins, and the
601 float128 round to odd built-in functions. Eliminate creating the
602 float128 round to odd built-in functions as special built-ins.
603 (rs6000_init_builtins): Eliminate special creation of the float128
604 round to odd built-in functions.
605 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
606 function insns to support both TFmode and KFmode variants.
607 (xsxsigqp_<mode>): Likewise.
608 (xsiexpqpf_<mode>): Likewise.
609 (xsiexpqp_<mode>): Likewise.
610 (xststdcqp_<mode>): Likewise.
611 (xststdcnegqp_<mode>): Likewise.
612 (xststdcqp_<mode>): Likewise.
614 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
616 * tree-ssa-threadupdate.c (compute_path_counts): Remove
617 unused path_in_freq_ptr parameter.
618 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
620 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
622 * ipa-inline.c (edge_badness): Dump sreal frequency.
623 (compute_inlined_call_time): Match natural implementaiton ...
624 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
625 forgotten division by CGRAPH_FREQ_BASE.
627 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
629 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
630 Solaris 11. Update comment.
631 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
633 * config/sol2.h (STARTFILE_SPEC): Likewise.
634 * configure: Regenerate.
636 2017-11-14 Carl Love <cel@us.ibm.com>
638 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
639 le_ and be_ prefixes to swap* variables. Remove
640 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
643 2017-11-14 Jason Merrill <jason@redhat.com>
645 Support GTY((cache)) on hash_map.
646 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
647 (ggc_cache_remove): Override it instead of ggc_mx.
648 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
649 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
650 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
651 (simple_cache_map_traits): Override maybe_mx.
652 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
653 (hash_map): Friend gt_cleare_cache.
654 (gt_cleare_cache): New.
655 * tree.h (tree_cache_traits): New hash_map traits class.
656 (tree_cache_map): New typedef.
658 2017-11-14 Richard Biener <rguenther@suse.de>
660 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
661 paramter and handling.
662 (cleanup_control_flow_bb): Likewise.
663 (cleanup_control_flow_pre): New helper performing a DFS walk
664 to call cleanup_control_flow_bb in PRE order.
665 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
666 via cleanup_control_flow_pre.
668 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
670 * config/aarch64/aarch64-simd.md
671 (aarch64_simd_bsl<mode>_internal): Remove DImode.
672 (*aarch64_simd_bsl<mode>_alt): Likewise.
673 (aarch64_simd_bsldi_internal): New.
674 (aarch64_simd_bsldi_alt): Likewise.
676 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
678 * tracer.c (better_p): Do not compare frequencies.
679 * reg-stack.c (better_edge): Likewise.
680 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
683 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
685 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
686 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
687 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
688 * lto-streamer-in.c (input_function): Use update_max_bb_count.
689 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
690 * predict.c (maybe_hot_frequency_p): Inline to ...
691 (maybe_hot_count_p): ... here; rewrite to counts.
692 (counts_to_freqs): Rename to ...
693 (update_max_bb_count): ... this one.
694 (expensive_function_p): Use counts.
695 (estimate_bb_frequencies): Update.
696 (rebuild_frequencies): Update.
697 * predict.h (counts_to_freqs): Rename to ...
698 (update_max_bb_count): ... this one.
699 * profile.c (compute_branch_probabilities): Add debug info
700 * tree-inline.c (expand_call_inline): Update debug info.
701 (optimize_inline_calls): Use update_max_bb_count..
702 (tree_function_versioning): Use update_max_bb_count..
703 * value-prof.c (gimple_value_profile_transformations):
704 Do not use update_max_bb_count.
706 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
708 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
709 always use frequencies.
711 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
713 * bb-reorder.c: Remove frequencies from comments.
714 (better_edge_p): Use profile counts.
715 (find_traces): Dump profile counts.
716 (rotate_loop): Use profile counts.
717 (find_traces_1_round): Likewise.
718 (connect_better_edge_p): Use counts instead of probabilities for
720 (copy_bb_p): Drop early check for non-0 frequency.
721 (sanitize_hot_paths): Update comments.
723 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
725 * ipa-split.c (struct split_point): Add count.
726 (consider_split): Do not compute incoming frequency; compute incoming
727 count and store it to split_point.
728 (split_function): Set count of the call to split part correctly.
730 2017-11-13 Carl Love <cel@us.ibm.com>
732 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
734 2017-11-13 Tom Tromey <tom@tromey.com>
736 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
738 2017-11-13 Carl Love <cel@us.ibm.com>
740 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
741 Add support for builtins:
742 unsigned int vec_first_{,miss}_match_{,or_eos}index,
743 vector {un,}signed {char,int,short},
744 vector {un,}signed {char,int,short}) arguments.
745 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
746 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
747 Add BU_P9V_AV_2 expansions for the builtins.
748 * config/rs6000/altivec.h (vec_first_match_index,
749 vec_first_mismatch_index, vec_first_match_or_eos_index,
750 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
751 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
752 new extern declaration.
753 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
754 * config/rs6000/vsx.md (first_match_index_<mode>,
755 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
756 first_mismatch_or_eos_index_<mode>): Add define expand.
757 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
758 * doc/extend.texi: Update the built-in documenation file for the new
761 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
763 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
764 into the min/max operations for _Float<N> and _Float<N>X types.
766 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
769 * dwarf2out.c (do_eh_frame): New static variable.
770 (dwarf2out_begin_prologue): Set it.
771 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
773 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
775 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
777 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
779 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
782 * debug.h (dwarf2out_do_eh_frame): Declare.
783 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
784 (dwarf2out_do_frame): Use it.
785 (dwarf2out_do_cfi_asm): Likewise.
786 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
787 (dwarf2out_assembly_start): Likewise.
788 (dwarf2out_begin_prologue): Fix comment.
789 * toplev.c (compile_file): Always call dwarf2out_frame_finish
790 if the target needs either debug or unwind DWARF2 info.
791 * lto-opts.c (lto_write_options): Do not save -fexceptions,
792 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
793 -ftrapping-math, -ftrapv and -fwrapv.
795 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
797 * cgraph.c (cgraph_edge::sreal_frequency): New function.
798 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
799 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
800 (estimate_edge_size_and_time): Likewise.
801 (ipa_merge_fn_summary_after_inlining): Likewise.
802 * ipa-inline.c (cgraph_freq_base_rec): Remove.
803 (compute_uninlined_call_time): Use sreal_frequency.
804 (compute_inlined_call_time): Likewise.
805 (ipa_inline): Do not initialize cgraph_freq_base_rec.
806 * profile-count.c: Include sreal.h.
807 (profile_count::to_sreal_scale): New.
808 * profile-count.h: Forward declare sreal.
809 (profile_count::to_sreal_scale): Declare.
811 2017-11-13 Nathan Sidwell <nathan@acm.org>
813 * diagnostic.c (maybe_line_and_column): New.
814 (diagnostic_get_location_text): Use it.
815 (diagnostic_report_current_module): Likewise.
816 (test_diagnostic_get_location_text): Add tests.
818 2017-11-13 Luis Machado <luis.machado@linaro.org>
820 * doc/md.texi (Specifying processor pipeline description): Fix
821 incorrect latency for the div instruction example.
823 2017-11-13 Jakub Jelinek <jakub@redhat.com>
825 PR tree-optimization/78821
826 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
827 that bit_not_p is the same.
828 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
829 (split_group): Count precisely bit_not_p bits in each statement.
830 (invert_op): New function.
831 (imm_store_chain_info::output_merged_store): Use invert_op to
832 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
833 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
835 2017-11-13 Martin Liska <mliska@suse.cz>
837 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
838 (struct source_info): Likewise.
839 (add_branch_counts): Likewise.
840 (add_line_counts): Likewise.
841 (function_summary): Likewise.
842 (output_intermediate_line): Likewise.
843 (generate_results): Likewise.
845 2017-11-13 Martin Liska <mliska@suse.cz>
847 * gcov.c (struct block_info): Remove typedef for block_t.
848 (struct line_info): Likewise.
849 (line_info::has_block): Likewise.
850 (EXIT_BLOCK): Likewise.
853 (get_cycles_count): Likewise.
854 (process_file): Likewise.
855 (read_graph_file): Likewise.
856 (solve_flow_graph): Likewise.
857 (find_exception_blocks): Likewise.
858 (add_line_counts): Likewise.
859 (accumulate_line_info): Likewise.
860 (output_line_details): Likewise.
862 2017-11-13 Martin Liska <mliska@suse.cz>
864 * gcov.c (struct arc_info): Remove typedef for arc_t.
865 (struct line_info): Likewise.
866 (add_branch_counts): Likewise.
867 (output_branch_count): Likewise.
868 (function_info::~function_info): Likewise.
870 (output_intermediate_line): Likewise.
871 (read_graph_file): Likewise.
872 (solve_flow_graph): Likewise.
873 (find_exception_blocks): Likewise.
874 (add_line_counts): Likewise.
875 (accumulate_line_info): Likewise.
876 (output_line_details): Likewise.
877 (output_function_details): Likewise.
879 2017-11-13 Martin Liska <mliska@suse.cz>
881 * gcov.c (struct function_info): Remove typedef for function_t.
882 (struct source_info): Likewise.
883 (source_info::get_functions_at_location): Likewise.
884 (solve_flow_graph): Likewise.
885 (find_exception_blocks): Likewise.
886 (add_line_counts): Likewise.
887 (output_intermediate_file): Likewise.
888 (process_file): Likewise.
889 (generate_results): Likewise.
890 (release_structures): Likewise.
891 (read_graph_file): Likewise.
892 (read_count_file): Likewise.
893 (accumulate_line_counts): Likewise.
894 (output_lines): Likewise.
896 2017-11-13 Martin Liska <mliska@suse.cz>
898 * gcov.c (function_info::function_info): Remove num_counts
899 and add vector<gcov_type>.
900 (function_info::~function_info): Use the vector.
901 (process_file): Likewise.
902 (read_graph_file): Likewise.
903 (read_count_file): Likewise.
904 (solve_flow_graph): Likewise.
906 2017-11-13 Martin Liska <mliska@suse.cz>
908 * gcov.c (function_info::is_artificial): New function.
909 (process_file): Erase all artificial early.
910 (generate_results): Skip as all artificial are already
913 2017-11-13 Martin Liska <mliska@suse.cz>
915 * gcov.c (read_graph_file): Store to global vector of functions.
916 (read_count_file): Iterate the vector.
917 (process_file): Likewise.
918 (generate_results): Likewise.
919 (release_structures): Likewise.
921 2017-11-13 Jakub Jelinek <jakub@redhat.com>
923 PR tree-optimization/82954
924 * gimple-ssa-store-merging.c
925 (imm_store_chain_info::coalesce_immediate_stores): If
926 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
928 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
930 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
931 Upddate call to ENDIAN_LANE_N.
932 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
933 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
934 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
935 and use aarch64_endian_lane_rtx.
936 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
938 2017-11-12 Tom de Vries <tom@codesourcery.com>
940 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
942 2017-11-12 Tom de Vries <tom@codesourcery.com>
944 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
946 2017-11-12 Tom de Vries <tom@codesourcery.com>
948 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
950 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
951 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
952 * defaults.h (ASM_OUTPUT_LABELREF): Same.
954 2017-11-11 Martin Sebor <msebor@redhat.com>
957 * doc/extend.texi (attribute nonstring): Remove spurious argument.
960 * prefic.c (translate_name): Replace strncpy with memcpy to
961 avoid -Wstringop-truncation.
963 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
965 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
967 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
969 * predict.c (maybe_hot_frequency_p): Do not use cfun.
971 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
973 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
976 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
978 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
979 enable generating XXBRH if the value is in a vector register.
980 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
981 value is in a vector register.
982 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
983 register to register bswap64's instead of doing the GPR sequence
984 used on previous machines.
985 (bswapdi2_xxbrd): New insn.
986 (bswapdi2_reg): Disallow on ISA 3.0.
987 (register to register bswap64 splitter): Do not split the insn on
988 ISA 3.0 systems that use XXBRD.
990 2017-11-10 Martin Sebor <msebor@redhat.com>
993 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
995 (find_subframework_file): Same.
997 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
999 * auto-profile.c (afdo_indirect_call): Drop frequency.
1000 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1001 (cgraph_node::create_edge): Drop frequency argument.
1002 (cgraph_node::create_indirect_edge): Drop frequency argument.
1003 (cgraph_edge::make_speculative): Drop frequency arguments.
1004 (cgraph_edge::resolve_speculation): Do not update frequencies
1005 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1006 (cgraph_node::dump): Check consistency in IPA mode.
1007 (cgraph_edge::maybe_hot_p): Use IPA counter.
1008 (cgraph_edge::verify_count_and_frequency): Rename to ...
1009 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1010 (cgraph_node::verify_node): Update.
1011 * cgraph.h (struct cgraph_edge): Drop frequency.
1012 (cgraph_edge::frequency): New function.
1013 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1015 (cgraph_edge::rebuild_edges): Likewise.
1016 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1017 (duplicate_thunk_for_node): Do not pass frequency.
1018 (cgraph_node::create_clone): Scale only counts.
1019 (cgraph_node::create_virtual_clone): Do not pass frequency.
1020 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1021 (cgraph_node::create_version_clone): Do not pass frequency.
1022 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1023 (cgraph_node::expand_thunk): Do not pass frequency.
1024 (cgraph_node::create_wrapper): Do not pass frequency.
1025 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1027 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1028 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1029 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1030 (gather_caller_stats): Use frequency function.
1031 (ipcp_cloning_candidate_p): Use frequency function.
1032 (ipcp_propagate_stage): Use frequency function.
1033 (get_info_about_necessary_edges): Use frequency function.
1034 (update_profiling_info): Update only IPA profile.
1035 (update_specialized_profile): Use frequency functoin.
1036 (perhaps_add_new_callers): Update only IPA profile.
1037 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1038 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1039 (dump_ipa_call_summary): Use frequency function.
1040 (estimate_edge_size_and_time): Use frequency function.
1041 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1042 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1043 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1044 (update_noncloned_counts): ... ths one; scale counts only.
1045 (clone_inlined_nodes): Do not scale frequency.
1046 (inline_call): Do not pass frequency.
1047 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1048 (compute_inlined_call_time): Use IPA profile.
1049 (want_inline_small_function_p): Use IPA profile.
1050 (want_inline_self_recursive_call_p): Use IPA profile.
1051 (edge_badness): Use IPA profile.
1052 (lookup_recursive_calls): Use IPA profile.
1053 (recursive_inlining): Do not pass frequency.
1054 (resolve_noninline_speculation): Do not update frequency.
1055 (inline_small_functions): Collect max of IPA profile.
1056 (dump_overall_stats): Dump IPA porfile.
1057 (dump_inline_stats): Dump IPA porfile.
1058 (ipa_inline): Collect IPA stats.
1059 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1060 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1061 (ipa_propagate_frequency): Use frequency function.
1062 (ipa_profile): Cleanup.
1063 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1064 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1065 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1066 (input_node): Do not stream frequency.
1067 (input_edge): Do not stream frequency.
1068 (merge_profile_summaries): Scale only IPA profiles.
1069 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1070 * predict.c (drop_profile): Do not recompute frequency.
1071 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1072 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1073 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1074 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1075 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1076 * tree-inline.c (copy_bb): Do not scale frequency.
1077 (expand_call_inline): Do not scale frequency.
1078 (tree_function_versioning): Do not scale frequency.
1079 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1081 2017-11-10 Julia Koval <julia.koval@intel.com>
1083 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1084 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1085 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1086 (_mm256_maskz_gf2p8affine_epi64_epi8)
1087 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1088 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1089 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1090 (__builtin_ia32_vgf2p8affineqb_v32qi)
1091 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1092 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1094 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1098 (arm_option_override): Refactor.
1099 (arm_option_reconfigure_globals): New.
1100 (arm_options_perform_arch_sanity_checks): New.
1101 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1103 (arm_options_perform_arch_sanity_checks): Likewise
1105 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1107 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1108 (power9-qpmul): New.
1109 * rs6000/rs6000.md ("type" attr): Add qmul.
1110 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1111 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1112 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1114 2017-11-10 Martin Sebor <msebor@redhat.com>
1117 * builtins.c (compute_objsize): Handle arrays that
1118 compute_builtin_object_size likes to fail for. Make extern.
1119 * builtins.h (compute_objsize): Declare.
1120 (check_strncpy_sizes): New function.
1121 (expand_builtin_strncpy): Call check_strncpy_sizes.
1122 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1123 -Wstringop-truncation.
1124 (gimple_fold_builtin_strncat): Same.
1125 * gimple.c (gimple_build_call_from_tree): Set call location.
1126 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1127 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1128 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1129 (handle_builtin_strlen): Use strlen_to_stridx.
1130 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1132 Use strlen_to_stridx.
1133 (pass_strlen::execute): Release strlen_to_stridx.
1134 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1135 (-Wstringop-truncation): Document new option.
1137 2017-11-10 Martin Liska <mliska@suse.cz>
1139 PR gcov-profile/82702
1140 * gcov.c (main): Handle intermediate files in a different
1142 (get_gcov_intermediate_filename): New function.
1143 (output_gcov_file): Remove support of intermediate files.
1144 (generate_results): Allocate intermediate file.
1145 (release_structures): Clean-up properly fn_end.
1146 (output_intermediate_file): Start iterating with line 1.
1148 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1150 PR tree-optimization/82929
1151 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1152 ops_swapped_p non-static data member.
1153 (store_immediate_info::store_immediate_info): Clear it.
1154 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1155 ops set ops_swapped_p.
1156 (count_multiple_uses): Handle ops_swapped_p.
1158 2017-11-10 Martin Liska <mliska@suse.cz>
1160 * coverage.c (coverage_init): Stream information about
1161 support of has_unexecuted_blocks.
1162 * doc/gcov.texi: Document that.
1163 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1164 * gcov.c (read_graph_file): Likewise.
1165 (output_line_beginning): Fix a small issue with
1168 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1170 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1171 reference of trivial component.
1173 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1176 * gimple-ssa-store-merging.c
1177 (pass_store_merging::terminate_all_aliasing_chains): For
1178 gimple_store_p stmts also call refs_output_dependent_p.
1180 PR rtl-optimization/82913
1181 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1184 2017-11-09 Jeff Law <law@redhat.com>
1186 * vr-values.h: New file with vr_values class.
1187 * tree-vrp.c: Include vr-values.h
1188 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1189 data objects into the vr_values class.
1190 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1191 (get_value_range): Make it a member function within vr_values class.
1192 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1193 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1194 (op_with_constant_singleton_value_range): Likewise.
1195 (extract_range_for_var_from_comparison_expr): Likewise.
1196 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1197 (extract_range_from_binary_expr): Likewise.
1198 (extract_range_from_unary_expr): Likewise.
1199 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1200 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1201 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1202 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1203 (compare_name_with_value, compare_names): Likewise.
1204 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1205 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1206 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1207 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1208 (extract_range_from_phi_node): Likewise.
1209 (simplify_truth_ops_using_ranges): Likewise.
1210 (simplify_div_or_mod_using_ranges): Likewise.
1211 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1212 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1213 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1214 (simplify_float_conversion_using_ranges): Likewise.
1215 (simplify_internal_call_using_ranges): Likewise.
1216 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1217 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1218 poitner via x_vr_values for calls into gimple folder.
1219 (vrp_initialize_lattice): Make this the vr_values ctor.
1220 (vrp_free_lattice): Make this the vr_values dtor.
1221 (set_vr_value): New function.
1222 (class vrp_prop): Add vr_values data member. Add various member
1223 functions as well as member functions that delegate to vr_values.
1224 (check_array_ref): Make a member function within vrp_prop class.
1225 (search_for_addr_array, vrp_initialize): Likewise.
1226 (vrp_finalize): Likewise. Revamp to avoid direct access to
1227 vr_value, values_propagated, etc.
1228 (check_array_bounds): Extract vrp_prop class instance pointer from
1229 walk info structure. Use it to call member functions.
1230 (check_all_array_refs): Make a member function within vrp_prop class.
1231 Smuggle class instance pointer via walk info structure.
1232 (x_vr_values): New local static.
1233 (vrp_valueize): Use x_vr_values to get class instance.
1234 (vr_valueize_1): Likewise.
1235 (class vrp_folder): Add vr_values data member. Add various member
1236 functions as well as member functions that delegate to vr_values.
1237 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1238 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1239 class instance from vr_values. Use it to call member functions.
1240 (vrp_dom_walker): Add vr_values data member.
1241 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1242 instance via x_vr_values.
1243 (identify_jump_threads): Accept vr_values as argument. Store
1244 it into the walker structure.
1245 (evrp_dom_walker): Add vr_values class data member. Add various
1247 (evrp_dom_walker::try_find_new_range): Use vr_values data
1248 member to access the memory allocator.
1249 (evrp_dom_walker::before_dom_children): Store vr_values class
1250 instance into the vrp_folder class.
1251 (evrp_dom_walker::push_value_range): Rework to avoid direct
1252 access to num_vr_values and vr_value.
1253 (evrp_dom_walker::pop_value_range): Likewise.
1254 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1255 Use vr_values to get to dump_all_value_ranges member function.
1256 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1257 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1258 Pass vr_values class instance down to identify_jump_threads.
1259 Remove call to vrp_free_lattice.
1260 (debug_all_value_ranges): Remove.
1262 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1263 (vrp_folder): Likewise.
1265 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1266 Get it from the existing bitmap instead.
1267 (vrp_intersect_ranges_1): Likewise.
1269 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1271 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1273 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1274 set bit_not_p to it.
1275 (imm_store_chain_info::coalesce_immediate_stores): Break group
1276 if bit_not_p is different.
1277 (count_multiple_uses, split_group,
1278 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1279 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1280 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1281 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1282 to store_immediate_info ctor.
1284 2017-11-09 Jim Wilson <jimw@sifive.com>
1286 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1287 (scan_prog_file): Likewise.
1289 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1291 * bb-reorder.c (max_entry_frequency): Remove.
1292 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1293 connect_traces, push_to_next_round_p): Remove prototypes.
1294 (find_traces_1_round): Use counts only.
1295 (push_to_next_round_p): Likewise.
1296 (find_traces): Likewise.
1297 (rotate_loop): Likewise.
1298 (find_traces_1_round): Likewise.
1299 (connect_traces): Likewise.
1300 (edge_order): Likewise.
1302 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1304 * config/arm/arm.c (output_return_instruction): Add comments to
1305 indicate requirement for cmse_nonsecure_entry return to account
1306 for the size of clearing instruction output here.
1307 (thumb_exit): Likewise.
1308 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1309 return in hardfloat mode.
1311 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1313 * config/rs6000/rs6000.c (machine_function): Add a bool,
1314 "toc_is_wrapped_separately".
1315 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1316 if it wasn't explicitly set or unset, we are optimizing for speed, and
1317 doing separate shrink-wrapping.
1318 (rs6000_get_separate_components): Enable the TOC component if
1319 saving the TOC register in the prologue.
1320 (rs6000_components_for_bb): Handle the TOC component.
1321 (rs6000_emit_prologue_components): Store the TOC register where needed.
1322 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1323 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1325 2017-11-09 Martin Jambor <mjambor@suse.cz>
1327 * ipa-param-manipulation.c: New file.
1328 * ipa-param-manipulation.h: Likewise.
1329 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1330 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1331 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1332 (ipa_parm_adjustment): Likewise.
1333 (ipa_parm_adjustment_vec): Likewise.
1334 (ipa_get_vector_of_formal_parms): Moved declaration to
1335 ipa-param-manipulation.h.
1336 (ipa_get_vector_of_formal_parm_types): Likewise.
1337 (ipa_modify_formal_parameters): Likewise.
1338 (ipa_modify_call_arguments): Likewise.
1339 (ipa_combine_adjustments): Likewise.
1340 (ipa_dump_param_adjustments): Likewise.
1341 (ipa_modify_expr): Likewise.
1342 (ipa_get_adjustment_candidate): Likewise.
1343 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1344 ipa-param-manipulation.c.
1345 (ipa_get_vector_of_formal_parm_types): Likewise.
1346 (ipa_modify_formal_parameters): Likewise.
1347 (ipa_modify_call_arguments): Likewise.
1348 (ipa_modify_expr): Likewise.
1349 (get_ssa_base_param): Likewise.
1350 (ipa_get_adjustment_candidate): Likewise.
1351 (index_in_adjustments_multiple_times_p): Likewise.
1352 (ipa_combine_adjustments): Likewise.
1353 (ipa_dump_param_adjustments): Likewise.
1354 * tree-sra.c: Also include ipa-param-manipulation.h
1355 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1358 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1359 Alan Hayward <alan.hayward@arm.com>
1360 David Sherwood <david.sherwood@arm.com>
1362 * doc/sourcebuild.texi (vect_masked_store): Document.
1364 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1365 Alan Hayward <alan.hayward@arm.com>
1366 David Sherwood <david.sherwood@arm.com>
1368 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1370 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1371 Alan Hayward <alan.hayward@arm.com>
1372 David Sherwood <david.sherwood@arm.com>
1374 * doc/sourcebuild.texi (vect_variable_length): Document.
1376 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1377 Alan Hayward <alan.hayward@arm.com>
1378 David Sherwood <david.sherwood@arm.com>
1380 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1382 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1383 Alan Hayward <alan.hayward@arm.com>
1384 David Sherwood <david.sherwood@arm.com>
1386 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1388 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1389 Alan Hayward <alan.hayward@arm.com>
1390 David Sherwood <david.sherwood@arm.com>
1392 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1393 previously undocumented selectors.
1394 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1396 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1398 * doc/rtl.texi (const_vector): Say that elements can be
1399 const_wide_ints too.
1400 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1401 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1402 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1403 * optabs.c (expand_vector_broadcast): Likewise.
1405 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1406 Alan Hayward <alan.hayward@arm.com>
1407 David Sherwood <david.sherwood@arm.com>
1409 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1410 scaled index even if the unscaled address was invalid.
1411 Don't increase the complexity of using a scale in that case.
1413 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1414 Alan Hayward <alan.hayward@arm.com>
1415 David Sherwood <david.sherwood@arm.com>
1417 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1418 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1419 UNITS_PER_WORD bytes.
1420 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1421 based on REGMODE_NATURAL_SIZE of the inner mode.
1422 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1423 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1424 * expmed.c (lowpart_bit_field_p): Divide the value up into
1425 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1426 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1427 whether something is likely to occupy more than one register.
1429 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1432 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1433 profile_count::adjust_for_ipa_scaling.
1434 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1435 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1437 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1439 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1441 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1442 (split_group): Add total_orig and total_new arguments, estimate the
1443 number of statements related to the store group without store merging
1444 and with store merging.
1445 (imm_store_chain_info::output_merged_store): Adjust split_group
1446 callers, punt if estimated number of statements with store merging
1447 is not smaller than estimated number of statements without it.
1449 (handled_load): Remove has_single_use checks.
1450 (pass_store_merging::process_store): Likewise.
1452 2017-11-09 Richard Biener <rguenther@suse.de>
1454 PR tree-optimization/82902
1455 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1457 2017-11-09 Martin Liska <mliska@suse.cz>
1460 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1463 2017-11-09 Martin Liska <mliska@suse.cz>
1465 PR tree-optimization/82669
1466 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1468 2017-11-09 Martin Liska <mliska@suse.cz>
1470 PR gcov-profile/48463
1471 * coverage.c (coverage_begin_function): Output also end locus
1472 of a function and information whether the function is
1474 * gcov-dump.c (tag_function): Parse and print the information.
1475 * gcov.c (INCLUDE_MAP): Add include.
1476 (INCLUDE_SET): Likewise.
1477 (struct line_info): Move earlier in the source file because
1478 of vector<line_info> in function_info structure.
1479 (line_info::line_info): Likewise.
1480 (line_info::has_block): Likewise.
1481 (struct source_info): Add new member index.
1482 (source_info::get_functions_at_location): New function.
1483 (function_info::group_line_p): New function.
1484 (output_intermediate_line): New function.
1485 (output_intermediate_file): Use the mentioned function.
1486 (struct function_start): New.
1487 (struct function_start_pair_hash): Likewise.
1488 (process_file): Add code that identifies group functions.
1489 Assign lines either to global or function scope.
1490 (generate_results): Skip artificial functions.
1491 (find_source): Assign index for each source file.
1492 (read_graph_file): Read new flag artificial and end_line.
1493 (add_line_counts): Assign it either to global of function scope.
1494 (accumulate_line_counts): Isolate core of the function to
1495 accumulate_line_info and call it for both function and global
1497 (accumulate_line_info): New function.
1498 (output_line_beginning): Fix GNU coding style.
1499 (print_source_line): New function.
1500 (output_line_details): Likewise.
1501 (output_function_details): Likewise.
1502 (output_lines): Iterate both source (global) scope and function
1504 (struct function_line_start_cmp): New class.
1505 * doc/gcov.texi: Reflect changes in documentation.
1507 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1510 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1511 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1512 and similarly for not instead of neg.
1514 2017-11-08 Andi Kleen <ak@linux.intel.com>
1516 * config/i386/i386.opt: Add -mforce-indirect-call.
1517 * config/i386/predicates.md: Check for flag_force_indirect_call.
1518 * doc/invoke.texi: Document -mforce-indirect-call
1520 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1522 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1524 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1525 config/riscv/riscv.c (predict.h): New include.
1526 (riscv_slow_unaligned_access_p): No longer static.
1527 (riscv_block_move_straight): Add require.
1528 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1530 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1533 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1534 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1535 nonimmediate_operand predicate for operand 1 instead of
1538 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1540 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1542 * config/aarch64/constraints.md (Uml): New constraint.
1543 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1546 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1548 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1549 of two vec_duplicates into a vec_concat.
1551 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1553 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1554 Simplify vec_merge of vec_duplicate and vec_concat.
1555 * config/aarch64/constraints.md (Utq): New constraint.
1556 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1559 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1561 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1562 Simplify vec_merge of vec_duplicate and const_vector.
1563 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1565 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1566 mode iterator. Update predicate on operand 1 to
1567 handle non-const_vec constants. Delete constraints.
1568 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1570 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1572 PR tree-optimization/78821
1573 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1575 (store_operand_info::store_operand_info): Initialize it to false.
1576 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1577 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1578 store in the group, and if chain_info is non-NULL, to ignore altogether
1580 (compatible_load_p): Fail if bit_not_p does not match.
1581 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1582 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1583 (pass_store_merging::process_store): Adjust
1584 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1585 call terminate_all_aliasing_chains newly when adding a store into
1586 a chain with non-NULL chain_info.
1588 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1590 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1592 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1594 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1596 (aarch64_layout_frame): Initialise emit_frame_chain.
1597 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1598 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1600 2017-11-08 Martin Liska <mliska@suse.cz>
1602 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1603 of gimple_call_internal_p.
1605 2017-11-07 Tom de Vries <tom@codesourcery.com>
1607 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1609 2017-11-07 Tom de Vries <tom@codesourcery.com>
1611 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1614 2017-11-08 Martin Liska <mliska@suse.cz>
1617 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1619 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1621 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1623 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1626 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1627 (*cmp<mode>_ccz_1): New insn with $k alternative.
1630 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1631 RTX_COMM_COMPARE as commutative as well.
1632 (ix86_binary_operator_ok): Formatting fix.
1633 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1634 *<code><mode>3<mask_name><round_saeonly_name>,
1635 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1636 *<s>mul<mode>3_highpart<mask_name>,
1637 *vec_widen_umult_even_v16si<mask_name>,
1638 *vec_widen_umult_even_v8si<mask_name>,
1639 *vec_widen_umult_even_v4si<mask_name>,
1640 *vec_widen_smult_even_v16si<mask_name>,
1641 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1642 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1643 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1644 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1645 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1646 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1647 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1648 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1649 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1650 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1651 ix86_binary_operator_ok. Formatting fixes.
1652 (*<plusminus_insn><mode>3<mask_name><round_name>,
1653 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1656 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1658 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1659 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1660 eq and ne if TARGET_ISEL.
1661 (cmp): New code_iterator.
1662 (UNS, UNSU_, UNSIK): New code_attrs.
1663 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1664 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1666 ("eq<mode>3"): ... this.
1667 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1669 ("ne<mode>3"): ... this.
1671 2017-11-07 Julia Koval <julia.koval@intel.com>
1674 * common/config/i386/i386-common.c
1675 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1676 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1677 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1678 * config/i386/i386.opt: Ditto.
1679 * config/i386/i386.c (ix86_target_string): Ditto.
1680 (ix86_option_override_internal): Ditto.
1681 (ix86_init_mpx_builtins): Move MPX to args2.
1682 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1683 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1684 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1685 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1686 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1687 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1688 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1690 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1693 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1694 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1695 (zero-extendsidi peephole2): Remove peephole.
1697 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1700 * stor-layout.c: Include attribs.h.
1701 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1702 explicit lookup of "aligned" attribute.
1704 2017-11-07 Andrew Waterman <andrew@sifive.com>
1706 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1707 (riscv_expand_block_move): Likewise.
1708 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1710 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1711 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1712 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1713 (riscv_adjust_block_mem): Likewise.
1714 (riscv_block_move_loop): Likewise.
1715 (riscv_expand_block_move): Likewise.
1716 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1718 2017-11-07 Michael Clark <michaeljclark@mac.com>
1720 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1721 (MUSL_DYNAMIC_LINKER): Likewise.
1723 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1725 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1726 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1729 2017-11-07 Richard Biener <rguenther@suse.de>
1731 * match.pd: Fix build.
1733 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1734 Jackson Woodruff <jackson.woodruff@arm.com>
1736 PR tree-optimization/71026
1737 * match.pd: Canonicalize negate in division.
1739 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1742 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1744 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1746 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1747 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1749 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1751 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1752 non-scalar integral types.
1753 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1754 (-(A-B), -(~A)): New transformations.
1756 2017-11-07 Tom de Vries <tom@codesourcery.com>
1758 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1759 semicolon after "do {} while (0)".
1760 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1761 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1762 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1763 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1764 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1766 2017-11-07 Tom de Vries <tom@codesourcery.com>
1768 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1769 after "do {} while (0)".
1770 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1771 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1772 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1773 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1774 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1776 2017-11-07 Tom de Vries <tom@codesourcery.com>
1779 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1781 (arm_rtx_costs_internal): Add missing semicolon after
1782 HANDLE_NARROW_SHIFT_ARITH call.
1784 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1786 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1787 disable isel if it was not set explicitly.
1789 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1791 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1793 (add_type_attribute) likewise.
1795 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1797 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1798 of struct ix86_frame.
1799 (ix86_initial_elimination_offset): Likewise.
1800 (ix86_expand_split_stack_prologue): Likewise.
1802 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1804 * tree-vrp.h (enum value_range_type): Update stale comment.
1806 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1807 Alan Hayward <alan.hayward@arm.com>
1808 David Sherwood <david.sherwood@arm.com>
1810 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1811 (aarch64_expand_vec_perm_const): Take the number of units too.
1812 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1813 (aarch64_expand_vec_perm_const): Likewise.
1814 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1815 (vec_perm<mode>): Update accordingly.
1817 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1818 Alan Hayward <alan.hayward@arm.com>
1819 David Sherwood <david.sherwood@arm.com>
1821 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1822 Take the number of units too.
1823 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1824 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1825 but check for a vector mode before rather than after the call.
1826 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1827 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1828 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1829 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1830 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1831 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1832 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1833 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1834 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1835 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1836 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1837 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1838 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1839 (aarch64_sqdmull2_n<mode>): Update accordingly.
1841 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1842 Alan Hayward <alan.hayward@arm.com>
1843 David Sherwood <david.sherwood@arm.com>
1845 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1846 the number of units too.
1847 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1848 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1849 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1850 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1851 (vec_store_lanesxi<mode>): Update accordingly.
1853 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1854 Alan Hayward <alan.hayward@arm.com>
1855 David Sherwood <david.sherwood@arm.com>
1857 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1858 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1859 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1860 of units rather than the mode.
1861 * config/aarch64/iterators.md (nunits): New mode attribute.
1862 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1863 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1864 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1865 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1866 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1867 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1868 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1869 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1870 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1871 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1872 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1873 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1874 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1875 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1876 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1877 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1878 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1879 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1880 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1881 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1882 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1883 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1884 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1885 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1886 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1887 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1888 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1889 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1890 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1891 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1892 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1893 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1894 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1895 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1896 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1897 (aarch64_simd_vec_setv2di): Likewise.
1899 2017-11-06 Carl Love <cel@us.ibm.com>
1901 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1903 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1904 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1906 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1908 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1909 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1910 Add power 8 macro expansions.
1911 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1912 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1913 power 8 instructions. (VSX_XXBR): Add iterator.
1915 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1917 * config/arm/arm.md (predicable_short_it): Change default to "no",
1918 improve documentation, remove uses that are identical to the default.
1919 (enabled_for_depr_it): Rename to enabled_for_short_it.
1920 * gcc/config/arm/arm-fixed.md (predicable_short_it):
1921 Remove default uses.
1922 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1923 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1924 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1925 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1927 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1930 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1931 float128 helper macros, which are no longer used after deleting
1932 the old 'q' built-in functions, and moving the round to odd
1933 built-in functions to being special built-in functions.
1934 (BU_FLOAT128_2): Likewise.
1935 (BU_FLOAT128_1_HW): Likewise.
1936 (BU_FLOAT128_2_HW): Likewise.
1937 (BU_FLOAT128_3_HW): Likewise.
1938 (FABSQ): Delete old 'q' built-in functions.
1939 (COPYSIGNQ): Likewise.
1940 (SQRTF128_ODD): Move round to odd built-in functions to be
1941 special built-in functions, so that we can handle
1942 -mabi=ieeelongdouble.
1943 (TRUNCF128_ODD): Likewise.
1944 (ADDF128_ODD): Likewise.
1945 (SUBF128_ODD): Likewise.
1946 (MULF128_ODD): Likewise.
1947 (DIVF128_ODD): Likewise.
1948 (FMAF128_ODD): Likewise.
1949 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1950 built-in names to 'f128'.
1951 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1952 old 'q' built-in functions, as the machine independent code for
1953 'f128' built-in functions handles this.
1954 (rs6000_expand_builtin): Add expansion for float128 round to odd
1955 functions, keying off on -mabi=ieeelongdouble of whether to use
1956 the KFmode or TFmode variant.
1957 (rs6000_init_builtins): Initialize the _Float128 round to odd
1959 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1960 _Float128 'q' built-in functions are now mapped into the new
1961 'f128' built-in functions.
1963 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1965 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1966 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1968 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1971 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1972 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1973 * toplev.c: Include "ipa-fnsummary.h".
1974 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1976 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1978 PR tree-optimization/82838
1979 * gimple-ssa-store-merging.c
1980 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1981 on a separate gimple_seq which is then appended to seq.
1983 2017-11-06 Jeff Law <law@redhat.com>
1986 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1987 (get_probe_interval): New functions.
1988 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1989 (ix86_adjust_stack_and_probe): Likewise.
1990 (output_adjust_stack_and_probe): Likewise.
1991 (ix86_emit_probe_stack_range): Likewise.
1992 (ix86_expand_prologue): Likewise.
1994 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1996 PR tree-optimization/82816
1997 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1998 if the modes of the two types are the same.
1999 (convert_plusminus_to_widen): Likewise.
2001 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2003 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2005 (usadv16qi): New define_expand.
2006 (usadv8hi): New define_expand.
2008 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2011 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2013 (inline_transform): Likewise.
2014 * predict.c (counts_to_freqs): Remove useless conditional.
2015 * profile-count.h (profile_count::apply_scale): Move sanity check.
2016 * tree-inline.c (copy_bb): Always scale.
2017 (copy_cfg_body): Likewise.
2019 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2022 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2024 (*cmp_ite0): Add enabled_for_depr_it attribute.
2025 (*cmp_ite1): Likewise.
2027 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2029 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2032 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2034 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2035 return the value by pointer.
2036 (extract_range_from_multiplicative_op_1): Update accordingly.
2037 Return as soon as an operation fails.
2039 2017-11-05 Tom de Vries <tom@codesourcery.com>
2042 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2043 (DEF_SANITIZER_BUILTIN): ... here.
2044 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2045 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2047 2017-11-05 Tom de Vries <tom@codesourcery.com>
2050 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2052 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2053 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2054 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2056 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2057 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2058 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2060 2017-11-05 Tom de Vries <tom@codesourcery.com>
2063 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2066 2017-11-04 Michael Clark <michaeljclark@mac.com>
2068 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2069 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2071 (*addsi3_extended): Likewise.
2072 (*addsi3_extended2): Likewise.
2073 (<optab>si3): Likewise.
2074 (<optab>di3): Likewise.
2075 (<optab><mode>3): Likewise.
2076 (<*optabe>si3_internal): Likewise.
2077 (zero_extendqi<SUPERQI:mode>2): Likewise.
2078 (*add<mode>hi3): Likewise.
2079 (*xor<mode>hi3): Likewise.
2080 (<optab>di3): Likewise.
2081 (*<optab>si3_extend): Likewise.
2082 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2083 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2084 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2086 2017-11-04 Andrew Waterman <andrew@sifive.com>
2088 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2089 TARGET_STRICT_ALIGN based upon -mtune argument.
2091 2017-11-04 Andrew Waterman <andrew@sifive.com>
2093 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2095 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2097 config/i386/i386.c (choose_basereg): Use optional scratch
2098 register and add assertion.
2099 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2100 needed, and don't allocate stack.
2101 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2102 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2103 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2105 2017-11-03 Jeff Law <law@redhat.com>
2107 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2108 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2109 to probe at the start of a noreturn function.
2111 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2113 PR tree-optimization/78821
2114 * gimple-ssa-store-merging.c: Update the file comment.
2115 (MAX_STORE_ALIAS_CHECKS): Define.
2116 (struct store_operand_info): New type.
2117 (store_operand_info::store_operand_info): New constructor.
2118 (struct store_immediate_info): Add rhs_code and ops data members.
2119 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2120 and op1r arguments to the ctor, initialize corresponding data members.
2121 (struct merged_store_group): Add load_align_base and load_align
2123 (merged_store_group::merged_store_group): Initialize them.
2124 (merged_store_group::do_merge): Update them.
2125 (merged_store_group::apply_stores): Pick the constant for
2126 encode_tree_to_bitpos from one of the two operands, or skip
2127 encode_tree_to_bitpos if neither operand is a constant.
2128 (class pass_store_merging): Add process_store method decl. Remove
2129 bool argument from terminate_all_aliasing_chains method decl.
2130 (pass_store_merging::terminate_all_aliasing_chains): Remove
2131 var_offset_p argument and corresponding handling.
2132 (stmts_may_clobber_ref_p): New function.
2133 (compatible_load_p): New function.
2134 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2135 if there is overlap and rhs_code is not INTEGER_CST. For
2136 non-overlapping stores terminate group if rhs is not mergeable.
2137 (get_alias_type_for_stmts): Change first argument from
2138 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2139 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2140 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2142 (get_location_for_stmts): Change first argument from
2143 auto_vec<gimple *> & to vec<gimple *> &.
2144 (struct split_store): Remove orig_stmts data member, add orig_stores.
2145 (split_store::split_store): Create orig_stores rather than orig_stmts.
2146 (find_constituent_stmts): Renamed to ...
2147 (find_constituent_stores): ... this. Change second argument from
2148 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2149 to info structures rather than the statements.
2150 (split_group): Rename ALLOW_UNALIGNED argument to
2151 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2152 it. Adjust find_constituent_stores caller.
2153 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2154 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2155 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2156 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2157 (mem_valid_for_store_merging): New function.
2158 (handled_load): New function.
2159 (pass_store_merging::process_store): New method.
2160 (pass_store_merging::execute): Use process_store method. Adjust
2161 terminate_all_aliasing_chains caller.
2163 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2165 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2166 Return true for more constants, symbols and label references.
2167 (aarch64_valid_floating_const): Remove unused function.
2169 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2172 * config/aarch64/aarch64.c (aarch64_layout_frame):
2173 Undo forcing of LR at bottom of frame.
2175 2017-11-03 Jeff Law <law@redhat.com>
2178 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2179 for int_registers_saved.
2181 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2182 extracted from tree-ssa-dom.c.
2183 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2184 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2185 (record_equivalences_from_incoming_edge): Add additional argument
2186 to single_pred_edge_ignoring_loop_edges call.
2187 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2188 (uncprop_dom_walker::before_dom_children): Add additional argument
2189 to single_pred_edge_ignoring_loop_edges call.
2190 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2191 single_pred_edge_ignoring_loop_edges rather than open coding.
2192 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2194 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2196 * match.pd (-(-A)): Rewrite.
2198 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2200 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2201 (rs6000_emit_int_cmove): New declaration.
2202 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2203 (rs6000_emit_sISEL): Delete.
2204 (rs6000_emit_int_cmove): Make non-static.
2205 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2206 instead of rs6000_emit_sISEL.
2208 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2210 * asan.c (create_cond_insert_point): Maintain profile.
2211 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2213 * basic-block.h (struct basic_block_def): Remove frequency.
2214 (EDGE_FREQUENCY): Use to_frequency
2215 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2217 (find_traces): Update to use to_frequency.
2218 (find_traces_1_round): Likewise; use only IPA counts.
2219 (bb_to_key): Likewise.
2220 (connect_traces): Use IPA counts only.
2221 (copy_bb_p): Update to use to_frequency.
2222 (fix_up_crossing_landing_pad): Likewise.
2223 (sanitize_hot_paths): Likewise.
2224 * bt-load.c (basic_block_freq): Likewise.
2225 * cfg.c (init_flow): Set count_max to uninitialized.
2226 (check_bb_profile): Remove frequencies; check counts.
2227 (dump_bb_info): Do not dump frequencies.
2228 (update_bb_profile_for_threading): Update counts only.
2229 (scale_bbs_frequencies_int): Likewise.
2230 (MAX_SAFE_MULTIPLIER): Remove.
2231 (scale_bbs_frequencies_gcov_type): Update counts only.
2232 (scale_bbs_frequencies_profile_count): Update counts only.
2233 (scale_bbs_frequencies): Update counts only.
2234 * cfg.h (struct control_flow_graph): Add count-max.
2235 (update_bb_profile_for_threading): Update prototype.
2236 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2237 (find_many_sub_basic_blocks): Likewise.
2238 * cfgcleanup.c (try_forward_edges): Likewise.
2239 (try_crossjump_to_edge): Likewise.
2240 * cfgexpand.c (expand_gimple_cond): Likewise.
2241 (expand_gimple_tailcall): Likewise.
2242 (construct_init_block): Likewise.
2243 (construct_exit_block): Likewise.
2244 * cfghooks.c (verify_flow_info): Check consistency of counts.
2245 (dump_bb_for_graph): Do not dump frequencies.
2246 (split_block_1): Do not update frequencies.
2247 (split_edge): Do not update frequencies.
2248 (make_forwarder_block): Do not update frequencies.
2249 (duplicate_block): Do not update frequencies.
2250 (account_profile_record): Do not update frequencies.
2251 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2252 for global heuristics.
2253 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2254 (expected_loop_iterations_unbounded): Use counts only.
2255 * cfgloopmanip.c (scale_loop_profile): Simplify.
2256 (create_empty_loop_on_edge): Simplify
2258 (duplicate_loop_to_header_edge): Simplify
2259 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2260 (update_br_prob_note): Take care of removing note when profile
2262 (relink_block_chain): Do not dump frequency.
2263 (rtl_account_profile_record): Use to_frequency.
2264 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2265 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2266 (cgraph_update_edges_for_call_stmt_node): Likewise.
2267 (cgraph_edge::verify_count_and_frequency): Update.
2268 (cgraph_node::verify_node): Temporarily disable frequency verification.
2269 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2270 to_cgraph_frequency.
2271 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2272 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2274 (cgraph_node::expand_thunk): Update profile.
2275 * except.c (dw2_build_landing_pads): Do not update frequency.
2276 * final.c (compute_alignments): Use to_frequency.
2277 (dump_basic_block_info): Do not dump frequency.
2278 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2279 (dump_gimple_bb_header): Do not dump frequency.
2280 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2282 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2283 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2284 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2285 (init_before_recovery): Do not update frequency.
2286 (sched_create_recovery_edges): Do not update frequency.
2287 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2288 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2289 (ipa_cp_c_finalize): Set max_count to uninitialized.
2290 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2291 (param_change_prob): Use counts.
2292 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2294 * ipa-split.c (consider_split): Use to_frequency.
2295 (split_function): Use to_frequency.
2296 * ira-build.c (loop_compare_func): Likewise.
2297 (mark_loops_for_removal): Likewise.
2298 (mark_all_loops_for_removal): Likewise.
2299 * loop-doloop.c (doloop_modify): Do not update frequency.
2300 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2302 * lto-streamer-in.c (input_function): Update count_max.
2303 * omp-expand.c (expand_omp_taskreg): Update count_max.
2304 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2305 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2306 (maybe_hot_count_p): Use ipa counts only.
2307 (maybe_hot_bb_p): Simplify.
2308 (maybe_hot_edge_p): Simplify.
2309 (probably_never_executed): Do not take frequency argument.
2310 (probably_never_executed_bb_p): Do not pass frequency.
2311 (probably_never_executed_edge_p): Likewise.
2312 (combine_predictions_for_bb): Check that profile is nonzero.
2313 (propagate_freq): Do not set frequency.
2314 (drop_profile): Simplify.
2315 (counts_to_freqs): Simplify.
2316 (expensive_function_p): Use to_frequency.
2317 (propagate_unlikely_bbs_forward): Simplify.
2318 (determine_unlikely_bbs): Simplify.
2319 (estimate_bb_frequencies): Add hack to silence graphite issues.
2320 (compute_function_frequency): Use ipa counts.
2321 (pass_profile::execute): Update.
2322 (rebuild_frequencies): Use counts only.
2323 (force_edge_cold): Use counts only.
2324 * profile-count.c (profile_count::dump): Dump new count types.
2325 (profile_count::differs_from_p): Check compatiblity.
2326 (profile_count::to_frequency): New function.
2327 (profile_count::to_cgraph_frequency): New function.
2328 * profile-count.h (struct function): Declare.
2329 (enum profile_quality): Add profile_guessed_local and
2330 profile_guessed_global0.
2331 (class profile_proability): Decrease number of bits to 29;
2332 update from_reg_br_prob_note and to_reg_br_prob_note.
2333 (class profile_count: Update comment; decrease number of bits
2334 to 61. Check compatibility.
2335 (profile_count::compatible_p): New private member function.
2336 (profile_count::ipa_p): New member function.
2337 (profile_count::operator<): Handle global zero correctly.
2338 (profile_count::operator>): Handle global zero correctly.
2339 (profile_count::operator<=): Handle global zero correctly.
2340 (profile_count::operator>=): Handle global zero correctly.
2341 (profile_count::nonzero_p): New member function.
2342 (profile_count::force_nonzero): New member function.
2343 (profile_count::max): New member function.
2344 (profile_count::apply_scale): Handle IPA scalling.
2345 (profile_count::guessed_local): New member function.
2346 (profile_count::global0): New member function.
2347 (profile_count::ipa): New member function.
2348 (profile_count::to_frequency): Declare.
2349 (profile_count::to_cgraph_frequency): Declare.
2350 * profile.c (OVERLAP_BASE): Delete.
2351 (compute_frequency_overlap): Delete.
2352 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2353 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2354 * sched-ebb.c (rank): Use counts only.
2355 * shrink-wrap.c (handle_simple_exit): Use counts only.
2356 (try_shrink_wrapping): Use counts only.
2357 (place_prologue_for_one_component): Use counts only.
2358 * tracer.c (find_best_predecessor): Use to_frequency.
2359 (find_trace): Use to_frequency.
2360 (tail_duplicate): Use to_frequency.
2361 * trans-mem.c (expand_transaction): Do not update frequency.
2362 * tree-call-cdce.c: Do not update frequency.
2363 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2364 (gimple_merge_blocks): Likewise.
2365 (gimple_split_edge): Likewise.
2366 (gimple_duplicate_sese_region): Likewise.
2367 (gimple_duplicate_sese_tail): Likewise.
2368 (move_sese_region_to_fn): Likewise.
2369 (gimple_account_profile_record): Likewise.
2370 (insert_cond_bb): Likewise.
2371 * tree-complex.c (expand_complex_div_wide): Likewise.
2372 * tree-eh.c (lower_resx): Update profile.
2373 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2375 (initialize_cfun): Do not initialize frequencies
2376 (freqs_to_counts): Delete.
2377 (copy_cfg_body): Ignore count parameter.
2378 (copy_body): Update.
2379 (expand_call_inline): Update count_max.
2380 (optimize_inline_calls): Update count_max.
2381 (tree_function_versioning): Update count_max.
2382 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2383 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2385 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2386 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2387 (try_peel_loop): Likewise.
2388 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2390 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2391 (tree_transform_and_unroll_loop): Do not use frequencies
2392 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2393 Use reliable prediction only.
2394 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2395 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2396 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2397 probability scaling.
2398 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2399 not update frequency
2400 (any_remaining_duplicated_blocks): Likewise.
2401 (update_profile): Likewise.
2402 (estimated_freqs_path): Delete.
2403 (freqs_to_counts_path): Delete.
2404 (clear_counts_path): Delete.
2405 (ssa_fix_duplicate_block_edges): Likewise.
2406 (duplicate_thread_path): Likewise.
2407 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2408 * tree-tailcall.c (decrease_profile): Do not update frequency.
2409 (eliminate_tail_call): Likewise.
2410 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2411 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2412 (optimize_mask_stores): Likewise.
2413 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2414 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2415 (ubsan_expand_ptr_ifn): Update profile.
2416 * value-prof.c (gimple_ic): Simplify.
2417 * value-prof.h (gimple_ic): Update prototype.
2418 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2419 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2421 (want_inline_self_recursive_call_p): Likewise.
2422 (resolve_noninline_speculation): Only cummulate defined counts.
2423 (inline_small_functions): Use nonzero_p.
2424 (ipa_inline): Do not access freed node.
2426 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2428 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2429 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2431 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2433 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2434 non-legitimate address.
2436 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2438 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2439 (*lt0_<mode>di, *lt0_<mode>si): New.
2441 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2443 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2444 TARGET_PAIRED_FLOAT.
2446 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2447 Jim Wilson <jim.wilson@linaro.org>
2449 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2450 * config/aarch64/aarch64-tune.md: Regenerated.
2451 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2452 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2454 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2456 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2457 (arc_expand_prologue): Restore blink for millicode.
2458 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2460 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2463 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2464 gen_vec_duplicate after forcing the scalar into a register.
2466 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2468 * combine (try_combine): Print the insns input to try_combine to the
2471 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2474 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2475 Remove second argument from aarch64_process_target_attr call.
2476 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2478 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2480 (aarch64_handle_attr_arch): Remove second argument.
2481 (aarch64_handle_attr_cpu): Ditto.
2482 (aarch64_handle_attr_tune): Ditto.
2483 (aarch64_handle_attr_isa_flags): Ditto.
2484 (aarch64_process_one_target_attr): Ditto.
2485 (aarch64_process_target_attr): Ditto.
2486 (aarch64_option_valid_attribute_p): Remove second argument.
2487 on aarch64_process_target_attr call.
2489 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2491 * diagnostic.c: Include "selftest-diagnostic.h".
2492 (selftest::assert_location_text): New function.
2493 (selftest::test_diagnostic_get_location_text): New function.
2494 (selftest::diagnostic_c_tests): Call it.
2496 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2498 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2499 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2500 (class selftest::test_diagnostic_context): Move to...
2501 * selftest-diagnostic.c: New file.
2502 * selftest-diagnostic.h: New file.
2504 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2506 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2507 offset range for FT32B.
2508 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2509 * config/ft32/ft32.md: Add TARGET_NOPM.
2510 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2511 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2513 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2515 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2517 2017-11-02 Jeff Law <law@redhat.com>
2519 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2520 virtual keyword on FINAL OVERRIDE members.
2522 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2523 virtuals together. Add virtual destructor.
2524 (substitute_and_fold_engine): Similarly.
2526 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2528 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2530 2017-11-02 Richard Biener <rguenther@suse.de>
2532 PR tree-optimization/82795
2533 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2535 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2537 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2539 (gcc_GAS_CHECK_FEATURE): Remove.
2540 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2541 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2542 * configure: Regenerate.
2544 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2546 * config/arc/arc.c (hwloop_optimize): Account for empty
2549 2017-11-02 Richard Biener <rguenther@suse.de>
2552 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2553 Truncate ARRAY_REF index and element size.
2555 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2557 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2559 2017-11-01 Jeff Law <law@redhat.com>
2561 * tree-ssa-ccp.c (ccp_folder): New class derived from
2562 substitute_and_fold_engine.
2563 (ccp_folder::get_value): New member function.
2564 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2565 (ccp_fold_stmt): Remove prototype.
2566 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2567 * tree-ssa-copy.c (copy_folder): New class derived from
2568 substitute_and_fold_engine.
2569 (copy_folder::get_value): Renamed from get_value.
2570 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2571 * tree-vrp.c (vrp_folder): New class derived from
2572 substitute_and_fold_engine.
2573 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2574 (vrp_folder::get_value): New member function.
2575 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2576 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2577 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2578 provide a class interface to folder/substitute routines.
2579 (ssa_prop_fold_stmt_fn): Remove typedef.
2580 (ssa_prop_get_value_fn): Likewise.
2581 (subsitute_and_fold): Remove prototype.
2582 (replace_uses_in): Likewise.
2583 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2584 Renamed from replace_uses_in. Call the virtual member function
2585 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2586 (substitute_and_fold_dom_walker): Remove initialization of
2587 data member entries for calbacks. Add substitute_and_fold_engine
2588 member and initialize it.
2589 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2590 member functions for get_value, replace_phi_args_in c
2591 replace_uses_in, and fold_stmt calls.
2592 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2593 substitute_and_fold. Remove assert. Update ctor call.
2595 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2596 (ssa_prop_visit_phi_fn): Likewise.
2597 (class ssa_propagation_engine): New class to provide an interface
2599 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2601 (ssa_prop_visit_phi): Likewise.
2602 (ssa_propagation_engine::simulate_stmt): Moved into class.
2603 Call visit_phi/visit_stmt from the class rather than via
2604 file scoped static variables.
2605 (ssa_propagation_engine::simulate_block): Moved into class.
2606 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2607 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2608 set file scoped statics for the visit_stmt/visit_phi callbacks.
2609 * tree-complex.c (complex_propagate): New class derived from
2610 ssa_propagation_engine.
2611 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2612 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2613 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2615 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2616 ssa_propagation_engine.
2617 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2618 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2619 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2620 * tree-ssa-copy.c (copy_prop): New class derived from
2621 ssa_propagation_engine.
2622 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2623 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2624 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2625 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2626 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2627 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2628 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2630 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2632 PR rtl-optimization/82778
2633 PR rtl-optimization/82597
2634 * compare-elim.c (struct comparison): Add in_a_setter field.
2635 (find_comparison_dom_walker::before_dom_children): Remove killed
2636 bitmap and df_simulate_find_defs call, instead walk the defs.
2637 Compute last_setter and initialize in_a_setter. Merge definitions
2638 with first initialization for a few variables.
2639 (try_validate_parallel): Use insn_invalid_p instead of
2640 recog_memoized. Return insn rather than just the pattern.
2641 (try_merge_compare): Fix up comment. Don't uselessly test if
2642 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2644 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2647 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2648 Alan Hayward <alan.hayward@arm.com>
2649 David Sherwood <david.sherwood@arm.com>
2651 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2652 aarch64_hard_regno_nregs to get the number of registers
2655 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2656 Alan Hayward <alan.hayward@arm.com>
2657 David Sherwood <david.sherwood@arm.com>
2659 * config/aarch64/constraints.md (Upl): Rename to...
2661 * config/aarch64/aarch64.md
2662 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2665 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2666 Alan Hayward <alan.hayward@arm.com>
2667 David Sherwood <david.sherwood@arm.com>
2669 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2670 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2673 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2674 Alan Hayward <alan.hayward@arm.com>
2675 David Sherwood <david.sherwood@arm.com>
2677 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2678 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2679 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2681 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2682 of a single element.
2683 * config/aarch64/iterators.md: Add a comment above the permute
2684 unspecs to say that they are generated directly by
2685 aarch64_expand_vec_perm_const.
2686 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2688 2017-11-01 Nathan Sidwell <nathan@acm.org>
2690 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2692 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2694 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2695 and medany code models, and describe what they do.
2697 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2699 Revert accidental duplicate:
2701 * combine.c (can_change_dest_mode): Reject changes in
2702 REGMODE_NATURAL_SIZE.
2704 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2706 PR rtl-optimization/64682
2707 PR rtl-optimization/69567
2708 PR rtl-optimization/69737
2709 PR rtl-optimization/82683
2710 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2711 register mentioned in the note, drop the note, unless it came from I3,
2712 in which case it should go to I3 again.
2714 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2716 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2717 and return false if not.
2718 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2720 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2722 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2723 range is known to be empty.
2725 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2726 Alan Hayward <alan.hayward@arm.com>
2727 David Sherwood <david.sherwood@arm.com>
2729 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2730 and CONST_VECTOR_NUNITS instead of computing the number of units from
2731 the byte sizes of the vector and element.
2732 (simplify_binary_operation_1): Likewise.
2733 (simplify_const_binary_operation): Likewise.
2734 (simplify_ternary_operation): Likewise.
2736 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2737 Alan Hayward <alan.hayward@arm.com>
2738 David Sherwood <david.sherwood@arm.com>
2740 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2741 (int_mem_offset): ...this new function.
2742 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2743 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2744 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
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 * lower-subreg.c (interesting_mode_p): New function.
2752 (compute_costs, find_decomposable_subregs, decompose_register)
2753 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2754 (resolve_clobber, dump_choices): Use it.
2756 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2757 Alan Hayward <alan.hayward@arm.com>
2758 David Sherwood <david.sherwood@arm.com>
2760 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2762 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2763 Alan Hayward <alan.hayward@arm.com>
2764 David Sherwood <david.sherwood@arm.com>
2766 * alias.c (find_base_value, find_base_term): Only process integer
2767 truncations. Check the precision rather than the size.
2769 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2770 Alan Hayward <alan.hayward@arm.com>
2771 David Sherwood <david.sherwood@arm.com>
2773 * machmode.h (is_narrower_int_mode): New function
2774 * optabs.c (expand_float, expand_fix): Use it.
2775 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2777 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2778 Alan Hayward <alan.hayward@arm.com>
2779 David Sherwood <david.sherwood@arm.com>
2781 * rtl.h (narrower_subreg_mode): New function.
2782 * ira-color.c (update_costs_from_allocno): Use it.
2784 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2785 Alan Hayward <alan.hayward@arm.com>
2786 David Sherwood <david.sherwood@arm.com>
2788 * optabs-query.h (convert_optab_p): New function, split out from...
2789 (convert_optab_handler): ...here.
2790 (widening_optab_handler): Delete.
2791 (find_widening_optab_handler): Remove permit_non_widening parameter.
2792 (find_widening_optab_handler_and_mode): Likewise. Provide an
2793 override that operates on mode class wrappers.
2794 * optabs-query.c (widening_optab_handler): Delete.
2795 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2796 parameter. Assert that the two modes are the same class and that
2797 the "from" mode is narrower than the "to" mode. Use
2798 convert_optab_handler instead of widening_optab_handler.
2799 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2800 instead of widening_optab_handler.
2801 * expr.c (expand_expr_real_2): Update calls to
2802 find_widening_optab_handler.
2803 * optabs.c (expand_widen_pattern_expr): Likewise.
2804 (expand_binop_directly): Take the insn_code as a parameter.
2805 (expand_binop): Only call find_widening_optab_handler for
2806 conversion optabs; use optab_handler otherwise. Update calls
2807 to find_widening_optab_handler and expand_binop_directly.
2808 Use convert_optab_handler instead of widening_optab_handler.
2809 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2810 find_widening_optab_handler and use scalar_mode rather than
2812 (convert_plusminus_to_widen): Likewise.
2814 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2815 Alan Hayward <alan.hayward@arm.com>
2816 David Sherwood <david.sherwood@arm.com>
2818 * machmode.h (fixed_size_mode): New class.
2819 * rtl.h (get_pool_mode): Return fixed_size_mode.
2820 * gengtype.c (main): Add fixed_size_mode.
2821 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2822 (get_raw_arg_mode): Likewise.
2823 * doc/tm.texi: Regenerate.
2824 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2825 * targhooks.c (default_get_reg_raw_mode): Likewise.
2826 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2827 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2828 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2829 (msp430_get_raw_result_mode): Likewise.
2830 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2831 * dbxout.c (dbxout_parms): Require fixed-size modes.
2832 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2833 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2834 * omp-low.c (lower_oacc_reductions): Likewise.
2835 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2836 (simplify_subreg): Update accordingly.
2837 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2838 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2839 that aren't fixed-size.
2840 (get_pool_mode): Return a fixed_size_mode.
2841 (output_constant_pool_2): Take a fixed_size_mode.
2843 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2844 Alan Hayward <alan.hayward@arm.com>
2845 David Sherwood <david.sherwood@arm.com>
2847 * doc/rtl.texi (vec_series): Document.
2848 (const): Say that the operand can be a vec_series.
2849 * rtl.def (VEC_SERIES): New rtx code.
2850 * rtl.h (const_vec_series_p_1): Declare.
2851 (const_vec_series_p): New function.
2852 * emit-rtl.h (gen_const_vec_series): Declare.
2853 (gen_vec_series): Likewise.
2854 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2855 (gen_vec_series): Likewise.
2856 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2857 * simplify-rtx.c (simplify_unary_operation): Handle negations
2859 (simplify_binary_operation_series): New function.
2860 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2861 (test_vector_ops_series): New function.
2862 (test_vector_ops): Call it.
2863 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2864 gen_const_vec_series.
2865 (altivec_lvsr): Likewise.
2866 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2868 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2869 Alan Hayward <alan.hayward@arm.com>
2870 David Sherwood <david.sherwood@arm.com>
2872 * doc/rtl.texi (const): Update description of address constants.
2873 Say that vector constants are allowed too.
2874 * common.md (E, F): Use CONSTANT_P instead of checking for
2876 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2877 checking for CONST_VECTOR.
2878 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2880 * expr.c (expand_expr_real_2): Check for vector modes instead
2881 of checking for CONST_VECTOR.
2882 * rtl.h (const_vec_p): New function.
2883 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2884 (unwrap_const_vec_duplicate): Handle them here too.
2886 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2887 David Malcolm <dmalcolm@redhat.com>
2888 Alan Hayward <alan.hayward@arm.com>
2889 David Sherwood <david.sherwood@arm.com>
2891 * rtl.h (vec_duplicate_p): New function.
2892 * selftest-rtl.c (assert_rtx_eq_at): New function.
2893 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2894 (assert_rtx_eq_at): Declare.
2895 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2896 * selftest-run-tests.c (selftest::run_tests): Call it.
2897 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2898 (simplify_unary_operation_1): Recursively handle vector duplicates.
2899 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2901 (simplify_subreg): Handle subregs of vector duplicates.
2902 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2903 (selftest::simplify_rtx_c_tests): New functions.
2905 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2906 Alan Hayward <alan.hayward@arm.com>
2907 David Sherwood <david.sherwood@arm.com>
2909 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2910 (gen_vec_duplicate): Likewise.
2911 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2913 (gen_const_vector): ...here.
2914 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2915 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2916 whose elements are all equal.
2917 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2918 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2919 (simplify_relational_operation): Likewise.
2920 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2922 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2923 (aarch64_expand_vector_init): Likewise.
2924 * config/arm/arm.c (neon_vdup_constant): Likewise.
2925 (neon_expand_vector_init): Likewise.
2926 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2927 (arm_block_set_unaligned_vect): Likewise.
2928 (arm_block_set_aligned_vect): Likewise.
2929 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2930 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2931 (expand_vec_perm_even_odd_pack): Likewise.
2932 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2933 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2934 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2935 gen_const_vec_duplicate.
2936 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2937 * config/mips/mips.c (mips_gen_const_int_vector): Use
2938 gen_const_vec_duplicate.
2939 (mips_expand_vector_init): Use CONST0_RTX.
2940 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2941 (define_split): Use gen_const_vec_duplicate.
2942 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2943 (define_split): Use gen_const_vec_duplicate.
2944 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2945 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2946 * config/spu/spu.c (spu_const): Likewise.
2948 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2949 Alan Hayward <alan.hayward@arm.com>
2950 David Sherwood <david.sherwood@arm.com>
2952 * combine.c (can_change_dest_mode): Reject changes in
2953 REGMODE_NATURAL_SIZE.
2955 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2957 * configure.ac (--enable-libssp): New.
2958 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2959 trying to determine target-specific default. Adjust indentation.
2960 * configure: Regenerated.
2961 * doc/install.texi (Configuration): Expand --disable-libssp
2964 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2966 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2969 2017-10-31 Martin Jambor <mjambor@suse.cz>
2972 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2974 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2976 * auto-profile.c (autofdo_source_profile::read): Use
2977 UNKNOWN_LOCATION rather than 0.
2978 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2979 (warning_at): ...this overload.
2980 (warning_at_rich_loc_n): Rename to...
2981 (warning_n): ...this overload.
2982 (error_at_rich_loc): Rename to...
2983 (error_at): ...this overload.
2984 (pedwarn_at_rich_loc): Rename to...
2985 (pedwarn): ...this overload.
2986 (permerror_at_rich_loc): Rename to...
2987 (permerror): ...this overload.
2988 (inform_at_rich_loc): Rename to...
2989 (inform): ...this overload.
2990 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2991 (diagnostic_n_impl_richloc): Rename to...
2992 (diagnostic_n_impl): ...this rich_location *-based decl.
2993 (inform_at_rich_loc): Rename to...
2994 (inform): ...this, and add an assertion.
2995 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2996 (warning_at_rich_loc): Rename to...
2997 (warning_at): ...this, and add an assertion.
2998 (warning_at_rich_loc_n): Rename to...
2999 (warning_n): ...this, and add an assertion.
3000 (warning_n): Update location_t-based implementation for removal of
3001 location_t-based diagnostic_n_impl.
3002 (pedwarn_at_rich_loc): Rename to...
3003 (pedwarn): ...this, and add an assertion.
3004 (permerror_at_rich_loc): Rename to...
3005 (permerror): ...this, and add an assertion.
3006 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3007 (error_at_rich_loc): Rename to...
3008 (error_at): ...this, and add an assertion.
3009 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3010 (driver::do_spec_on_infiles): Likewise.
3011 * substring-locations.c (format_warning_va): Update for renaming
3012 of inform_at_rich_loc.
3014 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3016 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3017 _Float<N>X built-in functions so that the variant without the
3018 "__builtin_" prefix is only enabled for the GNU C and Objective C
3019 languages when they are in non-strict ANSI/ISO mode.
3020 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3021 * target.def (floatn_builtin_p): Add a target hook to control
3022 whether _Float<N> and _Float<N>X built-in functions without the
3023 "__builtin_" prefix are enabled, and return true for C and
3024 Objective C in the default hook. Include langhooks.h in
3026 * targhooks.h (default_floatn_builtin_p): Likewise.
3027 * targhooks.c (default_floatn_builtin_p): Likewise.
3028 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3029 floatn_builtin_p target hook.
3030 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3032 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3033 Eric Botcazou <ebotcazou@adacore.com>
3035 PR rtl-optimization/81803
3036 * lra-constraints.c (curr_insn_transform): Also reload the whole
3037 register for a strict subreg no wider than a word if this is for
3038 a WORD_REGISTER_OPERATIONS target.
3040 2017-10-31 Jason Merrill <jason@redhat.com>
3042 * gdbinit.in: Skip over inlines from timevar.h.
3044 2017-10-31 Martin Liska <mliska@suse.cz>
3046 * doc/gcov.texi: Document new option.
3047 * gcov.c (print_usage): Likewise print it.
3048 (process_args): Support the argument.
3049 (format_count): New function.
3050 (format_gcov): Use the function.
3052 2017-10-31 Martin Liska <mliska@suse.cz>
3054 * gcov.c (struct name_map): do not use typedef.
3055 Define operator== and operator<.
3056 (name_search): Remove.
3057 (name_sort): Remove.
3058 (main): Do not allocate names.
3059 (process_file): Add vertical space.
3060 (generate_results): Use std::find.
3061 (release_structures): Do not release memory.
3062 (find_source): Use std::find.
3064 2017-10-31 Martin Liska <mliska@suse.cz>
3066 * gcov.c (struct line_info): Remove it's typedef.
3067 (line_info::line_info): Add proper ctor.
3068 (line_info::has_block): Do not use a typedef.
3069 (struct source_info): Do not use typedef.
3070 (circuit): Likewise.
3071 (get_cycles_count): Likewise.
3072 (output_intermediate_file): Iterate via vector iterator.
3073 (add_line_counts): Use std::vector methods.
3074 (accumulate_line_counts): Likewise.
3075 (output_lines): Likewise.
3077 2017-10-31 Martin Liska <mliska@suse.cz>
3079 * gcov.c (struct source_info): Remove typedef.
3080 (source_info::source_info): Add proper ctor.
3081 (accumulate_line_counts): Use struct, not it's typedef.
3082 (output_gcov_file): Likewise.
3083 (output_lines): Likewise.
3084 (main): Do not allocate an array.
3085 (output_intermediate_file): Use size of vector container.
3086 (process_file): Resize the vector.
3087 (generate_results): Do not preallocate, use newly added vector
3089 (release_structures): Do not release sources.
3090 (find_source): Use vector methods.
3091 (add_line_counts): Do not use typedef.
3093 2017-10-31 Martin Liska <mliska@suse.cz>
3095 * doc/gcov.texi: Document that.
3096 * gcov.c (add_line_counts): Mark lines with a non-executed
3098 (output_line_beginning): Handle such lines.
3099 (output_lines): Pass new argument.
3100 (output_intermediate_file): Print it in intermediate format.
3102 2017-10-31 Martin Liska <mliska@suse.cz>
3104 * color-macros.h: New file.
3105 * diagnostic-color.c: Factor out color related to macros to
3107 * doc/gcov.texi: Document -k option.
3108 * gcov.c (INCLUDE_STRING): Include string.h.
3109 (print_usage): Add -k option.
3110 (process_args): Parse it.
3111 (pad_count_string): New function.
3112 (output_line_beginning): Likewise.
3113 (DEFAULT_LINE_START): New macro.
3114 (output_lines): Support color output.
3116 2017-10-31 Martin Liska <mliska@suse.cz>
3118 PR gcov-profile/82633
3119 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3120 their interaction with GCOV infrastructure.
3121 * configure.ac: Add -fkeep-{inline,static}-functions to
3123 * configure: Regenerate.
3125 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3128 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3130 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3133 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3134 into a register if it does not fit into an immediate offset field.
3136 2017-10-31 Olivier Hainque <hainque@adacore.com>
3138 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3140 2017-10-31 Julia Koval <julia.koval@intel.com>
3142 * config.gcc: Add gfniintrin.h.
3143 * config/i386/gfniintrin.h: New.
3144 * config/i386/i386-builtin-types.def
3145 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3146 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3147 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3148 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3149 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3150 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3151 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3152 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3153 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3154 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3155 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3156 * config/i386/immintrin.h: Include gfniintrin.h.
3157 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3159 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3161 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3163 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3165 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3166 (arm_ashldi3_1bit): Remove pattern.
3167 (ashrdi3): Remove shift by 1 expansion.
3168 (arm_ashrdi3_1bit): Remove pattern.
3169 (lshrdi3): Remove shift by 1 expansion.
3170 (arm_lshrdi3_1bit): Remove pattern.
3171 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3172 cost of ashldi3 by 1.
3173 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3174 (<shift>di3_neon): Likewise.
3176 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3178 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3179 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3180 and (*aarch64_simd_mov<VQ:mode>).
3181 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3182 pattern alternative.
3183 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3184 attributes to match pattern alternative.
3186 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3188 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3189 * config/rs6000/emmintrin.h: New file.
3190 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3192 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3194 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3195 * (movdi_vfp_cortexa8): Remove pattern.
3197 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3199 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3201 (Specific, alpha*-dec-osf5.1): Remove.
3202 (Specific, mips-sgi-irix5): Remove.
3203 (Specific, mips-sgi-irix6): Remove.
3205 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3208 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3209 arguments to clear_bit_region_be.
3211 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3213 * gimplify.c: Include memmodel.h.
3215 2017-10-30 Martin Jambor <mjambor@suse.cz>
3217 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3218 condition whether loop should be executed at all.
3220 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3222 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3223 gimple folding of vec_madd() intrinsics.
3224 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3225 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3226 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3228 2017-10-30 Richard Biener <rguenther@suse.de>
3230 PR tree-optimization/82762
3232 2017-10-23 Richard Biener <rguenther@suse.de>
3234 PR tree-optimization/82129
3236 2017-08-01 Richard Biener <rguenther@suse.de>
3238 PR tree-optimization/81181
3239 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3240 (compute_antic): ... end of iteration here.
3242 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3244 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3245 -std=iso9899:2017 and -std=gnu17.
3246 * doc/standards.texi (C Language): Document C17 support.
3247 * doc/cpp.texi (Overview): Mention -std=c17.
3248 (Standard Predefined Macros): Document C11 and C17 values of
3249 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3250 * doc/extend.texi (Inline): Do not list individual options for
3251 standards newer than C99.
3252 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3254 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3257 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3259 * asan.c (asan_finish_file): Align asan globals array by shadow
3262 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3265 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3266 (struct store_immediate_info): Add bitregion_start and bitregion_end
3268 (store_immediate_info::store_immediate_info): Add brs and bre
3269 arguments and initialize bitregion_{start,end} from those.
3270 (struct merged_store_group): Add bitregion_start, bitregion_end,
3271 align_base and mask fields. Drop unnecessary struct keyword from
3272 struct store_immediate_info. Add do_merge method.
3273 (clear_bit_region_be): Use memset instead of loop storing zeros.
3274 (merged_store_group::do_merge): New method.
3275 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3276 stores as long as the surrounding bitregions have no gaps.
3277 (merged_store_group::merge_overlapping): Use do_merge.
3278 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3279 is byte aligned, rather than requiring that start and width are
3280 byte aligned. Drop unnecessary struct keyword from
3281 struct store_immediate_info. Allocate and populate also mask array.
3282 Make start of the arrays relative to bitregion_start rather than
3283 start and size them according to bitregion_{end,start} difference.
3284 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3285 struct store_immediate_info.
3286 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3287 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3288 struct keyword from struct store_immediate_info.
3289 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3290 between stores as long as the surrounding bitregions have no gaps.
3292 (struct split_store): Add orig non-static data member.
3293 (split_store::split_store): Initialize orig to false.
3294 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3295 if there is exactly a single original stmt. Change stmts argument
3296 to pointer from reference, if NULL, don't push anything to it. Add
3297 first argument, use it to optimize skipping over orig stmts that
3298 are known to be before bitpos already. Simplify.
3299 (split_group): Return unsigned int count how many stores are or
3300 would be needed rather than a bool. Add allow_unaligned argument.
3301 Change split_stores argument from reference to pointer, if NULL,
3302 only do a dry run computing how many stores would be produced.
3303 Rewritten algorithm to use both alignment and misalign if
3304 !allow_unaligned and handle bitfield stores with gaps.
3305 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3306 from bitregion_start instead of start. Compute allow_unaligned
3307 here, if true, do 2 split_group dry runs to compute which one
3308 produces fewer stores and prefer aligned if equal. Punt if
3309 new count is bigger or equal than original before emitting any
3310 statements, rather than during that. Remove no longer needed
3311 new_ssa_names tracking. Replace num_stmts with
3312 split_stores.length (). Use 32-bit stack allocated entries
3313 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3314 if possible. Handle bitfields with gaps.
3315 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3316 Compute bitregion_{start,end} for the stores and construct
3317 store_immediate_info with that. Formatting fixes.
3319 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3322 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3323 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3325 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3327 * gimplify.c: Include tm_p.h.
3329 * common.opt (gcoff): Re-add as ignored option.
3330 (gcoff1, gcoff2, gcoff3): Likewise.
3332 * Makefile.in (OBJS): Delete sdbout.o.
3333 (GTFILES): Delete $(srcdir)/sdbout.c.
3334 * debug.h: Delete sdb_debug_hooks.
3335 * final.c: Delete sdbout.h include.
3336 (final_scan_insn): Delete SDB_DEBUG check.
3337 (rest_of_clean_state): Likewise.
3338 * output.h: Delete sdb_begin_function_line.
3341 * toplev.c: Delete sdbout.h include.
3342 (process_options): Delete SDB_DEBUG check.
3343 * tree-core.h (tree_type_common): Delete pointer field of
3345 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3346 TYPE_SYMTAB_POINTER.
3347 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3348 (TYPE_SYMTAB_IS_POINTER): Delete.
3349 (TYPE_SYMTAB_IS_DIE): Renumber.
3350 * xcoffout.c: Refer to former sdbout.c file.
3351 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3353 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3354 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3355 (Debugging Options): Delete -gcoff.
3356 (-gstabs): Delete SDB reference.
3358 (-gcoff@var{level}): Delete.
3359 * doc/passes.texi (Debugging information output): Delete SDB and
3360 sdbout.c references.
3361 * doc/tm.texi: Regenerate.
3362 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3363 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3365 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3366 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3367 (SDB_DEBUGGING_INFO): Delete.
3368 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3369 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3370 * target.def (output_source_filename): Delete COFF reference.
3372 * common.opt (gcoff): Delete.
3373 (gxcoff+): Update Negative chain.
3374 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3376 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3377 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3378 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3380 (expand_function_start): Change sdb reference to past tense.
3381 (expand_function_end): Change sdb reference to past tense.
3382 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3383 * opts.c (debug_type_names): Delete coff entry.
3384 (common_handle_option): Delete OPT_gcoff case.
3385 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3387 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3388 * config/cris/cris.h: Delete SDB reference in comment.
3389 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3390 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3391 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3392 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3394 (ix86_expand_prologue): Likewise.
3395 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3396 * config/ia64/ia64.h: Likewise.
3397 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3398 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3400 * config/mmix/mmix.h: Likewise.
3401 * config/nds32/nds32.c: Likewise.
3402 * config/stormy/storym16.h: Likewise.
3403 * config/visium/visium.h: Likewise.
3404 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3406 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3408 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3409 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3410 FRAME_POINTER_REGNUM point at high end of local var area.
3412 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3414 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3415 Move comment around. Do not reset best_edge for a copiable
3416 destination if the copy would cause a partition change.
3417 (better_edge_p): Remove redundant check.
3419 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3421 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3423 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3425 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3426 for math functions that have _Float<N> and _Float<N>X variants.
3427 (mathfn_built_in_2): Add support for math functions that have
3428 _Float<N> and _Float<N>X variants.
3429 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3430 (expand_builtin_mathfn_ternary): Add support for fma with
3431 _Float<N> and _Float<N>X variants.
3432 (expand_builtin): Likewise.
3433 (fold_builtin_3): Likewise.
3434 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3435 create math function _Float<N> and _Float<N>X variants as external
3437 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3438 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3439 the __builtin_ prefix and if not strict ansi, without the prefix.
3440 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3441 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3442 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3443 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3444 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3445 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3446 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3447 function signatures for fma _Float<N> and _Float<N>X variants.
3448 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3449 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3450 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3451 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3452 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3453 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3454 * gencfn-macros.c (print_case_cfn): Add support for math functions
3455 that have _Float<N> and _Float<N>X variants.
3456 (print_define_operator_list): Likewise.
3457 (fltfn_suffixes): Likewise.
3459 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3460 for math functions that have _Float<N> and _Float<N>X variants.
3461 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3462 and _Float<N>X variants.
3463 (COPYSIGN): Likewise.
3466 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3467 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3469 (integer_valued_read_call_p): Likewise.
3470 * fold-const-call.c (fold_const_call_ss): Likewise.
3471 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3472 _Float<N> and _Float<N>X variants.
3473 (fold_const_call_ssss): Add support for fma _Float<N> and
3474 _Float<N>X variants.
3475 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3476 support for copysign and fma _Float<N> and _Float<N>X variants.
3477 (backprop::process_builtin_call_use): Likewise.
3478 * tree-call-cdce.c (can_test_argument_range); Add support for
3479 sqrt _Float<N> and _Float<N>X variants.
3480 (edom_only_function): Likewise.
3481 (get_no_error_domain): Likewise.
3482 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3483 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3484 copysign _Float<N> and _Float<N>X variants.
3485 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3486 handled by machine independent code.
3487 (FMAF128): Likewise.
3488 * doc/cpp.texi (Common Predefined Macros): Document defining
3489 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3490 fma _Float<N> and _Float<N>X variants.
3492 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3495 * config/i386/i386-modes.def (CCFPU): Remove definition.
3496 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3497 (ix86_cc_modes_compatible): Ditto.
3498 (ix86_expand_carry_flag_compare): Ditto.
3499 (ix86_expand_int_movcc): Ditto.
3500 (ix86_expand_int_addcc): Ditto.
3501 (ix86_reverse_condition): Ditto.
3502 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3503 Return true/false for unordered/ordered fp comparisons.
3504 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3505 (ix86_prepare_fp_compare_args): Update for rename.
3506 (ix86_expand_fp_compare): Update for rename. Generate unordered
3507 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3508 (ix86_expand_sse_compare_and_jump): Ditto.
3509 * config/i386/predicates.md (fcmov_comparison_operator):
3510 Remove CCFPU mode handling.
3511 (ix86_comparison_operator): Ditto.
3512 (ix86_carry_flag_operator): Ditto.
3513 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3514 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3515 (*cmpu<mode>_cc_i387): Ditto.
3516 (FPCMP): Remove mode iterator.
3517 (unord): Remove mode attribute.
3518 (unord_subst): New define_subst transformation
3519 (unord): New define_subst attribute.
3521 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3522 (*cmpi<unord>xf_i387): Ditto.
3523 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3524 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3525 using unord_subst transformation.
3526 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3527 (round_saeonly): Also handle CCFP mode.
3528 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3529 Remove UNSPEC_SAHF unspec handling.
3531 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3533 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3535 2017-10-27 Jeff Law <law@redhat.com>
3537 * gimple-ssa-sprintf.c: Include domwalk.h.
3538 (class sprintf_dom_walker): New class, derived from dom_walker.
3539 (sprintf_dom_walker::before_dom_children): New function.
3540 (struct call_info): Moved into sprintf_dom_walker class
3541 (compute_formath_length, handle_gimple_call): Likewise.
3542 (sprintf_length::execute): Call the dominator walker rather
3543 than walking the statements.
3545 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3546 gimple statement locations.
3547 (check_array_bounds): Corresponding changes. Get the statement's
3548 location directly from wi->stmt.
3550 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3553 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3555 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3557 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3558 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3560 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3563 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3564 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3565 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3566 maybe_get_pool_constant.
3567 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3570 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3572 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3573 2.26 caveat. Update gas and gld versions.
3574 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3577 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3579 * cgraph.h (set_malloc_flag): Declare.
3580 * cgraph.c (set_malloc_flag_1): New function.
3581 (set_malloc_flag): Likewise.
3582 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3583 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3585 (read_ipa_call_summary): Add support for reading is_return_callee.
3586 (write_ipa_call_summary): Stream is_return_callee.
3587 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3588 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3589 ipa-prop.h, ipa-fnsummary.h.
3590 (pure_const_names): Change to static.
3591 (malloc_state_e): Define.
3592 (malloc_state_names): Define.
3593 (funct_state_d): Add field malloc_state.
3594 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3595 (check_retval_uses): New function.
3596 (malloc_candidate_p): Likewise.
3597 (analyze_function): Add support for malloc attribute.
3598 (pure_const_write_summary): Stream malloc_state.
3599 (pure_const_read_summary): Add support for reading malloc_state.
3600 (dump_malloc_lattice): New function.
3601 (propagate_malloc): New function.
3602 (warn_function_malloc): New function.
3603 (ipa_pure_const::execute): Call propagate_malloc and
3604 ipa_free_fn_summary.
3605 (pass_local_pure_const::execute): Add support for malloc attribute.
3606 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3607 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3609 2017-10-27 Martin Liska <mliska@suse.cz>
3611 PR gcov-profile/82457
3612 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3613 for fork-like functions to be properly instrumented.
3615 2017-10-27 Richard Biener <rguenther@suse.de>
3618 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3619 info when we redirected EH.
3621 2017-10-26 Michael Collison <michael.collison@arm.com>
3623 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3625 (<optab>_trunchf<GPI:mode>2: New pattern.
3626 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3627 * config/aarch64/iterators.md (wv): New mode attribute.
3628 (vf, VF): New mode attributes.
3629 (vgp, VGP): New mode attributes.
3630 (s): Update attribute with SImode and DImode prefixes.
3632 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3634 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3635 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3636 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3637 (nios2_option_overide): Initialize it. Don't allow R0-relative
3638 addressing with PIC.
3639 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3640 (nios2_symbolic_constant_p): Likewise.
3641 (nios2_legitimate_address_p): Likewise.
3642 (nios2_r0rel_section_name_p): New.
3643 (nios2_symbol_ref_in_r0rel_data_p): New.
3644 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3645 (r0rel_constant_p): New.
3646 (nios2_print_operand_address): Handle r0rel_constant_p.
3647 (nios2_cdx_narrow_form_p): Likewise.
3648 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3649 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3650 (Nios II Options): Document -mr0rel-sec.
3652 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3654 * config/nios2/nios2.c: Include xregex.h.
3655 (nios2_gprel_sec_regex): New.
3656 (nios2_option_overide): Initialize it. Don't allow GP-relative
3657 addressing with PIC.
3658 (nios2_small_section_name_p): Check for regex match.
3659 * config/nios2/nios2.opt (mgprel-sec=): New option.
3660 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3661 (Nios II Options): Document -mgprel-sec.
3663 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3665 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3667 2017-10-26 Tom de Vries <tom@codesourcery.com>
3669 PR tree-optimization/82707
3670 * gimple.c (gimple_copy): Fix unsharing of
3671 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3673 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3675 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3676 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3677 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3678 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3679 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3680 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3681 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3682 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3683 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3684 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3685 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3686 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3687 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3688 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3689 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3690 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3691 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3693 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3695 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3697 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3698 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3699 warning to rs6000.c. Remove the Undocumented flag, since it has
3701 (-mabi=ibmlongdouble): Likewise.
3702 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3703 already set, set the default format for long double.
3704 (rs6000_debug_reg_global): Print whether long double is IBM or
3706 (rs6000_option_override_internal): Rework setting long double
3707 format. Only warn if the user is changing the long double default
3708 and they did not use -Wno-psabi.
3709 * doc/invoke.texi (PowerPC options): Update the documentation for
3710 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3712 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3713 Alan Hayward <alan.hayward@arm.com>
3714 David Sherwood <david.sherwood@arm.com>
3716 * rtl.h (wider_subreg_mode): New function.
3717 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3718 rather than an unsigned int *.
3719 * ira-color.c (regno_max_ref_width): Replace with...
3720 (regno_max_ref_mode): ...this new variable.
3721 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3722 Use wider_subreg_mode.
3723 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3724 rather than an unsigned int *.
3725 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3726 (process_alt_operands): Likewise.
3727 (invariant_p): Likewise.
3728 * lra-spills.c (assign_mem_slot): Likewise.
3729 (add_pseudo_to_slot): Likewise.
3730 * lra.c (collect_non_operand_hard_regs): Likewise.
3731 (add_regs_to_insn_regno_info): Likewise.
3732 * reload1.c (regno_max_ref_width): Replace with...
3733 (regno_max_ref_mode): ...this new variable.
3734 (reload): Update accordingly. Update call to
3735 ira_sort_regnos_for_alter_reg.
3736 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3737 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3738 (scan_paradoxical_subregs): Likewise.
3740 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3742 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3743 (aarch64_frame): Add emit_frame_chain boolean.
3744 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3745 Move eh_return case to aarch64_layout_frame.
3746 (aarch64_layout_frame): Initialize emit_frame_chain.
3747 (aarch64_expand_prologue): Use emit_frame_chain.
3749 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3751 * config/aarch64/aarch64.c (aarch64_layout_frame):
3752 Ensure LR is always stored at the bottom of the callee-saves.
3753 Remove rarely used frame layout which saves callee-saves at top of
3754 frame, so the store of LR can be used as a valid probe in all cases.
3756 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3758 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3759 Improve unaligned TImode/TFmode base/offset split.
3761 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3762 Alan Hayward <alan.hayward@arm.com>
3763 David Sherwood <david.sherwood@arm.com>
3765 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3766 * combine.c (find_single_use_1): Likewise.
3767 (expand_field_assignment): Likewise.
3768 (move_deaths): Likewise.
3769 * lra-constraints.c (simplify_operand_subreg): Likewise.
3770 (curr_insn_transform): Likewise.
3771 * lra.c (collect_non_operand_hard_regs): Likewise.
3772 (add_regs_to_insn_regno_info): Likewise.
3773 * rtlanal.c (reg_referenced_p): Likewise.
3774 (covers_regno_no_parallel_p): Likewise.
3776 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3778 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3779 Don't print any bits outside the precision of the value.
3780 * wide-int.cc (test_printing): Add some new tests.
3782 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3784 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3785 supports -xbrace_comment option.
3786 * configure: Regenerate.
3787 * config.in: Regenerate.
3788 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3789 (ASM_CPU_SPEC): Use it.
3791 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3793 * target.def (static_rtx_alignment): New hook.
3794 * targhooks.h (default_static_rtx_alignment): Declare.
3795 * targhooks.c (default_static_rtx_alignment): New function.
3796 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3797 * doc/tm.texi: Regenerate.
3798 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3799 instead of targetm.constant_alignment. Remove call to
3801 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3802 (cris_preferred_mininum_alignment): New function, split out from...
3803 (cris_constant_alignment): ...here.
3804 (cris_static_rtx_alignment): New function.
3805 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3807 (ix86_constant_alignment): ...here.
3808 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3809 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3810 (mmix_static_rtx_alignment): New function.
3811 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3812 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3814 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3817 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3818 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3820 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3822 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3823 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3824 mark as a sign-extending load.
3825 (local_pic_load_u): Define.
3827 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3830 * fold-const.c (operand_equal_for_comparison_p): Also return true
3831 if ARG0 is a simple variant of ARG1 with narrower precision.
3832 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3834 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3836 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3838 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3839 scatter_static, scatter_per_elt.
3840 * x86-tune-costs.h: Add new cost entries.
3842 2017-10-25 Richard Biener <rguenther@suse.de>
3844 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3845 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3846 class pass_fre): Move to ...
3847 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3848 class pass_fre): ... here and adjust for statistics.
3850 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3853 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3854 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3855 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3856 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3859 2017-10-25 Richard Biener <rguenther@suse.de>
3861 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3862 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3863 eliminate_push_avail, eliminate_insert): Move inside...
3864 (class eliminate_dom_walker): ... this class in preparation
3866 (fini_eliminate): Remove by merging with ...
3867 (eliminate): ... this function. Adjust for class changes.
3868 (pass_pre::execute): Remove fini_eliminate call.
3869 (pass_fre::execute): Likewise.
3871 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3874 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3875 (VPERMI2, VPERMI2I): New mode iterators.
3876 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3877 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3879 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3880 mode iterator. Remove 3 old define_insn patterns.
3881 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3882 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3883 VPERMI2 mode iterator, remove the other two expanders.
3884 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3885 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3886 instructions, remove the other two patterns.
3887 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3888 mode iterator, remove the other two patterns.
3889 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3890 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3891 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3892 and adjust argument order accordingly.
3893 (ix86_expand_vec_perm): Adjust caller.
3894 (expand_vec_perm_1): Likewise.
3895 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3896 (expand_vec_perm_vpermt2_vpshub2): ... this.
3897 (ix86_expand_vec_perm_const_1): Adjust caller.
3898 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3901 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3902 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3903 (vec_shl_<mode>): Remove unused expander.
3904 (avx512bw_<shift_insn><mode>3): New define_insn.
3905 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3906 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3908 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3911 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3914 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3916 PR rtl-optimization/82396
3917 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3918 (autopref_multipass_init): Simplify initialization.
3919 (autopref_rank_data): Simplify sort order.
3920 * gcc/sched-int.h (autopref_multipass_data_): Remove
3921 multi_mem_insn_p, min_offset and max_offset.
3923 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3926 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3927 Check special value of flag_omit_frame_pointer.
3928 (aarch64_can_eliminate): Likewise.
3929 (aarch64_override_options_after_change_1): Simplify handling of
3930 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3932 2017-10-24 Richard Biener <rguenther@suse.de>
3934 PR tree-optimization/82697
3935 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3936 zero for conditional load and unconditional store.
3938 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3940 * doc/install.texi: Document bootstrap-cet.
3942 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3945 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3946 ENDBR instruction at function entrance if function is only
3949 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3952 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3953 patterns to better describe from which operation the CF is computed.
3954 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3955 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3956 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3957 is 0, use _0 suffixed expanders instead of emitting a comparison
3960 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3962 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3963 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3965 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3968 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3969 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3970 * loop-iv.c (iv_get_reaching_def): Likewise.
3971 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3972 variable is promoted and the partition contains undefined values.
3974 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3976 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3978 (nios2_address_cost): Define.
3979 (nios2_legitimize_address): Recognize (exp + constant) directly.
3980 (TARGET_ADDRESS_COST): Define.
3982 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3984 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3985 (nios2_symbolic_memory_operand_p): Declare.
3986 (nios2_split_large_constant): Declare.
3987 (nios2_split_symbolic_memory_operand): Declare.
3988 * config/nios2/nios2.c: Adjust includes.
3989 (nios2_symbolic_constant_allowed): New.
3990 (nios2_symbolic_constant_p): New.
3991 (nios2_plus_symbolic_constant_p): New.
3992 (nios2_valid_addr_expr_p): Recognize addresses involving
3994 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3995 (nios2_symbolic_memory_operand_p): New.
3996 (nios2_large_constant_p): New.
3997 (nios2_split_large_constant): New.
3998 (nios2_split_plus_large_constant): New.
3999 (nios2_split_symbolic_memory_operand): New.
4000 (nios2_legitimize_address): Code refactoring. Handle addresses
4001 involving symbolic constants.
4002 (nios2_emit_move_sequence): Likewise.
4003 (nios2_print_operand): Improve error output.
4004 (nios2_print_operand_address): Handle LO_SUM.
4005 (nios2_cdx_narrow_form_p): Likewise.
4006 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4007 operands involving symbolic constants.
4008 (movhi_internal, movsi_internal): Likewise.
4009 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4010 (extendhisi2, extendqi<mode>2): Likewise.
4012 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4014 * tree-pass.h (PROP_rtl_split_insns): Define.
4015 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4017 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4019 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4021 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4024 * target.def (const_not_ok_for_debug_p): Default to
4025 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4026 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4027 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4028 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4029 which targetm.const_not_ok_for_debug_p returned true.
4030 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4032 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4034 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4035 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4036 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4038 (ix86_const_not_ok_for_debug_p): New function.
4039 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4040 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4042 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4045 * system.h: Conditionally include "unique-ptr.h" if
4046 INCLUDE_UNIQUE_PTR is defined.
4047 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4048 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4050 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4052 * config/rl78/rl78.md: New define_expand "subdi3".
4054 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4057 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4058 DF_REF_INSN if DF_REF_INSN_INFO is false.
4060 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4062 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4063 xmm_move instead of sse_move.
4064 (sse_store_index): New function.
4065 (ix86_register_move_cost): Be more sensible about mismatch stall;
4066 model AVX moves correctly; make difference between sse->integer and
4068 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4069 moves; make difference between SSE and AVX.
4070 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4071 and zmm_move. Increase size of sse load and store tables;
4072 add unaligned load and store tables; add ssemmx_to_integer.
4073 * x86-tune-costs.h: Update all entries according to real
4074 move latencies from Agner Fog's manual and chip documentation.
4076 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4079 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4080 * config/i386/constraints.md (Wf): New constraint.
4081 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4082 (cmp<dwi>_doubleword): Removed.
4083 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4084 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4085 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4086 expand with cmp<dwi>_doubleword. For LTU and GEU use
4087 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4089 * common.opt (gcolumn-info): Enable by default.
4090 * doc/invoke.texi (gcolumn-info): Document new default.
4092 2017-10-23 Richard Biener <rguenther@suse.de>
4094 PR tree-optimization/82672
4095 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4096 Fold the stmt if we propagated into it.
4098 2017-10-23 Richard Biener <rguenther@suse.de>
4100 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4101 (bitmap_remove_expr_from_set): ... this. All callers call this
4102 for non-constant values.
4103 (bitmap_set_subtract): Rename to...
4104 (bitmap_set_subtract_expressions): ... this. Adjust and
4106 (bitmap_set_contains_value): Remove superfluous check.
4107 (bitmap_set_replace_value): Inline into single caller ...
4108 (bitmap_value_replace_in_set): ... here and simplify.
4109 (dependent_clean): Merge into ...
4110 (clean): ... this using an overload. Adjust.
4111 (prune_clobbered_mems): Adjust.
4112 (compute_antic_aux): Likewise.
4113 (compute_partial_antic_aux): Likewise.
4115 2017-10-23 Richard Biener <rguenther@suse.de>
4117 PR tree-optimization/82129
4119 2017-08-01 Richard Biener <rguenther@suse.de>
4121 PR tree-optimization/81181
4122 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4123 (compute_antic): ... end of iteration here.
4125 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4127 * target.def (starting_frame_offset): New hook.
4128 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4129 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4130 * doc/tm.texi.in: Regenerate.
4131 * hooks.h (hook_hwi_void_0): Declare.
4132 * hooks.c (hook_hwi_void_0): New function.
4133 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4134 STARTING_FRAME_OFFSET.
4135 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4136 * reload1.c (reload): Likewise.
4137 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4138 instead of STARTING_FRAME_OFFSET.
4139 * function.c (try_fit_stack_local): Likewise.
4140 (assign_stack_local_1): Likewise
4141 (instantiate_virtual_regs): Likewise.
4142 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4143 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4144 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4145 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4146 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4147 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4148 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4149 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4150 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4151 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4152 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4153 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4154 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4155 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4156 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4157 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4158 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4159 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4160 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4161 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4162 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4163 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4164 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4165 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4166 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4167 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4168 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4169 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4170 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4171 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4172 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4173 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4174 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4175 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4176 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4177 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4178 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4179 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4180 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4181 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4182 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4183 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4184 return a HOST_WIDE_INT.
4185 (avr_builtin_setjmp_frame_value): Use it instead of
4186 STARTING_FRAME_OFFSET.
4187 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4188 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4189 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4191 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4192 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4193 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4194 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4195 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4196 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4197 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4198 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4199 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4200 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4201 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4202 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4204 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4205 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4206 * config/mips/mips.c (mips_compute_frame_info): Refer to
4207 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4208 (mips_starting_frame_offset): New function.
4209 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4210 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4211 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4212 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4213 and return a HOST_WIDE_INT.
4214 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4215 (mmix_initial_elimination_offset): Refer to
4216 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4217 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4218 * config/pa/pa.c (pa_starting_frame_offset): New function.
4219 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4220 (pa_expand_prologue): Likewise.
4221 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4222 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4223 !FRAME_GROWS_DOWNWARD handling to...
4224 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4225 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4226 !FRAME_GROWS_DOWNWARD handling to...
4227 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4228 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4229 !FRAME_GROWS_DOWNWARD handling to...
4230 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4231 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4233 (rs6000_starting_frame_offset): New function.
4234 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4235 !FRAME_GROWS_DOWNWARD handling to...
4236 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4237 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4238 !FRAME_GROWS_DOWNWARD handling to...
4239 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4240 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4241 !FRAME_GROWS_DOWNWARD handling to...
4242 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4243 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4244 (rs6000_starting_frame_offset): New function.
4245 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4246 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4247 * config/vax/vax.c (vax_starting_frame_offset): New function.
4248 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4249 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4250 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4251 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4252 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4253 * system.h (STARTING_FRAME_OFFSET): Poison.
4255 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4257 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4258 SCALAR_TYPE_MODE instead of TYPE_MODE.
4260 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4261 Alan Hayward <alan.hayward@arm.com>
4262 David Sherwood <david.sherwood@arm.com>
4264 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4266 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4267 Alan Hayward <alan.hayward@arm.com>
4268 David Sherwood <david.sherwood@arm.com>
4270 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4272 2017-10-23 Richard Biener <rguenther@suse.de>
4274 PR tree-optimization/82129
4275 * tree-ssa-pre.c (bitmap_set_and): Remove.
4276 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4277 canonicalizing expressions in the set to those with lowest
4278 ID rather than taking that from the first edge.
4280 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4282 * combine.c (rtx_equal_for_field_assignment_p): Use
4283 byte_lowpart_offset.
4285 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4286 Alan Hayward <alan.hayward@arm.com>
4287 David Sherwood <david.sherwood@arm.com>
4289 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4290 to a SUBREG_PROMOTED_VAR.
4292 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4293 Alan Hayward <alan.hayward@arm.com>
4294 David Sherwood <david.sherwood@arm.com>
4296 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4297 (expand_debug_source_expr): Likewise.
4298 * combine.c (combine_simplify_rtx): Likewise.
4299 * cse.c (fold_rtx): Likewise.
4300 * optabs.c (expand_float): Likewise.
4301 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4302 (simplify_binary_operation_1): Likewise.
4304 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4305 Alan Hayward <alan.hayward@arm.com>
4306 David Sherwood <david.sherwood@arm.com>
4308 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4309 (record_promoted_value): Likewise.
4310 * expr.c (expand_expr_real_2): Likewise.
4311 * ree.c (update_reg_equal_equiv_notes): Likewise.
4312 (combine_set_extension): Likewise.
4313 * rtlanal.c (low_bitmask_len): Likewise.
4314 * simplify-rtx.c (neg_const_int): Likewise.
4315 (simplify_binary_operation_1): Likewise.
4317 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4318 Alan Hayward <alan.hayward@arm.com>
4319 David Sherwood <david.sherwood@arm.com>
4321 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4322 * regcprop.c (maybe_mode_change): Likewise.
4323 * reload1.c (alter_reg): Likewise.
4325 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4327 * inchash.h (inchash::hash::add_wide_int): New function.
4328 * lto-streamer-out.c (hash_tree): Use it.
4330 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4332 * inchash.h (inchash::hash::add_wide_int): Rename to...
4333 (inchash::hash::add_hwi): ...this.
4334 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4335 (polymorphic_call_target_hasher::hash): Likewise.
4336 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4337 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4338 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4339 * lto-streamer-out.c (hash_tree): Likewise.
4340 * optc-save-gen.awk: Likewise.
4341 * tree.c (add_expr): Likewise.
4343 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4346 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4347 for ordered inequality comparisons even with TARGET_IEEE_FP.
4349 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4352 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4353 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4354 Expand with cmp<dwi>_doubleword.
4356 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4358 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4359 List CET intrinsics.
4360 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4361 specific to -fcf-protection option.
4363 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4365 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4366 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4367 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4368 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4369 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4370 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4371 (extra_objs): Add cet.o for Linux/x86 targets.
4372 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4373 * config/i386/cet.c: New file.
4374 * config/i386/cetintrin.h: Likewise.
4375 * config/i386/t-cet: Likewise.
4376 * config/i386/cpuid.h (bit_SHSTK): New.
4377 (bit_IBT): Likewise.
4378 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4379 pass IBT and SHSTK bits.
4380 * config/i386/i386-builtin-types.def
4381 (VOID_FTYPE_UNSIGNED_PVOID): New.
4382 (VOID_FTYPE_UINT64_PVOID): Likewise.
4383 * config/i386/i386-builtin.def: Add CET intrinsics.
4384 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4385 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4386 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4387 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4389 * config/i386/i386.c (rest_of_insert_endbranch): New.
4390 (pass_data_insert_endbranch): Likewise.
4391 (pass_insert_endbranch): Likewise.
4392 (make_pass_insert_endbranch): Likewise.
4393 (ix86_notrack_prefixed_insn_p): Likewise.
4394 (ix86_target_string): Add -mibt, -mshstk flags.
4395 (ix86_option_override_internal): Add flag_cf_protection
4397 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4398 (ix86_print_operand): Add 'notrack' prefix output.
4399 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4400 (ix86_expand_builtin): Expand CET intrinsics.
4401 (x86_output_mi_thunk): Add 'endbranch' instruction.
4402 * config/i386/i386.h (TARGET_IBT): New.
4403 (TARGET_IBT_P): Likewise.
4404 (TARGET_SHSTK): Likewise.
4405 (TARGET_SHSTK_P): Likewise.
4406 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4407 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4408 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4409 (builtin_setjmp_setup): New pattern.
4410 (builtin_longjmp): Likewise.
4411 (rdssp<mode>): Likewise.
4412 (incssp<mode>): Likewise.
4413 (saveprevssp): Likewise.
4414 (rstorssp): Likewise.
4415 (wrss<mode>): Likewise.
4416 (wruss<mode>): Likewise.
4417 (setssbsy): Likewise.
4418 (clrssbsy): Likewise.
4419 (nop_endbr): Likewise.
4420 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4422 * config/i386/immintrin.h: Include <cetintrin.h>.
4423 * config/i386/linux-common.h
4424 (file_end_indicate_exec_stack_and_cet): New prototype.
4425 (TARGET_ASM_FILE_END): New.
4427 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4429 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4430 latencies instead of having separate table; make difference between
4431 integer and float costs.
4432 * i386.h (processor_costs): Remove scalar_stmt_cost,
4433 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4434 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4436 * x86-tune-costs.h: Remove entries which has been removed in
4437 procesor_costs from all tables; make cond_taken_branch_cost
4438 and cond_not_taken_branch_cost COST_N_INSNS based.
4440 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4442 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4444 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4446 * config/i386/i386.md (isa): Remove fma_avx512f.
4447 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4448 <avx512>_fmadd_<mode>_mask3<round_name>,
4449 <avx512>_fmsub_<mode>_mask<round_name>,
4450 <avx512>_fmsub_<mode>_mask3<round_name>,
4451 <avx512>_fnmadd_<mode>_mask<round_name>,
4452 <avx512>_fnmadd_<mode>_mask3<round_name>,
4453 <avx512>_fnmsub_<mode>_mask<round_name>,
4454 <avx512>_fnmsub_<mode>_mask3<round_name>,
4455 <avx512>_fmaddsub_<mode>_mask<round_name>,
4456 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4457 <avx512>_fmsubadd_<mode>_mask<round_name>,
4458 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4459 (*vec_widen_umult_even_v16si<mask_name>,
4460 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4461 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4463 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4465 * extend.texi: Add 'nocf_check' documentation.
4466 * gimple.texi: Add second parameter to
4467 gimple_build_call_from_tree.
4468 * invoke.texi: Add -fcf-protection documentation.
4469 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4471 2017-10-20 Richard Biener <rguenther@suse.de>
4473 PR tree-optimization/82473
4474 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4475 the largest input type.
4477 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4479 * c-attribs.c (handle_nocf_check_attribute): New function.
4480 (c_common_attribute_table): Add 'nocf_check' handling.
4481 * gimple-parser.c: Add second argument NULL to
4482 gimple_build_call_from_tree.
4483 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4484 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4486 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4487 * common.opt: Add fcf-protection flag.
4488 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4489 * flag-types.h: Add enum cf_protection_level.
4490 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4491 Add 'nocf_check' attribute propagation to gimple call.
4492 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4493 (gimple_build_call_from_tree): Update prototype.
4494 (gimple_call_nocf_check_p): New function.
4495 (gimple_call_set_nocf_check): Likewise.
4496 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4497 * ipa-icf.c: Add nocf_check attribute in statement hash.
4498 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4499 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4500 * toplev.c (process_options): Add flag_cf_protection handling.
4502 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4504 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4506 2017-10-20 Richard Biener <rguenther@suse.de>
4508 PR tree-optimization/82603
4509 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4510 remove false predicated stores.
4512 2017-10-20 Richard Biener <rguenther@suse.de>
4514 * graphite-isl-ast-to-gimple.c
4515 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4516 Remove return value and simplify, dump copied stmt after lhs
4518 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4519 Reduce dump verbosity.
4520 (gsi_insert_earliest): Likewise.
4521 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4522 * graphite.c (print_global_statistics): Adjust dumping.
4523 (print_graphite_scop_statistics): Likewise.
4524 (print_graphite_statistics): Do not dump loops here.
4525 (graphite_transform_loops): But here.
4527 2017-10-20 Nicolas Roche <roche@adacore.com>
4529 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4530 * configure: Regenerate.
4532 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4535 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4536 functions when optimizing replace GIMPLE_RETURN stmts with
4537 calls to __builtin_unreachable ().
4540 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4541 for -fsanitize=thread link of executables.
4542 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4543 link of executables.
4546 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4548 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4549 define_insns for logical vector shifts to use VI248_AVX512BW
4550 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4551 condition, useless isa and prefix attributes. Change the first
4552 2 of these define_insns to ...
4553 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4554 define_insn for avx512vl.
4555 (<shift_insn><mode>3): ... and this, new define_insn without
4556 masking for non-avx512vl.
4559 * config/i386/sse.md (*andnot<mode>3,
4560 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4561 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4562 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4563 not applied use empty suffix even for TARGET_AVX512VL.
4564 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4565 is applied, supply evex,evex or evex,evex,evex instead of just
4568 2017-10-20 Julia Koval <julia.koval@intel.com>
4570 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4571 (OPTION_MASK_ISA_GFNI_UNSET): New.
4572 (ix86_handle_option): Handle OPT_mgfni.
4573 * config/i386/cpuid.h (bit_GFNI): New.
4574 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4575 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4576 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4577 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4578 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4579 * config/i386/i386.opt: Add mgfni.
4581 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4583 * config/msp430/msp430.c (msp430_option_override): Disable
4584 -fdelete-null-pointer-checks.
4585 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4587 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4589 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4590 of x87 and SSE instructions.
4592 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4594 * asan.c (create_cond_insert_point): Do not update edge count.
4595 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4596 (afdo_propagate_circuit): Likewise.
4597 (afdo_calculate_branch_prob): Likewise.
4598 (afdo_annotate_cfg): Likewise.
4599 * basic-block.h (struct edge_def): Remove count.
4600 (edge_def::count): New accessor.
4601 * bb-reorder.c (rotate_loop): Update.
4602 (find_traces_1_round): Update.
4603 (connect_traces): Update.
4604 (sanitize_hot_paths): Update.
4605 * cfg.c (unchecked_make_edge): Update.
4606 (make_single_succ_edge): Update.
4607 (check_bb_profile): Update.
4608 (dump_edge_info): Update.
4609 (update_bb_profile_for_threading): Update.
4610 (scale_bbs_frequencies_int): Update.
4611 (scale_bbs_frequencies_gcov_type): Update.
4612 (scale_bbs_frequencies_profile_count): Update.
4613 (scale_bbs_frequencies): Update.
4614 * cfganal.c (connect_infinite_loops_to_exit): Update.
4615 * cfgbuild.c (compute_outgoing_frequencies): Update.
4616 (find_many_sub_basic_blocks): Update.
4617 * cfgcleanup.c (try_forward_edges): Update.
4618 (try_crossjump_to_edge): Update
4619 * cfgexpand.c (expand_gimple_cond): Update
4620 (expand_gimple_tailcall): Update
4621 (construct_exit_block): Update
4622 * cfghooks.c (verify_flow_info): Update
4623 (redirect_edge_succ_nodup): Update
4624 (split_edge): Update
4625 (make_forwarder_block): Update
4626 (duplicate_block): Update
4627 (account_profile_record): Update
4628 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4629 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4630 * cfgloopmanip.c (scale_loop_profile): Update.
4632 (lv_adjust_loop_entry_edge): Update.
4633 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4634 (force_nonfallthru_and_redirect): Update.
4635 (purge_dead_edges): Update.
4636 (rtl_flow_call_edges_add): Update.
4637 * cgraphunit.c (init_lowered_empty_function): Update.
4638 (cgraph_node::expand_thunk): Update.
4639 * gimple-pretty-print.c (dump_probability): Update.
4640 (dump_edge_probability): Update.
4641 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4642 * haifa-sched.c (sched_create_recovery_edges): Update.
4643 * hsa-gen.c (convert_switch_statements): Update.
4644 * ifcvt.c (dead_or_predicable): Update.
4645 * ipa-inline-transform.c (inline_transform): Update.
4646 * ipa-split.c (split_function): Update.
4647 * ipa-utils.c (ipa_merge_profiles): Update.
4648 * loop-doloop.c (add_test): Update.
4649 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4650 * lto-streamer-in.c (input_cfg): Update.
4651 (input_function): Update.
4652 * lto-streamer-out.c (output_cfg): Update.
4653 * modulo-sched.c (sms_schedule): Update.
4654 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4655 * predict.c (maybe_hot_edge_p): Update.
4656 (unlikely_executed_edge_p): Update.
4657 (probably_never_executed_edge_p): Update.
4658 (dump_prediction): Update.
4659 (drop_profile): Update.
4660 (propagate_unlikely_bbs_forward): Update.
4661 (determine_unlikely_bbs): Update.
4662 (force_edge_cold): Update.
4663 * profile.c (compute_branch_probabilities): Update.
4664 * reg-stack.c (better_edge): Update.
4665 * shrink-wrap.c (handle_simple_exit): Update.
4666 * tracer.c (better_p): Update.
4667 * trans-mem.c (expand_transaction): Update.
4668 (split_bb_make_tm_edge): Update.
4669 * tree-call-cdce.c: Update.
4670 * tree-cfg.c (gimple_find_sub_bbs): Update.
4671 (gimple_split_edge): Update.
4672 (gimple_duplicate_sese_region): Update.
4673 (gimple_duplicate_sese_tail): Update.
4674 (gimple_flow_call_edges_add): Update.
4675 (insert_cond_bb): Update.
4676 (execute_fixup_cfg): Update.
4677 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4678 * tree-complex.c (expand_complex_div_wide): Update.
4679 * tree-eh.c (lower_resx): Update.
4680 (unsplit_eh): Update.
4681 (cleanup_empty_eh_move_lp): Update.
4682 * tree-inline.c (copy_edges_for_bb): Update.
4683 (freqs_to_counts): Update.
4684 (copy_cfg_body): Update.
4685 * tree-ssa-dce.c (remove_dead_stmt): Update.
4686 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4687 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4688 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4689 (unloop_loops): Update.
4690 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4691 * tree-ssa-loop-split.c (connect_loops): Update.
4692 (split_loop): Update.
4693 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4694 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4695 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4696 * tree-ssa-reassoc.c (branch_fixup): Update.
4697 * tree-ssa-tail-merge.c (replace_block_by): Update.
4698 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4699 (compute_path_counts): Update.
4700 (update_profile): Update.
4701 (recompute_probabilities): Update.
4702 (update_joiner_offpath_counts): Update.
4703 (estimated_freqs_path): Update.
4704 (freqs_to_counts_path): Update.
4705 (clear_counts_path): Update.
4706 (ssa_fix_duplicate_block_edges): Update.
4707 (duplicate_thread_path): Update.
4708 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4709 (case_bit_test_cmp): Update.
4710 (collect_switch_conv_info): Update.
4711 (gen_inbound_check): Update.
4712 (do_jump_if_equal): Update.
4713 (emit_cmp_and_jump_insns): Update.
4714 * tree-tailcall.c (decrease_profile): Update.
4715 (eliminate_tail_call): Update.
4716 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4717 (vect_do_peeling): Update.
4718 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4719 * ubsan.c (ubsan_expand_null_ifn): Update.
4720 (ubsan_expand_ptr_ifn): Update.
4721 * value-prof.c (gimple_divmod_fixed_value): Update.
4722 (gimple_mod_pow2): Update.
4723 (gimple_mod_subtract): Update.
4724 (gimple_ic): Update.
4725 (gimple_stringop_fixed_value): Update.
4727 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4730 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4732 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4734 PR rtl-optimization/82395
4735 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4736 based on non_spilled_static_chain_regno_p.
4738 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4740 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4741 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4742 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4744 2017-10-19 Martin Sebor <msebor@redhat.com>
4746 PR tree-optimization/82596
4747 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4749 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4751 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4752 (get_mem_refs_of_builtin_call): Likewise.
4753 * builtins.c (expand_builtin_apply): Adjust call to
4754 allocate_dynamic_stack_space.
4755 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4756 the third argument to allocate_dynamic_stack_space, otherwise -1.
4757 (expand_builtin): Deal with all alloca variants.
4758 (is_inexpensive_builtin): Likewise.
4759 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4760 * calls.c (special_function_p): Deal with all alloca variants.
4761 (initialize_argument_information): Adjust call to
4762 allocate_dynamic_stack_space.
4763 (expand_call): Likewise.
4764 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4765 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4766 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4767 use it for the stack usage computation.
4768 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4769 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4770 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4771 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4772 (in_loop_p): Remove first argument and useless check.
4773 (pass_walloca::execute): Remove useless test and adjust call to above.
4774 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4775 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4776 (gimplify_call_expr): Deal with all alloca variants.
4777 * hsa-gen.c (gen_hsa_alloca): Likewise.
4778 (gen_hsa_insns_for_call): Likewise.
4779 * ipa-pure-const.c (special_builtin_state): Likewise.
4780 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4781 * tree-object-size.c (alloc_object_size): Likewise.
4782 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4783 (call_may_clobber_ref_p_1): Likewise.
4784 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4785 (ccp_fold_stmt): Likewise.
4786 (optimize_stack_restore): Likewise.
4787 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4788 (mark_all_reaching_defs_necessary_1): Likewise.
4789 (propagate_necessity): Likewise.
4790 (eliminate_unnecessary_stmts): Likewise.
4791 * tree.c (build_common_builtin_nodes): Build
4792 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4793 (build_alloca_call_expr): New function.
4794 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4795 (CASE_BUILT_IN_ALLOCA): Likewise.
4796 (build_alloca_call_expr): Declare.
4797 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4799 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4802 * dwarf2out.c (new_die_raw): New static inline function.
4803 (new_die): Use it to create the DIE.
4804 (add_AT_external_die_ref): Likewise.
4805 (clone_die): Likewise.
4806 (clone_as_declaration): Likewise.
4807 (dwarf2out_vms_debug_main_pointer): Likewise.
4808 (base_type_die): Likewise. Remove early return for corner cases.
4809 Do not call add_pubtype on the DIE here.
4810 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4811 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4812 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4813 native order exists for base types, attach the DIE manually and call
4814 add_pubtype on it. Do not equate a reverse order DIE to the type.
4816 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4818 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4819 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4821 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4823 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4825 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4826 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4827 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4828 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4829 * opts.c (sanitizer_opts): Add builtin.
4830 * ubsan.c (instrument_builtin): New function.
4831 (pass_ubsan::execute): Call it.
4832 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4833 * doc/invoke.texi: Document -fsanitize=builtin.
4835 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4836 builtins, store max (log2 (align), 0) into uchar field instead of
4837 align into uptr field.
4838 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4839 store uchar 0 field instead of uptr 0 field.
4840 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4841 instead of passing one address of struct with 2 locations pass
4842 two addresses of structs with 1 location each.
4843 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4844 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4845 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4846 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4847 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4848 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4849 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4850 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4852 2017-10-19 Martin Liska <mliska@suse.cz>
4855 * file-find.c (remove_prefix): Remove.
4856 * file-find.h (remove_prefix): Likewise.
4857 * gcc-ar.c: Remove smartness of lookup.
4859 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4861 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4862 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4863 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4865 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4868 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4869 (setcc + and to xor + setcc): New peephole2.
4871 2017-10-19 Tom de Vries <tom@codesourcery.com>
4873 * doc/sourcebuild.texi (Test Directives, Variants of
4874 dg-require-support): Add dg-require-stack-size.
4876 2017-10-19 Martin Liska <mliska@suse.cz>
4879 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4880 that have a large alignment.
4881 (gimplify_target_expr): Likewise.
4883 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4885 PR rtl-optimization/82602
4886 * ira.c (rtx_moveable_p): Return false for volatile asm.
4888 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4891 * config/i386/i386-modes.def (CCGZ): New CC mode.
4892 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4893 * config/i386/predicates.md (ix86_comparison_operator):
4895 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4896 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4897 with double-word subtraction.
4898 (put_condition_code): Handle CCGZmode.
4900 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4902 * wide-int.cc (debug (const wide_int &)): New.
4903 (debug (const wide_int *)): New.
4904 (debug (const widest_int &)): New.
4905 (debug (const widest_int *)): New.
4907 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4910 * lra-constraints.c (curr_insn_transform): Use non-input operand
4911 instead of output one for matched reload.
4913 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4915 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4916 (tree-ssa-loop-ivopts.h): New header file.
4917 (struct builtin_info): New fields.
4918 (classify_builtin_1): Compute and record base and offset parts for
4919 memset builtin partition by calling strip_offset.
4920 (offset_cmp, fuse_memset_builtins): New functions.
4921 (finalize_partitions): Fuse adjacent memset partitions by calling
4923 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4924 Expose the interface.
4925 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4927 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4929 PR tree-optimization/82574
4930 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4931 that data reference must be executed exactly once per iteration
4932 against the outermost loop in nest.
4933 (classify_partition): Update call to above function.
4935 2017-10-18 Richard Biener <rguenther@suse.de>
4937 PR tree-optimization/82591
4938 * graphite.c (graphite_transform_loops): Move code gen message
4940 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4941 Here. Handle scop_to_isl_ast failing.
4942 (scop_to_isl_ast): Limit the number of ISL operations.
4944 2017-10-18 Richard Biener <rguenther@suse.de>
4946 * graphite-isl-ast-to-gimple.c
4947 (translate_isl_ast_to_gimple::set_rename): Simplify.
4948 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4949 (graphite_copy_stmts_from_block): ... here.
4950 (copy_bb_and_scalar_dependences): Simplify.
4951 (add_parameters_to_ivs_params): Canonicalize.
4952 (generate_entry_out_of_ssa_copies): Simplify.
4953 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4954 by passing in ISL dimension.
4955 (parameter_index_in_region_1): Rename to ...
4956 (parameter_index_in_region): ... this.
4957 (extract_affine): Adjust assert, pass down parameter index.
4958 (add_param_constraints): Use range-info when available.
4959 (build_scop_context): Adjust.
4960 * sese.c (new_sese_info): Adjust.
4961 (free_sese_info): Likewise.
4962 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4963 Remove unused typedefs.
4964 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4966 2017-10-18 Martin Liska <mliska@suse.cz>
4968 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4970 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4972 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4973 (s390_sched_init): Do not reset s390_sched_state if we entered the
4974 current basic block via a fallthru edge and all others are unlikely.
4976 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4978 * config/s390/s390.c (NUM_SIDES): New variable.
4979 (LONGRUNNING_THRESHOLD): New variable.
4980 (LATENCY_FACTOR): New variable.
4981 (s390_sched_score): Decrease score for long-running instructions on
4983 (s390_sched_variable_issue): Perform bookkeeping for long-running
4986 2017-10-18 Richard Biener <rguenther@suse.de>
4988 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4989 Simplify with removal of the parameter rename map.
4990 (set_rename): Likewise.
4991 (should_copy_to_new_region): Likewise.
4992 (graphite_copy_stmts_from_block): Likewise.
4993 (copy_bb_and_scalar_dependences): Remove initialization of
4994 unused copied_bb_map.
4996 (copy_internal_parameters): Likewise.
4997 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4998 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4999 Use INTEGRAL_TYPE_P.
5000 (parameter_index_in_region_1): Rename to ...
5001 (assign_parameter_index_in_region): ... this. Assert we have
5002 a parameter we handle.
5003 (scan_tree_for_params): Adjust.
5004 * sese.h (parameter_rename_map_t): Remove.
5005 (struct sese_info_t): Remove unused parameter_rename_map and
5006 copied_bb_map members.
5007 * sese.c (new_sese_info): Adjust.
5008 (free_sese_info): Likewise.
5010 2017-10-18 Martin Liska <mliska@suse.cz>
5013 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5014 on an abnormal edge.
5016 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5018 * doc/invoke.texi (ffunction-sections and fdata-sections):
5021 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5023 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5024 the use statement can throw internally.
5026 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5028 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5029 any RTX present on the RHS of a SET.
5030 * compare-elim.c (try_eliminate_compare): Restore comment.
5032 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5034 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5035 may be also called on tcc_type nodes.
5036 * langhooks.c (lhd_tree_size): Likewise.
5038 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5040 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5041 format_warning_at_substring.
5042 (maybe_warn): Convert source_range * param to a location_t. Pass
5043 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5044 (format_directive): Remove code to extract source_ranges and
5045 source_range * in favor of just a location_t.
5046 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5048 * substring-locations.c (format_warning_va): Convert
5049 source_range * param to a location_t.
5050 (format_warning_at_substring): Likewise.
5051 * substring-locations.h (format_warning_va): Likewise.
5052 (format_warning_at_substring): Likewise.
5054 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5056 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5058 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5059 and scatter/gather ops.
5061 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5062 vec_gather_load and vec_scatter_store.
5063 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5064 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5065 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5066 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5067 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5068 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5070 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5072 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5073 Detect FTST instruction and handle its register pops. Only pop
5074 second operand if can_pop_second_op is true.
5075 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5076 set can_pop_second_op to false in the compare_for_stack_reg call.
5078 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5079 output_fp_compare for stack register operands.
5080 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5081 instructions here. Do not emit stack register pops here. Assert
5082 that FCOMPP pops next to top stack register. Rewrite function.
5084 2017-10-17 Nathan Sidwell <nathan@acm.org>
5087 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5088 use DECL_ASSEMBLER_NAME_RAW.
5091 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5094 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5095 Wilco Dijkstra <wilco.dijkstra@arm.com>
5097 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5098 converstion to Pmode from the buf_addr.
5100 2017-10-17 Richard Biener <rguenther@suse.de>
5102 * graphite-dependences.c (scop_get_reads_and_writes): Change
5103 output parameters to references.
5105 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5107 PR 71026/tree-optimization
5108 * fold-const.c (distribute_real_division): Removed.
5109 (fold_binary_loc): Remove calls to distribute_real_divison.
5111 2017-10-17 Richard Biener <rguenther@suse.de>
5113 * graphite-scop-detection.c
5114 (scop_detection::stmt_has_simple_data_refs_p): Always use
5115 the full nest as region.
5116 (try_generate_gimple_bb): Likewise.
5117 * sese.c (scalar_evolution_in_region): Simplify now that
5118 SCEV can handle instantiation in regions.
5119 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5120 in the non-loop part of a function if requested.
5122 2017-10-17 Richard Biener <rguenther@suse.de>
5124 PR tree-optimization/82563
5125 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5127 (graphite_regenerate_ast_isl): Call it.
5128 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5130 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5132 PR tree-optimization/82549
5133 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5134 Formatting fixes. Instead of calling make_bit_field_ref with negative
5137 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5139 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5140 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5141 _mm_maskz_reduce_ss): New.
5142 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5143 __builtin_ia32_reducess_mask): Ditto..
5144 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5145 * config/i386/sse.md (reduces<mode>): Renamed to ...
5146 (reduces<mode><mask_scalar_name>): ... this.
5147 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5149 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5150 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5152 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5154 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5155 * selftest-run-tests.c (selftest::run_tests): Call
5156 selftest::unique_ptr_tests_cc_tests.
5157 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5158 * unique-ptr-tests.cc: New file.
5160 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5163 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5165 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5166 (make_hard_regno_born, make_hard_regno_dead): Update
5167 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5169 2017-10-16 Jeff Law <law@redhat.com>
5171 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5173 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5175 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5177 2017-10-16 Olivier Hainque <hainque@adacore.com>
5179 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5180 with_cpu if we were configured for an e500v2 target cpu name.
5182 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5184 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5185 * doc/invoke.texi: Document +nodsp as a valid extension for
5188 2017-10-16 Martin Liska <mliska@suse.cz>
5190 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5191 (test_set_range): Likewise.
5192 (test_range_functions): Rename to ...
5193 (test_bit_in_range): ... this.
5194 (sbitmap_c_tests): Add new test.
5196 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5198 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5200 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5201 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5203 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5205 * config/aarch64/aarch64-builtins.c
5206 (aarch64_types_quadopu_lane_qualifiers): New.
5207 (TYPES_QUADOPU_LANE): New.
5208 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5209 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5210 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5211 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5212 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5213 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5214 (Vdottype, DOTPROD): New.
5215 (sur): Add SDOT and UDOT.
5217 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5219 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5220 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5221 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5223 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5224 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5225 Enable TARGET_DOTPROD.
5226 (cortex-a75.cortex-a55): Likewise.
5227 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5229 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5231 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5232 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5234 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5236 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5237 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5238 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5239 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5240 * config/arm/types.md (neon_dot, neon_dot_q): New.
5241 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5243 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5245 * config/arm/arm.h (TARGET_DOTPROD): New.
5246 * config/arm/arm.c (arm_arch_dotprod): New.
5247 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5248 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5249 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5250 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5251 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5252 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5253 * doc/invoke.texi (armv8.2-a): Document dotprod
5255 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5257 * i386.c (ix86_vec_cost): New function.
5258 (ix86_rtx_costs): Handle vector operations better.
5259 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5260 * x86-tune-costs.h: Add new costs to all tables.
5262 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5264 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5266 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5267 divsd, sqrtss and sqrtsd
5268 * x86-tune-costs.h: Add new entries to all costs.
5269 (znver1_cost): Fix to match real instruction latencies.
5271 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5272 Michael Collison <michael.collison@arm.com>
5274 * compare-elim.c: Include emit-rtl.h.
5275 (can_merge_compare_into_arith): New function.
5276 (try_validate_parallel): Likewise.
5277 (try_merge_compare): Likewise.
5278 (try_eliminate_compare): Call the above when no previous clobber
5280 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5283 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5287 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5288 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5292 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5293 to be any operand_equal_p operands. For & (B - 1) require
5294 B to be power of 2. Recognize
5295 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5297 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5300 * optabs.c (expand_memory_blockage): Fix call of
5301 targetm.have_memory_blockage.
5303 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5306 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5307 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5308 objects to extra_objs instead of overwriting it.
5310 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5312 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5313 Use any_fp_register_operand as operand[3] predicate. Simplify
5314 equality test for operands[2] and operands[4] memory location.
5315 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5316 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5317 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5318 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5319 any_fp_register_operand as operand[1] predicate. Simplify
5320 equality test for operands[0] and operands[3] memory location.
5321 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5322 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5323 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5325 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5327 * target-insns.def: Add memory_blockage.
5328 * optabs.c (expand_memory_blockage): New function.
5329 (expand_asm_memory_barrier): Rename ...
5330 (expand_asm_memory_blockage): ... to this.
5331 (expand_mem_thread_fence): Call expand_memory_blockage
5332 instead of expand_asm_memory_barrier.
5333 (expand_mem_singnal_fence): Ditto.
5334 (expand_atomic_load): Ditto.
5335 (expand_atomic_store): Ditto.
5336 * doc/md.texi (Standard Pattern Names For Generation):
5337 Document memory_blockage instruction pattern.
5339 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5341 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5342 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5343 * config/rl78/rl78.md: New define_expand "adddi3".
5345 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5347 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5350 2017-10-13 Jeff Law <law@redhat.com>
5352 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5354 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5357 * internal-fn.c (expand_mul_overflow): If both operands have
5358 the same highpart of -1 or 0 and the topmost bit of lowpart
5359 is different, overflow is if res <= 0 rather than res < 0.
5361 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5363 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5364 TARGET_P9_VECTOR code for unaligned_load case.
5366 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5368 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5370 2017-10-13 Nathan Sidwell <nathan@acm.org>
5372 * tree-core.h (tree_contains_struct): Make bool.
5373 * tree.c (tree_contains_struct): Likewise.
5374 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5375 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5376 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5377 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5378 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5380 2017-10-13 Richard Biener <rguenther@suse.de>
5382 * graphite-isl-ast-to-gimple.c
5383 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5384 parameters and dominance check.
5385 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5386 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5387 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5388 Do not update SSA form here or do intermediate IL verification.
5389 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5390 (graphite_initialize): Remove check on the number of loops in
5391 the function and inline into graphite_transform_loops.
5392 (graphite_finalize): Inline into graphite_transform_loops.
5393 (graphite_transform_loops): Perform SSA update and IL verification
5395 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5397 2017-10-13 Richard Biener <rguenther@suse.de>
5399 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5400 graphite_expression_type_precision): Avoid global constructor
5402 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5403 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5404 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5405 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5406 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5407 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5409 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5412 * config/i386/i386.h (ix86_red_zone_size): New.
5413 * config/i386/i386.md (push peephole2s): Replace
5414 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5416 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5417 Alan Hayward <alan.hayward@arm.com>
5418 David Sherwood <david.sherwood@arm.com>
5420 * combine.c (can_change_dest_mode): Reject changes in
5421 REGMODE_NATURAL_SIZE.
5423 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5424 Alan Hayward <alan.hayward@arm.com>
5425 David Sherwood <david.sherwood@arm.com>
5427 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5428 (expand_debug_source_expr): Likewise.
5429 * combine.c (combine_simplify_rtx): Likewise.
5430 * cse.c (fold_rtx): Likewise.
5431 * fwprop.c (canonicalize_address): Likewise.
5432 * targhooks.c (default_shift_truncation_mask): Likewise.
5434 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5435 Alan Hayward <alan.hayward@arm.com>
5436 David Sherwood <david.sherwood@arm.com>
5438 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5439 (widened_mode): Likewise.
5440 (expand_unop): Likewise.
5441 * ree.c (transform_ifelse): Likewise.
5442 (merge_def_and_ext): Likewise.
5443 (combine_reaching_defs): Likewise.
5444 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5446 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5447 Alan Hayward <alan.hayward@arm.com>
5448 David Sherwood <david.sherwood@arm.com>
5450 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5451 * combine.c (gen_lowpart_for_combine): Likewise.
5452 * dwarf2out.c (rtl_for_decl_location): Likewise.
5453 * final.c (alter_subreg): Likewise.
5454 * rtlhooks.c (gen_lowpart_general): Likewise.
5455 (gen_lowpart_if_possible): Likewise.
5457 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5458 Alan Hayward <alan.hayward@arm.com>
5459 David Sherwood <david.sherwood@arm.com>
5461 * calls.c (expand_call): Use subreg_lowpart_offset.
5462 * cse.c (cse_insn): Likewise.
5463 * regcprop.c (copy_value): Likewise.
5464 (copyprop_hardreg_forward_1): Likewise.
5466 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5469 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5470 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5471 =Q constraints to +Q and into insn condition add check
5472 that operands[0] and operands[1] are equal.
5473 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5474 =Q constraints to +Q and into insn condition add check
5475 that operands[0] is equal to either operands[1] or operands[2].
5478 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5479 instead of handling MINUS_EXPR twice (once for each argument),
5480 canonicalize operand order and handle just once, use rtype where
5481 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5484 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5485 any values of __C while still being pattern recognizable as a simple
5488 2017-10-13 Richard Biener <rguenther@suse.de>
5490 PR tree-optimization/82451
5492 2017-10-02 Richard Biener <rguenther@suse.de>
5494 PR tree-optimization/82355
5495 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5496 a mapping for the enclosing loop but avoid generating one for
5498 (copy_bb_and_scalar_dependences): Remove premature codegen
5499 error on PHIs in blocks duplicated into multiple places.
5500 * graphite-scop-detection.c
5501 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5502 in the region use it as loop and nest to analyze the DR in.
5503 (try_generate_gimple_bb): Likewise.
5504 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5505 (add_loop_constraints): For blocks in a loop not in the region
5506 create a dimension with a single iteration.
5507 * sese.h (gbb_loop_at_index): Remove assert.
5509 * cfgloop.c (loop_preheader_edge): For the loop tree root
5510 return the single successor of the entry block.
5511 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5512 Reset the SCEV hashtable and niters.
5513 * graphite-scop-detection.c
5514 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5515 assert that we only have POLYNOMIAL_CHREC that vary in loops
5516 contained in the region.
5517 (scop_detection::graphite_can_represent_expr): Adjust.
5518 (scop_detection::stmt_has_simple_data_refs_p): For loops
5519 not in the region set loop to NULL. The nest is now the
5520 entry edge to the region.
5521 (try_generate_gimple_bb): Likewise.
5522 * sese.c (scalar_evolution_in_region): Adjust for
5523 instantiate_scev change.
5524 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5525 Make nest parameter the edge into the region.
5526 (create_data_ref): Likewise.
5527 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5528 entry edge into a region and adjust instantiate_scev calls.
5529 (create_data_ref): Likewise.
5530 (graphite_find_data_references_in_stmt): Likewise.
5531 (find_data_references_in_stmt): Pass the loop preheader edge
5532 from the nest argument.
5533 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5534 parameter the edge into the region.
5535 (instantiate_parameters): Use the loop preheader edge as entry.
5536 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5538 (get_instantiated_value_entry): Make instantiate_below parameter
5539 the edge into the region.
5540 (instantiate_scev_name): Likewise. Adjust dominance checks,
5541 when we cannot use loop-based instantiation instantiate by
5542 walking use-def chains.
5543 (instantiate_scev_poly): Adjust.
5544 (instantiate_scev_binary): Likewise.
5545 (instantiate_scev_convert): Likewise.
5546 (instantiate_scev_not): Likewise.
5547 (instantiate_array_ref): Remove.
5548 (instantiate_scev_3): Likewise.
5549 (instantiate_scev_2): Likewise.
5550 (instantiate_scev_1): Likewise.
5551 (instantiate_scev_r): Do not blindly handle N-operand trees.
5552 Do not instantiate array-refs. Handle all constants and invariants.
5553 (instantiate_scev): Make instantiate_below parameter
5554 the edge into the region.
5555 (resolve_mixers): Use the loop preheader edge for the region
5556 parameter to instantiate_scev_r.
5557 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5559 2017-10-13 Richard Biener <rguenther@suse.de>
5561 PR tree-optimization/82525
5562 * graphite-isl-ast-to-gimple.c
5563 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5565 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5566 Fail code generation when we cannot represent the isl integer.
5567 (binary_op_to_tree): Elide modulo operations that are no-ops
5568 in the type we code generate. Remove now superfluous code
5571 2017-10-13 Richard Biener <rguenther@suse.de>
5573 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5574 (scop_detection::harmful_loop_in_region): Remove premature
5575 IV type restriction.
5576 (scop_detection::graphite_can_represent_scev): We can handle
5577 pointer IVs just fine.
5579 2017-10-13 Alan Modra <amodra@gmail.com>
5581 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5582 "Clobbers and Scratch Registers". Add paragraph on
5583 alternative to clobbers for scratch registers and OpenBLAS
5586 2017-10-13 Alan Modra <amodra@gmail.com>
5588 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5589 example of a memory input for a string of known length. Move
5590 commentary out of table. Add a number of new examples
5591 covering array memory inputs.
5593 2017-10-12 Martin Liska <mliska@suse.cz>
5595 PR tree-optimization/82493
5596 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5597 (test_range_functions): New function.
5598 (sbitmap_c_tests): Likewise.
5599 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5600 * selftest.h (sbitmap_c_tests): New function.
5602 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5604 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5606 * config/rs6000/amo.h: Fix spacing issue.
5608 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5611 * config/i386/i386.md (*ashl<mode>3_mask_1,
5612 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5613 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5616 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5618 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5619 (profile_probability): Set max_probability
5620 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5622 * profile-count.c (profile_probability::differs_from_p): Do not
5623 rely on max_probaiblity == 10000
5625 2017-10-12 Jeff Law <law@redhat.com>
5627 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5630 2017-10-12 Martin Sebor <msebor@redhat.com>
5634 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5635 (handle_alias_pairs): Call it.
5636 * common.opt (-Wattribute-alias): New option.
5637 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5638 * doc/invoke.texi (-Wattribute-alias): Document.
5640 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5643 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5645 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5647 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5648 (make_hard_regno_born, make_hard_regno_dead): Update
5649 bb_killed_pseudos and bb_gen_pseudos.
5651 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5653 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5655 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5657 * config/alpha/alpha.c (alpha_split_conditional_move):
5658 Use std::swap instead of manually swapping.
5659 (alpha_stdarg_optimize_hook): Ditto.
5660 (alpha_canonicalize_comparison): Ditto.
5662 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5664 * tree-loop-distribution.c (struct builtin_info): New struct.
5665 (struct partition): Refactor fields into struct builtin_info.
5666 (partition_free): Free struct builtin_info.
5667 (build_size_arg_loc, build_addr_arg_loc): Delete.
5668 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5669 information from struct builtin_info.
5670 (find_single_drs): New function refactored from classify_partition.
5671 Also moved builtin validity checks to this function.
5672 (compute_access_range, alloc_builtin): New functions.
5673 (classify_builtin_st, classify_builtin_ldst): New functions.
5674 (classify_partition): Refactor code into functions find_single_drs,
5675 classify_builtin_st and classify_builtin_ldst.
5676 (distribute_loop): Don't do runtime alias check when distributing
5678 (find_seed_stmts_for_distribution): New function.
5679 (pass_loop_distribution::execute): Refactor code finding seed
5680 stmts into above function. Support distribution for the innermost
5681 two-level loop nest. Adjust dump information.
5683 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5685 * tree-loop-distribution.c: Adjust the general comment.
5686 (NUM_PARTITION_THRESHOLD): New macro.
5687 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5688 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5689 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5690 in call to build_partition_graph.
5691 (finalize_partitions): New parameter. Make loop distribution more
5692 conservative by fusing more partitions.
5693 (distribute_loop): Don't do runtime alias check in case of loop nest
5695 (find_seed_stmts_for_distribution): New function.
5696 (prepare_perfect_loop_nest): New function.
5697 (pass_loop_distribution::execute): Refactor code finding seed stmts
5698 and loop nest into above functions. Support loop nest distribution.
5699 Adjust dump information accordingly.
5701 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5703 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5704 and set PTYPE_SEQUENTIAL for merged partition.
5706 2017-10-12 Richard Biener <rguenther@suse.de>
5708 PR tree-optimization/69728
5710 2017-09-19 Richard Biener <rguenther@suse.de>
5712 PR tree-optimization/69728
5713 * graphite-sese-to-poly.c (schedule_error): New global.
5714 (add_loop_schedule): Handle empty domain by failing the
5716 (build_original_schedule): Handle schedule_error.
5718 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5719 domain by returning an unchanged schedule.
5721 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5723 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5724 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5726 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5728 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5731 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5734 * expr.c (store_field): Don't optimize away bitsize == 0 store
5735 from CALL_EXPR with addressable return type.
5737 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5739 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5740 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5741 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5742 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5743 TARGET_ISEL instead of TARGET_ISEL<sel>.
5745 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5747 * config/rs6000/rs6000.c
5748 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5750 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5752 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5754 (reg_or_cint_operand): Fix comment.
5755 (reg_or_zero_operand): New predicate.
5756 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5757 * config/rs6000/rs6000.c (output_isel): Delete.
5758 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5759 instead of reg_or_cint_operand. Output instruction directly (not via
5761 (isel_unsigned_<mode>): Ditto.
5762 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5763 gpc_reg_operand. Add an instruction alternative for this. Output
5764 instruction directly.
5765 (*isel_reversed_unsigned_<mode>): Ditto.
5767 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5769 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5770 (TARGET_CANONICALIZE_COMPARISON): Define.
5772 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5775 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5776 Check whether the dest is REG before adding REG_EQUIV note.
5778 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5781 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5783 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5784 (make_hard_regno_born, make_hard_regno_dead): Update
5785 bb_killed_pseudos and bb_gen_pseudos.
5787 2017-10-11 Nathan Sidwell <nathan@acm.org>
5789 * incpath.h (enum incpath_kind): Name enum, prefix values.
5790 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5791 * incpath.c (heads, tails): Use INC_MAX.
5792 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5793 (merge_include_chains, split_quote_chain,
5794 register_include_chains): Update incpath_kind names.
5795 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5796 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5798 (add_framework_path, darwin_register_objc_includes): Likewise.
5799 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5801 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5803 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5804 Do not use float_operator operator predicate.
5805 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5806 * config/i386/predicates.md (float_operator): Remove predicate.
5808 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5810 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5811 (*jccxf_i387): Ditto.
5812 (*jcc<mode>_i387): Ditto.
5813 (*jccu<mode>_i387): Ditto.
5814 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5815 (*jcc_*_i387 splitters): Remove.
5816 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5817 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5818 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5821 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5823 * profile-count.h (slow_safe_scale_64bit): New function.
5824 (safe_scale_64bit): New inline.
5825 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5826 * profile-count.c: Include wide-int.h
5827 (slow_safe_scale_64bit): New.
5829 2017-10-11 Nathan Sidwell <nathan@acm.org>
5831 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5832 HAS_DECL_ASSEMBLER_NAME_P.
5833 * gimple-expr.c (gimple_decl_printable_name: Check
5834 HAS_DECL_ASSEMBLER_NAME_P too.
5835 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5836 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5837 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5838 * passes.c (rest_of_decl_compilation): Check
5839 HAS_DECL_ASSEMBLER_NAME_P too.
5840 * recog.c (verify_changes): Likewise.
5841 * tree-pretty-print.c (dump_decl_name): Likewise.
5842 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5844 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5845 (SET_DECL_ASSEMBLER_NAME): Use it.
5846 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5847 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5848 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5850 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5852 * config.gcc (i386, x86_64): Add extra objects.
5853 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5854 (ix86_min_insn_size): Declare.
5855 (ix86_issue_rate): Declare.
5856 (ix86_adjust_cost): Declare.
5857 (ia32_multipass_dfa_lookahead): Declare.
5858 (ix86_macro_fusion_p): Declare.
5859 (ix86_macro_fusion_pair_p): Declare.
5860 (ix86_bd_has_dispatch): Declare.
5861 (ix86_bd_do_dispatch): Declare.
5862 (ix86_core2i7_init_hooks): Declare.
5863 (ix86_atom_sched_reorder): Declare.
5864 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5865 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5866 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5867 (rip_relative_addr_p): Rename to ...
5868 (ix86_rip_relative_addr_p): ... this one; export.
5869 (memory_address_length): Update.
5870 (ix86_issue_rate): Move to x86-tune-sched.c.
5871 (ix86_flags_dependent): Move to x86-tune-sched.c.
5872 (ix86_agi_dependent): Move to x86-tune-sched.c.
5873 (exact_dependency_1): Move to x86-tune-sched.c.
5874 (exact_store_load_dependency): Move to x86-tune-sched.c.
5875 (ix86_adjust_cost): Move to x86-tune-sched.c.
5876 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5877 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5878 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5879 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5880 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5881 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5882 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5883 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5884 (min_insn_size): Rename to ...
5885 (ix86_min_insn_size): ... this one; export.
5886 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5887 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5888 (core2i7_first_cycle_multipass_backtrack): Move to
5889 x86-tune-sched-core.c.
5890 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5891 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5892 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5893 (ix86_avoid_jump_mispredicts): Update.
5894 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5895 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5896 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5897 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5898 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5899 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5900 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5901 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5902 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5903 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5904 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5905 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5906 (BIG): Move to ix86-tune-sched-bd.c.
5907 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5908 (enum insn_path): Move to ix86-tune-sched-bd.c.
5909 (get_mem_group): Move to ix86-tune-sched-bd.c.
5910 (is_cmp): Move to ix86-tune-sched-bd.c.
5911 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5912 (is_branch): Move to ix86-tune-sched-bd.c.
5913 (is_prefetch): Move to ix86-tune-sched-bd.c.
5914 (init_window): Move to ix86-tune-sched-bd.c.
5915 (allocate_window): Move to ix86-tune-sched-bd.c.
5916 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5917 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5918 (process_end_window): Move to ix86-tune-sched-bd.c.
5919 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5920 (find_constant): Move to ix86-tune-sched-bd.c.
5921 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5922 (has_immediate): Move to ix86-tune-sched-bd.c.
5923 (get_insn_path): Move to ix86-tune-sched-bd.c.
5924 (get_insn_group): Move to ix86-tune-sched-bd.c.
5925 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5926 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5927 (add_insn_window): Move to ix86-tune-sched-bd.c.
5928 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5929 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5930 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5931 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5932 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5933 (do_dispatch): Move to ix86-tune-sched-bd.c.
5934 (has_dispatch): Move to ix86-tune-sched-bd.c.
5935 * config/i386/t-i386: Add new object files.
5936 * config/i386/x86-tune-costs.h: New file.
5937 * config/i386/x86-tune-sched-atom.c: New file.
5938 * config/i386/x86-tune-sched-bd.c: New file.
5939 * config/i386/x86-tune-sched-core.c: New file.
5940 * config/i386/x86-tune-sched.c: New file.
5942 2017-10-11 Liu Hao <lh_mouse@126.com>
5944 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5945 the generic version below instead.
5946 (should_colorize): Recognize Windows consoles as terminals
5948 * pretty-print.c [__MINGW32__] (write_all): New function.
5949 [__MINGW32__] (find_esc_head): Likewise.
5950 [__MINGW32__] (find_esc_terminator): Likewise.
5951 [__MINGW32__] (eat_esc_sequence): Likewise.
5952 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5954 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5957 2017-10-11 Richard Biener <rguenther@suse.de>
5959 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5960 Properly call analyze_scalar_evolution with the loop of the stmt.
5962 2017-10-11 Richard Biener <rguenther@suse.de>
5964 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5965 * tree-core.h (tree_base): Add chrec_var union member.
5966 * tree.h (CHREC_VAR): Remove.
5967 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5968 * tree-chrec.h (build_polynomial_chrec): Adjust.
5969 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5970 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5972 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5974 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5975 * match.pd: ... here.
5976 ((T) X == (T) Y): Relax condition.
5978 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5980 PR tree-optimization/82472
5981 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5983 (break_alias_scc_partitions): Update postorder number.
5985 2017-10-11 Martin Liska <mliska@suse.cz>
5988 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5990 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5992 2017-10-11 Martin Liska <mliska@suse.cz>
5994 * print-rtl.c (print_insn): Move declaration of idbuf
5995 to same scope as name.
5997 2017-10-11 Martin Liska <mliska@suse.cz>
6002 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6005 2017-10-11 Martin Liska <mliska@suse.cz>
6008 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6011 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6013 * config.gcc (default_gnu_indirect_function): Default to yes for
6014 arm*-*-linux* with glibc.
6016 2017-10-11 Richard Biener <rguenther@suse.de>
6018 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6019 default-defs and types we do not want to analyze.
6020 (interpret_loop_phi): Replace unreachable code with an assert.
6021 (compute_scalar_evolution_in_loop): Remove and inline ...
6022 (analyze_scalar_evolution_1): ... here, replacing condition with
6023 what makes the intent clearer. Remove handling of cases
6024 get_scalar_evolution now handles.
6026 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6028 PR rtl-optimization/81434
6029 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6030 comment for main loop. In sched_group_found if, also add checks for
6031 pass and min_cost_group.
6033 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6035 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6036 (rs6000_insn_cost): New function.
6037 * config/rs6000/rs6000.md (cost): New attribute.
6039 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6040 H.J. Lu <hongjiu.lu@intel.com>
6044 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6045 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6046 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6047 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6048 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6049 OPTION_MASK_ISA_MMX and some other bit set require both
6050 mmx and the ISAs without the mmx bit.
6051 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6052 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6053 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6054 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6055 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6056 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6057 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6058 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6059 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6060 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6061 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6062 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6063 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6064 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6065 Add OPTION_MASK_ISA_MMX.
6067 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6069 * config.gcc (armv7*-*-freebsd*): New target.
6070 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6072 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6074 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6075 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6078 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6080 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6081 a template parameter.
6082 (WIDE_INT_REF_FOR): Update accordingly.
6083 * tree.h (wi::int_traits <const_tree>): Delete.
6084 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6085 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6086 (wi::tree_to_wide_ref): New typedef.
6087 (wi::to_wide): New function.
6088 * calls.c (get_size_range): Use wi::to_wide when operating on
6090 * cgraph.c (cgraph_node::create_thunk): Likewise.
6091 * config/i386/i386.c (ix86_data_alignment): Likewise.
6092 (ix86_local_alignment): Likewise.
6093 * dbxout.c (stabstr_O): Likewise.
6094 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6095 * expr.c (const_vector_from_tree): Likewise.
6096 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6097 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6098 (fold_negate_expr_1, int_const_binop_1, const_binop)
6099 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6100 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6101 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6102 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6103 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6104 (fold_not_const, round_up_loc): Likewise.
6105 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6106 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6107 (alloca_call_type): Likewise.
6108 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6109 * godump.c (go_output_typedef): Likewise.
6110 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6111 * internal-fn.c (get_min_precision): Likewise.
6112 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6113 * ipa-polymorphic-call.c
6114 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6115 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6116 (ipa_modify_call_arguments): Likewise.
6117 * match.pd: Likewise.
6118 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6119 * print-tree.c (print_node_brief, print_node): Likewise.
6120 * stmt.c (expand_case): Likewise.
6121 * stor-layout.c (layout_type): Likewise.
6122 * tree-affine.c (tree_to_aff_combination): Likewise.
6123 * tree-cfg.c (group_case_labels_stmt): Likewise.
6124 * tree-data-ref.c (dr_analyze_indices): Likewise.
6125 (prune_runtime_alias_test_list): Likewise.
6126 * tree-dump.c (dequeue_and_dump): Likewise.
6127 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6128 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6129 * tree-pretty-print.c (dump_generic_node): Likewise.
6130 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6131 (simple_iv_with_niters): Likewise.
6132 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6133 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6134 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6135 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6136 (refine_value_range_using_guard, number_of_iterations_ne_max)
6137 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6138 (get_cst_init_from_scev, record_nonwrapping_iv)
6139 (scev_var_range_cant_overflow): Likewise.
6140 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6141 * tree-ssa-pre.c (compute_avail): Likewise.
6142 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6143 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6144 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6145 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6146 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6147 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6148 (dump_case_nodes, try_switch_expansion): Likewise.
6149 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6150 (vect_do_peeling): Likewise.
6151 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6152 * tree-vect-stmts.c (vectorizable_load): Likewise.
6153 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6154 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6155 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6156 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6157 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6158 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6159 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6160 (evrp_dom_walker::before_dom_children): Likewise.
6161 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6162 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6163 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6164 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6165 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6166 (get_range_pos_neg): Likewise.
6167 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6168 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6169 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6170 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6171 * config/avr/avr.c (avr_fold_builtin): Likewise.
6172 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6173 * config/msp430/msp430.c (msp430_attr): Likewise.
6174 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6175 * config/powerpcspe/powerpcspe-c.c
6176 (altivec_resolve_overloaded_builtin): Likewise.
6177 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6178 (rs6000_expand_ternop_builtin): Likewise.
6179 * config/rs6000/rs6000-c.c
6180 (altivec_resolve_overloaded_builtin): Likewise.
6181 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6182 (rs6000_expand_ternop_builtin): Likewise.
6183 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6185 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6187 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6188 when copying loop nest with only one inner loop.
6190 2017-10-10 Richard Biener <rguenther@suse.de>
6192 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6193 blocks if SCEV is active.
6194 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6196 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6197 (scev_initialize): Assert we are not yet initialized.
6199 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6201 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6202 inner loop's exit stmt by making it always exit the loop, otherwise
6203 we would generate an infinite empty loop.
6205 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6207 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6208 renaming variables in new preheader if it's deleted.
6210 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6212 * tree-loop-distribution.c (struct partition): Remove unused field
6213 loops of the structure.
6214 (partition_alloc, partition_free): Ditto.
6215 (build_rdg_partition_for_vertex): Ditto.
6217 2017-10-09 Jeff Law <law@redhat.com>
6219 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6220 return type to match prototype and documentation.
6222 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6224 * config/rs6000/rs6000.c (processor_costs): Move to ...
6225 * config/rs6000/rs6000.h: ... here.
6226 (rs6000_cost): Declare.
6228 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6230 * except.c (setjmp_fn): New global variable.
6231 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6232 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6233 if DONT_USE_BUILTIN_SETJMP is defined.
6235 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6237 * target.def (insn_cost): New hook.
6238 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6239 * doc/tm.texi: Regenerate.
6240 * rtlanal.c (insn_cost): Use the new hook.
6242 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6244 * combine.c (combine_validate_cost): Compute the new insn_cost,
6245 not just pattern_cost.
6246 (try_combine): Adjust comment.
6248 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6250 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6252 * combine.c (uid_insn_cost): Adjust comment.
6253 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6255 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6256 * dse.c (find_shift_sequence): Ditto.
6257 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6258 (bb_valid_for_noce_process_p): Use pattern_cost.
6259 * rtl.h (insn_rtx_cost): Delete.
6260 (pattern_cost): New prototype.
6261 (insn_cost): New prototype.
6262 * rtlanal.c (insn_rtx_cost): Rename to...
6263 (pattern_cost): ... this.
6266 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6268 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6269 (*jcc<mode>_0_r_i387): Ditto.
6270 (*jccxf_r_i387): Ditto.
6271 (*jcc<mode>_r_i387): Ditto.
6272 (*jccu<mode>_r_i387): Ditto.
6273 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6274 (*jcc): Rename from *jcc_1.
6276 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6278 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6279 deferred rescans after the lvx/stvx recombination pre-pass.
6281 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6283 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6284 memory operation instruction support.
6285 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6286 (rs6000-ibm-aix[789]*): Likewise.
6287 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6288 Document new functions.
6290 2017-10-09 Richard Biener <rguenther@suse.de>
6292 PR tree-optimization/82397
6293 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6294 equality only for semantically equal trees.
6296 2017-10-09 Richard Biener <rguenther@suse.de>
6298 PR tree-optimization/82449
6299 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6300 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6301 allow constant addresses.
6302 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6305 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6307 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6310 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6313 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6316 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6319 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6321 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6324 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6325 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6327 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6329 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6330 (WI_BINARY_PREDICATE_RESULT): Likewise.
6331 (wi::binary_traits::operator_result): New type.
6332 (wi::binary_traits::predicate_result): Likewise.
6333 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6334 (generic_wide_int::operator==, generic_wide_int::operator!=)
6335 (generic_wide_int::operator&, generic_wide_int::and_not)
6336 (generic_wide_int::operator|, generic_wide_int::or_not)
6337 (generic_wide_int::operator^, generic_wide_int::operator+
6338 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6340 (operator~, unary operator-, operator==, operator!=, operator&)
6341 (operator|, operator^, operator+, binary operator-, operator*): New
6343 * expr.c (get_inner_reference): Use wi::bit_and_not.
6344 * fold-const.c (fold_binary_loc): Likewise.
6345 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6346 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6347 (bit_value_binop): Likewise.
6348 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6349 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6350 (extract_range_from_binary_expr_1): Likewise.
6351 (masked_increment): Likewise.
6352 (simplify_bit_ops_using_ranges): Likewise.
6354 2017-10-09 Martin Jambor <mjambor@suse.cz>
6357 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6358 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6359 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6361 (hsa_fixup_mov_insn_type): New function.
6362 (hsa_op_with_type::get_in_type): Use it.
6363 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6364 immediates in an assert.
6365 (hsa_op_with_type::extend_int_to_32bit): New method.
6366 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6367 types. Convert to dest type if necessary.
6368 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6369 (reg_for_gimple_ssa): Pass false as min32int to
6370 hsa_type_for_scalar_tree_type.
6371 (gen_hsa_addr): Fixup type when creating addresable temporary.
6372 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6373 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6374 necessary. Call hsa_fixup_mov_insn_type.
6375 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6376 extend operands and convert to dest type if necessary.
6377 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6378 to dest type if necessary.
6379 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6380 if conversion nt necessary and size matches.
6381 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6382 to dest type if necessary.
6383 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6384 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6386 (gen_hsa_clrsb): Likewise.
6387 (gen_hsa_ffs): Likewise.
6388 (gen_hsa_divmod): Extend operands and convert to dest type if
6390 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6392 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6394 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6395 Remove empty default arguments. Use a brace block as output
6397 (conditional return): Ditto.
6399 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6400 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6402 (group_ending_nop): Ditto.
6403 (doloop_end): Ditto.
6404 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6405 (splitters for those): Ditto.
6407 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6409 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6410 a conditional jump (and the compare for it) so that pc_rtx is the
6412 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6413 for the deleted and renamed ctr<mode>_internal[234] patterns.
6414 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6415 Delete second conditional return pattern.
6416 (ctr<mode>_internal2): Delete this second bdnz pattern.
6417 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6418 (ctr<mode>_internal4): Delete this second bdz pattern.
6420 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6422 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6423 (always_initialized_rtx_for_ssa_name_p): New predicate.
6424 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6425 (finish_out_of_ssa): Free new field of SA.
6426 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6427 * tree-ssa-coalesce.c: Include tree-ssa.h.
6428 (get_parm_default_def_partitions): Remove extern keyword.
6429 (get_undefined_value_partitions): New function.
6430 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6431 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6433 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6435 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6437 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6439 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6441 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6442 for targets that preffer 128bit.
6444 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6446 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6448 2017-10-08 Olivier Hainque <hainque@adacore.com>
6450 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6451 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6453 (thumb_set_return_address): Likewise.
6455 2017-10-08 Olivier Hainque <hainque@adacore.com>
6457 * common/config/arm/arm-common.c (arm_except_unwind_info):
6458 Handle DWARF2_UNWIND_INFO.
6460 2017-10-07 Michael Collison <michael.collison@arm.com>
6462 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6465 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6467 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6468 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6469 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6470 defined, force the creation of a new block for a dispatch label.
6472 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6474 * invoke.texi (Wsuggest-attribute=cold): Document.
6475 * common.opt (Wsuggest-attribute=cold): New
6476 * ipa-pure-const.c (warn_function_cold): New function.
6477 * predict.c (compute_function_frequency): Use it.
6478 * predict.h (warn_function_cold): Declare.
6480 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6482 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6485 2017-10-06 Martin Liska <mliska@suse.cz>
6487 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6488 keyword for member functions.
6489 (struct sanopt_tree_couple): New struct.
6490 (struct sanopt_tree_couple_hash): New function.
6491 (struct sanopt_ctx): Add new hash_map.
6492 (has_dominating_ubsan_ptr_check): New function.
6493 (record_ubsan_ptr_check_stmt): Likewise.
6494 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6495 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6496 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6498 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6501 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6502 aarch64_simd_valid_immediate on CONST_VECTORs.
6503 (aarch64_reg_or_bic_imm): Likewise.
6505 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6507 PR rtl-optimization/82396
6508 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6510 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6512 * graphite-dependences.c (scop_get_reads): Move code to...
6513 (scop_get_must_writes): Move code to...
6514 (scop_get_may_writes): Move code to...
6515 (scop_get_reads_and_writes): ... here.
6516 (scop_get_dependences): Call scop_get_reads_and_writes.
6518 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6520 PR tree-optimization/82434
6521 * fold-const.h (can_native_encode_type_p,
6522 can_native_encode_string_p): Remove.
6523 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6524 don't encode anything, just return what would be otherwise returned.
6525 (native_encode_fixed, native_encode_complex, native_encode_vector):
6527 (native_encode_string): Likewise. Inline by hand
6528 can_native_encode_string_p.
6529 (can_native_encode_type_p): Remove.
6530 (can_native_encode_string_p): Remove.
6531 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6532 STRING_CSTs using can_native_encode_string_p, test all
6533 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6534 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6535 argument from native_encode_expr.
6536 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6537 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6540 2017-10-06 Richard Biener <rguenther@suse.de>
6542 PR tree-optimization/82397
6543 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6544 operand_equal_p but rely on data_ref_compare_tree for detecting
6546 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6547 to match up with dr_group_sort_cmp.
6549 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6552 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6554 * config/s390/s390-builtin-types.def: Regenerate.
6556 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6559 * config/s390/s390-builtin-types.def: Regenerate.
6560 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6561 Change flag from B_VXE to B_VX.
6562 (s390_vec_min_dbl): Remove B_VXE flag.
6564 2017-10-06 Richard Biener <rguenther@suse.de>
6566 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6567 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6568 translate_isl_ast_to_gimple::is_valid_rename,
6569 translate_isl_ast_to_gimple::get_rename,
6570 translate_isl_ast_to_gimple::get_def_bb_for_const,
6571 translate_isl_ast_to_gimple::get_new_name,
6572 translate_isl_ast_to_gimple::collect_all_ssa_names,
6573 translate_isl_ast_to_gimple::copy_loop_phi_args,
6574 translate_isl_ast_to_gimple::collect_all_ssa_names,
6575 translate_isl_ast_to_gimple::copy_loop_phi_args,
6576 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6577 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6578 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6579 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6580 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6581 translate_isl_ast_to_gimple::copy_cond_phi_args,
6582 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6583 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6584 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6585 translate_isl_ast_to_gimple::rename_uses,
6586 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6587 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6588 (set_rename_for_each_def): Likewise.
6589 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6590 here. Handle rewriting SCEV analyzable uses here.
6591 (copy_bb_and_scalar_dependences): Generate code for PHI
6593 (graphite_regenerate_ast_isl): Adjust.
6594 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6595 (add_write, add_read): New functions.
6596 (build_cross_bb_scalars_def): Use it and simplify.
6597 (build_cross_bb_scalars_use): Likewise.
6598 (graphite_find_cross_bb_scalar_vars): Inline into...
6599 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6600 simulating out-of-SSA. Compute liveout and add dependencies.
6601 (build_scops): Force an empty entry block.
6602 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6604 (sese_build_liveouts): Declare.
6605 (sese_trivially_empty_bb_p): Likewise.
6606 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6607 compute liveout and debug_liveout.
6608 (sese_bad_liveouts_use): Remove.
6609 (sese_reset_debug_liveouts_bb): Likewise.
6610 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6611 (sese_build_liveouts): Build liveout and debug_liveout and store
6613 (new_sese_info): Adjust.
6614 (free_sese_info): Likewise.
6615 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6616 do not build liveout here.
6617 (move_sese_in_condition): Adjust region entry.
6618 (scev_analyzable_p): Match up with chrec_apply requirements.
6619 (sese_trivially_empty_bb_p): New.
6620 * tree-into-ssa.c (get_reaching_def): Properly support generating
6621 default-defs for incremental rewrite of anonymous names.
6623 2017-10-06 Richard Biener <rguenther@suse.de>
6625 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6626 precision do not perform modulo reduction.
6628 2017-10-06 Richard Biener <rguenther@suse.de>
6630 PR tree-optimization/82436
6631 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6632 conservatively choose the vectorization factor when checking
6633 whether we can perform the required load permutation.
6634 (vect_transform_slp_perm_load): Assert when we may not fail.
6636 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6638 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6639 message for incompatible -msdata=* and -mcall-* options.
6641 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6643 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6644 rate for post-reload scheduling.
6646 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6648 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6650 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6652 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6653 to improve monte carlo in scimark.
6655 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6657 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6658 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6659 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6660 pentium4_cost, nocona_cost): Set reassociation width to 1.
6661 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6662 width to 2 for fp operations and 1 otherwise.
6663 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6665 (atom_cost): Set reassociation width to 2.
6666 (slm_cost, generic_cost): Set fp reassociation width
6667 to 2 and 1 otherwise.
6668 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6669 (core_cost): Set fp reassociation width to 4 and vector to 2.
6670 (ix86_reassociation_width): Rewrite using cost table; special case
6671 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6672 and TARGET_AVX128_OPTIMAL.
6673 * config/i386/i386.h (processor_costs): Add
6674 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6675 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6676 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6677 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6678 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6679 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6681 2017-10-05 Nathan Sidwell <nathan@acm.org>
6683 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6685 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6687 * config/arm/arm.c (arm_test_fpu_data): New.
6688 (arm_run_selftests): Call arm_test_fpu_data.
6690 2017-10-04 Nathan Sidwell <nathan@acm.org>
6692 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6694 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6696 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6698 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6699 check type for aarch64_simd_valid_immediate.
6700 (aarch64_output_simd_mov_immediate): Update prototype.
6701 (aarch64_simd_valid_immediate): Update prototype.
6702 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6703 support for ORR-immediate.
6704 (and<mode>3): modified pattern to add support for BIC-immediate.
6705 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6706 now checks for valid immediate for BIC and ORR based on new enum
6708 (aarch64_output_simd_mov_immediate): Function now used to output
6709 BIC/ORR imm as well based on new enum argument.
6710 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6712 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6713 (aarch64_reg_or_bic_imm): Likewise.
6715 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6717 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6718 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6720 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6724 PR rtl-optimization/82396
6725 * haifa-sched.c (autopref_multipass_init): Simplify
6727 (autopref_rank_data): Simplify sort order.
6728 * sched-int.h (autopref_multipass_data_): Remove
6729 multi_mem_insn_p, min_offset and max_offset.
6731 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6733 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6735 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6737 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6740 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6742 * doc/sourcebuild.texi: Document vect_long_mult.
6744 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6746 PR tree-optimization/82413
6747 * fold-const.c (build_range_check): Use widest_int when comparing
6748 the maximum ETYPE value with HIGH.
6750 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6752 PR rtl-optimization/82396
6753 * haifa-sched.c (autopref_multipass_init): Simplify
6755 (autopref_rank_data): Simplify sort order.
6756 * sched-int.h (autopref_multipass_data_): Remove
6757 multi_mem_insn_p, min_offset and max_offset.
6759 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6761 PR tree-optimization/82381
6762 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6763 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6766 PR tree-optimization/82374
6767 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6768 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6769 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6770 current_function_decl to the new decl.
6772 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6774 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6775 helper macro for IEEE float128 hardware built-in functions.
6776 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6778 (TRUNCF128_ODD): Likewise.
6779 (ADDF128_ODD): Likewise.
6780 (SUBF128_ODD): Likewise.
6781 (MULF128_ODD): Likewise.
6782 (DIVF128_ODD): Likewise.
6783 (FMAF128_ODD): Likewise.
6784 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6785 UNSPEC_TRUNC_ROUND_TO_ODD.
6786 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6787 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6788 floating point round to odd instructions.
6789 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6790 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6791 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6792 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6793 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6794 (trunc<mode>sf2_hw): Change the truncate with round to odd
6795 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6796 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6797 to odd hardware instructions.
6798 (sub<mode>3_odd): Likewise.
6799 (mul<mode>3_odd): Likewise.
6800 (div<mode>3_odd): Likewise.
6801 (sqrt<mode>2_odd): Likewise.
6802 (fma<mode>4_odd): Likewise.
6803 (fms<mode>4_odd): Likewise.
6804 (nfma<mode>4_odd): Likewise.
6805 (nfms<mode>4_odd): Likewise.
6806 (trunc<mode>df2_odd): Change the truncate with round to odd
6807 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6809 * doc/extend.texi (PowerPC built-in functions): Update documentation
6810 for existing IEEE float128-bit built-in functions. Add built-in
6811 functions that generate the IEEE 128-bit floating point round to
6814 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6816 PR rtl-optimization/77729
6817 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6818 to (X&(C1&~C2))|C2 transformations.
6820 2017-10-03 Martin Jambor <mjambor@suse.cz>
6822 PR tree-optimization/82363
6823 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6824 mismatch, mark lacc written regardless of racc.
6826 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6828 PR tree-optimization/82381
6829 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6830 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6831 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6832 fallthrough into reassoc_stmt_dominates_stmt_p.
6835 * combine.c (combine_instructions): Don't combine in unreachable
6838 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6841 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6842 function to not use the have_cpu variable. Do not set cpu_index,
6843 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6845 (rs6000_valid_attribute_p): Remove duplicate initializations of
6846 old_optimize and func_optimize.
6847 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6848 (rs6000_activate_target_options): Make global.
6849 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6852 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6854 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6855 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6856 Set *poffset to 0 and *psize and *pmax_size to -1 if
6857 *poffset + *psize overflows in HOST_WIDE_INT.
6859 PR tree-optimization/82387
6860 PR tree-optimization/82388
6861 PR tree-optimization/82389
6862 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6863 instead of live_bytes non-NULL.
6865 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6868 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6871 2017-10-02 Richard Biener <rguenther@suse.de>
6873 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6874 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6875 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6877 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6879 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6880 requested precision matches the type's.
6881 * calls.c (alloc_max_size): Calculate the new candidate size as
6882 a widest_int and use wi::to_widest when comparing it with the
6883 current candidate size.
6884 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6885 zero rather than integer_zero_node.
6886 * match.pd: Check for a no-op conversion before using wi::add
6887 rather than after. Use tree_to_uhwi when summing small shift
6888 counts into an unsigned int.
6890 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6891 Alan Hayward <alan.hayward@arm.com>
6892 David Sherwood <david.sherwood@arm.com>
6895 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6896 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6897 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6898 POINTER_AND_FP_REGS.
6900 2017-10-02 Richard Biener <rguenther@suse.de>
6902 PR tree-optimization/82355
6903 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6904 a mapping for the enclosing loop but avoid generating one for
6906 (copy_bb_and_scalar_dependences): Remove premature codegen
6907 error on PHIs in blocks duplicated into multiple places.
6908 * graphite-scop-detection.c
6909 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6910 in the region use it as loop and nest to analyze the DR in.
6911 (try_generate_gimple_bb): Likewise.
6912 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6913 (add_loop_constraints): For blocks in a loop not in the region
6914 create a dimension with a single iteration.
6915 * sese.h (gbb_loop_at_index): Remove assert.
6917 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6919 * omp-expand.c (adjust_context_scope): New function.
6920 (expand_parallel_call): Call adjust_context_scope.
6922 2017-10-01 Jeff Law <law@redhat.com>
6924 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6925 dom_opt_dom_walker class with direct access to private members.
6926 Add comments. Call test_for_singularity.
6927 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6928 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6929 m_dummy_cond anymore.
6930 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6932 (pass_dominator:execute): Build the dummy_cond here and pass it
6933 to the dom_opt_dom_walker ctor.
6934 (test_for_singularity): New function.
6936 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6937 Maya Rashish <coypu@sdf.org>
6939 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6940 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6941 (alpha*-*-netbsd*) Use nbsd_tm_file.
6942 (arm*-*-netbsdelf*) Likewise.
6943 (i[34567]86-*-netbsdelf*) Likewise.
6944 (x86_64-*-netbsd*) Likewise.
6945 (mips*-*-netbsd*) Likewise.
6946 (powerpc-*-netbsd*) Likewise.
6947 (sh*-*-netbsd*) Likewise.
6948 (sparc-*-netbsdelf*) Likewise.
6949 (sparc64-*-netbsd*) Likewise.
6950 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6952 (vax-*-netbsdelf*) Likewise.
6953 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6954 (UINT_FAST8_TYPE) Likewise.
6955 (INT_FAST16_TYPE) Check CHAR_FAST16.
6956 (UINT_FAST16_TYPE) Likewise.
6958 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6961 * config/i386/i386.md
6962 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6963 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6964 *divmodsi4_zext_2): New define_insn_and_split.
6965 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6966 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6967 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6968 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6969 New define_insn_and_split.
6970 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6971 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6972 operands[1] having DImode when mode is SImode.
6974 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6975 always SImode for DIV and MOD in REG_EQUAL notes.
6977 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6980 * match.pd: Fix handling of NaNs in pattern.
6982 2017-09-29 Jeff Law <law@redhat.com>
6984 * sbitmap.c (bitmap_bit_in_range_p): New function.
6985 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6986 * tree-ssa-dse.c (live_bytes_read): New function.
6987 (dse_classify_store): Ignore reads of dead bytes.
6989 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6990 typos and whitespace errors.
6991 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6992 * config/s390/s390.c (s390_emit_prologue): Likewise.
6994 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6997 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6998 with a symbol for LRA.
7000 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7002 PR rtl-optimization/82338
7003 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7005 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7007 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7008 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7009 (qsort_chk): Declare.
7010 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7011 (qsort_chk): New function.
7013 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7015 PR tree-optimization/82337
7016 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7017 phi definition if the PHI result appears in an abnormal PHI.
7018 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7019 basis appears in an abnormal PHI.
7021 2017-09-29 Richard Biener <rguenther@suse.de>
7023 * graphite-isl-ast-to-gimple.c
7024 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7025 (binary_op_to_tree): Use it.
7026 (get_rename_from_scev): Likewise.
7027 (copy_loop_phi_nodes): Likewise.
7028 (copy_bb_and_scalar_dependences): Likewise.
7029 (translate_pending_phi_nodes): Likewise.
7031 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7034 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7035 for movabsq $(i32 << shift), r64.
7037 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7039 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7040 index when encoding %esp as %rsp to avoid 0x67 prefix.
7042 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7044 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7045 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7047 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7049 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7050 Extensions with more than 16 double VFP registers.
7051 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7052 to_clear_mask and all code related to it. Replace the remaining
7053 entry by a sbitmap and adapt code accordingly.
7055 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7057 * brig-builtins.def: Change pure attributes to const.
7059 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7061 * config.gcc (default_gnu_indirect_function): Default to yes for
7062 sparc*-*-linux* with glibc.
7064 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7066 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7067 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7068 when creating .init_array and .fini_array sections with priority
7071 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7074 * config/aarch64/aarch64.c
7075 (aarch64_builtin_support_vector_misalignment): Always return false
7076 when misalignment is unknown.
7078 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7080 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7081 this function to return false if the definition used by the swap
7082 instruction is artificial, or if the memory address from which the
7083 constant value is loaded is not represented by a base address held
7084 in a register or if the base address register is a frame or stack
7085 pointer. Additionally, return false if the base address of the
7086 loaded constant is a SYMBOL_REF but is not considered to be a
7088 (replace_swapped_load_constant): New function.
7089 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7090 loaded constant vector with a load of a swapped constant vector.
7092 2017-09-27 Carl Love <cel@us.ibm.com>
7094 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7095 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7096 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7099 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7101 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7102 first, always call autopref_rank_data otherwise.
7104 2017-09-27 Richard Biener <rguenther@suse.de>
7106 * graphite-scop-detection.c (find_scop_parameters): Move
7107 loop bound handling ...
7108 (gather_bbs::before_dom_children): ... here, avoiding the need
7109 to build scop_info->loop_nest.
7110 (record_loop_in_sese): Remove.
7111 * sese.h (sese_info_t::loop_nest): Remove.
7112 * sese.c (new_sese_info): Do not allocate loop_nest.
7113 (free_sese_info): Do not free loop_nest.
7115 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7118 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7119 lhs from calls if the lhs has addressable type.
7121 2017-09-27 Richard Biener <rguenther@suse.de>
7123 * graphite.h (scop::max_alias_set): New member.
7124 * graphite-scop-detection.c: Remove references to non-existing
7125 --param in comments.
7126 (build_alias_sets): Record the maximum alias set used for drs.
7127 (build_scops): Support zero as unlimited for
7128 --param graphite-max-arrays-per-scop.
7129 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7131 (build_poly_sr_1): ... here. Compute alias set based on the
7132 maximum alias set used for drs rather than
7133 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7135 2017-09-27 Richard Biener <rguenther@suse.de>
7137 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7138 --param loop-block-tile-size=0 to disable tiling.
7140 2017-09-27 Richard Biener <rguenther@suse.de>
7142 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7143 (graphite-max-nb-scop-params): Document special value zero.
7144 * domwalk.h (dom_walker::STOP): New symbolical constant.
7145 (dom_walker::dom_walker): Add optional parameter for bb to
7147 (dom_walker::~dom_walker): Declare.
7148 (dom_walker::before_dom_children): Document STOP return value.
7149 (dom_walker::m_user_bb_to_rpo): New member.
7150 (dom_walker::m_bb_to_rpo): Likewise.
7151 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7152 mapping here if not provided by the user.
7153 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7154 provided by the user.
7155 (dom_walker::STOP): Define.
7156 (dom_walker::walk): Do not compute bb to RPO mapping here.
7157 Support STOP return value from before_dom_children to stop
7159 * graphite-optimize-isl.c (optimize_isl): If the schedule
7160 is the same still generate code if -fgraphite-identity
7161 or -floop-parallelize-all are given.
7162 * graphite-scop-detection.c: Include cfganal.h.
7163 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7165 (gather_bbs::before_dom_children): Return STOP for BBs
7167 (build_scops): Compute bb to RPO mapping and pass it to
7168 the domwalk. Treat --param graphite-max-nb-scop-params=0
7169 as not limiting the number of params.
7170 * graphite.c (graphite_initialize): Remove limit on the
7171 number of basic-blocks in a function.
7172 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7173 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7174 default value of 10.
7176 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7178 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7179 Adjust code to eliminate needing to do the shift right 32-bits
7180 operation after XSCVDPSPN.
7182 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7184 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7185 ((X / Y) != 0 -> X >= Y): Likewise.
7187 2017-09-26 Carl Love <cel@us.ibm.com>
7189 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7190 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7191 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7192 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7193 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7194 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7195 definitions and overloading.
7196 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7197 statement for P9V_BUILTIN_XST_LEN_R.
7198 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7199 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7200 define_expand and define_insn for the instructions and builtins.
7201 * doc/extend.texi: Update the built-in documentation file for the new
7203 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7204 define_insn for the instructions
7206 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7209 * gcc/config/netbsd-protos.h: New file.
7210 * gcc/config/netbsd.c: New file.
7211 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7212 * gcc/config/t-netbsd: New file.
7213 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7214 (tmake_file) Add t-netbsd.
7215 (extra_objs) Add netbsd.o.
7217 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7221 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7223 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7225 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7226 sign extension from a vector register to a GPR by doing a 32-bit
7227 direct move and then an EXTSW.
7228 (extendsi<mode>2 splitter): Likewise.
7229 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7230 right or vector extract after doing XSCVDPSPN. Use
7231 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7233 (movdi_from_sf_zero_ext): Likewise.
7234 (reload_gpr_from_vsxsf): Likewise.
7235 (p8_mfvsrd_4_disf): Delete, no longer used.
7236 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7237 then needing to move the SFmode to a GPR to use the XSCVDPSP
7238 instruction instead of FRSP and XSCVDPSPN.
7239 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7240 it is adjacent to the other XSCVSPDP insns.
7241 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7242 SFmode to be in traditional Altivec registers.
7243 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7244 (vsx_xscvspdpn): Likewise.
7245 (vsx_xscvspdpn_scalar): Likewise.
7247 2017-09-26 Martin Jambor <mjambor@suse.cz>
7249 * tree-sra.c (compare_access_positions): Put integral types first,
7250 stabilize sorting of integral types, remove conditions putting
7251 non-full-precision integers last.
7252 (sort_and_splice_var_accesses): Disable scalarization if a
7253 non-integert would be represented by a non-full-precision integer.
7255 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7257 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7258 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7259 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7260 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7261 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7262 conditionals inside the function instead of around it. Call
7263 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7264 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7266 2017-09-26 Richard Biener <rguenther@suse.de>
7268 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7270 (scop_detection::build_scop_breadth): ... this. Removed.
7271 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7272 (scop_detection::harmful_stmt_in_bb): Likewise.
7273 (scop_detection::graphite_can_represent_stmt): Likewise.
7274 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7275 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7276 (scop_detection::can_represent_loop_1): ... this. Removed.
7277 (scop_detection::harmful_loop_in_region): Simplify after inlining
7278 the above and remove more quadraticness.
7279 (build_scops): Adjust.
7280 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7283 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7286 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7287 REGNO (base) == SP_REG if base is a REG.
7290 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7291 if it is different SSA_NAME.
7292 (optimize_range_tests_cmp_bitwise): New function.
7293 (optimize_range_tests): Call it.
7295 2017-09-26 Richard Biener <rguenther@suse.de>
7297 PR tree-optimization/82321
7298 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7299 for the def being inside the loop.
7301 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7303 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7305 * config/s390/s390-builtins.def: Fix constraint on op4.
7307 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7309 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7310 independent expanders.
7311 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7312 ("vec_ordered", "vec_unordered"): New expanders.
7314 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7316 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7319 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7321 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7322 vec_unpacks_lo_v16qi.
7323 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7325 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7327 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7328 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7329 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7331 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7333 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7335 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7337 ("*vec_slb<mode>"): New insn pattern.
7338 ("vec_shr_<mode>"): New expander.
7339 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7340 and force the shift count operand to V16QImode.
7341 ("vec_srb<mode>"): Set shift count mode to V16QI.
7343 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7345 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7346 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7347 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7349 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7352 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7354 2017-09-26 Richard Biener <rguenther@suse.de>
7356 PR tree-optimization/82320
7357 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7360 2017-09-25 Jeff Law <law@redhat.com>
7362 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7363 prototype for new argument.
7364 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7365 mostly extracted from rs6000_emit_allocate_stack.
7366 (rs6000_emit_probe_stack_range_stack_clash): New function.
7367 (rs6000_emit_allocate_stack): Call
7368 rs6000_emit_probe_stack_range_stack_clash as needed.
7369 (rs6000_emit_probe_stack_range): Add additional argument
7370 to call to gen_probe_stack_range{si,di}.
7371 (output_probe_stack_range): New.
7372 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7373 (output_probe_stack_range_stack_clash): New.
7374 (rs6000_emit_prologue): Emit notes into dump file as requested.
7375 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7376 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7377 Add additional operand and pass it to output_probe_stack_range.
7379 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7381 PR tree-optimization/82163
7382 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7383 (checking_verify_loop_closed_ssa): New parameter.
7384 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7385 (check_loop_closed_ssa_stmt): Delete.
7386 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7387 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7388 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7391 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7393 * brig-builtins.def: Treat HSAIL barrier builtins as
7394 setjmp/longjump style functions.
7396 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7398 * target.def (constant_alignment): New hook.
7399 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7400 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7401 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7402 * doc/tm.texi: Regenerate.
7403 * targhooks.h (default_constant_alignment): Declare.
7404 (constant_alignment_word_strings): Likewise.
7405 * targhooks.c (default_constant_alignment): New function.
7406 (constant_alignment_word_strings): Likewise.
7407 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7408 instead of CONSTANT_ALIGNMENT.
7409 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7410 (force_const_mem): Likewise.
7411 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7412 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7413 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7414 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7415 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7417 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7418 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7419 constant_alignment_word_strings.
7420 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7421 (CONSTANT_ALIGNMENT): Likewise.
7422 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7423 (arm_constant_alignment): New function.
7424 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7425 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7426 constant_alignment_word_strings.
7427 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7428 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7429 constant_alignment_word_strings.
7430 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7431 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7432 (cris_constant_alignment): New function.
7433 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7434 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7435 (epiphany_constant_alignment): New function.
7436 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7437 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7438 constant_alignment_word_strings.
7439 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7440 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7441 constant_alignment_word_strings.
7442 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7443 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7444 constant_alignment_word_strings.
7445 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7446 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7447 * config/i386/i386.c (ix86_constant_alignment): Make static.
7448 Use the same interface as the target hook.
7449 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7450 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7451 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7452 constant_alignment_word_strings.
7453 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7454 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7455 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7456 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7457 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7458 constant_alignment_word_strings.
7459 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7460 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7461 constant_alignment_word_strings.
7462 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7463 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7464 constant_alignment_word_strings.
7465 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7466 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7468 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7469 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7470 * config/mips/mips.c (mips_constant_alignment): New function.
7471 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7472 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7473 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7474 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7475 (mmix_constant_alignment): Make static. Use the same interface
7477 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7478 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7479 constant_alignment_word_strings.
7480 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7481 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7482 constant_alignment_word_strings.
7483 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7484 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7485 constant_alignment_word_strings.
7486 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7487 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7488 (rs6000_constant_alignment): New function.
7489 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7490 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7491 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7492 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7493 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7494 (rs6000_constant_alignment): New function.
7495 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7496 * config/s390/s390.c (s390_constant_alignment): New function.
7497 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7498 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7499 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7500 constant_alignment_word_strings.
7501 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7502 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7503 (sparc_constant_alignment): New function.
7504 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7505 * config/spu/spu.c (spu_constant_alignment): New function.
7506 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7507 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7508 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7509 constant_alignment_word_strings.
7510 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7511 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7512 constant_alignment_word_strings.
7513 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7514 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7515 constant_alignment_word_strings.
7516 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7517 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7518 (visium_constant_alignment): New function.
7519 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7520 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7521 (xtensa_constant_alignment): New function.
7522 * system.h (CONSTANT_ALIGNMENT): Poison.
7524 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7526 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7527 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7528 (rs6000_builtin_valid_without_lhs): New helper function.
7529 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7530 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7532 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7534 * target.h (vec_perm_indices): Use unsigned short rather than
7536 (auto_vec_perm_indices): Likewise.
7537 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7538 Use unsigned int rather than unsigned char.
7539 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7541 2017-09-25 Richard Biener <rguenther@suse.de>
7543 * cfgloop.h (sort_sibling_loops): Declare.
7544 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7545 (sort_sibling_loops): New function sorting the sibling loop list
7547 * graphite.c (graphite_transform_loops): Sort sibling loops.
7549 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7551 * target.def (vec_perm_const_ok): Change sel parameter to
7553 * optabs-query.c (can_vec_perm_p): Update accordingly.
7554 * doc/tm.texi: Regenerate.
7555 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7556 auto_vec_perm_indices and remove separate nelt field.
7557 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7558 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7559 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7560 (aarch64_expand_vec_perm_const): Update accordingly.
7561 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7562 to vec_perm_indices.
7563 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7564 auto_vec_perm_indices and remove separate nelt field.
7565 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7566 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7567 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7569 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7570 to vec_perm_indices.
7571 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7572 sel to vec_perm_indices.
7573 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7574 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7575 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7577 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7580 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7583 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7584 on the FUNCTION_DECL function context if it has a DIE that is a
7587 2017-09-25 Richard Biener <rguenther@suse.de>
7589 PR tree-optimization/82285
7590 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7593 2017-09-25 Tom de Vries <tom@codesourcery.com>
7597 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7600 2017-09-25 Richard Biener <rguenther@suse.de>
7602 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7603 ISL errors other than isl_error_quota happen. Dump if the
7604 schedule is the same.
7605 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7606 errors instead of aborting inside ISL.
7608 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7611 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7612 of libgcc_eh for m64.
7613 * config/i386/darwin64.h: Likewise.
7615 2017-09-25 Richard Biener <rguenther@suse.de>
7618 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7619 attribute for incomplete types nor twice for complete ones.
7621 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7624 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7625 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7628 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7631 * config/i386/i386.opt (mprefer-avx256): Use
7632 ix86_target_flags variable.
7633 * config/i386/i386.c (ix86_target_string): Move
7634 -mprefer-avx256 to flag2_opts.
7636 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7639 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7640 and x != -1 | y != -1 into (x & y) != -1.
7642 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7644 * config.gcc: Add new case statement to set
7645 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7646 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7647 s390x-*-linux* case statements. Added aarch64 to the list of
7648 supported architectures.
7650 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7652 PR tree-optimization/82289
7653 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7654 STMT_VINFO_RELEVANT_P.
7656 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7657 Alan Hayward <alan.hayward@arm.com>
7658 David Sherwood <david.sherwood@arm.com>
7660 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7661 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7662 (extract_range_from_binary_expr_1): Don't call
7663 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7665 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7666 Alan Hayward <alan.hayward@arm.com>
7667 David Sherwood <david.sherwood@arm.com>
7669 * target.def (preferred_vector_alignment): New hook.
7670 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7672 * doc/tm.texi: Regenerate.
7673 * targhooks.h (default_preferred_vector_alignment): Declare.
7674 * targhooks.c (default_preferred_vector_alignment): New function.
7675 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7677 (DR_TARGET_ALIGNMENT): New macro.
7678 (aligned_access_p): Update commentary.
7679 (vect_known_alignment_in_bytes): New function.
7680 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7682 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7683 Calculate the misalignment based on the target alignment rather than
7685 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7686 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7687 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7688 the target alignment, rather than masking the element misalignment
7689 with the number of elements in a vector. Also use the target
7690 alignment when calculating the maximum number of peels.
7691 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7692 instead of TYPE_ALIGN_UNIT.
7693 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7694 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7695 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7696 (vect_create_data_ref_ptr): Likewise.
7697 (vect_setup_realignment): Realign by ANDing with
7698 -DR_TARGET_MISALIGNMENT.
7699 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7700 the number of peels based on DR_TARGET_ALIGNMENT.
7701 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7702 with the guaranteed alignment boundary when deciding whether
7704 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7705 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7706 (ensure_base_align): Remove stmt_info parameter. Get the
7707 target base alignment from DR_TARGET_ALIGNMENT.
7708 (vectorizable_store): Update call accordingly. Interpret
7709 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7711 (vectorizable_load): Likewise.
7713 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7714 Alan Hayward <alan.hayward@arm.com>
7715 David Sherwood <david.sherwood@arm.com>
7717 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7718 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7719 (vect_enhance_data_refs_alignment): Likewise.
7721 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7723 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7724 error. Only quit immediately if parsing is complete.
7725 (BEGIN): Initialize fatal_err and parse_done.
7726 (begin fpu, end fpu): Check number of arguments.
7727 (begin arch, end arch): Likewise.
7728 (begin cpu, end cpu): Likewise.
7729 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7730 (optalias): Likewise.
7732 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7734 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7735 * config/arm/arm-isa.h: Delete. Move definitions to ...
7736 * arm-cpus.in: ... here. Use new feature and fgroup values.
7737 * config/arm/arm.c (arm_option_override): Use lower case for feature
7739 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7740 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7741 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7743 (print_isa_bits_for): New function.
7744 (gen_isa): New function.
7745 (gen_comm_data): Use print_isa_bits_for.
7746 (define feature): New keyword.
7747 (define fgroup): New keyword.
7748 * config/arm/t-arm (TM_H): Remove.
7749 (GTM_H): Add arm-isa.h.
7750 (arm-isa.h): Add rule to generate file.
7751 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7752 case for feature bit names.
7754 2017-09-22 Richard Biener <rguenther@suse.de>
7756 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7758 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7759 print of no dependency loops ...
7760 * graphite.c (graphite_transform_loops): ... here.
7761 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7763 (same_close_phi_node, remove_duplicate_close_phi,
7764 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7765 (canonicalize_loop_closed_ssa): ... here and simplify.
7766 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7767 (optimize_isl): Use dump_printf_loc to tell when we stopped
7768 optimizing because of an ISL timeout.
7770 2017-09-22 Richard Biener <rguenther@suse.de>
7772 PR tree-optimization/82291
7773 * tree-if-conv.c (predicate_mem_writes): Make sure to
7774 remove writes in blocks predicated with false.
7776 2017-09-22 Richard Biener <rguenther@suse.de>
7778 * sese.c: Include cfganal.h.
7779 (if_region_set_false_region): Remove.
7780 (create_if_region_on_edge): Likewise.
7781 (move_sese_in_condition): Re-implement without destroying
7784 2017-09-22 Richard Biener <rguenther@suse.de>
7786 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7787 Verify both BBs contain loop PHI nodes before dispatching to
7789 (graphite_regenerate_ast_isl): Do not recompute dominators,
7790 do not verify three times. Restructure for clarity.
7791 * graphite-scop-detection.c (same_close_phi_node,
7792 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7793 defined_in_loop_p, canonicalize_loop_closed_ssa,
7794 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7795 checking and SSA rewrite, move to ...
7796 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7797 (graphite_initialize): Do not pass in ctx, do not reset the
7798 SCEV cache, compute only dominators.
7799 (graphite_transform_loops): Allocate ISL ctx after
7800 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7801 Maintain post-dominators only around build_scops.
7802 * sese.c (if_region_set_false_region): Make static. Free
7803 and recompute dominators.
7804 (move_sese_in_condition): Assert we don't get called with
7805 post-dominators computed.
7806 * sese.h (if_region_set_false_region): Remove.
7808 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7810 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7811 mode attribute for TARGET_AVX512VL.
7813 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7815 * config/i386/i386.opt (mprefer-avx256): New option.
7816 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7818 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7819 for TARGET_PREFER_AVX256.
7820 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7822 2017-09-21 Jeff Law <law@redhat.com>
7824 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7825 Fix dump output if the only stack space is for pushed registers.
7827 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7829 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7832 2017-09-21 Martin Sebor <msebor@redhat.com>
7835 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7836 code (in C++) or code that triggers warnings.
7838 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7840 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7842 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7844 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7845 * sched-rgn.c: Ditto.
7846 * sel-sched-ir.c: Ditto.
7848 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7850 * toplev.h (set_random_seed): Adjust return type.
7851 * toplev.c (init_local_tick): Move eager initialization of random_seed
7852 to get_random_seed. Adjust comment.
7853 (init_random_seed): Inline to get_random_seed, delete.
7854 (get_random_seed): Initialize random_seed lazily.
7855 (set_random_seed): Do not return previous value.
7856 (print_switch_value): Do not call get_random_seed.
7858 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7860 * cgraph.c (delete_function_version): New, broken out from...
7861 (cgraph_node::delete_function_version): ...here. Rename to
7862 cgraph_node::delete_function_version_by_decl. Update all uses.
7863 (cgraph_node::remove): Call delete_function_version.
7865 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7868 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7869 VAR_DECLs to which addressable non-volatile parameters are mapped
7870 and for id->retvar after the return value assignment. Clear
7871 id->retval and id->retbnd after inlining.
7873 2017-09-21 Richard Biener <rguenther@suse.de>
7875 PR tree-optimization/82276
7876 PR tree-optimization/82244
7877 * tree-vrp.c (build_assert_expr_for): Set
7878 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7880 (remove_range_assertions): Revert earlier change.
7882 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7885 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7887 2017-09-21 Richard Biener <rguenther@suse.de>
7889 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7890 Restore valid IL after code generation errors.
7891 * graphite.c (graphite_transform_loops): Diagnose code
7892 generation issues as MSG_MISSED_OPTIMIZATION and continue
7893 with processing SCOPs.
7895 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7896 Alan Hayward <alan.hayward@arm.com>
7897 David Sherwood <david.sherwood@arm.com>
7899 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7900 rather than choosing between plus_constant and gen_rtx_<CODE>.
7901 * expr.c (emit_push_insn): Likewise.
7902 (expand_expr_real_2): Likewise.
7904 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7905 Alan Hayward <alan.hayward@arm.com>
7906 David Sherwood <david.sherwood@arm.com>
7908 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7910 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7911 Alan Hayward <alan.hayward@arm.com>
7912 David Sherwood <david.sherwood@arm.com>
7914 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7915 calling tree_to_uhwi.
7917 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7918 Alan Hayward <alan.hayward@arm.com>
7919 David Sherwood <david.sherwood@arm.com>
7921 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7922 INTEGER_CST rather than a negative test for ADDR_EXPR.
7924 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7925 Alan Hayward <alan.hayward@arm.com>
7926 David Sherwood <david.sherwood@arm.com>
7928 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7929 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7931 2017-09-21 Richard Biener <rguenther@suse.de>
7933 PR tree-optimization/71351
7934 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7935 graphite_create_new_loop_guard): Remove, fold remaining parts
7937 (translate_isl_ast_node_for): ... here and simplify.
7939 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7942 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7943 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7944 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7945 alternative always use QI mode, for -Os imov (=R,R) alternative
7946 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7949 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7950 Jeff Law <law@redhat.com>
7952 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7953 (allocate_stack_space): New function, partially extracted from
7955 (s390_emit_prologue): Track offset to most recent stack probe.
7956 Code to allocate space moved into allocate_stack_space.
7957 Dump actions when no stack is allocated.
7958 (s390_prologue_plus_offset): New function.
7959 (s390_emit_stack_probe): Likewise.
7961 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7963 * common.opt (Wa, Wl, Wp, g, gz=): Add
7965 (gno-column-info): Remove.
7966 (gcolumn-info): Drop RejectNegative.
7968 (gno-record-gcc-switches): Remove.
7969 (grecord-gcc-switches): Drop RejectNegative.
7970 (gno-split-dwarf): Remove.
7971 (gsplit-dwarf): Drop RejectNegative.
7972 (gno-strict-dwarf): Remove.
7973 (gstrict-dwarf): Drop RejectNegative.
7974 * config/darwin.opt (gfull, gused): Add RejectNegative.
7975 * dwarf2out.c (gen_producer_string): Drop
7976 gno-record-gcc-switches handler.
7977 * optc-gen.awk: Add g to prefixes with negative forms.
7978 * opts-common.c (remapping_prefix_p): New.
7979 (find_opt): Check it.
7980 (generate_canonical_option): Test g prefix.
7981 (option_map): Add -gno- mapping.
7982 (add_misspelling_candidates): Check remapping_prefix_p.
7984 2017-09-20 Jeff Law <law@redhat.com>
7986 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7987 thinko in stack clash protection support.
7989 * explow.c (compute_stack_clash_protection_loop_data): Use
7990 CONST_INT_P instead of explicit test. Verify object is a
7991 CONST_INT_P before looking at INTVAL.
7992 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7993 instead of explicit test.
7995 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7998 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7999 address instead of to r1 and r11.
8001 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8003 * config.gcc: Support "knm".
8004 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8005 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8007 * config/i386/i386.c (m_KNM): Define.
8008 (processor_target_table): Add "knm".
8010 (ix86_option_override_internal): Add "knm".
8011 (ix86_issue_rate): Add PROCESSOR_KNM.
8012 (ix86_adjust_cost): Ditto.
8013 (ia32_multipass_dfa_lookahead): Ditto.
8014 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8015 (fold_builtin_cpu): Add M_INTEL_KNM.
8016 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8017 (processor_type): Add PROCESSOR_KNM.
8018 * config/i386/x86-tune.def: Add m_KNM.
8019 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8021 2017-09-20 Richard Biener <rguenther@suse.de>
8023 PR tree-optimization/80213
8024 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8025 are allowed in empty BBs as well.
8026 (canonicalize_loop_closed_ssa): Also look for other complex
8028 (scop_detection::get_sese): Include the loop-closed PHI block
8030 (scop_detection::merge_sese): Remove code adding extra blocks.
8031 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8032 (build_scops): Assert the final returned scop is invalid.
8034 2017-09-20 Richard Biener <rguenther@suse.de>
8036 PR tree-optimization/82264
8037 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8039 (vn_phi_lookup): Likewise.
8040 (vn_phi_insert): Likewise.
8042 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8044 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8045 that fits into uhwi or shwi, add DW_AT_const_value regardless
8046 of early_dwarf without going through RTL, using add_AT_unsigned
8049 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8050 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8051 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8052 (DEBUG_MACRO_SECTION): Likewise.
8053 (DEBUG_LTO_MACRO_SECTION): Likewise.
8054 (DEBUG_STR_DWO_SECTION): Likewise.
8055 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8056 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8057 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8058 (DEBUG_LTO_LINE_STR_SECTION): Define.
8059 (init_sections_and_labels): Initialize debug_line_str_section
8060 variable. Initialize debug_loc_section for -gdwarf-5 to
8061 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8063 2017-09-20 Richard Biener <rguenther@suse.de>
8065 * graphite-sese-to-poly.c (extract_affine): Properly handle
8066 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8068 2017-09-20 Richard Biener <rguenther@suse.de>
8070 PR tree-optimization/81373
8071 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8072 Force SESE live-out defs to be handled even if they are
8075 2017-09-19 Jeff Law <law@redhat.com>
8077 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8078 nothing for stack adjustments with REG_STACK_CHECK.
8079 * sched-deps.c (parse_add_or_inc): Reject insns with
8080 REG_STACK_CHECK from dependency breaking.
8081 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8082 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8083 * reg-notes.def (STACK_CHECK): New note.
8085 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8086 (ix86_expand_prologue): Dump stack clash info as needed.
8087 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8089 * function.c (dump_stack_clash_frame_info): New function.
8090 * function.h (dump_stack_clash_frame_info): Prototype.
8091 (enum stack_clash_probes): New enum.
8093 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8094 flag_stack_clash_protection.
8095 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8096 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8097 (arm_frame_pointer_required): Likewise.
8098 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8099 (ia64_expand_prologue): Likewise.
8100 * config/mips/mips.c (mips_expand_prologue): Likewise.
8101 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8102 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8103 (sparc_flat_expand_prologue): Likewise.
8104 * config/spu/spu.c (spu_expand_prologue): Likewise.
8106 * explow.c: Include "params.h".
8107 (anti_adjust_stack_and_probe_stack_clash): New function.
8108 (get_stack_check_protect): Likewise.
8109 (compute_stack_clash_protection_loop_data): Likewise.
8110 (emit_stack_clash_protection_loop_start): Likewise.
8111 (emit_stack_clash_protection_loop_end): Likewise.
8112 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8113 Use anti_adjust_stack_and_probe_stack_clash.
8114 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8115 (emit_stack_clash_protection_loop_start): Likewise.
8116 (emit_stack_clash_protection_loop_end): Likewise.
8117 * rtl.h (get_stack_check_protect): Prototype.
8118 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8119 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8120 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8122 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8124 * doc/tm.texi: Rebuilt.
8125 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8126 get_stack_check_protect.
8127 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8128 * config/arm/arm.c (arm_expand_prologue): Likewise.
8129 (arm_frame_pointer_required): Likewise.
8130 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8131 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8132 * config/mips/mips.c (mips_expand_prologue): Likewise.
8133 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8134 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8135 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8136 (sparc_flat_expand_prologue): Likewise.
8138 * common.opt (-fstack-clash-protection): New option.
8139 * flag-types.h (enum stack_check_type): Note difference between
8140 -fstack-check= and -fstack-clash-protection.
8141 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8142 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8143 * toplev.c (process_options): Issue warnings/errors for cases
8144 not handled with -fstack-clash-protection.
8145 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8146 (-fstack-check): Note additional problem with -fstack-check=generic.
8147 Note that -fstack-check is primarily for Ada and refer users
8148 to -fstack-clash-protection for stack-clash-protection.
8149 Document new params for stack clash protection.
8151 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8153 * config/i386/i386.c (ix86_split_long_move): Do not handle
8154 address used for LEA in a special way.
8156 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8158 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8160 2017-09-19 Martin Sebor <msebor@redhat.com>
8163 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8164 of incompatible types.
8166 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8168 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8169 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8170 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8171 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8173 2017-09-19 Richard Biener <rguenther@suse.de>
8175 PR tree-optimization/82244
8176 * tree-vrp.c (remove_range_assertions): Do not propagate
8177 a constant to abnormals but replace the assert with a copy.
8179 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8181 PR rtl-optimization/57878
8182 PR rtl-optimization/68988
8183 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8184 avoidance test involving non_reload_pseudos. Move frequency test
8185 below the general fragmentation avoidance test.
8187 2017-09-19 Richard Biener <rguenther@suse.de>
8189 PR tree-optimization/69728
8190 * graphite-sese-to-poly.c (schedule_error): New global.
8191 (add_loop_schedule): Handle empty domain by failing the
8193 (build_original_schedule): Handle schedule_error.
8195 2017-09-19 Richard Biener <rguenther@suse.de>
8197 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8198 Do not iterate to sibling loops but only to siblings of inner
8201 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8204 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8205 that the registers are different.
8207 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8209 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8210 to processor_model and "amdfam17h" to arch_names_table.
8211 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8213 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8216 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8218 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8219 Alan Hayward <alan.hayward@arm.com>
8220 David Sherwood <david.sherwood@arm.com>
8222 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8224 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8226 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8227 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8228 vect_schedule_slp_instance.
8229 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8230 Update call to vect_slp_analyze_node_operations. Simplify return
8232 (vect_slp_analyze_bb_1): Update call accordingly.
8233 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8234 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8235 (vect_schedule_slp): Update call accordingly.
8237 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8238 Alan Hayward <alan.hayward@arm.com>
8239 David Sherwood <david.sherwood@arm.com>
8241 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8242 with types that aren't in fact scalar.
8244 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8246 * tree-vect-slp.c (vect_record_max_nunits): New function,
8248 (vect_build_slp_tree_1): ...here.
8249 (vect_build_slp_tree_2): Call it for phis too.
8251 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8253 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8254 to vect_get_vec_def_for_operand when getting the mask operand.
8256 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8257 Alan Hayward <alan.hayward@arm.com>
8258 David Sherwood <david.sherwood@arm.com>
8260 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8263 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8264 Alan Hayward <alan.hayward@arm.com>
8265 David Sherwood <david.sherwood@arm.com>
8267 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8268 calculation for vector booleans.
8270 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8271 Alan Hayward <alan.hayward@arm.com>
8272 David Sherwood <david.sherwood@arm.com>
8274 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8276 (vect_transform_stmt): ...here.
8277 (vect_analyze_stmt): Use it instead of calling
8278 vectorizable_live_operation directly.
8280 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8282 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8283 non-SIMT targets in acc vector loops.
8285 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8287 * configure.ac: Add arc and check if assembler supports gdwarf2.
8288 * configure: Regenerate.
8290 2017-09-18 Richard Biener <rguenther@suse.de>
8292 PR tree-optimization/82220
8293 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8294 epilogue niters from the min_profitable_iters compute.
8296 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8299 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8300 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8301 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8303 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8306 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8307 switching to a new text section.
8309 2017-09-18 Richard Biener <rguenther@suse.de>
8311 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8313 (build_alias_set): Reject aliases with no access function.
8315 2017-09-18 Richard Biener <rguenther@suse.de>
8317 PR tree-optimization/79622
8318 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8320 (build_cross_bb_scalars_use): Likewise.
8322 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8324 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8326 2017-09-18 Alan Modra <amodra@gmail.com>
8329 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8330 stack_pointer_rtx for count 0. Update comments. Break up
8331 large rtl expression.
8333 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8335 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8336 Increase to 20 bytes.
8337 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8338 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8339 or avx version of the stub.
8341 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8344 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8345 compute the minimum stack alignment. Also update preferred stack
8346 boundary for leaf functions.
8348 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8350 PR tree-optimization/82228
8351 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8354 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8356 * common/config/nds32/nds32-common.c
8357 (nds32_option_optimization_table): Refine formatting.
8358 (nds32_option_optimization_table): Use -fsched-pressure and
8359 -fomit-frame-pointer for specific optimization level.
8361 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8363 * config/nds32/nds32.c: Refine formatting and comments.
8364 * config/nds32/nds32.h: Likewise.
8365 * config/nds32/nds32.md: Likewise.
8366 * config/nds32/nds32-cost.c: Likewise.
8367 * config/nds32/nds32-isr.c: Likewise.
8368 * config/nds32/nds32-md-auxiliary.c: Likewise.
8369 * config/nds32/nds32-multiple.md: Likewise.
8370 * config/nds32/nds32-predicates.c: Likewise.
8372 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8373 Jakub Jelinek <jakub@redhat.com>
8375 Add support for -std=c++2a.
8376 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8378 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8380 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8383 * doc/extend.texi (Common Function Attributes): Add
8384 references to ARM, AArch64, and S/390 specific attributes.
8385 (Function Specific Option Pragmas): Add AArch64 and S/390
8386 to list of back ends that support the target pragma.
8388 2017-09-15 Nathan Sidwell <nathan@acm.org>
8390 * doc/standards.texi: Fix C++17 description. Update URLs for
8393 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8395 * common.opt (Wcast-align=strict): New warning option.
8396 * doc/invoke.texi: Document -Wcast-align=strict.
8398 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8400 * cgraph.h (cgraph_thunk_info): Add comments.
8401 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8402 assert for VIRTUAL_* arguments stricter.
8404 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8406 PR tree-optimization/71026
8407 * match.pd: Move RDIV patterns from fold-const.c
8408 * fold-const.c (distribute_real_division): Removed.
8409 (fold_binary_loc): Remove calls to distribute_real_divison.
8411 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8413 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8414 c++1z and gnu++1z as deprecated. Change other references to
8415 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8416 Change -Wc++1z-compat to -Wc++17-compat.
8417 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8418 * dwarf2out.c (highest_c_language): Handle C++17.
8419 (gen_compile_unit_die): Likewise.
8421 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8423 PR rtl-optimization/82192
8424 * combine.c (make_extraction): Don't look through non-paradoxical
8425 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8428 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8429 Alan Hayward <alan.hayward@arm.com>
8430 David Sherwood <david.sherwood@arm.com>
8432 * target.def (function_arg_offset): New hook.
8433 * targhooks.h (default_function_arg_offset): Declare.
8434 * targhooks.c (default_function_arg_offset): New function.
8435 * function.c (locate_and_pad_parm): Use
8436 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8437 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8438 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8439 * doc/tm.texi: Regenerate.
8440 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8441 * config/spu/spu.c (spu_function_arg_offset): New function.
8442 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8443 * system.h (FUNCTION_ARG_OFFSET): Poison.
8445 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8446 Alan Hayard <alan.hayward@arm.com>
8447 David Sherwood <david.sherwood@arm.com>
8449 * target.def (truly_noop_truncation): New hook.
8450 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8451 than TRULY_NOOP_TRUNCATION.
8452 * hooks.h (hook_bool_uint_uint_true): Declare.
8453 * hooks.c (hook_bool_uint_uint_true): New function.
8454 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8455 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8456 * doc/tm.texi: Regenerate.
8457 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8458 rather than TRULY_NOOP_TRUNCATION in comments.
8459 (simplify_comparison): Likewise.
8460 (record_truncated_value): Likewise.
8461 * expmed.c (extract_bit_field_1): Likewise.
8462 (extract_split_bit_field): Likewise.
8463 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8464 instead of TRULY_NOOP_TRUNCATION.
8465 * function.c (assign_parm_setup_block): Likewise.
8466 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8467 * rtlhooks.c: Include target.h.
8468 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8469 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8470 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8471 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8472 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8473 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8474 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8475 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8476 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8477 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8478 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8479 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8480 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8481 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8482 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8483 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8484 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8485 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8486 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8487 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8488 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8489 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8490 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8491 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8492 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8493 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8494 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8495 rather than TRULY_NOOP_TRUNCATION in comments.
8496 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8497 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8498 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8499 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8500 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8501 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8502 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8503 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8504 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8505 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8506 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8507 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8508 rather than TRULY_NOOP_TRUNCATION in comments.
8509 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8510 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8511 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8512 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8513 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8514 TRULY_NOOP_TRUNCATION condition.
8515 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8516 (TRULY_NOOP_TRUNCATION): Delete.
8517 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8518 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8519 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8520 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8521 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8522 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8523 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8524 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8525 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8526 rather than TRULY_NOOP_TRUNCATION in comments.
8527 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8528 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8529 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8530 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8531 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8532 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8534 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8537 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8538 (*cmp_ior): Likewise.
8539 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8540 (*ior_scc_scc_cmp): Likewise.
8541 (*and_scc_scc): Likewise.
8542 (*and_scc_scc_cmp): Likewise.
8544 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8545 Alan Hayard <alan.hayward@arm.com>
8546 David Sherwood <david.sherwood@arm.com>
8548 * target.def (can_change_mode_class): New hook.
8549 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8550 (hard_regno_nregs): Likewise.
8551 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8552 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8553 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8554 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8555 (LOAD_EXTEND_OP): Update accordingly.
8556 * doc/tm.texi: Regenerate.
8557 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8558 CANNOT_CHANGE_MODE_CLASS.
8559 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8560 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8561 * combine.c (simplify_set): Update accordingly.
8562 * emit-rtl.c (validate_subreg): Likewise.
8563 * recog.c (general_operand): Likewise.
8564 * regcprop.c (mode_change_ok): Likewise.
8565 * reload1.c (choose_reload_regs): Likewise.
8566 (inherit_piecemeal_p): Likewise.
8567 * rtlanal.c (simplify_subreg_regno): Likewise.
8568 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8569 instead of CANNOT_CHANGE_MODE_CLASS.
8570 (reload_cse_simplify_operands): Likewise.
8571 * reload.c (push_reload): Use targetm.can_change_mode_class
8572 instead of CANNOT_CHANGE_MODE_CLASS.
8573 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8574 REG_CANNOT_CHANGE_MODE_P.
8575 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8576 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8577 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8578 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8579 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8580 (arm_can_change_mode_class): New function.
8581 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8582 than CANNOT_CHANGE_MODE_CLASS in comments.
8583 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8584 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8585 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8586 (ix86_can_change_mode_class): ...this new function, inverting the
8587 sense of the return value.
8588 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8589 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8590 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8591 (ia64_can_change_mode_class): New function.
8592 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8593 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8594 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8595 (m32c_can_change_mode_class): ...this new function, inverting the
8596 sense of the return value.
8597 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8598 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8599 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8600 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8601 (mips_can_change_mode_class): ...this new function, inverting the
8602 sense of the return value.
8603 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8604 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8605 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8606 (msp430_can_change_mode_class): New function.
8607 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8608 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8609 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8610 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8611 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8612 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8613 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8614 (pa_cannot_change_mode_class): Replace with...
8615 (pa_can_change_mode_class): ...this new function, inverting the
8616 sense of the return value.
8617 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8618 than CANNOT_CHANGE_MODE_CLASS in comments.
8619 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8620 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8621 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8622 (pdp11_cannot_change_mode_class): Replace with...
8623 (pdp11_can_change_mode_class): ...this new function, inverting the
8624 sense of the return value.
8625 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8626 * config/powerpcspe/powerpcspe-protos.h
8627 (rs6000_cannot_change_mode_class_ptr): Delete.
8628 * config/powerpcspe/powerpcspe.c
8629 (rs6000_cannot_change_mode_class_ptr): Delete.
8630 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8631 (rs6000_option_override_internal): Assign to
8632 targetm.can_change_mode_class instead of
8633 rs6000_cannot_change_mode_class_ptr.
8634 (rs6000_cannot_change_mode_class): Replace with...
8635 (rs6000_can_change_mode_class): ...this new function, inverting the
8636 sense of the return value.
8637 (rs6000_debug_cannot_change_mode_class): Replace with...
8638 (rs6000_debug_can_change_mode_class): ...this new function.
8639 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8640 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8641 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8642 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8643 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8645 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8646 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8647 (rs6000_option_override_internal): Assign to
8648 targetm.can_change_mode_class instead of
8649 rs6000_cannot_change_mode_class_ptr.
8650 (rs6000_cannot_change_mode_class): Replace with...
8651 (rs6000_can_change_mode_class): ...this new function, inverting the
8652 sense of the return value.
8653 (rs6000_debug_cannot_change_mode_class): Replace with...
8654 (rs6000_debug_can_change_mode_class): ...this new function.
8655 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8656 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8657 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8658 (s390_can_change_mode_class): ...this new function, inverting the
8659 sense of the return value.
8660 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8661 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8662 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8663 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8664 (sh_cannot_change_mode_class): Replace with...
8665 (sh_can_change_mode_class): ...this new function, inverting the
8666 sense of the return value.
8667 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8668 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8669 (sparc_can_change_mode_class): New function.
8670 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8671 * config/spu/spu.c (spu_can_change_mode_class): New function.
8672 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8673 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8674 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8675 (visium_can_change_mode_class): New function.
8676 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8678 2017-09-15 Richard Biener <rguenther@suse.de>
8680 PR tree-optimization/82217
8681 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8682 but not undefined case.
8684 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8687 * postreload.c (reload_cse_simplify_operands): Skip
8688 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8690 2017-09-15 Richard Biener <rguenther@suse.de>
8692 PR tree-optimization/68823
8693 * graphite-scop-detection.c (build_alias_set): If we have a
8694 possible dependence check whether we can handle them by just
8695 looking at the DRs DR_ACCESS_FNs.
8696 (build_scops): If build_alias_set fails, fail the SCOP.
8698 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8700 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8701 to support float128 built-in functions that require the ISA 3.0
8703 (BU_FLOAT128_3_HW): Likewise.
8704 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8706 (FMAF128): Likewise.
8708 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8709 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8710 floating point instructions.
8711 (rs6000_invalid_builtin): Likewise.
8712 (rs6000_builtin_mask_names): Likewise.
8713 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8714 (RS6000_BTM_FLOAT128_HW): Likewise.
8715 (RS6000_BTM_COMMON): Likewise.
8716 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8718 * doc/extend.texi (RS/6000 built-in functions): Document the
8719 IEEE 128-bit floating point square root and fused multiply-add
8722 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8724 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8725 reg (r2) isn't in the set of registers defined in the prologue.
8727 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8728 Alan Hayward <alan.hayward@arm.com>
8729 David Sherwood <david.sherwood@arm.com>
8731 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8732 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8733 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8734 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8735 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8737 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8738 max_vectorization_factor.
8739 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8741 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8742 Alan Hayward <alan.hayward@arm.com>
8743 David Sherwood <david.sherwood@arm.com>
8745 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8746 (vect_worthwhile_without_simd_p): Declare.
8747 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8748 (vectorizable_reduction): Use it.
8749 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8750 (vectorizable_operation): Likewise.
8752 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8753 Alan Hayward <alan.hayward@arm.com>
8754 David Sherwood <david.sherwood@arm.com>
8756 * tree-vectorizer.h (vect_get_num_copies): New function.
8757 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8758 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8759 (vectorizable_induction): Likewise.
8760 (vectorizable_live_operation): Likewise.
8761 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8762 (vectorizable_bswap): Likewise.
8763 (vectorizable_call): Likewise.
8764 (vectorizable_conversion): Likewise.
8765 (vectorizable_assignment): Likewise.
8766 (vectorizable_shift): Likewise.
8767 (vectorizable_operation): Likewise.
8768 (vectorizable_store): Likewise.
8769 (vectorizable_load): Likewise.
8770 (vectorizable_condition): Likewise.
8771 (vectorizable_comparison): Likewise.
8772 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8774 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8775 Alan Hayward <alan.hayward@arm.com>
8776 David Sherwood <david.sherwood@arm.com>
8778 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8779 of vect_init_vector.
8781 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8782 Alan Hayward <alan.hayward@arm.com>
8783 David Sherwood <david.sherwood@arm.com>
8785 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8786 an inline wrapper that provides a location.
8787 (gimple_build_vector): Likewise.
8788 * gimple-fold.c (gimple_build_vector_from_val): New function.
8789 (gimple_build_vector): Likewise.
8790 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8791 functions to build the initial value. Always return a gimple value.
8792 (get_initial_defs_for_reduction): Likewise. Only compute
8794 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8795 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8796 (vectorizable_induction): Use gimple_build_vector rather than
8799 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8800 Alan Hayward <alan.hayward@arm.com>
8801 David Sherwood <david.sherwood@arm.com>
8803 * target.h (vec_perm_indices): New typedef.
8804 (auto_vec_perm_indices): Likewise.
8805 * optabs-query.h: Include target.h
8806 (can_vec_perm_p): Take a vec_perm_indices *.
8807 * optabs-query.c (can_vec_perm_p): Likewise.
8808 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8809 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8810 * tree-vect-generic.c (lower_vec_perm): Likewise.
8811 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8812 (vect_grouped_load_supported): Likewise.
8813 (vect_shift_permute_load_chain): Likewise.
8814 (vect_permute_store_chain): Use auto_vec_perm_indices.
8815 (vect_permute_load_chain): Likewise.
8816 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8817 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8818 Update uses of can_vec_perm_p.
8819 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8820 mode with a number of elements. Take a vec_perm_indices *.
8821 (vect_create_epilog_for_reduction): Update accordingly.
8822 Use auto_vec_perm_indices.
8823 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8824 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8825 (vect_transform_slp_perm_load): Likewise.
8826 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8827 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8828 (vect_gen_perm_mask_checked): Likewise.
8829 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8830 (vect_gen_perm_mask_checked): Likewise.
8831 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8832 (vectorizable_store): Likewise.
8833 (vectorizable_load): Likewise.
8834 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8835 (vectorizable_bswap): Likewise.
8837 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8838 Alan Hayward <alan.hayward@arm.com>
8839 David Sherwood <david.sherwood@arm.com>
8841 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8842 * tree.c (build_vector): Likewise.
8843 (build_vector_from_ctor): Update accordingly.
8844 (build_vector_from_val): Likewise.
8845 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8846 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8847 * tree-vect-generic.c (add_rshift): Likewise.
8848 (expand_vector_divmod): Likewise.
8849 (optimize_vector_constructor): Likewise.
8850 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8851 (vect_transform_slp_perm_load): Likewise.
8852 (vect_schedule_slp_instance): Likewise.
8853 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8854 (vectorizable_call): Likewise.
8855 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8856 * expmed.c (make_tree): Likewise.
8857 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8858 a vector passed to build_vector.
8859 (fold_convert_const): Likewise.
8860 (exact_inverse): Likewise.
8861 (fold_ternary_loc): Likewise.
8862 (fold_relational_const): Likewise.
8863 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8864 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8865 (const_unop): Likewise. Store the reduction accumulator in a
8866 variable rather than an array.
8867 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8868 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8869 the new vector, rather than constructing it after the input arrays.
8870 (native_interpret_vector): Use auto_vec<tree> when building
8871 a vector passed to build_vector. Add elements in order.
8872 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8873 auto_vec<tree> when building a vector passed to build_vector.
8874 (vect_create_epilog_for_reduction): Likewise.
8875 (vectorizable_induction): Likewise.
8876 (get_initial_def_for_reduction): Likewise. Fix indentation of
8878 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8880 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8881 passed to build_vector.
8883 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8884 Alan Hayward <alan.hayward@arm.com>
8885 David Sherwood <david.sherwood@arm.com>
8887 * tree-core.h (tree_base::u): Add an "nelts" field.
8888 (tree_vector): Use VECTOR_CST_NELTS as the length.
8889 * tree.c (tree_size): Likewise.
8890 (make_vector): Initialize VECTOR_CST_NELTS.
8891 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8892 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8893 TYPE_VECTOR_SUBPARTS.
8894 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8895 as the number of units, setting it from VECTOR_CST_NELTS.
8896 (const_vector_from_tree): Likewise.
8897 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8898 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8899 (fold_negate_expr_1): Likewise.
8900 (fold_convert_const): Likewise.
8901 (const_binop): Likewise. Differentiate the number of output and
8903 (const_unop): Likewise.
8904 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8905 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8906 in cases that did the opposite.
8908 2017-09-14 Richard Biener <rguenther@suse.de>
8910 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8913 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8915 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8917 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8920 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8921 if and where to split a bb, except for splitting before debug insn
8922 sequences followed by non-label real insn. Delete debug insns
8923 in between basic blocks.
8925 * combine.c (make_compound_operation_int): Formatting fixes.
8927 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8928 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8929 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8930 * config/sol2.h (LINK_EH_SPEC): Likewise.
8931 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8932 * config/s390/linux.h (LINK_SPEC): Likewise.
8933 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8934 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8935 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8936 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8937 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8938 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8939 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8940 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8942 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8944 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8946 (ENDFILE_LINUX_SPEC): Likewise.
8947 (LINK_EH_SPEC): Likewise.
8948 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8949 (LINK_OS_LINUX_SPEC32): Likewise.
8950 (LINK_OS_LINUX_SPEC64): Likewise.
8951 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8952 (LINK_OS_LINUX_SPEC): Likewise.
8954 2017-09-13 Martin Liska <mliska@suse.cz>
8957 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8958 CASE_HIGH is NULL_TREE.
8960 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8961 Alan Hayward <alan.hayward@arm.com>
8962 David Sherwood <david.sherwood@arm.com>
8964 * target.def (secondary_memory_needed): New hook.
8965 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8966 instead of SECONDARY_MEMORY_NEEDED.
8967 (secondary_memory_needed_mode): Likewise.
8968 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8969 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8970 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8971 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8972 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8973 * doc/tm.texi: Regenerate.
8974 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8975 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8976 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8977 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8978 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8979 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8980 mode argument first and change the reg_class arguments to reg_class_t.
8981 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8982 Make static. Update the call to inline_secondary_memory_needed.
8983 (ix86_register_move_cost): Update the call to
8984 inline_secondary_memory_needed.
8985 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8986 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8988 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8989 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8991 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8992 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8993 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8994 and match hook interface. Add comment from mips.h.
8995 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8996 * config/mmix/mmix.md (truncdfsf2): Refer to
8997 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8999 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9000 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9001 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9002 (pa_secondary_memory_needed): New function.
9003 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9004 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9005 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9006 (pdp11_secondary_memory_needed): Make static and match hook interface.
9007 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9008 * config/powerpcspe/powerpcspe-protos.h
9009 (rs6000_secondary_memory_needed_ptr): Delete.
9010 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9012 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9013 (rs6000_option_override_internal): Assign to
9014 targetm.secondary_memory_needed rather than
9015 rs6000_secondary_memory_needed_ptr.
9016 (rs6000_secondary_memory_needed): Match hook interface.
9017 (rs6000_debug_secondary_memory_needed): Likewise.
9018 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9019 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9020 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9021 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9022 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9023 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9025 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9026 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9027 (rs6000_option_override_internal): Assign to
9028 targetm.secondary_memory_needed rather than
9029 rs6000_secondary_memory_needed_ptr.
9030 (rs6000_secondary_memory_needed): Match hook interface.
9031 (rs6000_debug_secondary_memory_needed): Likewise.
9032 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9033 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9034 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9035 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9036 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9037 (sparc_secondary_memory_needed): New function.
9038 * lra-constraints.c (check_and_process_move): Refer to
9039 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9041 (curr_insn_transform): Likewise.
9042 (process_alt_operands): Use targetm.secondary_memory_needed
9043 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9044 (check_secondary_memory_needed_p): Likewise.
9045 (choose_split_class): Likewise.
9046 * reload.c: Unconditionally include code that was previously
9047 conditional on SECONDARY_MEMORY_NEEDED.
9048 (push_secondary_reload): Use targetm.secondary_memory_needed
9049 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9050 (push_reload): Likewise.
9051 * reload1.c: Unconditionally include code that was previously
9052 conditional on SECONDARY_MEMORY_NEEDED.
9053 (choose_reload_regs): Use targetm.secondary_memory_needed
9054 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9055 (gen_reload): Likewise.
9056 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9058 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9059 Alan Hayward <alan.hayward@arm.com>
9060 David Sherwood <david.sherwood@arm.com>
9062 * target.def (secondary_memory_needed_mode): New hook:
9063 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9064 * targhooks.h (default_secondary_memory_needed_mode): New function.
9065 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9066 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9067 * doc/tm.texi: Regenerate.
9068 * lra-constraints.c (check_and_process_move): Use
9069 targetm.secondary_memory_needed_mode instead of
9070 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9071 (curr_insn_transform): Likewise.
9072 * reload.c (get_secondary_mem): Likewise.
9073 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9074 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9076 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9077 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9078 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9079 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9080 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9082 * config/powerpcspe/powerpcspe-protos.h
9083 (rs6000_secondary_memory_needed_mode): Delete.
9084 * config/powerpcspe/powerpcspe.c
9085 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9086 (rs6000_secondary_memory_needed_mode): Make static.
9087 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9088 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9090 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9092 (rs6000_secondary_memory_needed_mode): Make static.
9093 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9094 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9095 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9096 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9097 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9099 (sparc_secondary_memory_needed_mode): New function.
9100 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9102 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9104 * config/aarch64/constraints.md (Umq): New constraint.
9105 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9107 (mov<mode>): Update condition.
9109 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9111 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9112 when bitposition is the same.
9114 2017-09-13 Richard Biener <rguenther@suse.de>
9116 * dwarf2out.c (output_die_symbol): Remove.
9117 (output_die): Do not output a DIEs symbol.
9119 2017-09-13 Richard Biener <rguenther@suse.de>
9122 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9123 default-def to avoid breaking iterator update with the weird
9124 interaction with cgraph_update_edges_for_call_stmt_node.
9126 2017-09-13 Richard Biener <rguenther@suse.de>
9128 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9129 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9130 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9131 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9133 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9135 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9136 Disable pc relative literal load irrespective of
9137 TARGET_FIX_ERR_A53_84341 for default.
9139 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9141 * config/sparc/sparc.c (output_return): Output the source location of
9142 the insn in the delay slot, if any.
9143 (output_sibcall): Likewise.
9145 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9148 * common.opt (-static-pie): New alias.
9149 (shared): Negate static-pie.
9150 (-no-pie): Update help text.
9152 (static-pie): New option.
9153 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9154 -static-pie support.
9155 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9156 (LINK_EH_SPEC): Likewise.
9157 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9158 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9159 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9160 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9161 (init_gcc_specs): Likewise.
9162 (init_spec): Likewise.
9163 (display_help): Update help message for -pie.
9164 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9167 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9169 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9170 (movdi_aarch64): Likewise.
9171 (movti_aarch64): Likewise.
9173 2017-09-12 Simon Wright <simon@pushface.org>
9176 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9177 calculation of the minor version, always output as 0.
9179 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9182 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9183 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9184 on it early, rather than manual conversion late. For
9185 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9186 instead of performing manual conversion.
9188 2017-09-12 Carl Love <cel@us.ibm.com>
9190 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9191 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9193 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9194 VMULOSW): Add definitions.
9195 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9196 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9197 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9198 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9199 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9201 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9203 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9205 (movti_aarch64): Likewise.
9206 (movdf_aarch64): Likewise.
9207 (movtf_aarch64): Likewise.
9208 (load_pairdi): Likewise.
9209 (store_pairdi): Likewise.
9210 (load_pairdf): Likewise.
9211 (store_pairdf): Likewise.
9212 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9213 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9214 (ldr_got_small_<mode>): Likewise.
9215 (ldr_got_small_28k_<mode>): Likewise.
9216 (ldr_got_tiny): Likewise.
9217 * config/aarch64/iterators.md (ldst_sz): New.
9218 (ldpstp_sz): Likewise.
9219 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9221 (thunderx_load): Split load_8 to load_16.
9222 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9224 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9225 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9226 (xgene1_store_pair): Split store_8 to store_16.
9227 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9228 (falkor_st_0_st_sd): Split store_8 to store_16.
9230 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9232 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9233 and store1/2/3/4 to store_4/8/12/16.
9234 * config/aarch64/aarch64.md: Update for rename.
9235 * config/arm/arm.md: Likewise.: Likewise.
9236 * config/arm/arm.c: Likewise.
9237 * config/arm/thumb1.md: Likewise.
9238 * config/arm/thumb2.md: Likewise.
9239 * config/arm/vfp.md: Likewise.
9240 * config/arm/arm-generic.md: Likewise.
9241 * config/arm/arm1020e.md: Likewise.
9242 * config/arm/arm1026ejs.md: Likewise.
9243 * config/arm/arm1136jfs.md: Likewise.
9244 * config/arm/arm926ejs.md: Likewise.
9245 * config/arm/cortex-a15.md: Likewise.
9246 * config/arm/cortex-a17.md: Likewise.
9247 * config/arm/cortex-a5.md: Likewise.
9248 * config/arm/cortex-a53.md: Likewise.
9249 * config/arm/cortex-a57.md: Likewise.
9250 * config/arm/cortex-a7.md: Likewise.
9251 * config/arm/cortex-a8.md: Likewise.
9252 * config/arm/cortex-a9.md: Likewise.
9253 * config/arm/cortex-m4.md: Likewise.
9254 * config/arm/cortex-m7.md: Likewise.
9255 * config/arm/cortex-r4.md: Likewise.
9256 * config/arm/exynos-m1.md: Likewise.
9257 * config/arm/fa526.md: Likewise.
9258 * config/arm/fa606te.md: Likewise.
9259 * config/arm/fa626te.md: Likewise.
9260 * config/arm/fa726te.md: Likewise.
9261 * config/arm/fmp626.md: Likewise.
9262 * config/arm/iwmmxt.md: Likewise.
9263 * config/arm/ldmstm.md: Likewise.
9264 * config/arm/marvell-pj4.md: Likewise.
9265 * config/arm/xgene1.md: Likewise.
9266 * config/aarch64/thunderx.md: Likewise.
9267 * config/aarch64/thunderx2t99.md: Likewise.
9268 * config/aarch64/falkor.md: Likewise.
9270 2017-09-12 Martin Liska <mliska@suse.cz>
9272 * attribs.c (private_lookup_attribute): New function.
9273 * attribs.h (private_lookup_attribute): Declared here.
9274 (lookup_attribute): Called from this place.
9276 2017-09-12 Richard Biener <rguenther@suse.de>
9278 PR tree-optimization/82157
9279 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9280 stmts with side-effects.
9282 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9283 Alan Hayward <alan.hayward@arm.com>
9284 David Sherwood <david.sherwood@arm.com>
9286 * target.def (hard_regno_nregs): New hook.
9287 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9288 * targhooks.h (default_hard_regno_nregs): Declare.
9289 * targhooks.c (default_hard_regno_nregs): New function.
9290 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9291 (TARGET_HARD_REGNO_NREGS): ...this hook.
9292 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9293 (CLASS_MAX_NREGS): Likewise.
9294 * doc/tm.texi: Regenerate.
9295 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9296 instead of HARD_REGNO_NREGS.
9297 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9298 HARD_REGNO_NREGS in the comment.
9299 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9300 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9301 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9302 Return an unsigned int.
9303 (TARGET_HARD_REGNO_NREGS): Redefine.
9304 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9305 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9306 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9307 (arc_hard_regno_nregs): New function.
9308 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9309 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9310 (arm_hard_regno_nregs): New function.
9311 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9312 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9313 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9314 (TARGET_HARD_REGNO_NREGS): Redefine.
9315 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9316 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9317 (HARD_REGNO_NREGS): Delete.
9318 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9319 (cr16_hard_regno_nregs): New function.
9320 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9321 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9322 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9323 (cris_hard_regno_nregs): New function.
9324 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9325 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9326 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9327 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9328 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9329 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9330 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9331 (frv_hard_regno_nregs): Make static. Take and return an
9333 (frv_class_max_nregs): Remove outdated copy of documentation.
9334 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9335 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9336 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9337 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9338 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9339 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9340 (TARGET_HARD_REGNO_NREGS): Redefine.
9341 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9342 (CLASS_MAX_NREGS): Update comment.
9343 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9344 (ia64_hard_regno_nregs): New function.
9345 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9346 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9347 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9348 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9349 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9351 (m32c_hard_regno_nregs): Likewise. Make static.
9352 (TARGET_HARD_REGNO_NREGS): Redefine.
9353 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9354 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9355 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9356 (m68k_hard_regno_nregs): New function.
9357 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9358 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9359 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9360 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9361 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9362 Take and return an unsigned int.
9363 (TARGET_HARD_REGNO_NREGS): Redefine.
9364 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9365 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9366 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9367 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9368 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9369 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9370 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9371 (msp430_hard_regno_nregs): Make static. Take and return an
9373 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9374 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9375 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9376 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9377 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9378 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9379 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9380 (TARGET_HARD_REGNO_NREGS): Redefine.
9381 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9382 (PA_HARD_REGNO_NREGS): ...this.
9383 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9384 (PA_HARD_REGNO_NREGS): ...this.
9385 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9386 (pa_hard_regno_nregs): New function.
9387 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9388 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9389 (pdp11_hard_regno_nregs): New function.
9390 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9391 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9392 (rs6000_hard_regno_nregs_hook): New function.
9393 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9394 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9395 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9396 Take and return an unsigned int. Move earlier in file.
9397 (TARGET_HARD_REGNO_NREGS): Redefine.
9398 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9399 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9400 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9401 (rl78_hard_regno_nregs): Make static. Take and return an
9403 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9404 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9405 (rs6000_hard_regno_nregs_hook): New function.
9406 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9407 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9408 (TARGET_HARD_REGNO_NREGS): Redefine.
9409 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9410 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9411 instead of HARD_REGNO_NREGS.
9412 (s390_hard_regno_nregs): New function.
9413 (s390_hard_regno_mode_ok): Add comment from s390.h.
9414 (TARGET_HARD_REGNO_NREGS): Redefine.
9415 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9416 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9417 (sh_hard_regno_nregs): New function.
9418 (sh_pass_in_reg_p): Use it.
9419 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9420 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9421 (sparc_hard_regno_nregs): New function.
9422 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9423 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9424 (spu_function_arg_advance): Use it, supplying a valid register number.
9425 (TARGET_HARD_REGNO_NREGS): Redefine.
9426 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9427 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9428 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9429 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9430 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9431 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9432 (CLASS_MAX_NREGS): Remove copy of old documentation.
9433 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9434 (visium_hard_regno_nregs): New function.
9435 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9436 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9437 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9438 xtensa_hard_regno_nregs): New function.
9439 * system.h (HARD_REGNO_NREGS): Poison.
9441 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9443 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9444 hard_regno_nregs instead of HARD_REGNO_NREGS.
9445 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9446 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9447 (c6x_expand_epilogue): Likewise.
9448 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9449 (frv_read_iacc_argument): Likewise.
9450 * config/sh/sh.c: Include regs.h.
9451 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9452 (regs_used): Likewise.
9453 (output_stack_adjust): Likewise.
9454 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9455 * expmed.c: Include regs.h.
9456 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9457 * ree.c: Include regs.h.
9458 (combine_reaching_defs): Use hard_regno_nregs instead of
9460 (add_removable_extension): Likewise.
9462 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9464 * regs.h (hard_regno_nregs): Turn into a function.
9465 (end_hard_regno): Update accordingly.
9466 * caller-save.c (setup_save_areas): Likewise.
9467 (save_call_clobbered_regs): Likewise.
9468 (replace_reg_with_saved_mem): Likewise.
9469 (insert_restore): Likewise.
9470 (insert_save): Likewise.
9471 * combine.c (can_change_dest_mode): Likewise.
9472 (move_deaths): Likewise.
9473 (distribute_notes): Likewise.
9474 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9475 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9476 (rs6000_split_multireg_move): Likewise.
9477 (rs6000_register_move_cost): Likewise.
9478 (rs6000_memory_move_cost): Likewise.
9479 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9480 (rs6000_split_multireg_move): Likewise.
9481 (rs6000_register_move_cost): Likewise.
9482 (rs6000_memory_move_cost): Likewise.
9483 * cselib.c (cselib_reset_table): Likewise.
9484 (cselib_lookup_1): Likewise.
9485 * emit-rtl.c (set_mode_and_regno): Likewise.
9486 * function.c (aggregate_value_p): Likewise.
9487 * ira-color.c (setup_profitable_hard_regs): Likewise.
9488 (check_hard_reg_p): Likewise.
9489 (calculate_saved_nregs): Likewise.
9490 (assign_hard_reg): Likewise.
9491 (improve_allocation): Likewise.
9492 (calculate_spill_cost): Likewise.
9493 * ira-emit.c (modify_move_list): Likewise.
9494 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9495 (ira_hard_reg_in_set_p): Likewise.
9496 * ira.c (setup_reg_mode_hard_regset): Likewise.
9497 (clarify_prohibited_class_mode_regs): Likewise.
9498 (check_allocation): Likewise.
9499 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9500 (lra_setup_reg_renumber): Likewise.
9501 (setup_try_hard_regno_pseudos): Likewise.
9502 (spill_for): Likewise.
9503 (assign_hard_regno): Likewise.
9504 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9505 * lra-constraints.c (in_class_p): Likewise.
9506 (lra_constraint_offset): Likewise.
9507 (simplify_operand_subreg): Likewise.
9508 (lra_constraints): Likewise.
9509 (split_reg): Likewise.
9510 (split_if_necessary): Likewise.
9511 (invariant_p): Likewise.
9512 (inherit_in_ebb): Likewise.
9513 * lra-lives.c (process_bb_lives): Likewise.
9514 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9515 (get_hard_regs): Likewise.
9516 (do_remat): Likewise.
9517 * lra-spills.c (assign_spill_hard_regs): Likewise.
9518 * mode-switching.c (create_pre_exit): Likewise.
9519 * postreload.c (reload_combine_recognize_pattern): Likewise.
9520 * recog.c (peep2_find_free_register): Likewise.
9521 * regcprop.c (kill_value_regno): Likewise.
9522 (set_value_regno): Likewise.
9523 (copy_value): Likewise.
9524 (maybe_mode_change): Likewise.
9525 (find_oldest_value_reg): Likewise.
9526 (copyprop_hardreg_forward_1): Likewise.
9527 * regrename.c (check_new_reg_p): Likewise.
9528 (regrename_do_replace): Likewise.
9529 * reload.c (push_reload): Likewise.
9530 (combine_reloads): Likewise.
9531 (find_dummy_reload): Likewise.
9532 (operands_match_p): Likewise.
9533 (find_reloads): Likewise.
9534 (find_equiv_reg): Likewise.
9535 (reload_adjust_reg_for_mode): Likewise.
9536 * reload1.c (count_pseudo): Likewise.
9537 (count_spilled_pseudo): Likewise.
9538 (find_reg): Likewise.
9539 (clear_reload_reg_in_use): Likewise.
9540 (free_for_value_p): Likewise.
9541 (allocate_reload_reg): Likewise.
9542 (choose_reload_regs): Likewise.
9543 (reload_adjust_reg_for_temp): Likewise.
9544 (emit_reload_insns): Likewise.
9545 (delete_output_reload): Likewise.
9546 * rtlanal.c (subreg_get_info): Likewise.
9547 * sched-deps.c (sched_analyze_reg): Likewise.
9548 * sel-sched.c (init_regs_for_mode): Likewise.
9549 (mark_unavailable_hard_regs): Likewise.
9550 (choose_best_reg_1): Likewise.
9551 (verify_target_availability): Likewise.
9552 * valtrack.c (dead_debug_insert_temp): Likewise.
9553 * var-tracking.c (track_loc_p): Likewise.
9554 (emit_note_insn_var_location): Likewise.
9555 * varasm.c (make_decl_rtl): Likewise.
9556 * reginfo.c (choose_hard_reg_mode): Likewise.
9557 (init_reg_modes_target): Refer directly to
9558 this_target_regs->x_hard_regno_nregs.
9560 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9562 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9563 instead of hard_regno_nregs.
9565 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9567 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9568 end_hard_regno instead of hard_regno_nregs.
9569 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9570 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9571 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9572 * ira-color.c (improve_allocation): Likewise.
9573 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9574 * lra-lives.c (mark_regno_live): Likewise.
9575 (mark_regno_dead): Likewise.
9576 * lra-remat.c (operand_to_remat): Likewise.
9577 * lra.c (collect_non_operand_hard_regs): Likewise.
9578 * postreload.c (reload_combine_note_store): Likewise.
9579 (move2add_valid_value_p): Likewise.
9580 * reload.c (regno_clobbered_p): Likewise.
9582 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9584 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9586 * config/v850/v850.c (v850_reorg): Likewise.
9587 * reload.c (refers_to_regno_for_reload_p): Likewise.
9588 (find_equiv_reg): Likewise.
9589 * reload1.c (reload_reg_reaches_end_p): Likewise.
9591 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9593 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9595 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9596 * config/arm/arm.c (output_move_neon): Likewise.
9597 (arm_attr_length_move_neon): Likewise.
9598 (neon_split_vcombine): Likewise.
9599 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9600 (c6x_mark_reg_written): Likewise.
9601 (c6x_dwarf_register_span): Likewise.
9602 * config/i386/i386.c (ix86_save_reg): Likewise.
9603 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9604 (rws_access_reg): Likewise.
9605 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9606 * mode-switching.c (create_pre_exit): Likewise.
9607 * ree.c (combine_reaching_defs): Likewise.
9608 (add_removable_extension): Likewise.
9609 * regcprop.c (find_oldest_value_reg): Likewise.
9610 (copyprop_hardreg_forward_1): Likewise.
9611 * reload.c (reload_inner_reg_of_subreg): Likewise.
9612 (push_reload): Likewise.
9613 (combine_reloads): Likewise.
9614 (find_dummy_reload): Likewise.
9615 (reload_adjust_reg_for_mode): Likewise.
9616 * reload1.c (find_reload_regs): Likewise.
9617 (forget_old_reloads_1): Likewise.
9618 (reload_reg_free_for_value_p): Likewise.
9619 (reload_adjust_reg_for_temp): Likewise.
9620 (emit_reload_insns): Likewise.
9621 (delete_output_reload): Likewise.
9622 * sel-sched.c (choose_best_reg_1): Likewise.
9623 (choose_best_pseudo_reg): Likewise.
9625 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9626 Alan Hayward <alan.hayward@arm.com>
9627 David Sherwood <david.sherwood@arm.com>
9629 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9630 * target.def (slow_unaligned_access): New hook.
9631 * targhooks.h (default_slow_unaligned_access): Declare.
9632 * targhooks.c (default_slow_unaligned_access): New function.
9633 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9634 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9635 * doc/tm.texi: Regenerate.
9636 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9637 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9638 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9640 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9641 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9643 (rs6000_slow_unaligned_access): New function.
9644 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9645 (expand_block_compare): Likewise.
9646 (expand_strn_compare): Likewise.
9647 (rs6000_rtx_costs): Likewise.
9648 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9649 (riscv_slow_unaligned_access): Likewise.
9650 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9651 (riscv_slow_unaligned_access_p): ...this and make static.
9652 (riscv_option_override): Update accordingly.
9653 (riscv_slow_unaligned_access): New function.
9654 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9655 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9656 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9657 (rs6000_slow_unaligned_access): New function.
9658 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9659 (rs6000_rtx_costs): Likewise.
9660 * config/rs6000/rs6000-string.c (expand_block_compare)
9661 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9662 of SLOW_UNALIGNED_ACCESS.
9663 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9664 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9665 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9666 of SLOW_UNALIGNED_ACCESS.
9667 * expmed.c (simple_mem_bitfield_p): Likewise.
9668 * expr.c (alignment_for_piecewise_move): Likewise.
9669 (emit_group_load_1): Likewise.
9670 (emit_group_store): Likewise.
9671 (copy_blkmode_from_reg): Likewise.
9672 (emit_push_insn): Likewise.
9673 (expand_assignment): Likewise.
9674 (store_field): Likewise.
9675 (expand_expr_real_1): Likewise.
9676 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9677 * lra-constraints.c (simplify_operand_subreg): Likewise.
9678 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9679 * gimple-ssa-store-merging.c: Likewise in block comment at start
9681 * tree-ssa-strlen.c: Include target.h.
9682 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9683 of SLOW_UNALIGNED_ACCESS.
9684 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9686 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9688 PR rtl-optimization/82185
9689 * expmed.c (emit_store_flag_int): Only test tem if it has been
9692 2017-09-12 Richard Biener <rguenther@suse.de>
9695 * match.pd ((FTYPE) N CMP CST): Fix typo.
9697 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9699 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9701 (mips_near_type_p): Add 'short_call' attribute as a synonym
9703 * doc/extend.texi (short_call): Document new function attribute.
9705 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9708 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9709 assertion check that in the condition.
9710 (get_atomic_generic_size): Likewise. Before testing if parameter
9711 has pointer type, if it has array type, call for C++
9712 default_conversion to perform array-to-pointer conversion.
9714 2017-09-12 Richard Biener <rguenther@suse.de>
9716 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9717 for operations we cannot scalarize.
9719 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9721 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9722 vectors heap vectors. Clean up comments.
9723 Make visited_bbs a reference.
9724 (profitable_jump_thread_path): Make GC
9725 vectors heap vectors. Clean up comments.
9727 (convert_and_register_jump_thread_path): Make GC vectors heap
9729 (check_subpath_and_update_thread_path): Same. Clean up comments.
9730 Make visited_bbs a reference.
9731 (handle_phi): Abstract common code to to
9732 register_jump_thread_path_if_profitable.
9733 Rename VAR_BB to DEF_BB.
9735 Make GC vectors heap vectors.
9736 Make visited_bbs a reference.
9737 (handle_assignment): Same.
9738 (register_jump_thread_path_if_profitable): New.
9739 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9741 Make GC vectors heap vectors. Clean up comments.
9742 Make visited_bbs a reference.
9743 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9744 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9747 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9750 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9751 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9753 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9755 Revert r251800 and r251799.
9757 2017-09-11 Martin Jambor <mjambor@suse.cz>
9760 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9761 arguments in advance.
9762 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9763 use it to find predecessor edges.
9764 (naive_outof_ssa): Collect vector of predecessors.
9766 2017-09-08 Jason Merrill <jason@redhat.com>
9768 PR c++/70029 - ICE with ref-qualifier and -flto
9769 * langhooks.h (struct lang_hooks_for_types): Add
9770 copy_lang_qualifiers.
9771 * attribs.c (build_type_attribute_qual_variant): Use it.
9772 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9774 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9775 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9777 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9780 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9781 (*mulsi3_sp64): New instruction.
9782 (mulsi3): New expander.
9784 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9786 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9788 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9790 * sancov.c: Include memmodel.h.
9792 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9795 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9798 2017-09-07 Carl Love <cel@us.ibm.com>
9800 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9801 the sldi instruction.
9803 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9805 * sancov.c: Include tm_p.h.
9807 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9810 * output.h (switch_to_other_text_partition): New declaration.
9811 * varasm.c (switch_to_other_text_partition): New function.
9812 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9813 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9814 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9815 to the other text partition before emitting LCL label and switch back
9816 after emitting the word after it.
9818 2017-09-07 Richard Biener <rguenther@suse.de>
9820 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9821 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9822 critical edges here, after loop init.
9823 (pass_data_pre): Remove PROP_no_crit_edges flags.
9824 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9825 for valueization of call args to avoid leaking VN_TOP.
9826 (visit_use): Assert we do not visit default defs.
9827 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9828 Use error_mark_node to more easily detect leaking VN_TOP.
9829 All default-defs are varying, not VN_TOP. Mark them visited.
9830 (run_scc_vn): Make code match comment.
9832 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9834 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9835 OPTION_MASK_FLOAT128_KEYWORD.
9836 (POWERPC_MASKS): Likewise.
9837 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9838 support for the -mfloat128-type option, and make -mfloat128
9839 default on PowerPC Linux systems. Define or undefine
9840 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9841 Define __float128 to be __ieee128 if IEEE 128-bit support is
9842 enabled, or undefine it.
9843 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9844 Delete defining __FLOAT128_TYPE__.
9845 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9846 -mfloat128-type option and make -mfloat128 default on PowerPC
9848 (TARGET_FLOAT128_TYPE): Likewise.
9849 (-mfloat128-type): Likewise.
9850 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9851 Delete the -mfloat128-type option and make -mfloat128 default on
9852 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9853 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9854 128-bit floating point is enabled. Change tests from using
9855 -mfloat128-type to -mfloat128.
9856 (rs6000_mangle_type): Use the correct mangling for the __float128
9857 type even if normal long double is restricted to 64-bits.
9858 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9860 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9861 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9862 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9864 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9865 documentation for -mfloat128.
9867 2017-09-06 Olivier Hainque <hainque@adacore.com>
9869 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9871 2017-09-06 Wish Wu <wishwu007@gmail.com>
9872 Jakub Jelinek <jakub@redhat.com>
9874 * asan.c (initialize_sanitizer_builtins): Add
9875 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9876 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9877 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9878 BT_FN_VOID_UINT64_PTR variables.
9879 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9880 (BT_FN_VOID_UINT16_UINT16): Likewise.
9881 (BT_FN_VOID_UINT32_UINT32): Likewise.
9882 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9883 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9884 (BT_FN_VOID_UINT64_PTR): Likewise.
9885 * common.opt (flag_sanitize_coverage): New variable.
9886 (fsanitize-coverage=trace-pc): Remove.
9887 (fsanitize-coverage=): Add.
9888 * flag-types.h (enum sanitize_coverage_code): New enum.
9889 * fold-const.c (fold_range_test): Disable non-short-circuit
9890 optimization if flag_sanitize_coverage.
9891 (fold_truth_andor): Likewise.
9892 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9893 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9894 (coverage_sanitizer_opts): New array.
9895 (get_closest_sanitizer_option): Add OPTS argument, handle also
9896 OPT_fsanitize_coverage_.
9897 (parse_sanitizer_options): Adjusted to also handle
9898 OPT_fsanitize_coverage_.
9899 (common_handle_option): Add OPT_fsanitize_coverage_.
9900 * sancov.c (instrument_comparison, instrument_switch): New function.
9901 (sancov_pass): Add trace-cmp support.
9902 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9903 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9904 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9905 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9906 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9907 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9908 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9909 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9910 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9911 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9913 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9915 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9916 error. Only quit immediately if parsing is complete.
9917 (BEGIN): Initialize fatal_err and parse_done.
9918 (begin fpu, end fpu): Check number of arguments.
9919 (begin arch, end arch): Likewise.
9920 (begin cpu, end cpu): Likewise.
9921 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9922 (optalias): Likewise.
9924 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9926 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9927 * config/arm/arm-isa.h: Delete. Move definitions to ...
9928 * arm-cpus.in: ... here. Use new feature and fgroup values.
9929 * config/arm/arm.c (arm_option_override): Use lower case for feature
9931 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9932 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9933 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9935 (print_isa_bits_for): New function.
9936 (gen_isa): New function.
9937 (gen_comm_data): Use print_isa_bits_for.
9938 (define feature): New keyword.
9939 (define fgroup): New keyword.
9940 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9941 (arm-isa.h): Add rule to generate file.
9942 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9943 case for feature bit names.
9945 2017-09-06 Richard Biener <rguenther@suse.de>
9947 * tree-ssa-pre.c (NECESSARY): Remove.
9948 (create_expression_by_pieces): Do not touch pass-local flags.
9949 (insert_into_preds_of_block): Likewise.
9950 (do_pre_regular_insertion): Likewise.
9951 (eliminate_insert): Likewise.
9952 (eliminate_dom_walker::before_dom_children): Likewise.
9953 (fini_eliminate): Do not look at inserted_exprs.
9954 (mark_operand_necessary): Remove.
9955 (remove_dead_inserted_code): Replace with simple work-list
9956 algorithm based on inserted_exprs and SSA uses.
9957 (pass_pre::execute): Re-order fini_eliminate and
9958 remove_dead_inserted_code.
9960 2017-09-06 Olivier Hainque <hainque@adacore.com>
9962 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9963 for VxWorks 7. Adjust surrounding comments.
9965 2017-09-06 Richard Biener <rguenther@suse.de>
9967 * gimple-ssa-strength-reduction.c
9968 (find_candidates_dom_walker::before_dom_children): Also allow
9971 2017-09-06 Richard Biener <rguenther@suse.de>
9973 PR tree-optimization/82108
9974 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9975 for gap in the non-permutation SLP case.
9977 2017-09-06 Martin Jambor <mjambor@suse.cz>
9979 PR tree-optimization/82078
9980 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9981 add_access_to_work_queue...
9982 (build_accesses_from_assign): ...here.
9983 (propagate_all_subaccesses): Make sure racc is the group
9984 representative, if there is one.
9986 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9989 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9992 2017-09-06 Richard Biener <rguenther@suse.de>
9994 * gimple-ssa-strength-reduction.c
9995 (find_candidates_dom_walker::before_doom_children): Use a
9996 type and not a mode check.
9998 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10001 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10003 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10005 2017-09-05 Jeff Law <law@redhat.com>
10007 PR tree-optimization/64910
10008 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10009 cases where we have 3 or more operands.
10011 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10013 PR middle-end/81768
10014 * omp-low.c (lower_omp_for): Recompute tree invariant if
10015 gimple_omp_for_initial/final is ADDR_EXPR.
10017 PR middle-end/81768
10018 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10019 into gimple val before gimplification fo the COND_EXPR.
10021 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10023 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10024 REGION_COPY argument.
10025 (thread_through_all_blocks): Remove unused argument to
10026 duplicate_thread_path.
10028 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10029 Alan Hayward <alan.hayward@arm.com>
10030 David Sherwood <david.sherwood@arm.com>
10032 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10033 Take a scalar_mode rather than a machine_mode.
10034 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10035 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10036 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10037 (aarch64_gen_adjusted_ldpstp): Likewise.
10038 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10040 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10041 Alan Hayward <alan.hayward@arm.com>
10042 David Sherwood <david.sherwood@arm.com>
10044 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10045 Take a scalar_int_mode instead of a machine_mode.
10046 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10047 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10048 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10049 (aarch64_simd_attr_length_rglist): Delete.
10050 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10051 a scalar_int_mode instead of a machine_mode.
10052 (aarch64_add_offset): Likewise.
10053 (aarch64_internal_mov_immediate): Likewise
10054 (aarch64_add_constant_internal): Likewise.
10055 (aarch64_add_constant): Likewise.
10056 (aarch64_movw_imm): Likewise.
10057 (aarch64_rtx_arith_op_extract_p): Likewise.
10058 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10059 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10060 Remove assert that the mode isn't a vector.
10061 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10062 (aarch64_expand_mov_immediate): Update calls after above changes.
10063 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10064 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10065 (aarch64_move_imm): Likewise.
10066 (aarch64_can_const_movi_rtx_p): Likewise.
10067 (aarch64_strip_extend): Likewise.
10068 (aarch64_extr_rtx_p): Likewise.
10069 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10070 a CONST_INT when the mode parameter is VOIDmode.
10071 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10073 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10075 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10076 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10077 (bitwise_type_for_mode): Update accordingly.
10079 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10081 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10082 * stor-layout.c (mode_for_size_tree): Likewise.
10083 (mode_for_array): Update accordingly.
10084 (layout_decl): Likewise.
10085 (compute_record_mode): Likewise. Only set the mode once.
10087 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10089 * target.def (get_mask_mode): Change return type to opt_mode.
10091 * doc/tm.texi: Regenerate.
10092 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10093 * targhooks.c (default_get_mask_mode): Likewise.
10094 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10095 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10096 targetm.get_mask_mode.
10097 * tree.c (build_truth_vector_type): Likewise.
10099 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10101 * machmode.h (mode_for_vector): Return an opt_mode.
10102 * stor-layout.c (mode_for_vector): Likewise.
10103 (mode_for_int_vector): Update accordingly.
10104 (layout_type): Likewise.
10105 * config/i386/i386.c (emit_memmov): Likewise.
10106 (ix86_expand_set_or_movmem): Likewise.
10107 (ix86_expand_vector_init): Likewise.
10108 (ix86_get_mask_mode): Likewise.
10109 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10111 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10112 * expmed.c (extract_bit_field_1): Likewise.
10113 * expr.c (expand_expr_real_2): Likewise.
10114 * optabs-query.c (can_vec_perm_p): Likewise.
10115 (can_vec_mask_load_store_p): Likewise.
10116 * optabs.c (expand_vec_perm): Likewise.
10117 * targhooks.c (default_get_mask_mode): Likewise.
10118 * tree-vect-stmts.c (vectorizable_store): Likewise.
10119 (vectorizable_load): Likewise.
10120 (get_vectype_for_scalar_type_and_size): Likewise.
10122 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10124 * machmode.h (mode_for_int_vector): New function.
10125 * stor-layout.c (mode_for_int_vector): Likewise.
10126 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10127 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10128 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10129 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10130 (s390_expand_vcond): Likewise.
10132 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10134 * machmode.h (opt_machine_mode): New type.
10135 (opt_mode<T>): Allow construction from anything that can be
10137 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10138 (mode_for_size): Return an opt_machine_mode.
10139 * stor-layout.c (mode_for_size): Likewise.
10140 (mode_for_size_tree): Update call accordingly.
10141 (bitwise_mode_for_mode): Likewise.
10142 (make_fract_type): Likewise.
10143 (make_accum_type): Likewise.
10144 * caller-save.c (replace_reg_with_saved_mem): Update call
10146 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10147 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10148 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10149 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10150 * expmed.c (extract_bit_field_1): Likewise.
10151 * reload.c (get_secondary_mem): Likewise.
10152 * varasm.c (assemble_integer): Likewise.
10153 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10156 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10158 * machmode.h (decimal_float_mode_for_size): New function.
10159 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10160 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10161 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10162 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10163 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10164 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10166 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10168 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10169 (get_builtin_sync_mode): Likewise.
10170 (expand_ifn_atomic_compare_exchange): Likewise.
10171 (expand_builtin_atomic_clear): Likewise.
10172 (expand_builtin_atomic_test_and_set): Likewise.
10173 (fold_builtin_atomic_always_lock_free): Likewise.
10174 * calls.c (compute_argument_addresses): Likewise.
10175 (emit_library_call_value_1): Likewise.
10176 (store_one_arg): Likewise.
10177 * combine.c (combine_instructions): Likewise.
10178 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10179 * config/arm/arm.c (arm_function_value): Likewise.
10180 (aapcs_allocate_return_reg): Likewise.
10181 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10182 * config/i386/i386.c (construct_container): Likewise.
10183 (ix86_gimplify_va_arg): Likewise.
10184 (ix86_expand_sse_cmp): Likewise.
10185 (emit_memmov): Likewise.
10186 (emit_memset): Likewise.
10187 (expand_small_movmem_or_setmem): Likewise.
10188 (ix86_expand_pextr): Likewise.
10189 (ix86_expand_pinsr): Likewise.
10190 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10191 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10193 * config/mips/mips.c (mips_function_value_1) Likewise.
10194 (mips_block_move_straight): Likewise.
10195 (mips_expand_ins_as_unaligned_store): Likewise.
10196 * config/powerpcspe/powerpcspe.c
10197 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10198 (rs6000_darwin64_record_arg_flush): Likewise.
10199 * config/rs6000/rs6000.c
10200 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10201 (rs6000_darwin64_record_arg_flush): Likewise.
10202 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10203 (sparc_function_value_1): Likewise.
10204 * config/spu/spu.c (adjust_operand): Likewise.
10205 (spu_emit_branch_or_set): Likewise.
10206 (arith_immediate_p): Likewise.
10207 * emit-rtl.c (gen_lowpart_common): Likewise.
10208 * expr.c (expand_expr_real_1): Likewise.
10209 * function.c (assign_parm_setup_block): Likewise.
10210 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10211 * reload1.c (alter_reg): Likewise.
10212 * stor-layout.c (mode_for_vector): Likewise.
10213 (layout_type): Likewise.
10215 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10217 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10218 (spu_convert_move): Likewise.
10219 * lower-subreg.c (resolve_simple_move): Likewise.
10221 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10224 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10225 define_insn to a define_expand.
10226 (altivec_vsum2sws_direct): New define_insn.
10227 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10229 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10231 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10234 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10239 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10240 realign stack if stack alignment needed is less than incoming
10243 2017-09-05 Marek Polacek <polacek@redhat.com>
10246 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10249 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10251 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10253 2017-09-05 Richard Biener <rguenther@suse.de>
10255 PR tree-optimization/82084
10256 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10258 2017-09-05 Richard Biener <rguenther@suse.de>
10260 PR tree-optimization/82102
10261 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10263 2017-09-05 Martin Liska <mliska@suse.cz>
10265 PR tree-optimization/82032
10266 * tree-cfg.c (generate_range_test): New function.
10267 * tree-cfg.h (generate_range_test): Declared here.
10268 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10269 (cleanup_control_expr_graph): Use it.
10270 * tree-switch-conversion.c (try_switch_expansion): Remove
10272 (emit_case_nodes): Use generate_range_test.
10274 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10277 * config/i386/i386.md (*<btsc><mode>_mask): Add
10278 TARGET_USE_BT to insn constraint.
10279 (*btr<mode>_mask): Ditto.
10281 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10283 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10284 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10286 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10289 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10290 TARGET_NEON and TARGET_IWMMXT.
10291 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10292 TARGET_NEON and TARGET_IWMMXT.
10293 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10295 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10297 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10298 (ix86_rewrite_tls_address): Ditto.
10299 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10300 (ix86_rewrite_tls_address_1): Ditto.
10301 (ix86_rewrite_tls_address): Ditto.
10302 * config/i386/predicates.md (tls_address_pattern): New predicate.
10303 * config/i386/i386.md (TLS address splitter): New splitter.
10305 2017-09-04 Richard Biener <rguenther@suse.de>
10307 PR tree-optimization/82084
10308 * fold-const.h (can_native_encode_string_p): Declare.
10309 * fold-const.c (can_native_encode_string_p): Factor out from ...
10310 (native_encode_string): ... here.
10311 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10312 vectorizing stores from constants we later cannot handle.
10314 2017-09-04 Marek Polacek <polacek@redhat.com>
10317 * doc/invoke.texi: Update -Wtautological-compare documentation.
10319 2017-09-04 Jeff Law <law@redhat.com>
10321 PR tree-optimization/64910
10322 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10323 swap the first and last operand if the last is a constant.
10325 2017-09-04 Marek Polacek <polacek@redhat.com>
10328 * convert.c (do_narrow): When sanitizing signed integer overflows,
10329 bail out for signed types.
10330 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10332 2017-09-04 Richard Biener <rguenther@suse.de>
10334 PR tree-optimization/82060
10335 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10336 Move devirtualization after stmt folding and before EH/AB/noreturn
10337 cleanup to get the stmt refs canonicalized. Use a bool instead
10338 of gimple_modified_p since that doesn't work for NOPs. Schedule
10339 NOPs generated by folding for removal.
10341 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10342 Alan Hayward <alan.hayward@arm.com>
10343 David Sherwood <david.sherwood@arm.com>
10345 * coretypes.h (pad_direction): New enum.
10346 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10347 (FUNCTION_ARG_PADDING): Likewise.
10348 * target.def (function_arg_padding): New hook.
10349 * targhooks.h (default_function_arg_padding): Declare.
10350 * targhooks.c (default_function_arg_padding): New function.
10351 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10352 (TARGET_FUNCTION_ARG_PADDING): ...this.
10353 * doc/tm.texi: Regenerate.
10354 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10355 instead of direction.
10356 (compute_argument_addresses): Likewise.
10357 (load_register_parameters): Likewise.
10358 (emit_library_call_value_1): Likewise.
10359 (store_one_arg): Use targetm.calls.function_arg_padding instead
10360 of FUNCTION_ARG_PADDING.
10361 (must_pass_in_stack_var_size_or_pad): Likewise.
10362 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10363 (emit_group_store): Likewise.
10364 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10365 instead of FUNCTION_ARG_PADDING.
10366 (emit_push_insn): Likewise, and propagate enum change throughout
10368 * function.h (direction): Delete.
10369 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10371 * function.c (assign_parm_find_stack_rtl): Likewise.
10372 (assign_parm_setup_block_p): Likewise.
10373 (assign_parm_setup_block): Likewise.
10374 (gimplify_parameters): Likewise.
10375 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10376 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10378 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10379 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10380 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10381 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10382 (aarch64_function_arg_padding): ...this new function.
10383 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10384 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10385 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10386 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10387 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10388 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10389 (arm_pad_arg_upward): Replace with...
10390 (arm_function_arg_padding): ...this new function.
10391 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10393 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10394 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10395 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10396 (ia64_hpux_function_arg_padding): Replace with...
10397 (ia64_function_arg_padding): ...this new function. Use pad_direction
10398 instead of direction. Check for TARGET_HPUX.
10399 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10400 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10401 (iq2000_function_arg_padding): New function.
10402 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10403 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10404 (mips_function_arg_padding): ...this new function.
10405 (mips_pad_reg_upward): Update accordingly.
10406 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10407 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10408 targetm.calls.function_arg_padding.
10409 (FUNCTION_ARG_PADDING): Delete.
10410 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10411 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10412 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10413 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10414 (nios2_block_reg_padding): Return pad_direction instead of direction.
10415 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10416 instead of direction.
10417 (nios2_function_arg_padding): Likewise. Make static.
10418 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10419 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10420 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10421 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10422 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10423 (pa_function_arg_padding): Make static. Return pad_direction instead
10425 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10426 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10427 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10428 instead of direction. Use targetm.calls.function_arg_padding.
10429 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10430 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10431 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10432 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10433 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10435 (function_arg_padding): Rename to...
10436 (rs6000_function_arg_padding): ...this. Make static. Return
10437 pad_direction instead of direction.
10438 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10439 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10440 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10441 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10442 instead of direction. Use targetm.calls.function_arg_padding.
10443 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10444 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10445 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10446 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10447 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10448 (function_arg_padding): Rename to...
10449 (rs6000_function_arg_padding): ...this. Make static. Return
10450 pad_direction instead of direction.
10451 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10452 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10453 * config/s390/s390.c (s390_function_arg_padding): New function.
10454 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10455 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10456 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10457 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10458 (function_arg_padding): Rename to...
10459 (sparc_function_arg_padding): ...this. Make static. Return
10460 pad_direction instead of direction.
10461 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10462 * config/spu/spu.c (spu_function_arg_padding): New function.
10463 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10464 * system.h (FUNCTION_ARG_PADDING): Poison.
10466 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10467 Alan Hayward <alan.hayward@arm.com>
10468 David Sherwood <david.sherwood@arm.com>
10470 * target.def (modes_tieable_p): New hook.
10471 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10472 (TARGET_MODES_TIEABLE_P): ...this.
10473 * doc/tm.texi.in: Regenerate.
10474 * hooks.h (hook_bool_mode_mode_true): Declare.
10475 * hooks.c (hook_bool_mode_mode_true): New function.
10476 * combine.c (subst): Use targetm.modes_tieable_p instead of
10478 * dse.c (find_shift_sequence): Likewise.
10479 * expmed.c (extract_low_bits): Likewise.
10480 * lower-subreg.c: Include target.h.
10481 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10483 * rtlanal.c (rtx_cost): Likewise.
10484 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10485 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10486 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10487 (TARGET_MODES_TIEABLE_P): Redefine.
10488 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10489 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10490 (TARGET_MODES_TIEABLE_P): Redefine.
10491 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10492 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10493 (arc_modes_tieable_p): New function.
10494 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10495 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10496 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10497 (arm_modes_tieable_p): Make static.
10498 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10499 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10500 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10501 (TARGET_MODES_TIEABLE_P): Redefine.
10502 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10503 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10504 (TARGET_MODES_TIEABLE_P): Redefine.
10505 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10506 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10507 (cr16_modes_tieable_p): New function.
10508 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10509 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10510 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10511 (TRULY_NOOP_TRUNCATION): Update comment.
10512 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10513 (TRULY_NOOP_TRUNCATION): Update comment.
10514 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10515 (frv_modes_tieable_p): New function.
10516 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10517 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10518 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10519 (TARGET_MODES_TIEABLE_P): Redefine.
10520 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10521 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10522 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10523 (TARGET_MODES_TIEABLE_P): Redefine.
10524 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10525 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10526 (ia64_modes_tieable_p): New function.
10527 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10528 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10529 (iq2000_modes_tieable_p): New function.
10530 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10531 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10532 (lm32_modes_tieable_p): New function.
10533 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10534 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10535 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10536 (TARGET_MODES_TIEABLE_P): Redefine.
10537 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10538 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10539 (m32r_modes_tieable_p): New function.
10540 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10541 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10542 (m68k_modes_tieable_p): New function.
10543 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10544 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10545 (mcore_modes_tieable_p): New function.
10546 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10547 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10549 (TARGET_MODES_TIEABLE_P): Redefine.
10550 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10551 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10552 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10553 (TARGET_MODES_TIEABLE_P): Redefine.
10554 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10555 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10556 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10557 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10558 (mn10300_modes_tieable_p): ...this and make static.
10559 (TARGET_MODES_TIEABLE_P): Redefine.
10560 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10561 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10562 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10563 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10564 (msp430_modes_tieable_p): Make static.
10565 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10566 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10567 (TARGET_MODES_TIEABLE_P): Redefine.
10568 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10569 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10570 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10571 (TARGET_MODES_TIEABLE_P): Redefine.
10572 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10573 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10574 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10575 (TARGET_MODES_TIEABLE_P): Redefine.
10576 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10577 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10578 (pdp11_modes_tieable_p): New function.
10579 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10580 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10581 (rs6000_modes_tieable_p): New function.
10582 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10583 * config/powerpcspe/powerpcspe.md: Update comment.
10584 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10585 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10586 (TARGET_MODES_TIEABLE_P): Redefine.
10587 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10588 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10589 (rl78_modes_tieable_p): New function.
10590 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10591 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10592 (rs6000_modes_tieable_p): New function.
10593 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10594 * config/rs6000/rs6000.md: Update comment.
10595 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10596 * config/rx/rx.c (rx_modes_tieable_p): New function.
10597 (TARGET_MODES_TIEABLE_P): Redefine.
10598 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10599 * config/s390/s390.c (s390_modes_tieable_p): New function.
10600 (TARGET_MODES_TIEABLE_P): Redefine.
10601 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10602 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10603 (sh_modes_tieable_p): New function.
10604 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10605 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10606 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10607 (sparc_modes_tieable_p): Make static.
10608 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10609 * config/spu/spu.c (spu_modes_tieable_p): New function.
10610 (TARGET_MODES_TIEABLE_P): Redefine.
10611 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10612 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10613 (TARGET_MODES_TIEABLE_P): Redefine.
10614 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10615 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10616 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10617 * config/v850/v850.c (v850_modes_tieable_p): New function.
10618 (TARGET_MODES_TIEABLE_P): Redefine.
10619 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10620 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10621 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10622 (visium_modes_tieable_p): New function.
10623 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10624 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10625 (xtensa_modes_tieable_p): New function.
10626 * system.h (MODES_TIEABLE_P): Poison.
10628 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10629 Alan Hayward <alan.hayward@arm.com>
10630 David Sherwood <david.sherwood@arm.com>
10632 * target.def (hard_regno_mode_ok): New hook.
10633 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10634 (TARGET_HARD_REGNO_MODE_OK): ...this.
10635 * doc/tm.texi.in: Regenerate.
10636 * hooks.h (hook_bool_uint_mode_true): Declare.
10637 * hooks.c (hook_bool_uint_mode_true): New function.
10638 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10639 HARD_REGNO_MODE_OK.
10640 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10641 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10642 instead of HARD_REGNO_MODE_OK.
10643 * caller-save.c: Include target.h.
10644 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10645 HARD_REGNO_MODE_OK.
10646 * combine.c (can_combine_p): Likewise.
10647 (combinable_i3pat): Likewise.
10648 (can_change_dest_mode): Likewise.
10649 * expr.c (init_expr_target): Likewise.
10650 (convert_move): Likewise.
10651 (convert_modes): Likewise.
10652 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10653 (setup_prohibited_mode_move_regs): Likewise.
10654 * ira.h (target_ira): Likewise.
10655 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10656 * lra-constraints.c (process_alt_operands): Likewise.
10657 (split_reg): Likewise.
10658 * recog.c (peep2_find_free_register): Likewise.
10659 * ree.c (combine_reaching_defs): Likewise.
10660 * regcprop.c (maybe_mode_change): Likewise.
10661 * reginfo.c (init_reg_sets_1): Likewise.
10662 (choose_hard_reg_mode): Likewise.
10663 (simplifiable_subregs): Likewise.
10664 * regrename.c (check_new_reg_p): Likewise.
10665 * reload.c (find_valid_class): Likewise.
10666 (find_valid_class_1): Likewise.
10667 (reload_inner_reg_of_subreg): Likewise.
10668 (push_reload): Likewise.
10669 (combine_reloads): Likewise.
10670 (find_dummy_reload): Likewise.
10671 (find_reloads): Likewise.
10672 * reload1.c (find_reg): Likewise.
10673 (set_reload_reg): Likewise.
10674 (allocate_reload_reg): Likewise.
10675 (choose_reload_regs): Likewise.
10676 (reload_adjust_reg_for_temp): Likewise.
10677 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10678 (simplify_subreg_regno): Likewise.
10679 * sel-sched.c (init_regs_for_mode): Likewise.
10680 * varasm.c (make_decl_rtl): Likewise.
10681 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10682 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10683 HARD_REGNO_MODE_OK.
10684 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10685 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10686 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10687 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10688 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10689 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10690 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10691 (arc_mode_class): Delete.
10692 (HARD_REGNO_MODE_OK): Delete.
10693 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10694 (arc_hard_regno_mode_ok): Rename old array to...
10695 (arc_hard_regno_mode_ok_modes): ...this.
10696 (arc_conditional_register_usage): Update accordingly.
10697 (arc_mode_class): Make static.
10698 (arc_hard_regno_mode_ok): New function.
10699 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10700 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10701 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10702 (arm_hard_regno_mode_ok): Make static.
10703 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10704 HARD_REGNO_MODE_OK.
10705 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10706 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10707 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10709 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10710 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10711 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10712 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10713 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10714 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10715 * config/bfin/predicates.md (valid_reg_operand): Use
10716 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10717 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10718 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10719 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10720 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10721 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10722 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10723 (cr16_hard_regno_mode_ok): Make static and return a bool.
10724 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10725 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10726 (cris_hard_regno_mode_ok): New function.
10727 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10728 (epiphany_mode_class): Delete.
10729 (HARD_REGNO_MODE_OK): Delete.
10730 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10731 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10732 (hard_regno_mode_ok): Rename to...
10733 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10734 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10735 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10736 HARD_REGNO_MODE_OK.
10737 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10738 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10739 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10740 (frv_hard_regno_mode_ok): Make static and return a bool.
10741 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10742 HARD_REGNO_MODE_OK.
10743 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10744 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10745 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10746 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10748 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10749 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10750 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10751 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10753 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10754 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10755 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10756 (ia64_hard_regno_mode_ok): New function.
10757 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10758 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10759 (iq2000_hard_regno_mode_ok): New function.
10760 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10761 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10762 (lm32_hard_regno_mode_ok): New function.
10763 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10764 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10765 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10766 instead of HARD_REGNO_MODE_OK.
10767 (m32c_hard_regno_ok): Rename to...
10768 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10769 (m32c_cannot_change_mode_class): Update accordingly.
10770 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10771 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10772 (m32r_mode_class): Delete.
10773 (HARD_REGNO_MODE_OK): Delete.
10774 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10775 (m32r_hard_regno_mode_ok): Rename to...
10776 (m32r_hard_regno_modes): ...this.
10777 (m32r_mode_class): Make static.
10778 (m32r_hard_regno_mode_ok): New function.
10779 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10780 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10781 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10782 (m68k_hard_regno_mode_ok): Make static.
10783 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10784 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10785 (mcore_hard_regno_mode_ok): New function.
10786 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10787 (HARD_REGNO_MODE_OK): Delete.
10788 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10790 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10791 (microblaze_hard_regno_mode_ok): New function.
10792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10793 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10794 (mips_hard_regno_mode_ok): Delete.
10795 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10796 (mips_hard_regno_mode_ok_p): ...this and make static.
10797 (mips_hard_regno_mode_ok_p): Rename to...
10798 (mips_hard_regno_mode_ok_uncached): ...this.
10799 (mips_hard_regno_mode_ok): New function.
10800 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10801 of HARD_REGNO_MODE_OK.
10802 (mips_option_override): Update after above name changes.
10803 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10804 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10805 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10806 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10807 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10808 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10809 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10810 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10811 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10812 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10813 (msp430_hard_regno_mode_ok): Make static and return a bool.
10814 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10815 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10816 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10818 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10819 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10820 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10821 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10822 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10823 (PA_HARD_REGNO_MODE_OK): ...this
10824 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10825 (PA_HARD_REGNO_MODE_OK): ...this.
10826 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10827 (pa_hard_regno_mode_ok): New function.
10828 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10829 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10830 (pdp11_hard_regno_mode_ok): New function.
10831 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10832 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10834 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10837 (rs6000_hard_regno_mode_ok): Rename to...
10838 (rs6000_hard_regno_mode_ok_uncached): ...this.
10839 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10840 (rs6000_hard_regno_mode_ok): New function.
10841 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10842 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10843 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10844 (riscv_hard_regno_mode_ok): ...this and make static.
10845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10846 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10847 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10848 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10849 (rl78_hard_regno_mode_ok): Make static and return bool.
10850 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10851 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10853 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10854 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10855 (rs6000_hard_regno_mode_ok): Rename to...
10856 (rs6000_hard_regno_mode_ok_uncached): ...this.
10857 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10858 (rs6000_hard_regno_mode_ok): New function.
10859 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10860 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10861 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10862 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10863 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10864 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10866 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10867 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10868 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10869 (sh_hard_regno_mode_ok): Make static.
10870 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10871 instead of HARD_REGNO_MODE_OK.
10872 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10873 (sparc_mode_class): Delete.
10874 (HARD_REGNO_MODE_OK): Delete.
10875 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10876 (hard_regno_mode_classes): Make static.
10877 (sparc_mode_class): Likewise.
10878 (sparc_hard_regno_mode_ok): New function.
10879 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10880 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10881 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10884 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10885 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10886 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10887 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10888 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10889 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10890 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10891 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10892 (visium_hard_regno_mode_ok): New function.
10893 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10894 instead of HARD_REGNO_MODE_OK.
10895 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10896 (HARD_REGNO_MODE_OK): Delete.
10897 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10898 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10899 (xtensa_option_override): Update accordingly.
10900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10901 (xtensa_hard_regno_mode_ok): New function.
10902 * system.h (HARD_REGNO_MODE_OK): Poison.
10904 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10905 Alan Hayward <alan.hayward@arm.com>
10906 David Sherwood <david.sherwood@arm.com>
10908 * target.def (hard_regno_call_part_clobbered): New hook.
10909 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10910 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10911 * doc/tm.texi: Regenerate.
10912 * hooks.h (hook_bool_uint_mode_false): Declare.
10913 * hooks.c (hook_bool_uint_mode_false): New function.
10914 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10915 * cselib.c (cselib_process_insn): Use
10916 targetm.hard_regno_call_part_clobbered instead of
10917 HARD_REGNO_CALL_PART_CLOBBERED.
10918 * ira-conflicts.c (ira_build_conflicts): Likewise.
10919 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10920 * lra-constraints.c (need_for_call_save_p): Likewise.
10921 * lra-lives.c: Include target.h.
10922 (check_pseudos_live_through_calls): Use
10923 targetm.hard_regno_call_part_clobbered instead of
10924 HARD_REGNO_CALL_PART_CLOBBERED.
10925 * regcprop.c: Include target.h.
10926 (copyprop_hardreg_forward_1): Use
10927 targetm.hard_regno_call_part_clobbered instead of
10928 HARD_REGNO_CALL_PART_CLOBBERED.
10929 * reginfo.c (choose_hard_reg_mode): Likewise.
10930 * regrename.c (check_new_reg_p): Likewise.
10931 * reload.c (find_equiv_reg): Likewise.
10932 * reload1.c (emit_reload_insns): Likewise.
10933 * sched-deps.c (deps_analyze_insn): Likewise.
10934 * sel-sched.c (init_regs_for_mode): Likewise.
10935 (mark_unavailable_hard_regs): Likewise.
10936 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10937 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10938 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10940 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10941 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10942 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10944 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10946 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10947 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10948 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10950 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10951 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10952 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10954 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10955 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10957 * config/powerpcspe/powerpcspe.c
10958 (rs6000_hard_regno_call_part_clobbered): New function.
10959 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10960 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10961 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10963 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10964 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10965 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10967 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10968 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10969 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10971 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10972 Alan Hayward <alan.hayward@arm.com>
10973 David Sherwood <david.sherwood@arm.com>
10975 * rtl.h (subreg_memory_offset): Declare.
10976 * emit-rtl.c (subreg_memory_offset): New function.
10977 * expmed.c (store_bit_field_1): Use it.
10978 * expr.c (undefined_operand_subword_p): Likewise.
10979 * simplify-rtx.c (simplify_subreg): Likewise.
10981 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10983 PR rtl-optimization/57448
10986 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10987 using atomic_load pattern.
10988 (expand_atomic_store): Likewise.
10990 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10993 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10994 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10995 handling. Use replace_call_with_value with NULL instead of
10996 gsi_replace, unlink_stmt_vdef and release_defs.
10998 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11001 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11003 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11006 * rtl.h (emit_library_call_value_1): Declare.
11007 (emit_library_call): Replace declaration with a series of overloads.
11008 Remove the parameter count argument.
11009 (emit_library_call_value): Likewise.
11010 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11011 with an "rtx_mode_t *".
11012 (emit_library_call_value): Delete.
11013 (emit_library_call): Likewise.
11014 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11015 (asan_emit_allocas_unpoison): Likewise.
11016 * builtins.c (expand_builtin_powi): Likewise.
11017 (expand_asan_emit_allocas_unpoison): Likewise.
11018 * cfgexpand.c (expand_main_function): Likewise.
11019 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11020 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11021 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11022 * config/arm/arm.c (arm_trampoline_init): Likewise.
11023 (arm_call_tls_get_addr): Likewise.
11024 (arm_expand_divmod_libfunc): Likewise.
11025 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11026 (smulsi3_highpart): Likewise.
11027 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11028 (c6x_expand_compare): Likewise.
11029 (c6x_expand_movmem): Likewise.
11030 * config/frv/frv.c (frv_trampoline_init): Likewise.
11031 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11032 (ix86_expand_divmod_libfunc): Likewise.
11033 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11034 (ia64_expand_compare): Likewise.
11035 (ia64_profile_hook): Likewise.
11036 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11037 (nonlocal_goto): Likewise.
11038 (restore_stack_nonlocal): Likewise.
11039 * config/m32r/m32r.c (block_move_call): Likewise.
11040 (m32r_trampoline_init): Likewise.
11041 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11042 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11043 (m68k_call_m68k_read_tp): Likewise.
11044 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11045 (microblaze_expand_divide): Likewise.
11046 * config/mips/mips.h (mips_args): Likewise.
11047 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11048 (MIPS_ICACHE_SYNC): Likewise.
11049 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11050 (nios2_trampoline_init): Likewise.
11051 * config/pa/pa.c (hppa_tls_call): Likewise.
11052 (pa_trampoline_init): Likewise.
11053 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11054 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11055 (expand_strn_compare): Likewise.
11056 (rs6000_generate_compare): Likewise.
11057 (rs6000_expand_float128_convert): Likewise.
11058 (output_profile_hook): Likewise.
11059 (rs6000_trampoline_init): Likewise.
11060 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11061 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11062 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11063 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11064 (rs6000_generate_compare): Likewise.
11065 (rs6000_expand_float128_convert): Likewise.
11066 (output_profile_hook): Likewise.
11067 (rs6000_trampoline_init): Likewise.
11068 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11069 * config/sh/sh.c (sh_trampoline_init): Likewise.
11070 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11071 (sparc_emit_float_lib_cmp): Likewise.
11072 (sparc32_initialize_trampoline): Likewise.
11073 (sparc64_initialize_trampoline): Likewise.
11074 (sparc_profile_hook): Likewise.
11075 * config/spu/spu.c (ea_load_store): Likewise.
11076 * config/spu/spu.md (floatunssidf2): Likewise.
11077 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11078 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11079 * config/visium/visium.c (expand_block_move_4): Likewise.
11080 (expand_block_move_2): Likewise.
11081 (expand_block_move_1): Likewise.
11082 (expand_block_set_4): Likewise.
11083 (expand_block_set_2): Likewise.
11084 (expand_block_set_1): Likewise.
11085 (visium_trampoline_init): Likewise.
11086 (visium_profile_hook): Likewise.
11087 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11088 (xtensa_setup_frame_addresses): Likewise.
11089 (xtensa_trampoline_init): Likewise.
11090 * except.c (sjlj_emit_function_enter): Likewise.
11091 (sjlj_emit_function_exit): Likewise.
11092 * explow.c (allocate_dynamic_stack_space): Likewise.
11093 (probe_stack_range): Likewise.
11094 * expr.c (convert_mode_scalar): Likewise.
11095 * optabs.c (expand_binop): Likewise.
11096 (expand_twoval_binop_libfunc): Likewise.
11097 (expand_unop): Likewise.
11098 (prepare_cmp_insn): Likewise.
11099 (prepare_float_lib_cmp): Likewise.
11100 (expand_float): Likewise.
11101 (expand_fix): Likewise.
11102 (expand_fixed_convert): Likewise.
11103 (maybe_emit_sync_lock_test_and_set): Likewise.
11104 (expand_atomic_compare_and_swap): Likewise.
11105 (expand_mem_thread_fence): Likewise.
11106 (expand_atomic_fetch_op): Likewise.
11108 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11110 * doc/generic.texi (OpenACC): Adjust URL.
11111 * doc/invoke.texi (C Dialect Options): Ditto.
11113 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11115 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11116 predicate for operand 1. Add (m,<S>) constraint.
11117 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11118 Prevent memory operand 1 with register operand 2.
11120 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11122 PR rtl-optimization/82024
11123 * combine.c (try_combine): If the combination result is a PARALLEL,
11124 and we only need to retain the SET in there that would be placed
11125 at I2, check that we can place that at I3 instead, before doing so.
11127 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11130 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11132 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11133 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11136 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11137 Jeff Law <law@redhat.com>
11139 * varasm.c (bss_initializer_p): Do not put constants into .bss
11140 (categorize_decl_for_section): Handle bss_initializer_p returning
11141 false when DECL_INITIAL is NULL.
11143 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11146 * config/s390/s390.c (s390_can_inline_p): New function.
11148 2017-09-01 Jeff Law <law@redhat.com>
11150 PR tree-optimization/82052
11151 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11152 Always initialize the returned slot after a hash table miss
11153 when INSERT is true.
11155 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11157 * config/s390/s390.md (mem_signal_fence): Remove.
11158 * doc/md.texi (mem_signal_fence): Remove.
11159 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11161 * target-insns.def (mem_signal_fence): Remove.
11163 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11166 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11169 * asan.c (create_odr_indicator): Strip name encoding from assembler
11170 name before appending it after __odr_asan_.
11172 2017-09-01 Martin Liska <mliska@suse.cz>
11174 PR tree-optimization/82059
11175 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11176 frequency only when an edge is redirected.
11178 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11180 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11181 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11182 (arc_conditional_register_usage): Remove ARC600 lp_count
11184 (arc_file_start): Emit Tag_ARC_CPU_variation.
11185 (arc_can_use_doloop_p): New conditions to use ZOLs.
11186 (hwloop_fail): New function.
11187 (hwloop_optimize): Likewise.
11188 (hwloop_pattern_reg): Likewise.
11189 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11190 (arc_reorg_loops): New function, calls reorg_loops.
11191 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11192 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11194 (arc_loop_hazard): Remove function, functionality moved into
11196 (arc_hazard): Remove arc_loop_hazard call.
11197 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11198 into hwloop_optimize.
11199 (arc_label_align): Remove ZOL handling.
11200 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11201 * config/arc/arc.md (doloop_begin): Remove pattern.
11202 (doloop_begin_i): Likewise.
11203 (doloop_end_i): Likewise.
11204 (doloop_fallback): Likewise.
11205 (doloop_fallback_m): Likewise.
11206 (doloop_end): Reimplement expand.
11207 (arc_lp): New pattern for LP instruction.
11208 (loop_end): New pattern.
11209 (loop_fail): Likewise.
11210 (decrement_and_branch_until_zero): Likewise.
11211 * config/arc/arc.opt (mlpc-width): New option.
11212 * doc/invoke.texi (mlpc-width): Document option.
11214 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11216 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11217 (arc_ccfsm_advance): Fix checking for delay slots.
11218 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11220 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11222 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11224 (movhi_insn): Update store instruction constraint which are saving
11225 6-bit short immediates.
11226 (movsi_insn): Consider also short scaled load operations.
11227 (zero_extendhisi2_i): Use Usd constraint instead of T.
11228 (extendhisi2_i): Add q constraint.
11229 (arc_clzsi2): Add type and length attributes.
11230 (arc_ctzsi2): Likewise.
11231 * config/arc/constraints.md (Usc): Update constraint, the
11232 assembler can parse two relocations for a single instruction.
11234 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11236 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11237 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11239 2017-08-31 Olivier Hainque <hainque@adacore.com>
11241 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11242 match as powerpc-wrs-vxworks*.
11244 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11246 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11247 register constraint for by-element operand.
11248 (aarch64_mls_elt_merge<mode>): Likewise.
11250 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11252 * config/arc/arc.c (arc_can_follow_jump): Check for short
11255 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11257 * config.gcc: Use g.opt for arc.
11258 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11259 functionality moved to ...
11260 (legitimate_scaled_address_p): New function, ...here.
11261 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11262 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11263 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11265 (arc_override_options): Handle G option.
11266 (arc_output_pic_addr_const): Correct function definition.
11267 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11268 (arc_decl_anon_ns_mem_p): Delete.
11269 (arc_in_small_data_p): Overhaul this function to take into
11270 consideration the value given via G option.
11271 (arc_rewrite_small_data_1): Renamed and corrected old
11272 arc_rewrite_small_data function.
11273 (arc_rewrite_small_data): New function.
11274 (small_data_pattern): Don't use pic_offset_table_rtx.
11275 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11276 * config/arc/simdext.md (movmisalignv2hi): Use
11277 prepare_move_operands function.
11279 (movmisalign*): Likewise.
11280 * doc/invoke.texi (ARC options): Document -G option.
11282 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11284 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11286 * config/arc/arc.c (arc_print_operand): Output scalled address for
11287 sdata whenever is possible.
11288 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11289 load/stores are available.
11290 (compact_sda_memory_operand): Check for the alignment required by
11291 code density instructions.
11292 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11294 * config/arc/constraints.md (Usd): Update constraint.
11295 (Us0): New constraint.
11296 (Usc): Update constraint.
11298 2017-08-31 Richard Biener <rguenther@suse.de>
11300 PR middle-end/82054
11301 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11302 function only once.
11304 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11306 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11307 Resize type_signature.
11309 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11310 Alan Hayward <alan.hayward@arm.com>
11311 David Sherwood <david.sherwood@arm.com>
11313 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11314 subregs whose inner modes can be stored in GPRs.
11315 (aarch64_classify_index): Likewise.
11317 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11318 Alan Hayward <alan.hayward@arm.com>
11319 David Sherwood <david.sherwood@arm.com>
11321 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11322 (V_INT_EQUIV): ...this.
11323 (v_cmp_result): Rename to...
11324 (v_int_equiv): ...this.
11325 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11326 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11327 (copysign<mode>3): Likewise.
11328 (aarch64_simd_bsl<mode>_internal): Likewise.
11329 (aarch64_simd_bsl<mode>): Likewise.
11330 (vec_cmp<mode><mode>): Likewise.
11331 (vcond<mode><mode>): Likewise.
11332 (vcond<v_cmp_mixed><mode>): Likewise.
11333 (vcondu<mode><v_cmp_mixed>): Likewise.
11334 (aarch64_cm<optab><mode>): Likewise.
11335 (aarch64_cmtst<mode>): Likewise.
11336 (aarch64_fac<optab><mode>): Likewise.
11337 (vec_perm_const<mode>): Likewise.
11338 (vcond_mask_<mode><v_cmp_result>): Rename to...
11339 (vcond_mask_<mode><v_int_equiv>): ...this.
11340 (vec_cmp<mode><v_cmp_result>): Rename to...
11341 (vec_cmp<mode><v_int_equiv>): ...this.
11343 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11344 Alan Hayward <alan.hayward@arm.com>
11345 David Sherwood <david.sherwood@arm.com>
11347 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11349 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11350 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11351 (UNSPEC_LD4_DREG): New unspecs.
11352 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11353 (aarch64_ld2<mode>_dreg_be): Replace with...
11354 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11356 (aarch64_ld3<mode>_dreg_le)
11357 (aarch64_ld3<mode>_dreg_be): Replace with...
11358 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11360 (aarch64_ld4<mode>_dreg_le)
11361 (aarch64_ld4<mode>_dreg_be): Replace with...
11362 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11365 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11367 PR tree-optimization/81987
11368 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11369 insert an initializer in a location not dominated by the stride
11372 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11374 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11375 on the entire header of the finally block in the fallthru case.
11377 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11379 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11381 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11383 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11384 rs6000_emit_move_from_cr and call renamed function.
11385 (rs6000_emit_prologue): Call renamed functions.
11386 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11387 movesi_from_cr, remove volatile CRs.
11389 2017-08-30 Jon Beniston <jon@beniston.com>
11390 Richard Biener <rguenther@suse.de>
11392 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11393 of VECTOR_MODE_P check.
11394 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11395 element vector types.
11397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11399 * df.h (df_read_modify_subreg_p): Remove in favor of...
11400 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11401 const_rtx instead of an rtx.
11402 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11403 * df-problems.c (df_word_lr_mark_ref): Likewise.
11404 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11405 (mark_pseudo_reg_dead): Likewise.
11406 (mark_ref_dead): Likewise.
11407 * reginfo.c (init_subregs_of_mode): Likewise.
11408 * sched-deps.c (sched_analyze_1): Likewise.
11409 * df-scan.c (df_def_record_1): Likewise.
11410 (df_uses_record): Likewise.
11411 (df_read_modify_subreg_p): Remove in favor of...
11412 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11413 const_rtx instead of an rtx.
11415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11416 Alan Hayward <alan.hayward@arm.com>
11417 David Sherwood <david.sherwood@arm.com>
11419 * rtl.h (partial_subreg_p): New function.
11420 * caller-save.c (save_call_clobbered_regs): Use it.
11421 * calls.c (expand_call): Likewise.
11422 * combine.c (combinable_i3pat): Likewise.
11423 (simplify_set): Likewise.
11424 (make_extraction): Likewise.
11425 (make_compound_operation_int): Likewise.
11426 (gen_lowpart_or_truncate): Likewise.
11427 (force_to_mode): Likewise.
11428 (make_field_assignment): Likewise.
11429 (reg_truncated_to_mode): Likewise.
11430 (record_truncated_value): Likewise.
11431 (move_deaths): Likewise.
11432 * cse.c (record_jump_cond): Likewise.
11433 (cse_insn): Likewise.
11434 * cselib.c (cselib_lookup_1): Likewise.
11435 * expmed.c (extract_bit_field_using_extv): Likewise.
11436 * function.c (assign_parm_setup_reg): Likewise.
11437 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11438 * ira-build.c (create_insn_allocnos): Likewise.
11439 * lra-coalesce.c (merge_pseudos): Likewise.
11440 * lra-constraints.c (match_reload): Likewise.
11441 (simplify_operand_subreg): Likewise.
11442 (curr_insn_transform): Likewise.
11443 * lra-lives.c (process_bb_lives): Likewise.
11444 * lra.c (new_insn_reg): Likewise.
11445 (lra_substitute_pseudo): Likewise.
11446 * regcprop.c (mode_change_ok): Likewise.
11447 (maybe_mode_change): Likewise.
11448 (copyprop_hardreg_forward_1): Likewise.
11449 * reload.c (push_reload): Likewise.
11450 (find_reloads): Likewise.
11451 (find_reloads_subreg_address): Likewise.
11452 * reload1.c (alter_reg): Likewise.
11453 (eliminate_regs_1): Likewise.
11454 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11456 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11458 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11459 back to if statements, including unpack.
11461 2017-08-30 Martin Liska <mliska@suse.cz>
11463 PR inline-asm/82001
11464 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11466 (func_checker::compare_asm_inputs_outputs): ... this function.
11467 (func_checker::compare_gimple_asm): Use the function to compare
11468 also ASM constrains.
11469 * ipa-icf-gimple.h: Rename the function.
11471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11472 Alan Hayward <alan.hayward@arm.com>
11473 David Sherwood <david.sherwood@arm.com>
11475 * coretypes.h (complex_mode): New type.
11476 * gdbhooks.py (build_pretty_printer): Handle it.
11477 * machmode.h (complex_mode): New class.
11478 (complex_mode::includes_p): New function.
11479 (is_complex_int_mode): Likewise.
11480 (is_complex_float_mode): Likewise.
11481 * genmodes.c (get_mode_class): Handle complex mode classes.
11482 * function.c (expand_function_end): Use is_complex_int_mode.
11484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11485 Alan Hayward <alan.hayward@arm.com>
11486 David Sherwood <david.sherwood@arm.com>
11488 * coretypes.h (scalar_mode_pod): New typedef.
11489 * gdbhooks.py (build_pretty_printer): Handle it.
11490 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11491 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11492 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11493 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11494 as_a <scalar_mode>.
11496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11497 Alan Hayward <alan.hayward@arm.com>
11498 David Sherwood <david.sherwood@arm.com>
11500 * machmode.h (mode_for_vector): Take a scalar_mode instead
11502 * stor-layout.c (mode_for_vector): Likewise.
11503 * explow.c (promote_mode): Use as_a <scalar_mode>.
11504 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11507 Alan Hayward <alan.hayward@arm.com>
11508 David Sherwood <david.sherwood@arm.com>
11510 * target.def (preferred_simd_mode): Take a scalar_mode
11511 instead of a machine_mode.
11512 * targhooks.h (default_preferred_simd_mode): Likewise.
11513 * targhooks.c (default_preferred_simd_mode): Likewise.
11514 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11515 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11516 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11517 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11518 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11519 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11520 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11521 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11523 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11524 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11525 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11526 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11527 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11528 * doc/tm.texi: Regenerate.
11529 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11533 Alan Hayward <alan.hayward@arm.com>
11534 David Sherwood <david.sherwood@arm.com>
11536 * target.def (scalar_mode_supported_p): Take a scalar_mode
11537 instead of a machine_mode.
11538 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11539 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11540 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11541 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11542 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11543 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11544 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11545 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11546 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11547 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11548 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11549 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11550 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11551 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11553 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11554 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11555 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11556 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11557 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11559 * doc/tm.texi: Regenerate.
11561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11562 Alan Hayward <alan.hayward@arm.com>
11563 David Sherwood <david.sherwood@arm.com>
11565 * coretypes.h (opt_scalar_mode): New typedef.
11566 * gdbhooks.py (build_pretty_printers): Handle it.
11567 * machmode.h (mode_iterator::get_2xwider): Add overload for
11569 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11571 * expr.c (convert_mode_scalar): Likewise.
11572 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11573 * optabs.c (expand_float): Likewise.
11574 (expand_fix): Likewise.
11575 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11578 Alan Hayward <alan.hayward@arm.com>
11579 David Sherwood <david.sherwood@arm.com>
11581 * optabs.c (expand_float): Explicitly check for scalars before
11582 using a branching expansion.
11583 (expand_fix): Likewise.
11585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11586 Alan Hayward <alan.hayward@arm.com>
11587 David Sherwood <david.sherwood@arm.com>
11589 * expr.c (convert_mode): Split scalar handling out into...
11590 (convert_mode_scalar): ...this new function. Treat the modes
11593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11594 Alan Hayward <alan.hayward@arm.com>
11595 David Sherwood <david.sherwood@arm.com>
11597 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11599 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11602 Alan Hayward <alan.hayward@arm.com>
11603 David Sherwood <david.sherwood@arm.com>
11605 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11606 rather than a machine_mode.
11607 (fixed_from_string): Likewise.
11608 (fixed_convert): Likewise.
11609 (fixed_convert_from_int): Likewise.
11610 (fixed_convert_from_real): Likewise.
11611 (real_convert_from_fixed): Likewise.
11612 * fixed-value.c (fixed_from_double_int): Likewise.
11613 (fixed_from_string): Likewise.
11614 (fixed_convert): Likewise.
11615 (fixed_convert_from_int): Likewise.
11616 (fixed_convert_from_real): Likewise.
11617 (real_convert_from_fixed): Likewise.
11618 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11621 Alan Hayward <alan.hayward@arm.com>
11622 David Sherwood <david.sherwood@arm.com>
11624 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11625 of separate mode class checks. Do not allow vector modes here.
11626 (immed_wide_int_const): Use as_a <scalar_mode>.
11627 * explow.c (trunc_int_for_mode): Likewise.
11628 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11629 (wi::shwi): Likewise.
11630 (wi::min_value): Likewise.
11631 (wi::max_value): Likewise.
11632 * dwarf2out.c (loc_descriptor): Likewise.
11633 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11634 for CONST_WIDE_INT.
11636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11637 Alan Hayward <alan.hayward@arm.com>
11638 David Sherwood <david.sherwood@arm.com>
11640 * tree.h (SCALAR_TYPE_MODE): New macro.
11641 * expr.c (expand_expr_addr_expr_1): Use it.
11642 (expand_expr_real_2): Likewise.
11643 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11644 (fold_convert_const_fixed_from_int): Likewise.
11645 (fold_convert_const_fixed_from_real): Likewise.
11646 (native_encode_fixed): Likewise
11647 (native_encode_complex): Likewise
11648 (native_encode_vector): Likewise.
11649 (native_interpret_fixed): Likewise.
11650 (native_interpret_real): Likewise.
11651 (native_interpret_complex): Likewise.
11652 (native_interpret_vector): Likewise.
11653 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11654 (simd_clone_adjust_argument_types): Likewise.
11655 (simd_clone_init_simd_arrays): Likewise.
11656 (simd_clone_adjust): Likewise.
11657 * stor-layout.c (layout_type): Likewise.
11658 * tree.c (build_minus_one_cst): Likewise.
11659 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11660 * tree-inline.c (estimate_move_cost): Likewise.
11661 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11662 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11663 (vectorizable_reduction): Likewise.
11664 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11665 (vect_recog_mixed_size_cond_pattern): Likewise.
11666 (check_bool_pattern): Likewise.
11667 (adjust_bool_pattern): Likewise.
11668 (search_type_for_mask_1): Likewise.
11669 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11670 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11671 (vectorizable_load): Likewise.
11672 (vectorizable_store): Likewise.
11673 * ubsan.c (ubsan_encode_value): Likewise.
11674 * varasm.c (output_constant): Likewise.
11676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11677 Alan Hayward <alan.hayward@arm.com>
11678 David Sherwood <david.sherwood@arm.com>
11680 * coretypes.h (scalar_mode): New class.
11681 * machmode.h (scalar_mode): Likewise.
11682 (scalar_mode::includes_p): New function.
11683 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11684 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11685 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11686 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11687 * expmed.c (store_bit_field_1): Likewise.
11688 (extract_bit_field_1): Likewise.
11689 * expr.c (write_complex_part): Likewise.
11690 (read_complex_part): Likewise.
11691 (emit_move_complex_push): Likewise.
11692 (expand_expr_real_2): Likewise.
11693 * function.c (assign_parm_setup_reg): Likewise.
11694 (assign_parms_unsplit_complex): Likewise.
11695 * optabs.c (expand_binop): Likewise.
11696 * rtlanal.c (subreg_get_info): Likewise.
11697 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11698 * varasm.c (output_constant_pool_2): Likewise.
11700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11701 Alan Hayward <alan.hayward@arm.com>
11702 David Sherwood <david.sherwood@arm.com>
11704 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11706 (expmed_mult_highpart_optab): Likewise.
11707 (expmed_mult_highpart): Likewise.
11709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11710 Alan Hayward <alan.hayward@arm.com>
11711 David Sherwood <david.sherwood@arm.com>
11713 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11714 instead of a machine_mode.
11715 (builtin_memset_read_str): Likewise.
11716 * builtins.c (c_readstr): Likewise.
11717 (builtin_memcpy_read_str): Likewise.
11718 (builtin_strncpy_read_str): Likewise.
11719 (builtin_memset_read_str): Likewise.
11720 (builtin_memset_gen_str): Likewise.
11721 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11722 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11723 instead of a machine_mode.
11724 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11726 (make_extraction): Likewise.
11727 (try_widen_shift_mode): Take and return scalar_int_modes instead
11729 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11730 a scalar_int_mode instead of a machine_mode.
11731 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11732 (avr_addr_space_pointer_mode): Likewise.
11733 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11734 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11735 (msp430_unwind_word_mode): Likewise.
11736 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11737 (spu_addr_space_pointer_mode): Likewise.
11738 (spu_addr_space_address_mode): Likewise.
11739 (spu_libgcc_cmp_return_mode): Likewise.
11740 (spu_libgcc_shift_count_mode): Likewise.
11741 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11742 (rl78_addr_space_pointer_mode): Likewise.
11743 (fl78_unwind_word_mode): Likewise.
11744 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11746 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11747 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11748 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11749 (mips_valid_pointer_mode): Likewise.
11750 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11751 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11752 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11754 (ft32_addr_space_address_mode): Likewise.
11755 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11756 scalar_int_mode instead of a machine_mode.
11757 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11759 (m32c_addr_space_address_mode): Likewise.
11760 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11761 (rs6000_eh_return_filter_mode): Likewise.
11762 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11763 (rs6000_eh_return_filter_mode): Likewise.
11764 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11765 (s390_libgcc_shift_count_mode): Likewise.
11766 (s390_unwind_word_mode): Likewise.
11767 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11769 * target.def (mode_rep_extended): Likewise.
11770 (valid_pointer_mode): Likewise.
11771 (addr_space.valid_pointer_mode): Likewise.
11772 (eh_return_filter_mode): Return a scalar_int_mode rather than
11774 (libgcc_cmp_return_mode): Likewise.
11775 (libgcc_shift_count_mode): Likewise.
11776 (unwind_word_mode): Likewise.
11777 (addr_space.pointer_mode): Likewise.
11778 (addr_space.address_mode): Likewise.
11779 * doc/tm.texi: Regenerate.
11780 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11782 (do_jump): Use scalar_int_mode for local variables.
11783 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11784 rather than a machine_mode.
11785 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11786 (scompare_loc_descriptor_wide): Likewise.
11787 (scompare_loc_descriptor_narrow): Likewise.
11788 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11790 * except.c (sjlj_emit_dispatch_table): Likewise.
11791 (expand_builtin_eh_copy_values): Likewise.
11792 * explow.c (convert_memory_address_addr_space_1): Likewise.
11793 Take a scalar_int_mode rather than a machine_mode.
11794 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11795 than a machine_mode.
11796 (memory_address_addr_space): Use scalar_int_mode for local variables.
11797 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11798 rather than a machine_mode.
11799 * expmed.c (mask_rtx): Likewise.
11800 (init_expmed_one_conv): Likewise.
11801 (expand_mult_highpart_adjust): Likewise.
11802 (extract_high_half): Likewise.
11803 (expmed_mult_highpart_optab): Likewise.
11804 (expmed_mult_highpart): Likewise.
11805 (expand_smod_pow2): Likewise.
11806 (expand_sdiv_pow2): Likewise.
11807 (emit_store_flag_int): Likewise.
11808 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11810 (extract_low_bits): Likewise.
11811 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11813 * expr.c (pieces_addr::adjust): Likewise.
11814 (can_store_by_pieces): Likewise.
11815 (store_by_pieces): Likewise.
11816 (clear_by_pieces_1): Likewise.
11817 (expand_expr_addr_expr_1): Likewise.
11818 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11819 (expand_expr_real_1): Likewise.
11820 (try_casesi): Likewise.
11821 * final.c (shorten_branches): Likewise.
11822 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11823 type of "mode" to machine_mode.
11824 * internal-fn.c (expand_arith_overflow_result_store): Take a
11825 scalar_int_mode rather than a machine_mode.
11826 (expand_mul_overflow): Use scalar_int_mode for local variables.
11827 * loop-doloop.c (doloop_modify): Likewise.
11828 (doloop_optimize): Likewise.
11829 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11830 than a machine_mode.
11831 (expand_doubleword_shift_condmove): Likewise.
11832 (expand_doubleword_shift): Likewise.
11833 (expand_doubleword_clz): Likewise.
11834 (expand_doubleword_popcount): Likewise.
11835 (expand_doubleword_parity): Likewise.
11836 (expand_absneg_bit): Use scalar_int_mode for local variables.
11837 (prepare_float_lib_cmp): Likewise.
11838 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11839 rather than a machine_mode.
11840 (convert_memory_address_addr_space): Likewise.
11841 (get_mode_bounds): Likewise.
11842 (get_address_mode): Return a scalar_int_mode rather than a
11844 * rtlanal.c (get_address_mode): Likewise.
11845 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11846 than a machine_mode.
11847 * targhooks.c (default_mode_rep_extended): Likewise.
11848 (default_valid_pointer_mode): Likewise.
11849 (default_addr_space_valid_pointer_mode): Likewise.
11850 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11851 than a machine_mode.
11852 (default_libgcc_cmp_return_mode): Likewise.
11853 (default_libgcc_shift_count_mode): Likewise.
11854 (default_unwind_word_mode): Likewise.
11855 (default_addr_space_pointer_mode): Likewise.
11856 (default_addr_space_address_mode): Likewise.
11857 * targhooks.h (default_eh_return_filter_mode): Likewise.
11858 (default_libgcc_cmp_return_mode): Likewise.
11859 (default_libgcc_shift_count_mode): Likewise.
11860 (default_unwind_word_mode): Likewise.
11861 (default_addr_space_pointer_mode): Likewise.
11862 (default_addr_space_address_mode): Likewise.
11863 (default_mode_rep_extended): Take a scalar_int_mode rather than
11865 (default_valid_pointer_mode): Likewise.
11866 (default_addr_space_valid_pointer_mode): Likewise.
11867 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11869 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11870 rather than a machine_mode.
11871 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11872 for local variables.
11873 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11874 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11875 than a machine_mode.
11877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11878 Alan Hayward <alan.hayward@arm.com>
11879 David Sherwood <david.sherwood@arm.com>
11881 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11882 the mode argument to scalar_int_mode.
11883 (do_jump_by_parts_zero_rtx): Likewise.
11884 (do_jump_by_parts_equality_rtx): Likewise.
11885 (do_jump_by_parts_greater): Take a mode argument.
11886 (do_jump_by_parts_equality): Likewise.
11887 (do_jump_1): Update calls accordingly.
11889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11890 Alan Hayward <alan.hayward@arm.com>
11891 David Sherwood <david.sherwood@arm.com>
11893 * is-a.h (safe_dyn_cast): New function.
11894 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11895 (jump_table_for_label): Likewise.
11896 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11897 instead of an rtx_insn *.
11898 (shorten_branches): Use dyn_cast instead of LABEL_P and
11899 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11900 rtx_jump_table_data::get_data_mode.
11901 (final_scan_insn): Likewise.
11903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11904 Alan Hayward <alan.hayward@arm.com>
11905 David Sherwood <david.sherwood@arm.com>
11907 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11908 trying to combine a full-register integer set with a subreg
11911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11912 Alan Hayward <alan.hayward@arm.com>
11913 David Sherwood <david.sherwood@arm.com>
11915 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11916 that is always either address_mode or pointer_mode.
11918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11919 Alan Hayward <alan.hayward@arm.com>
11920 David Sherwood <david.sherwood@arm.com>
11922 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11923 when the two are known to be equal.
11925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11926 Alan Hayward <alan.hayward@arm.com>
11927 David Sherwood <david.sherwood@arm.com>
11929 * simplify-rtx.c (simplify_const_unary_operation): Use
11930 is_a <scalar_int_mode> instead of checking for a nonzero
11931 precision. Forcibly convert op_mode to a scalar_int_mode
11932 in that case. More clearly differentiate the operand and
11933 result modes and use the former when deciding what the value
11934 of a count-bits operation should be. Use is_int_mode instead
11935 of checking for a MODE_INT. Remove redundant check for whether
11936 this mode has a zero precision.
11938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11939 Alan Hayward <alan.hayward@arm.com>
11940 David Sherwood <david.sherwood@arm.com>
11942 * optabs.c (widen_leading): Change the type of the mode argument
11943 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11944 (widen_bswap): Likewise.
11945 (expand_parity): Likewise.
11946 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11947 (expand_ffs): Likewise.
11948 (epand_unop): Check for scalar integer modes before calling the
11951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11952 Alan Hayward <alan.hayward@arm.com>
11953 David Sherwood <david.sherwood@arm.com>
11955 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11957 (expand_expr_real_1): Update call accordingly.
11959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11960 Alan Hayward <alan.hayward@arm.com>
11961 David Sherwood <david.sherwood@arm.com>
11963 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11964 value_mode arguments. Use scalar_int_mode internally.
11965 (store_bit_field_1): Rename the new integer mode from imode
11966 to op0_mode and use it instead of GET_MODE (op0). Update calls
11967 to store_split_bit_field, store_bit_field_using_insv and
11968 store_fixed_bit_field.
11969 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11970 Use scalar_int_mode internally. Use a bit count rather than a mode
11971 when calculating the largest bit size for get_best_mode.
11972 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11973 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11974 Remove assertion that OP0 has a scalar integer mode.
11975 (store_split_bit_field): Add op0_mode and value_mode arguments.
11976 Update calls to extract_fixed_bit_field.
11977 (extract_bit_field_using_extv): Add an op0_mode argument.
11978 Use scalar_int_mode internally.
11979 (extract_bit_field_1): Rename the new integer mode from imode to
11980 op0_mode and use it instead of GET_MODE (op0). Update calls to
11981 extract_split_bit_field, extract_bit_field_using_extv and
11982 extract_fixed_bit_field.
11983 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11984 to extract_split_bit_field and extract_fixed_bit_field_1.
11985 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11986 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11987 on the target mode.
11988 (extract_split_bit_field): Add an op0_mode argument. Update call
11989 to extract_fixed_bit_field.
11991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11992 Alan Hayward <alan.hayward@arm.com>
11993 David Sherwood <david.sherwood@arm.com>
11995 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11996 * explow.c (hard_function_value): Likewise.
11997 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11998 convert_to_mode call outside the loop.
11999 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12000 for the mode iterator. Require the mode specified by max_pieces
12002 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12004 (copy_blkmode_to_reg): Likewise.
12005 (set_storage_via_setmem): Likewise.
12006 * optabs.c (prepare_cmp_insn): Likewise.
12007 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12008 * stor-layout.c (finish_bitfield_representative): Likewise.
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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12015 * expr.c (convert_move): Use them.
12016 (convert_modes): Likewise.
12017 (store_expr_with_bounds): Likewise.
12019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12020 Alan Hayward <alan.hayward@arm.com>
12021 David Sherwood <david.sherwood@arm.com>
12023 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12024 parameter for the mode of "x". Remove the "known_x", "known_mode"
12025 and "known_ret" arguments. Change the type of the mode argument
12026 to scalar_int_mode.
12027 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12028 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12029 (reg_num_sign_bit_copies_for_combine): Likewise.
12030 * rtlanal.c (nonzero_bits1): Likewise.
12031 (num_sign_bit_copies1): Likewise.
12032 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12033 (reg_num_sign_bit_copies_general): Likewise.
12034 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12035 (reg_nonzero_bits_general): Likewise.
12037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12038 Alan Hayward <alan.hayward@arm.com>
12039 David Sherwood <david.sherwood@arm.com>
12041 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12042 than in subroutines. Return 1 for non-integer modes.
12043 (cached_num_sign_bit_copies): Change the type of the mode parameter
12044 to scalar_int_mode.
12045 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12046 classes. Handle CONST_INT_P first and then check whether X also
12047 has a scalar integer mode. Check the same thing for inner registers
12048 of a SUBREG and for values that are being extended or truncated.
12050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12051 Alan Hayward <alan.hayward@arm.com>
12052 David Sherwood <david.sherwood@arm.com>
12054 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12055 in subroutines. Return the mode mask for non-integer modes.
12056 (cached_nonzero_bits): Change the type of the mode parameter
12057 to scalar_int_mode.
12058 (nonzero_bits1): Likewise. Remove early exit for other mode
12059 classes. Handle CONST_INT_P first and then check whether X
12060 also has a scalar integer mode.
12062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12063 Alan Hayward <alan.hayward@arm.com>
12064 David Sherwood <david.sherwood@arm.com>
12066 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12067 Return a scalar_int_mode and assert that the size is greater than
12069 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12070 (op_by_pieces_d::op_by_pieces_d): Likewise.
12071 (op_by_pieces_d::run): Likewise.
12072 (can_store_by_pieces): Likewise.
12074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12075 Alan Hayward <alan.hayward@arm.com>
12076 David Sherwood <david.sherwood@arm.com>
12078 * combine.c (extract_left_shift): Add a mode argument and update
12080 (make_compound_operation_int): Change the type of the mode parameter
12081 to scalar_int_mode and update the call to extract_left_shift.
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 * combine.c (simplify_and_const_int): Change the type of the mode
12088 parameter to scalar_int_mode.
12089 (simplify_and_const_int_1): Likewise. Update recursive call.
12091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12092 Alan Hayward <alan.hayward@arm.com>
12093 David Sherwood <david.sherwood@arm.com>
12095 * combine.c (simplify_compare_const): Check that the mode is a
12096 scalar_int_mode (rather than VOIDmode) before testing its
12098 (simplify_comparison): Move COMPARISON_P handling out of the
12099 loop and restrict the latter part of the loop to scalar_int_modes.
12100 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12101 and when considering SUBREG_REGs. Use is_int_mode instead of
12102 checking GET_MODE_CLASS against MODE_INT.
12104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12105 Alan Hayward <alan.hayward@arm.com>
12106 David Sherwood <david.sherwood@arm.com>
12108 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12109 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12110 shift_unit_mode and for modes involved in scalar shifts.
12112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12113 Alan Hayward <alan.hayward@arm.com>
12114 David Sherwood <david.sherwood@arm.com>
12116 * combine.c (force_int_to_mode): New function, split out from...
12117 (force_to_mode): ...here. Keep xmode up-to-date and use it
12118 instead of GET_MODE (x).
12120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12121 Alan Hayward <alan.hayward@arm.com>
12122 David Sherwood <david.sherwood@arm.com>
12124 * optabs-query.h (extraction_insn::struct_mode): Change type to
12125 opt_scalar_int_mode and update comment.
12126 (extraction_insn::field_mode): Change type to scalar_int_mode.
12127 (extraction_insn::pos_mode): Likewise.
12128 * combine.c (make_extraction): Update accordingly.
12129 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12130 (get_optab_extraction_insn): Likewise.
12131 * recog.c (simplify_while_replacing): Likewise.
12132 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12133 parameter to opt_scalar_int_mode.
12135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12136 Alan Hayward <alan.hayward@arm.com>
12137 David Sherwood <david.sherwood@arm.com>
12139 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12140 to a scalar_int_mode instead of a machine_mode.
12141 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12142 (get_best_mode): Return a boolean and use a pointer argument to store
12143 the selected mode. Replace the limit mode parameter with a bit limit.
12144 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12145 for the values returned by bit_field_mode_iterator::next_mode.
12146 (store_bit_field): Update call to get_best_mode.
12147 (store_fixed_bit_field): Likewise.
12148 (extract_fixed_bit_field): Likewise.
12149 * expr.c (optimize_bitfield_assignment_op): Likewise.
12150 * fold-const.c (optimize_bit_field_compare): Likewise.
12151 (fold_truth_andor_1): Likewise.
12152 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12153 Update for new type of m_mode.
12154 (get_best_mode): As above.
12156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12157 Alan Hayward <alan.hayward@arm.com>
12158 David Sherwood <david.sherwood@arm.com>
12160 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12161 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12162 (store_bit_field): Check is_a <scalar_int_mode> before calling
12163 strict_volatile_bitfield_p.
12164 (extract_bit_field): 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 * target.def (cstore_mode): Return a scalar_int_mode.
12171 * doc/tm.texi: Regenerate.
12172 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12173 * targhooks.h (default_cstore_mode): Likewise.
12174 * targhooks.c (default_cstore_mode): Likewise, using a forced
12176 * expmed.c (emit_cstore): Expect the target of the cstore to be
12179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12180 Alan Hayward <alan.hayward@arm.com>
12181 David Sherwood <david.sherwood@arm.com>
12183 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12185 (niter_desc): Likewise mode.
12186 (iv_analyze): Add a mode parameter.
12188 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12189 and change its type to scalar_int_mode.
12190 * loop-iv.c: Update commentary at head of file.
12191 (iv_constant): Pass the mode paraeter before the rtx it describes
12192 and change its type to scalar_int_mode. Remove VOIDmode handling.
12193 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12194 (iv_extend): Likewise.
12195 (shorten_into_mode): Likewise.
12196 (iv_add): Use scalar_int_mode.
12197 (iv_mult): Likewise.
12198 (iv_shift): Likewise.
12199 (canonicalize_iv_subregs): Likewise.
12200 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12201 it describes and change its mode to scalar_int_mode. Also change
12202 the type of the returned inner_mode to scalar_int_mode.
12203 (get_biv_step): Likewise, turning outer_mode from a pointer
12204 into a direct parameter. Update call to get_biv_step_1.
12205 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12206 iv_constant and get_biv_step.
12207 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12208 and change its type to scalar_int_mode. Don't initialise iv->mode
12209 to VOIDmode and remove later checks for its still being VOIDmode.
12210 Update calls to iv_analyze_op and iv_analyze_expr. Check
12211 is_a <scalar_int_mode> when changing the mode under consideration.
12212 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12213 Update call to iv_analyze_expr.
12214 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12215 inner register is not also a scalar_int_mode. Update call to
12217 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12218 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12219 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12220 separate mode class checks. Update calls to iv_analyze. Remove
12221 fix-up of VOIDmodes after iv_analyze_biv.
12222 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12223 don't have a scalar_int_mode. Update call to biv_p.
12225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12226 Alan Hayward <alan.hayward@arm.com>
12227 David Sherwood <david.sherwood@arm.com>
12229 * cfgexpand.c (convert_debug_memory_address): Use
12230 as_a <scalar_int_mode>.
12231 * combine.c (expand_compound_operation): Likewise.
12232 (make_extraction): Likewise.
12233 (change_zero_ext): Likewise.
12234 (simplify_comparison): Likewise.
12235 * cse.c (cse_insn): Likewise.
12236 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12237 (mem_loc_descriptor): Likewise.
12238 (loc_descriptor): Likewise.
12239 * expmed.c (init_expmed_one_mode): Likewise.
12240 (synth_mult): Likewise.
12241 (emit_store_flag_1): Likewise.
12242 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12243 of a comparison with size.
12244 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12245 (reduce_to_bit_field_precision): Likewise.
12246 * function.c (expand_function_end): Likewise.
12247 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12248 * loop-doloop.c (doloop_modify): Likewise.
12249 * optabs.c (expand_binop): Likewise.
12250 (expand_unop): Likewise.
12251 (expand_copysign_absneg): Likewise.
12252 (prepare_cmp_insn): Likewise.
12253 (maybe_legitimize_operand): Likewise.
12254 * recog.c (const_scalar_int_operand): Likewise.
12255 * rtlanal.c (get_address_mode): Likewise.
12256 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12257 (simplify_cond_clz_ctz): Likewise.
12258 * tree-nested.c (get_nl_goto_field): Likewise.
12259 * tree.c (build_vector_type_for_mode): Likewise.
12260 * var-tracking.c (use_narrower_mode): Likewise.
12262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12263 Alan Hayward <alan.hayward@arm.com>
12264 David Sherwood <david.sherwood@arm.com>
12266 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12267 * builtins.c (expand_builtin_signbit): Use it.
12268 * cfgexpand.c (expand_debug_expr): Likewise.
12269 * dojump.c (do_jump): Likewise.
12270 (do_compare_and_jump): Likewise.
12271 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12272 * expmed.c (make_tree): Likewise.
12273 * expr.c (expand_expr_real_2): Likewise.
12274 (expand_expr_real_1): Likewise.
12275 (try_casesi): Likewise.
12276 * fold-const-call.c (fold_const_call_ss): Likewise.
12277 * fold-const.c (unextend): Likewise.
12278 (extract_muldiv_1): Likewise.
12279 (fold_single_bit_test): Likewise.
12280 (native_encode_int): Likewise.
12281 (native_encode_string): Likewise.
12282 (native_interpret_int): Likewise.
12283 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12284 * internal-fn.c (expand_addsub_overflow): Likewise.
12285 (expand_neg_overflow): Likewise.
12286 (expand_mul_overflow): Likewise.
12287 (expand_arith_overflow): Likewise.
12288 * match.pd: Likewise.
12289 * stor-layout.c (layout_type): Likewise.
12290 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12291 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12292 * tree-ssanames.c (get_range_info): Likewise.
12293 * tree-switch-conversion.c (array_value_type) Likewise.
12294 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12295 (vect_recog_divmod_pattern): Likewise.
12296 (vect_recog_mixed_size_cond_pattern): Likewise.
12297 * tree-vrp.c (extract_range_basic): Likewise.
12298 (simplify_float_conversion_using_ranges): Likewise.
12299 * tree.c (int_fits_type_p): Likewise.
12300 * ubsan.c (instrument_bool_enum_load): Likewise.
12301 * varasm.c (mergeable_string_section): Likewise.
12302 (narrowing_initializer_constant_valid_p): Likewise.
12303 (output_constant): Likewise.
12305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12306 Alan Hayward <alan.hayward@arm.com>
12307 David Sherwood <david.sherwood@arm.com>
12309 * machmode.h (NARROWEST_INT_MODE): New macro.
12310 * expr.c (alignment_for_piecewise_move): Use it instead of
12311 GET_CLASS_NARROWEST_MODE (MODE_INT).
12312 (push_block): Likewise.
12313 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12315 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * postreload.c (move2add_valid_value_p): Change the type of the
12322 mode parameter to scalar_int_mode.
12323 (move2add_use_add2_insn): Add a mode parameter and use it instead
12325 (move2add_use_add3_insn): Likewise.
12326 (reload_cse_move2add): Update accordingly.
12328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12329 Alan Hayward <alan.hayward@arm.com>
12330 David Sherwood <david.sherwood@arm.com>
12332 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12334 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12335 * optabs.c (expand_unop): Likewise.
12337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12338 Alan Hayward <alan.hayward@arm.com>
12339 David Sherwood <david.sherwood@arm.com>
12341 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12342 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12343 (popcount_loc_descriptor): Likewise.
12344 (bswap_loc_descriptor): Likewise.
12345 (rotate_loc_descriptor): Likewise.
12346 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12347 calling the functions above.
12349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12350 Alan Hayward <alan.hayward@arm.com>
12351 David Sherwood <david.sherwood@arm.com>
12353 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12355 (try_combine): Likewise.
12356 (simplify_if_then_else): Likewise.
12357 * cse.c (cse_insn): Likewise.
12358 * dwarf2out.c (mem_loc_descriptor): Likewise.
12359 * emit-rtl.c (gen_lowpart_common): Likewise.
12360 * simplify-rtx.c (simplify_truncation): Likewise.
12361 (simplify_binary_operation_1): Likewise.
12362 (simplify_const_relational_operation): Likewise.
12363 (simplify_ternary_operation): Likewise.
12364 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12367 Alan Hayward <alan.hayward@arm.com>
12368 David Sherwood <david.sherwood@arm.com>
12370 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12371 * reload.c (push_reload): Likewise.
12372 (find_reloads): Likewise.
12374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12375 Alan Hayward <alan.hayward@arm.com>
12376 David Sherwood <david.sherwood@arm.com>
12378 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12379 (make_compound_operation_int): Likewise.
12380 (change_zero_ext): Likewise.
12381 * expr.c (convert_move): Likewise.
12382 (convert_modes): Likewise.
12383 * fwprop.c (forward_propagate_subreg): Likewise.
12384 * loop-iv.c (get_biv_step_1): Likewise.
12385 * optabs.c (widen_operand): Likewise.
12386 * postreload.c (move2add_valid_value_p): Likewise.
12387 * recog.c (simplify_while_replacing): Likewise.
12388 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12389 (simplify_binary_operation_1): Likewise. Remove redundant
12390 mode equality check.
12392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12393 Alan Hayward <alan.hayward@arm.com>
12394 David Sherwood <david.sherwood@arm.com>
12396 * combine.c (combine_simplify_rtx): Add checks for
12397 is_a <scalar_int_mode>.
12398 (simplify_if_then_else): Likewise.
12399 (make_field_assignment): Likewise.
12400 (simplify_comparison): Likewise.
12401 * ifcvt.c (noce_try_bitop): Likewise.
12402 * loop-invariant.c (canonicalize_address_mult): Likewise.
12403 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12406 Alan Hayward <alan.hayward@arm.com>
12407 David Sherwood <david.sherwood@arm.com>
12409 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12410 is_a <scalar_int_mode> instead of != BLKmode.
12412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12413 Alan Hayward <alan.hayward@arm.com>
12414 David Sherwood <david.sherwood@arm.com>
12416 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12417 instead of != VOIDmode.
12418 * combine.c (if_then_else_cond): Likewise.
12419 (change_zero_ext): Likewise.
12420 * dwarf2out.c (mem_loc_descriptor): Likewise.
12421 (loc_descriptor): Likewise.
12422 * rtlanal.c (canonicalize_condition): Likewise.
12423 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12426 Alan Hayward <alan.hayward@arm.com>
12427 David Sherwood <david.sherwood@arm.com>
12429 * simplify-rtx.c (simplify_binary_operation_1): Use
12430 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12433 Alan Hayward <alan.hayward@arm.com>
12434 David Sherwood <david.sherwood@arm.com>
12436 * wide-int.h (int_traits<unsigned char>) New class.
12437 (int_traits<unsigned short>) Likewise.
12438 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12439 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12441 * combine.c (set_nonzero_bits_and_sign_copies): Use
12442 is_a <scalar_int_mode>.
12443 (find_split_point): Likewise.
12444 (combine_simplify_rtx): Likewise.
12445 (simplify_logical): Likewise.
12446 (expand_compound_operation): Likewise.
12447 (expand_field_assignment): Likewise.
12448 (make_compound_operation): Likewise.
12449 (extended_count): Likewise.
12450 (change_zero_ext): Likewise.
12451 (simplify_comparison): Likewise.
12452 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12453 (ucompare_loc_descriptor): Likewise.
12454 (minmax_loc_descriptor): Likewise.
12455 (mem_loc_descriptor): Likewise.
12456 (loc_descriptor): Likewise.
12457 * expmed.c (init_expmed_one_mode): Likewise.
12458 * lra-constraints.c (lra_constraint_offset): Likewise.
12459 * optabs.c (prepare_libcall_arg): Likewise.
12460 * postreload.c (move2add_note_store): Likewise.
12461 * reload.c (operands_match_p): Likewise.
12462 * rtl.h (load_extend_op): Likewise.
12463 * rtlhooks.c (gen_lowpart_general): Likewise.
12464 * simplify-rtx.c (simplify_truncation): Likewise.
12465 (simplify_unary_operation_1): Likewise.
12466 (simplify_binary_operation_1): Likewise.
12467 (simplify_const_binary_operation): Likewise.
12468 (simplify_const_relational_operation): Likewise.
12469 (simplify_subreg): Likewise.
12470 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12471 * var-tracking.c (adjust_mems): Likewise.
12472 (prepare_call_arguments): Likewise.
12474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12475 Alan Hayward <alan.hayward@arm.com>
12476 David Sherwood <david.sherwood@arm.com>
12478 * machmode.h (is_int_mode): New fuction.
12479 * combine.c (find_split_point): Use it.
12480 (combine_simplify_rtx): Likewise.
12481 (simplify_if_then_else): Likewise.
12482 (simplify_set): Likewise.
12483 (simplify_shift_const_1): Likewise.
12484 (simplify_comparison): Likewise.
12485 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12486 * cse.c (notreg_cost): Likewise.
12487 (cse_insn): Likewise.
12488 * cselib.c (cselib_lookup_1): Likewise.
12489 * dojump.c (do_jump_1): Likewise.
12490 (do_compare_rtx_and_jump): Likewise.
12491 * dse.c (get_call_args): Likewise.
12492 * dwarf2out.c (rtl_for_decl_init): Likewise.
12493 (native_encode_initializer): Likewise.
12494 * expmed.c (emit_store_flag_1): Likewise.
12495 (emit_store_flag): Likewise.
12496 * expr.c (convert_modes): Likewise.
12497 (store_field): Likewise.
12498 (expand_expr_real_1): Likewise.
12499 * fold-const.c (fold_read_from_constant_string): Likewise.
12500 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12501 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12502 * optabs.c (expand_binop): Likewise.
12503 (expand_unop): Likewise.
12504 (expand_abs_nojump): Likewise.
12505 (expand_one_cmpl_abs_nojump): Likewise.
12506 * simplify-rtx.c (mode_signbit_p): Likewise.
12507 (val_signbit_p): Likewise.
12508 (val_signbit_known_set_p): Likewise.
12509 (val_signbit_known_clear_p): Likewise.
12510 (simplify_relational_operation_1): Likewise.
12511 * tree.c (vector_type_mode): Likewise.
12513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12514 Alan Hayward <alan.hayward@arm.com>
12515 David Sherwood <david.sherwood@arm.com>
12517 * machmode.h (smallest_mode_for_size): Fix formatting.
12518 (smallest_int_mode_for_size): New function.
12519 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12520 instead of smallest_mode_for_size.
12521 * combine.c (make_extraction): Likewise.
12522 * config/arc/arc.c (arc_expand_movmem): Likewise.
12523 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12524 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12525 * config/s390/s390.c (s390_expand_insv): Likewise.
12526 * config/sparc/sparc.c (assign_int_registers): Likewise.
12527 * config/spu/spu.c (spu_function_value): Likewise.
12528 (spu_function_arg): Likewise.
12529 * coverage.c (get_gcov_type): Likewise.
12530 (get_gcov_unsigned_t): Likewise.
12531 * dse.c (find_shift_sequence): Likewise.
12532 * expmed.c (store_bit_field_1): Likewise.
12533 * expr.c (convert_move): Likewise.
12534 (store_field): Likewise.
12535 * internal-fn.c (expand_arith_overflow): Likewise.
12536 * optabs-query.c (get_best_extraction_insn): Likewise.
12537 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12538 * stor-layout.c (layout_type): Likewise.
12539 (initialize_sizetypes): Likewise.
12540 * targhooks.c (default_get_mask_mode): Likewise.
12541 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12544 Alan Hayward <alan.hayward@arm.com>
12545 David Sherwood <david.sherwood@arm.com>
12547 * machmode.h (opt_mode::else_blk): New function.
12548 (int_mode_for_mode): Declare.
12549 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12550 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12552 * cfgexpand.c (expand_debug_expr): Likewise.
12553 * combine.c (gen_lowpart_or_truncate): Likewise.
12554 (gen_lowpart_for_combine): Likewise.
12555 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12556 * config/avr/avr.c (avr_to_int_mode): Likewise.
12557 (avr_out_plus_1): Likewise.
12558 (avr_out_plus): Likewise.
12559 (avr_out_round): Likewise.
12560 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12561 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12562 (s390_expand_vcond): Likewise.
12563 * config/spu/spu.c (spu_split_immediate): Likewise.
12564 (spu_expand_mov): Likewise.
12565 * dse.c (get_stored_val): Likewise.
12566 * expmed.c (store_bit_field_1): Likewise.
12567 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12569 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12570 (extract_low_bits): Likewise.
12571 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12572 handling rather than repeating the check.
12573 (emit_group_store): Likewise.
12574 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12575 * optabs.c (expand_absneg_bit): Likewise.
12576 (expand_copysign_absneg): Likewise.
12577 (expand_copysign_bit): Likewise.
12578 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12579 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12580 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12581 * var-tracking.c (prepare_call_arguments): Likewise.
12582 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12583 int_mode_for_mode instead of mode_for_size.
12584 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12587 Alan Hayward <alan.hayward@arm.com>
12588 David Sherwood <david.sherwood@arm.com>
12590 * machmode.h (int_mode_for_size): New function.
12591 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12592 instead of mode_for_size.
12593 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12595 * combine.c (expand_field_assignment): Use int_mode_for_size
12596 instead of mode_for_size.
12597 (make_extraction): Likewise.
12598 (simplify_shift_const_1): Likewise.
12599 (simplify_comparison): Likewise.
12600 * dojump.c (do_jump): Likewise.
12601 * dwarf2out.c (mem_loc_descriptor): Likewise.
12602 * emit-rtl.c (init_derived_machine_modes): Likewise.
12603 * expmed.c (flip_storage_order): Likewise.
12604 (convert_extracted_bit_field): Likewise.
12605 * expr.c (copy_blkmode_from_reg): Likewise.
12606 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12607 * internal-fn.c (expand_mul_overflow): Likewise.
12608 * lower-subreg.c (simple_move): Likewise.
12609 * optabs-libfuncs.c (init_optabs): Likewise.
12610 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12611 * tree.c (vector_type_mode): Likewise.
12612 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12613 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12614 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12615 * tree-vect-stmts.c (vectorizable_load): Likewise.
12616 (vectorizable_store): Likewise.
12618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12619 Alan Hayward <alan.hayward@arm.com>
12620 David Sherwood <david.sherwood@arm.com>
12622 * coretypes.h (pod_mode): New type.
12623 (scalar_int_mode_pod): New typedef.
12624 * machmode.h (pod_mode): New class.
12625 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12626 * genmodes.c (emit_mode_int_n): Update accordingly.
12627 * lower-subreg.h (target_lower_subreg): Change type to
12628 scalar_int_mode_pod.
12629 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12630 scalar_int_mode_pod.
12632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12633 Alan Hayward <alan.hayward@arm.com>
12634 David Sherwood <david.sherwood@arm.com>
12636 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12637 machine_mode to scalar_int_mode.
12638 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12639 (rs6000_option_override_internal): Remove cast to int.
12640 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12641 machine_mode to scalar_int_mode.
12642 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12643 (rs6000_option_override_internal): Remove cast to int.
12644 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12645 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12647 * config/s390/s390.c (s390_expand_builtin): Likewise.
12648 * coretypes.h (scalar_int_mode): New type.
12649 (opt_scalar_int_mode): New typedef.
12650 * machmode.h (scalar_int_mode): New class.
12651 (scalar_int_mode::includes_p): New function.
12652 (byte_mode): Change type to scalar_int_mode.
12653 (word_mode): Likewise.
12654 (ptr_mode): Likewise.
12655 * emit-rtl.c (byte_mode): Likewise.
12656 (word_mode): Likewise.
12657 (ptr_mode): Likewise.
12658 (init_derived_machine_modes): Update accordingly.
12659 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12660 and MODE_PARTIAL_INT.
12661 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12662 opt_scalar_int_mode.
12664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12665 Alan Hayward <alan.hayward@arm.com>
12666 David Sherwood <david.sherwood@arm.com>
12668 * target.def (libgcc_floating_mode_supported_p): Take a
12670 * doc/tm.texi: Regenerate.
12671 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12673 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12674 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12678 Alan Hayward <alan.hayward@arm.com>
12679 David Sherwood <david.sherwood@arm.com>
12681 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12682 * doc/tm.texi: Regenerate.
12683 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12684 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12685 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12686 * targhooks.h (default_floatn_mode): Likewise.
12687 * targhooks.c (default_floatn_mode): Likewise.
12688 * tree.c (build_common_tree_nodes): Update accordingly.
12690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12691 Alan Hayward <alan.hayward@arm.com>
12692 David Sherwood <david.sherwood@arm.com>
12694 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12695 (mode_iterator::iterate_p): Likewise.
12696 (mode_iterator::get_wider): Likewise.
12697 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12700 Alan Hayward <alan.hayward@arm.com>
12701 David Sherwood <david.sherwood@arm.com>
12703 * coretypes.h (opt_scalar_float_mode): New typedef.
12704 * machmode.h (float_mode_for_size): New function.
12705 * emit-rtl.c (double_mode): Delete.
12706 (init_emit_once): Use float_mode_for_size.
12707 * stor-layout.c (layout_type): Likewise.
12708 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
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 * output.h (assemble_real): Take a scalar_float_mode.
12715 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12716 * config/arm/arm.md (consttable_4): Likewise.
12717 (consttable_8): Likewise.
12718 (consttable_16): Likewise.
12719 * config/mips/mips.md (consttable_float): Likewise.
12720 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12721 * varasm.c (assemble_real): Take a scalar_float_mode.
12722 (output_constant_pool_2): Update accordingly.
12723 (output_constant): Likewise.
12725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12726 Alan Hayward <alan.hayward@arm.com>
12727 David Sherwood <david.sherwood@arm.com>
12729 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12730 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12731 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12732 (native_encode_real): Likewise.
12733 (native_interpret_real): Likewise.
12734 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12735 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12738 Alan Hayward <alan.hayward@arm.com>
12739 David Sherwood <david.sherwood@arm.com>
12741 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12742 <scalar_float_mode>. Simplify.
12743 (gen_extend_conv_libfunc): Likewise.
12745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12746 Alan Hayward <alan.hayward@arm.com>
12747 David Sherwood <david.sherwood@arm.com>
12749 * coretypes.h (scalar_float_mode): New type.
12750 * machmode.h (mode_traits::from_int): Use machine_mode if
12751 USE_ENUM_MODES is defined.
12752 (is_a): New function.
12754 (dyn_cast): Likewise.
12755 (scalar_float_mode): New class.
12756 (scalar_float_mode::includes_p): New function.
12757 (is_float_mode): Likewise.
12758 * gdbhooks.py (MachineModePrinter): New class.
12759 (build_pretty_printer): Use it for scalar_float_mode.
12760 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12761 (format_helper::format_helper): Turn into a template.
12762 * genmodes.c (get_mode_class): New function.
12763 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12764 or machine_mode if none.
12765 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12766 as_a <scalar_float_mode>.
12767 * dwarf2out.c (mem_loc_descriptor): Likewise.
12768 (insert_float): Likewise.
12769 (add_const_value_attribute): Likewise.
12770 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12771 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12772 (expand_unop): Update accordingly.
12773 (expand_abs_nojump): Likewise.
12774 (expand_copysign_absneg): Take a scalar_float_mode.
12775 (expand_copysign_bit): Likewise.
12776 (expand_copysign): Update accordingly.
12778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12779 Alan Hayward <alan.hayward@arm.com>
12780 David Sherwood <david.sherwood@arm.com>
12782 * coretypes.h (opt_mode): New class.
12783 * machmode.h (opt_mode): Likewise.
12784 (opt_mode::else_void): New function.
12785 (opt_mode::require): Likewise.
12786 (opt_mode::exists): Likewise.
12787 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12788 (GET_MODE_2XWIDER_MODE): Likewise.
12789 (mode_iterator::get_wider): Update accordingly.
12790 (mode_iterator::get_2xwider): Likewise.
12791 (mode_iterator::get_known_wider): Likewise, turning into a template.
12792 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12793 forcing a wider mode to exist.
12794 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12795 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12796 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12797 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12798 * lower-subreg.c (init_lower_subreg): Likewise.
12799 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12800 on the final iteration.
12801 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12802 a wider mode exists before asking for a move pattern.
12803 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12804 forcing a wider mode to exist.
12805 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12806 returning false if no such mode exists.
12807 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12808 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12809 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12810 Avoid checking for a MODE_INT if we already know the mode is not a
12812 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12813 forcing a wider mode to exist.
12814 (expmed_mult_highpart_optab): Likewise.
12815 (expmed_mult_highpart): Likewise.
12816 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12818 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12819 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12820 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12821 * internal-fn.c (expand_mul_overflow): Update use of
12822 GET_MODE_2XWIDER_MODE.
12823 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12824 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12825 GET_MODE_WIDER_MODE.
12826 (convert_plusminus_to_widen): Likewise.
12827 * tree-switch-conversion.c (array_value_type): Likewise.
12828 * var-tracking.c (emit_note_insn_var_location): Likewise.
12829 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12830 Return false inside rather than outside the loop if no wider mode
12832 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12833 and GET_MODE_2XWIDER_MODE
12834 (can_compare_p): Use else_void.
12835 * gdbhooks.py (OptMachineModePrinter): New class.
12836 (build_pretty_printer): Use it for opt_mode.
12838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12839 Alan Hayward <alan.hayward@arm.com>
12840 David Sherwood <david.sherwood@arm.com>
12842 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12843 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12846 Alan Hayward <alan.hayward@arm.com>
12847 David Sherwood <david.sherwood@arm.com>
12849 * machmode.h (mode_traits): New structure.
12850 (get_narrowest_mode): New function.
12851 (mode_iterator::start): Likewise.
12852 (mode_iterator::iterate_p): Likewise.
12853 (mode_iterator::get_wider): Likewise.
12854 (mode_iterator::get_known_wider): Likewise.
12855 (mode_iterator::get_2xwider): Likewise.
12856 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12857 (FOR_EACH_MODE): Likewise.
12858 (FOR_EACH_MODE_FROM): Likewise.
12859 (FOR_EACH_MODE_UNTIL): Likewise.
12860 (FOR_EACH_WIDER_MODE): Likewise.
12861 (FOR_EACH_2XWIDER_MODE): Likewise.
12862 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12863 * combine.c (simplify_comparison): Likewise
12864 * config/i386/i386.c (type_natural_mode): Likewise.
12865 * cse.c (cse_insn): Likewise.
12866 * dse.c (find_shift_sequence): Likewise.
12867 * emit-rtl.c (init_derived_machine_modes): Likewise.
12868 (init_emit_once): Likewise.
12869 * explow.c (hard_function_value): Likewise.
12870 * expmed.c (extract_fixed_bit_field_1): Likewise.
12871 (extract_bit_field_1): Likewise.
12872 (expand_divmod): Likewise.
12873 (emit_store_flag_1): Likewise.
12874 * expr.c (init_expr_target): Likewise.
12875 (convert_move): Likewise.
12876 (alignment_for_piecewise_move): Likewise.
12877 (widest_int_mode_for_size): Likewise.
12878 (emit_block_move_via_movmem): Likewise.
12879 (copy_blkmode_to_reg): Likewise.
12880 (set_storage_via_setmem): Likewise.
12881 (compress_float_constant): Likewise.
12882 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12883 * optabs-query.c (get_best_extraction_insn): Likewise.
12884 * optabs.c (expand_binop): Likewise.
12885 (expand_twoval_unop): Likewise.
12886 (expand_twoval_binop): Likewise.
12887 (widen_leading): Likewise.
12888 (widen_bswap): Likewise.
12889 (expand_parity): Likewise.
12890 (expand_unop): Likewise.
12891 (prepare_cmp_insn): Likewise.
12892 (prepare_float_lib_cmp): Likewise.
12893 (expand_float): Likewise.
12894 (expand_fix): Likewise.
12895 (expand_sfix_optab): Likewise.
12896 * postreload.c (move2add_use_add2_insn): Likewise.
12897 * reg-stack.c (reg_to_stack): Likewise.
12898 * reginfo.c (choose_hard_reg_mode): Likewise.
12899 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12900 * stor-layout.c (mode_for_size): Likewise.
12901 (smallest_mode_for_size): Likewise.
12902 (mode_for_vector): Likewise.
12903 (finish_bitfield_representative): Likewise.
12904 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12905 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12906 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12907 * var-tracking.c (prepare_call_arguments): Likewise.
12909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12910 Alan Hayward <alan.hayward@arm.com>
12911 David Sherwood <david.sherwood@arm.com>
12913 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12914 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12915 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12916 * machmode.h (mode_size): Move earlier in file.
12917 (mode_precision): Likewise.
12918 (mode_inner): Likewise.
12919 (mode_nunits): Likewise.
12920 (mode_unit_size): Likewise.
12921 (unit_unit_precision): Likewise.
12922 (mode_wider): Likewise.
12923 (mode_2xwider): Likewise.
12924 (machine_mode): New class.
12925 (mode_to_bytes): New function.
12926 (mode_to_bits): Likewise.
12927 (mode_to_precision): Likewise.
12928 (mode_to_inner): Likewise.
12929 (mode_to_unit_size): Likewise.
12930 (mode_to_unit_precision): Likewise.
12931 (mode_to_nunits): Likewise.
12932 (GET_MODE_SIZE): Use mode_to_bytes.
12933 (GET_MODE_BITSIZE): Use mode_to_bits.
12934 (GET_MODE_PRECISION): Use mode_to_precision.
12935 (GET_MODE_INNER): Use mode_to_inner.
12936 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12937 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12938 (GET_MODE_NUNITS): Use mode_to_nunits.
12939 * system.h (ALWAYS_INLINE): New macro.
12940 * config/powerpcspe/powerpcspe-c.c
12941 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12942 int for arg1_mode and arg2_mode.
12944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12945 Alan Hayward <alan.hayward@arm.com>
12946 David Sherwood <david.sherwood@arm.com>
12948 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12949 Prefix mode names with E_ in case statements.
12950 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12951 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12952 (aarch64_split_simd_move): Likewise.
12953 (aarch64_gen_storewb_pair): Likewise.
12954 (aarch64_gen_loadwb_pair): Likewise.
12955 (aarch64_gen_store_pair): Likewise.
12956 (aarch64_gen_load_pair): Likewise.
12957 (aarch64_get_condition_code_1): Likewise.
12958 (aarch64_constant_pool_reload_icode): Likewise.
12959 (get_rsqrte_type): Likewise.
12960 (get_rsqrts_type): Likewise.
12961 (get_recpe_type): Likewise.
12962 (get_recps_type): Likewise.
12963 (aarch64_gimplify_va_arg_expr): Likewise.
12964 (aarch64_simd_container_mode): Likewise.
12965 (aarch64_emit_load_exclusive): Likewise.
12966 (aarch64_emit_store_exclusive): Likewise.
12967 (aarch64_expand_compare_and_swap): Likewise.
12968 (aarch64_gen_atomic_cas): Likewise.
12969 (aarch64_emit_bic): Likewise.
12970 (aarch64_emit_atomic_swap): Likewise.
12971 (aarch64_emit_atomic_load_op): Likewise.
12972 (aarch64_evpc_trn): Likewise.
12973 (aarch64_evpc_uzp): Likewise.
12974 (aarch64_evpc_zip): Likewise.
12975 (aarch64_evpc_ext): Likewise.
12976 (aarch64_evpc_rev): Likewise.
12977 (aarch64_evpc_dup): Likewise.
12978 (aarch64_gen_ccmp_first): Likewise.
12979 (aarch64_gen_ccmp_next): Likewise.
12980 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12981 (alpha_emit_xfloating_libcall): Likewise.
12982 (emit_insxl): Likewise.
12983 (alpha_arg_type): Likewise.
12984 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12985 (arc_preferred_simd_mode): Likewise.
12986 (arc_secondary_reload): Likewise.
12987 (get_arc_condition_code): Likewise.
12988 (arc_print_operand): Likewise.
12989 (arc_legitimate_constant_p): Likewise.
12990 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12991 * config/arc/arc.md (casesi_load): Likewise.
12992 (casesi_compact_jump): Likewise.
12993 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12994 (cc_use_register): Likewise.
12995 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12996 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12997 (arm_init_iwmmxt_builtins): Likewise.
12998 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12999 (neon_expand_vector_init): Likewise.
13000 (arm_attr_length_move_neon): Likewise.
13001 (maybe_get_arm_condition_code): Likewise.
13002 (arm_emit_vector_const): Likewise.
13003 (arm_preferred_simd_mode): Likewise.
13004 (arm_output_iwmmxt_tinsr): Likewise.
13005 (thumb1_output_casesi): Likewise.
13006 (thumb2_output_casesi): Likewise.
13007 (arm_emit_load_exclusive): Likewise.
13008 (arm_emit_store_exclusive): Likewise.
13009 (arm_expand_compare_and_swap): Likewise.
13010 (arm_evpc_neon_vuzp): Likewise.
13011 (arm_evpc_neon_vzip): Likewise.
13012 (arm_evpc_neon_vrev): Likewise.
13013 (arm_evpc_neon_vtrn): Likewise.
13014 (arm_evpc_neon_vext): Likewise.
13015 (arm_validize_comparison): Likewise.
13016 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13017 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13018 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13019 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13020 (c6x_preferred_simd_mode): Likewise.
13021 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13022 (epiphany_rtx_costs): Likewise.
13023 * config/epiphany/predicates.md (proper_comparison_operator):
13025 * config/frv/frv.c (condexec_memory_operand): Likewise.
13026 (frv_emit_move): Likewise.
13027 (output_move_single): Likewise.
13028 (output_condmove_single): Likewise.
13029 (frv_hard_regno_mode_ok): Likewise.
13030 (frv_matching_accg_mode): Likewise.
13031 * config/h8300/h8300.c (split_adds_subs): Likewise.
13032 (h8300_rtx_costs): Likewise.
13033 (h8300_print_operand): Likewise.
13034 (compute_mov_length): Likewise.
13035 (output_logical_op): Likewise.
13036 (compute_logical_op_length): Likewise.
13037 (compute_logical_op_cc): Likewise.
13038 (h8300_shift_needs_scratch_p): Likewise.
13039 (output_a_shift): Likewise.
13040 (compute_a_shift_length): Likewise.
13041 (compute_a_shift_cc): Likewise.
13042 (expand_a_rotate): Likewise.
13043 (output_a_rotate): Likewise.
13044 * config/i386/i386.c (classify_argument): Likewise.
13045 (function_arg_advance_32): Likewise.
13046 (function_arg_32): Likewise.
13047 (function_arg_64): Likewise.
13048 (function_value_64): Likewise.
13049 (ix86_gimplify_va_arg): Likewise.
13050 (ix86_legitimate_constant_p): Likewise.
13051 (put_condition_code): Likewise.
13052 (split_double_mode): Likewise.
13053 (ix86_avx256_split_vector_move_misalign): Likewise.
13054 (ix86_expand_vector_logical_operator): Likewise.
13055 (ix86_split_idivmod): Likewise.
13056 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13057 (ix86_build_const_vector): Likewise.
13058 (ix86_build_signbit_mask): Likewise.
13059 (ix86_match_ccmode): Likewise.
13060 (ix86_cc_modes_compatible): Likewise.
13061 (ix86_expand_branch): Likewise.
13062 (ix86_expand_sse_cmp): Likewise.
13063 (ix86_expand_sse_movcc): Likewise.
13064 (ix86_expand_int_sse_cmp): Likewise.
13065 (ix86_expand_vec_perm_vpermi2): Likewise.
13066 (ix86_expand_vec_perm): Likewise.
13067 (ix86_expand_sse_unpack): Likewise.
13068 (ix86_expand_int_addcc): Likewise.
13069 (ix86_split_to_parts): Likewise.
13070 (ix86_vectorize_builtin_gather): Likewise.
13071 (ix86_vectorize_builtin_scatter): Likewise.
13072 (avx_vpermilp_parallel): Likewise.
13073 (inline_memory_move_cost): Likewise.
13074 (ix86_tieable_integer_mode_p): Likewise.
13075 (x86_maybe_negate_const_int): Likewise.
13076 (ix86_expand_vector_init_duplicate): Likewise.
13077 (ix86_expand_vector_init_one_nonzero): Likewise.
13078 (ix86_expand_vector_init_one_var): Likewise.
13079 (ix86_expand_vector_init_concat): Likewise.
13080 (ix86_expand_vector_init_interleave): Likewise.
13081 (ix86_expand_vector_init_general): Likewise.
13082 (ix86_expand_vector_set): Likewise.
13083 (ix86_expand_vector_extract): Likewise.
13084 (emit_reduc_half): Likewise.
13085 (ix86_emit_i387_round): Likewise.
13086 (ix86_mangle_type): Likewise.
13087 (ix86_expand_round_sse4): Likewise.
13088 (expand_vec_perm_blend): Likewise.
13089 (canonicalize_vector_int_perm): Likewise.
13090 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13091 (expand_vec_perm_1): Likewise.
13092 (expand_vec_perm_interleave3): Likewise.
13093 (expand_vec_perm_even_odd_pack): Likewise.
13094 (expand_vec_perm_even_odd_1): Likewise.
13095 (expand_vec_perm_broadcast_1): Likewise.
13096 (ix86_vectorize_vec_perm_const_ok): Likewise.
13097 (ix86_expand_vecop_qihi): Likewise.
13098 (ix86_expand_mul_widen_hilo): Likewise.
13099 (ix86_expand_sse2_abs): Likewise.
13100 (ix86_expand_pextr): Likewise.
13101 (ix86_expand_pinsr): Likewise.
13102 (ix86_preferred_simd_mode): Likewise.
13103 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13104 * config/i386/sse.md (*andnot<mode>3): Likewise.
13105 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13106 (*<code><mode>3): Likewise.
13107 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13108 (ia64_expand_atomic_op): Likewise.
13109 (ia64_arg_type): Likewise.
13110 (ia64_mode_to_int): Likewise.
13111 (ia64_scalar_mode_supported_p): Likewise.
13112 (ia64_vector_mode_supported_p): Likewise.
13113 (expand_vec_perm_broadcast): Likewise.
13114 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13115 (iq2000_function_arg_advance): Likewise.
13116 (iq2000_function_arg): Likewise.
13117 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13118 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13119 (m68k_libcall_value): Likewise.
13120 (m68k_function_value): Likewise.
13121 (sched_attr_op_type): Likewise.
13122 * config/mcore/mcore.c (mcore_output_move): Likewise.
13123 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13125 (microblaze_function_arg): Likewise.
13126 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13127 (mips_print_operand): Likewise.
13128 (mips_mode_ok_for_mov_fmt_p): Likewise.
13129 (mips_vector_mode_supported_p): Likewise.
13130 (mips_preferred_simd_mode): Likewise.
13131 (mips_expand_vpc_loongson_even_odd): Likewise.
13132 (mips_expand_vec_unpack): Likewise.
13133 (mips_expand_vi_broadcast): Likewise.
13134 (mips_expand_vector_init): Likewise.
13135 (mips_expand_vec_reduc): Likewise.
13136 (mips_expand_msa_cmp): Likewise.
13137 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13138 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13139 (cc_flags_for_mode): Likewise.
13140 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13141 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13142 (nds32_output_casesi_pc_relative): Likewise.
13143 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13144 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13145 (nvptx_gen_unpack): Likewise.
13146 (nvptx_gen_pack): Likewise.
13147 (nvptx_gen_shuffle): Likewise.
13148 (nvptx_gen_wcast): Likewise.
13149 (nvptx_preferred_simd_mode): Likewise.
13150 * config/pa/pa.c (pa_secondary_reload): Likewise.
13151 * config/pa/predicates.md (base14_operand): Likewise.
13152 * config/powerpcspe/powerpcspe-c.c
13153 (altivec_resolve_overloaded_builtin): Likewise.
13154 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13156 (rs6000_preferred_simd_mode): Likewise.
13157 (output_vec_const_move): Likewise.
13158 (rs6000_expand_vector_extract): Likewise.
13159 (rs6000_split_vec_extract_var): Likewise.
13160 (reg_offset_addressing_ok_p): Likewise.
13161 (rs6000_legitimate_offset_address_p): Likewise.
13162 (rs6000_legitimize_address): Likewise.
13163 (rs6000_emit_set_const): Likewise.
13164 (rs6000_const_vec): Likewise.
13165 (rs6000_emit_move): Likewise.
13166 (spe_build_register_parallel): Likewise.
13167 (rs6000_darwin64_record_arg_recurse): Likewise.
13168 (swap_selector_for_mode): Likewise.
13169 (spe_init_builtins): Likewise.
13170 (paired_init_builtins): Likewise.
13171 (altivec_init_builtins): Likewise.
13172 (do_load_for_compare): Likewise.
13173 (rs6000_generate_compare): Likewise.
13174 (rs6000_expand_float128_convert): Likewise.
13175 (emit_load_locked): Likewise.
13176 (emit_store_conditional): Likewise.
13177 (rs6000_output_function_epilogue): Likewise.
13178 (rs6000_handle_altivec_attribute): Likewise.
13179 (rs6000_function_value): Likewise.
13180 (emit_fusion_gpr_load): Likewise.
13181 (emit_fusion_p9_load): Likewise.
13182 (emit_fusion_p9_store): Likewise.
13183 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13184 (fusion_gpr_mem_load): Likewise.
13185 (fusion_addis_mem_combo_load): Likewise.
13186 (fusion_addis_mem_combo_store): Likewise.
13187 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13188 (fusion_gpr_mem_load): Likewise.
13189 (fusion_addis_mem_combo_load): Likewise.
13190 (fusion_addis_mem_combo_store): Likewise.
13191 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13193 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13194 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13195 (rs6000_preferred_simd_mode): Likewise.
13196 (output_vec_const_move): Likewise.
13197 (rs6000_expand_vector_extract): Likewise.
13198 (rs6000_split_vec_extract_var): Likewise.
13199 (reg_offset_addressing_ok_p): Likewise.
13200 (rs6000_legitimate_offset_address_p): Likewise.
13201 (rs6000_legitimize_address): Likewise.
13202 (rs6000_emit_set_const): Likewise.
13203 (rs6000_const_vec): Likewise.
13204 (rs6000_emit_move): Likewise.
13205 (rs6000_darwin64_record_arg_recurse): Likewise.
13206 (swap_selector_for_mode): Likewise.
13207 (paired_init_builtins): Likewise.
13208 (altivec_init_builtins): Likewise.
13209 (rs6000_expand_float128_convert): Likewise.
13210 (emit_load_locked): Likewise.
13211 (emit_store_conditional): Likewise.
13212 (rs6000_output_function_epilogue): Likewise.
13213 (rs6000_handle_altivec_attribute): Likewise.
13214 (rs6000_function_value): Likewise.
13215 (emit_fusion_gpr_load): Likewise.
13216 (emit_fusion_p9_load): Likewise.
13217 (emit_fusion_p9_store): Likewise.
13218 * config/rx/rx.c (rx_gen_move_template): Likewise.
13219 (flags_from_mode): Likewise.
13220 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13221 (s390_slb_comparison): Likewise.
13222 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13223 (s390_vector_mode_supported_p): Likewise.
13224 (s390_cc_modes_compatible): Likewise.
13225 (s390_match_ccmode_set): Likewise.
13226 (s390_canonicalize_comparison): Likewise.
13227 (s390_emit_compare_and_swap): Likewise.
13228 (s390_branch_condition_mask): Likewise.
13229 (s390_rtx_costs): Likewise.
13230 (s390_secondary_reload): Likewise.
13231 (__SECONDARY_RELOAD_CASE): Likewise.
13232 (s390_expand_cs): Likewise.
13233 (s390_preferred_simd_mode): Likewise.
13234 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13235 * config/sh/sh.c (sh_print_operand): Likewise.
13236 (dump_table): Likewise.
13237 (sh_secondary_reload): Likewise.
13238 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13239 * config/sh/sh.md (casesi_worker_1): Likewise.
13240 (casesi_worker_2): Likewise.
13241 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13242 (fcc_comparison_operator): Likewise.
13243 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13244 (emit_soft_tfmode_cvt): Likewise.
13245 (sparc_preferred_simd_mode): Likewise.
13246 (output_cbranch): Likewise.
13247 (sparc_print_operand): Likewise.
13248 (sparc_expand_vec_perm_bmask): Likewise.
13249 (vector_init_bshuffle): Likewise.
13250 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13251 (spu_vector_mode_supported_p): Likewise.
13252 (spu_expand_insv): Likewise.
13253 (spu_emit_branch_or_set): Likewise.
13254 (spu_handle_vector_attribute): Likewise.
13255 (spu_builtin_splats): Likewise.
13256 (spu_builtin_extract): Likewise.
13257 (spu_builtin_promote): Likewise.
13258 (spu_expand_sign_extend): Likewise.
13259 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13260 (tilegx_simd_int): Likewise.
13261 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13262 (tilepro_simd_int): Likewise.
13263 * config/v850/v850.c (const_double_split): Likewise.
13264 (v850_print_operand): Likewise.
13265 (ep_memory_offset): Likewise.
13266 * config/vax/vax.c (vax_rtx_costs): Likewise.
13267 (vax_output_int_move): Likewise.
13268 (vax_output_int_add): Likewise.
13269 (vax_output_int_subtract): Likewise.
13270 * config/visium/predicates.md (visium_branch_operator): Likewise.
13271 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13272 (visium_print_operand_address): Likewise.
13273 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13274 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13275 (xtensa_expand_conditional_branch): Likewise.
13276 (xtensa_copy_incoming_a7): Likewise.
13277 (xtensa_output_literal): Likewise.
13278 * dfp.c (decimal_real_maxval): Likewise.
13279 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13282 Alan Hayward <alan.hayward@arm.com>
13283 David Sherwood <david.sherwood@arm.com>
13285 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13286 (mode_nunits_inline): Likewise.
13287 (mode_inner_inline): Likewise.
13288 (mode_unit_size_inline): Likewise.
13289 (mode_unit_precision_inline): Likewise.
13290 (emit_insn_modes_h): Likewise. Also emit a #define of the
13292 (emit_mode_wider): Add an E_ prefix to mode names.
13293 (emit_mode_complex): Likewise.
13294 (emit_mode_inner): Likewise.
13295 (emit_mode_adjustments): Likewise.
13296 (emit_mode_int_n): Likewise.
13297 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13298 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13299 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13300 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13301 (CRC32_BUILTIN, ENTRY): Likewise.
13302 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13303 (aarch64_pop_regs): Likewise.
13304 (aarch64_process_components): Likewise.
13305 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13306 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13307 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13308 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13309 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13310 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13312 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13313 (pcmpistr): Likewise.
13314 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13315 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13316 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13318 * config/rl78/rl78.c (mduc_regs): Likewise.
13319 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13320 (htm_expand_builtin): Likewise.
13321 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13322 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13323 * config/xtensa/xtensa.c (print_operand): Likewise.
13324 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13325 (NUM_MODE_VECTOR_INT): Likewise.
13326 * genoutput.c (null_operand): Likewise.
13327 (output_operand_data): Likewise.
13328 * genrecog.c (print_parameter_value): Likewise.
13329 * lra.c (debug_operand_data): Likewise.
13331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13333 * dwarf2out.c (scompare_loc_descriptor_wide)
13334 (scompare_loc_descriptor_narrow): New functions, split out from...
13335 (scompare_loc_descriptor): ...here.
13336 * expmed.c (emit_store_flag_int): New function, split out from...
13337 (emit_store_flag): ...here.
13339 2017-08-30 Richard Biener <rguenther@suse.de>
13341 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13342 (dwarf2out_early_finish): Move setting of AT_pubnames from
13343 early debug output to early finish.
13345 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13347 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13348 and -mdata-region to the assembler.
13350 2017-08-30 Richard Biener <rguenther@suse.de>
13352 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13354 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13355 (dwarf2out_early_global_decl): Only generate a DIE for the
13356 abstract origin if it doesn't already exist or is a declaration DIE.
13357 (resolve_addr): Do not add the linkage name twice when
13358 generating a stub DIE for the DW_TAG_GNU_call_site target.
13360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13362 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13363 Use machine_mode rather than int for arg1_mode.
13365 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13368 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13369 that the second argument of the built-in functions to unpack
13370 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13371 switch statement instead a lot of if statements.
13372 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13373 Allow 64-bit values to be in Altivec registers as well as
13374 traditional floating point registers.
13375 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13377 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13379 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13380 MAX_REGS_PER_ADDRESS == 1.
13382 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13384 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13385 * config/i386/i386.c (ix86_option_override_internal): Simplify
13386 setting of opts->x_flag_entry.
13388 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13389 Jakub Jelinek <jakub@redhat.com>
13390 Richard Biener <rguenther@suse.de>
13392 PR tree-optimization/81503
13393 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13394 folded constant fits in the target type; reorder tests for clarity.
13396 2017-08-29 Martin Liska <mliska@suse.cz>
13398 * passes.def: Include pass_lower_switch.
13399 * stmt.c (dump_case_nodes): Remove and move to
13400 tree-switch-conversion.
13401 (case_values_threshold): Likewise.
13402 (expand_switch_as_decision_tree_p): Likewise.
13403 (emit_case_decision_tree): Likewise.
13404 (expand_case): Likewise.
13405 (balance_case_nodes): Likewise.
13406 (node_has_low_bound): Likewise.
13407 (node_has_high_bound): Likewise.
13408 (node_is_bounded): Likewise.
13409 (emit_case_nodes): Likewise.
13410 (struct simple_case_node): New struct.
13411 (add_case_node): Remove.
13412 (emit_case_dispatch_table): Use vector instead of case_list.
13413 (reset_out_edges_aux): Remove.
13414 (compute_cases_per_edge): Likewise.
13415 (expand_case): Build list of simple_case_node.
13416 (expand_sjlj_dispatch_table): Use it.
13417 * tree-switch-conversion.c (struct case_node): Moved from
13418 stmt.c and adjusted.
13419 (emit_case_nodes): Likewise.
13420 (node_has_low_bound): Likewise.
13421 (node_has_high_bound): Likewise.
13422 (node_is_bounded): Likewise.
13423 (case_values_threshold): Likewise.
13424 (reset_out_edges_aux): Likewise.
13425 (compute_cases_per_edge): Likewise.
13426 (add_case_node): Likewise.
13427 (dump_case_nodes): Likewise.
13428 (balance_case_nodes): Likewise.
13429 (expand_switch_as_decision_tree_p): Likewise.
13430 (emit_jump): Likewise.
13431 (emit_case_decision_tree): Likewise.
13432 (try_switch_expansion): Likewise.
13433 (do_jump_if_equal): Likewise.
13434 (emit_cmp_and_jump_insns): Likewise.
13435 (fix_phi_operands_for_edge): New function.
13436 (record_phi_operand_mapping): Likewise.
13437 (class pass_lower_switch): New pass.
13438 (pass_lower_switch::execute): New function.
13439 (make_pass_lower_switch): Likewise.
13440 (conditional_probability):
13441 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13442 * tree-pass.h: Add make_pass_lower_switch.
13444 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13447 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13450 2017-08-29 Richard Biener <rguenther@suse.de>
13452 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13453 we do not add a DW_AT_inline attribute twice.
13454 (gen_subprogram_die): Remove code setting DW_AT_inline on
13455 DECL_ABSTRACT_P nodes.
13457 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13459 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13460 calls to internal functions.
13461 (gimplify_modify_expr): Likewise.
13462 * tree-call-cdce.c (use_internal_fn): Likewise.
13463 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13464 (convert_to_divmod): Set the nothrow flag.
13465 * tree-if-conv.c (predicate_mem_writes): Likewise.
13466 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13467 (vectorizable_call): Likewise.
13468 (vectorizable_store): Likewise.
13469 (vectorizable_load): Likewise.
13470 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13471 (vect_recog_mask_conversion_pattern): Likewise.
13473 2017-08-29 Martin Liska <mliska@suse.cz>
13476 * gcc.c (driver_handle_option): Add new argument.
13477 * opts-common.c (handle_option): Pass
13478 target_option_override_hook.
13479 * opts-global.c (lang_handle_option): Add new option.
13480 (set_default_handlers): Add new argument.
13481 (decode_options): Likewise.
13482 * opts.c (target_handle_option): Likewise.
13483 (common_handle_option): Call target_option_override_hook.
13484 * opts.h (struct cl_option_handler_func): Add hook for
13485 target option override.
13486 (struct cl_option_handlers): Likewise.
13487 (set_default_handlers): Add new argument.
13488 (decode_options): Likewise.
13489 (common_handle_option): Likewise.
13490 (target_handle_option): Likewise.
13491 * toplev.c (toplev::main): Pass targetm.target_option.override
13494 2017-08-29 Richard Biener <rguenther@suse.de>
13495 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13497 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13498 life to the active subtree.
13500 2017-08-28 Jeff Law <law@redhat.com>
13502 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13503 derive_equivalences.
13504 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13505 Code moved into....
13506 (edge_info::derive_equivalences): New private member function
13508 * tree-ssa-dom.c (class edge_info): Changed from a struct
13509 to a class. Add ctor/dtor, methods and data members.
13510 (edge_info::edge_info): Renamed from allocate_edge_info.
13511 Initialize additional members.
13512 (edge_info::~edge_info): New.
13513 (free_dom_edge_info): Delete the edge info.
13514 (record_edge_info): Use new class & associated member functions.
13515 Tighten forms for testing for edge equivalences.
13516 (record_temporary_equivalences): Iterate over the simple
13517 equivalences rather than assuming there's only one per edge.
13518 (cprop_into_successor_phis): Iterate over the simple
13519 equivalences rather than assuming there's only one per edge.
13520 (optimize_stmt): Use operand_equal_p rather than pointer
13521 equality for mini-DSE code.
13523 2017-08-28 Nathan Sidwell <nathan@acm.org>
13525 * gcc.c (execute): Fold SIGPIPE handling into switch
13526 statement. Adjust internal error message.
13528 2017-08-28 Richard Biener <rguenther@suse.de>
13531 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13532 Do nothing for removed DIEs.
13534 2017-08-28 Richard Biener <rguenther@suse.de>
13536 PR tree-optimization/81977
13537 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13540 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13543 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13544 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13545 using targetm.gen_mem_thread_fence.
13547 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13550 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13551 predicate to register_operand. Reorder operands.
13552 (*btr<mode>): Ditto.
13553 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13554 (*btr<mode>_mask): Ditto.
13556 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13558 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13559 * config/rs6000/xmmintrin.h: New file.
13560 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13562 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13565 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13566 parameter and_insn and return it.
13567 (recombine_lvx_pattern): Insert a copy to ensure availability of
13568 the base register of the copied masking operation at the point of
13569 the instruction replacement.
13570 (recombine_stvx_pattern): Likewise.
13572 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13574 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13575 undocumented switches.
13576 (-mpower9-dform-vector): Likewise.
13577 (-mpower9-dform): Likewise.
13578 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13579 comments to delete references to -mpower9-dform* switches.
13580 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13581 Delete reference to -mpower9-dform* switches, test for
13582 -mpower9-vector instead.
13583 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13584 (OTHER_P9_VECTOR_MASKS): Likewise.
13585 (POWERPC_MASKS): Likewise.
13586 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13587 tests against -mpower9-dform* to -mpower9-vector. Delete code
13588 that checked for -mpower9-dform* consistancy with other options.
13589 Add test for -mpower9-misc to enable other power9 switches.
13590 (rs6000_init_hard_regno_mode_ok): Likewise.
13591 (rs6000_option_override_internal): Likewise.
13592 (rs6000_emit_prologue): Likewise.
13593 (rs6000_emit_epilogue): Likewise.
13594 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13595 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13596 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13598 (emit_fusion_p9_store): Likewise.
13599 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13600 resetting these macros if the assembler does not support ISA 3.0
13602 (TARGET_P9_DFORM_VECTOR): Likewise.
13603 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13604 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13606 2017-08-25 Alan Modra <amodra@gmail.com>
13608 PR rtl-optimization/81747
13609 * cse.c (cse_extended_basic_block): Don't attempt to record
13610 equivalences for degenerate conditional branches that branch
13611 to their fall-through.
13613 2017-08-24 Martin Sebor <msebor@redhat.com>
13615 PR middle-end/81908
13616 * gimple-fold.c (size_must_be_zero_p): New function.
13617 (gimple_fold_builtin_memory_op): Call it.
13619 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13621 * config/rs6000/mm_malloc.h: New file.
13623 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13625 PR tree-optimization/81913
13626 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13627 analysis when either IVs in condition can wrap.
13629 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13631 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13632 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13634 2017-08-24 Richard Biener <rguenther@suse.de>
13637 * targhooks.c (default_target_can_inline_p): Properly
13638 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13639 is present and always compare.
13640 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13641 infer -mfpmath=sse from TARGET_SSE_P.
13642 (ix86_can_inline_p): Properly use target_option_default_node when
13643 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13645 2017-08-24 Richard Biener <rguenther@suse.de>
13648 * dwarf2out.c (output_die): Handle flag_generate_offload like
13650 (output_comp_unit): Likewise.
13651 (gen_array_type_die): Likewise.
13652 (dwarf2out_early_finish): Likewise.
13653 (note_variable_value_in_expr): Likewise.
13654 (dwarf2out_finish): Likewise. Adjust assert.
13655 * cgraphunit.c (symbol_table::compile): Move setting of
13656 flag_generate_offload earlier ...
13657 (symbol_table::finalize_compilation_unit): ... here, before
13658 early debug finalization.
13660 2017-08-24 Richard Biener <rguenther@suse.de>
13662 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13663 and ipa-fnsummary.h.
13664 (ix86_can_inline_p): When ix86_fpmath flags do not match
13665 check whether the callee uses FP math at all.
13667 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13669 PR middle-end/81931
13670 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13671 instead of TYPE_PRECISION.
13673 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13674 Alan Hayward <alan.hayward@arm.com>
13675 David Sherwood <david.sherwood@arm.com>
13677 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13679 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13681 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13683 (ix86_frame::stack_realign_allocate): New field.
13684 (struct machine_frame_state): Modify comments.
13685 (machine_frame_state::sp_realigned_fp_end): New field.
13686 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13687 layout calculation.
13688 (sp_valid_at): Add assertion to assure no attempt to access invalid
13689 offset of a realigned stack.
13690 (fp_valid_at): Likewise.
13691 (choose_baseaddr): Modify comments.
13692 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13693 ix86_expand_prologue.
13694 (ix86_expand_prologue): Modify stack realignment and allocation.
13695 (ix86_expand_epilogue): Modify comments.
13696 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13697 avx2_runtime, avx512f, and avx512f_runtime.
13699 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13701 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13702 (mstackrealign): Do not init to -1.
13703 * config/i386/i386.c (ix86_option_override_internal):
13704 Check opts_set, not opts when setting default value of
13705 opts->x_ix86_force_align_arg_pointer.
13707 2017-08-23 Richard Biener <rguenther@suse.de>
13709 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13710 lang_hooks.decl_printable_name.
13711 * print-rtl-function.c (print_rtx_function): Likewise.
13712 * tree-pretty-print.c (dump_function_header): Likewise.
13714 2017-08-23 Richard Biener <rguenther@suse.de>
13717 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13718 -g0 at compile-time.
13720 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13722 PR middle-end/19706
13723 * doc/sourcebuild.texi (Other hardware attributes):
13726 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13728 PR middle-end/19706
13729 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13730 Fix single-use check.
13732 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13734 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13736 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13738 * doc/install.texi: Modify to add more details on running selected
13741 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13743 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13744 is combined with -mabi=ms.
13745 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13748 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13750 PR tree-optimization/81488
13751 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13752 and cached_basis fields.
13753 (MAX_SPREAD): New constant.
13754 (alloc_cand_and_find_basis): Initialize new fields.
13755 (clear_visited): New function.
13756 (create_phi_basis_1): Rename from create_phi_basis, set visited
13757 and cached_basis fields.
13758 (create_phi_basis): New wrapper function.
13759 (phi_add_costs_1): Rename from phi_add_costs, add spread
13760 parameter, set visited field, short-circuit when limits reached.
13761 (phi_add_costs): New wrapper function.
13762 (record_phi_increments_1): Rename from record_phi_increments, set
13764 (record_phi_increments): New wrapper function.
13765 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13766 (phi_incr_cost): New wrapper function.
13767 (all_phi_incrs_profitable_1): Rename from
13768 all_phi_incrs_profitable, set visited field.
13769 (all_phi_incrs_profitable): New wrapper function.
13771 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13772 Alan Hayward <alan.hayward@arm.com>
13773 David Sherwood <david.sherwood@arm.com>
13775 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13776 that takes the outer and inner modes.
13777 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13778 comparison as the canonical test for a paradoxical subreg.
13779 * combine.c (simplify_set): Use paradoxical_subreg_p.
13780 (make_extraction): Likewise.
13781 (force_to_mode): Likewise.
13782 (rtx_equal_for_field_assignment_p): Likewise.
13783 (gen_lowpart_for_combine): Likewise.
13784 (simplify_comparison): Likewise.
13785 * cse.c (equiv_constant): Likewise.
13786 * expmed.c (store_bit_field_1): Likewise.
13787 * final.c (alter_subreg): Likewise.
13788 * fwprop.c (propagate_rtx): Likewise.
13789 (forward_propagate_subreg): Likewise.
13790 * ira-conflicts.c (ira_build_conflicts): Likewise.
13791 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13792 * lra-constraints.c (curr_insn_transform): Likewise.
13793 (split_reg): Likewise.
13794 * lra-eliminations.c (move_plus_up): Likewise.
13795 (lra_eliminate_regs_1): Likewise.
13796 * recog.c (general_operand): Likewise.
13797 * ree.c (combine_reaching_defs): Likewise.
13798 * reload.c (push_reload): Likewise.
13799 (find_reloads): Likewise.
13800 * reload1.c (elimination_effects): Likewise.
13801 (compute_reload_subreg_offset): Likewise.
13802 (choose_reload_regs): Likewise.
13803 * rtlanal.c (subreg_lsb_1): Likewise.
13804 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13805 (simplify_subreg): Likewise.
13806 * var-tracking.c (track_loc_p): Likewise.
13807 * emit-rtl.c (byte_lowpart_offset): Likewise.
13808 (paradoxical_subreg_p): Delete out-of-line definition.
13810 2017-08-22 Jeff Law <law@redhat.com>
13812 PR tree-optimization/81741
13813 PR tree-optimization/71947
13814 * tree-ssa-dom.c: Include tree-inline.h.
13815 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13816 equivalences if one is more expensive to compute than the other.
13817 * tree-ssa-scopedtables.h (class const_or_copies): Make
13818 record_const_or_copy_raw method private.
13819 (class avail_exprs_stack): New method simplify_binary_operation.
13820 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13821 avail_exprs_stack::simplify_binary_operation as needed.
13822 (avail_exprs_stack::simplify_binary_operation): New function.
13824 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13826 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13827 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13828 (DOT_SYMBOLS): Likewise.
13829 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13830 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13831 (RS6000_ABI_NAME): Likewise.
13832 (TARGET_CMODEL): Likewise.
13833 (TOC_SECTION_ASM_OP): Likewise.
13834 (SET_CMODEL): New macro.
13835 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13837 2017-08-22 Richard Biener <rguenther@suse.de>
13839 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13840 to allow for free-lang-data replacements similar to verify_type_variant.
13842 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13843 Alan Hayward <alan.hayward@arm.com>
13844 David Sherwood <david.sherwood@arm.com>
13846 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13847 VOIDmode for the LABEL_REF.
13849 2017-08-22 Richard Biener <rguenther@suse.de>
13851 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13853 2017-08-22 Richard Biener <rguenther@suse.de>
13855 * common.opt (feliminate-dwarf2-dups): Ignore.
13856 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13857 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13858 same_die_p_wrap, compute_section_prefix,
13859 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13860 (comdat_symbol_id, comdat_symbol_number): Likewise.
13861 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13863 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13864 (output_die): Mark unreachable path unreachable.
13865 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13866 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13867 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13868 (dwarf2out_early_finish): Likewise.
13870 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13872 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13874 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13877 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13878 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13879 (avr_attribute_table) <io, io_low, address>: Initialize
13880 .decl_required with true.
13882 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13884 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13885 undocumented debugging options.
13886 (-mvsx-scalar-double): Likewise.
13887 (-mallow-df-permute): Likewise.
13888 (-mvectorize-builtins): Likewise.
13889 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13890 (rs6000_builtin_vectorized_function): Likewise.
13891 (rs6000_builtin_md_vectorized_function): Likewise.
13892 (rs6000_opt_vars): Likewise.
13894 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13897 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13898 (*btrq_imm): Rename from *btrq.
13899 (*btcq_imm): Rename from *btcq.
13900 (btsc): New code attribute.
13901 (*<btsc><mode>): New insn pattern.
13902 (*btr<mode>): Ditto.
13903 (*<btsc><mode>_mask): New insn_and_split pattern.
13904 (*btr<mode>_mask): Ditto.
13906 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13907 Alan Hayward <alan.hayward@arm.com>
13908 David Sherwood <david.sherwood@arm.com>
13910 * function.c (pad_below): Simplify padding calculation.
13912 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13913 Alan Hayward <alan.hayward@arm.com>
13914 David Sherwood <david.sherwood@arm.com>
13916 * target.def (function_prologue): Remove frame size argument.
13917 (function_epilogue): Likewise.
13918 * doc/tm.texi: Regenerate.
13919 * final.c (final_start_function): Update call to function_prologue.
13920 (final_end_function): Update call to function_epilogue.
13921 (default_function_pro_epilogue): Remove frame size argument.
13922 * output.h (default_function_pro_epilogue): Likewise.
13923 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13924 (arm_output_function_prologue): Likewise.
13925 * config/frv/frv.c (frv_function_prologue): Likewise.
13926 (frv_function_epilogue): Likewise.
13927 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13928 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13929 (ia64_output_function_epilogue): Likewise.
13930 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13931 (m32r_output_function_epilogue): Likewise.
13932 * config/microblaze/microblaze.c (microblaze_function_prologue)
13933 (microblaze_function_epilogue): Likewise.
13934 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13935 (mips_output_function_epilogue): Likewise.
13936 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13937 (mmix_target_asm_function_epilogue): Likewise.
13938 * config/msp430/msp430.c (msp430_start_function): Likewise.
13939 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13940 (nds32_asm_function_epilogue): Likewise.
13941 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13942 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13943 (pa_output_function_epilogue): Likewise.
13944 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13945 (rs6000_output_function_epilogue): Likewise.
13946 * config/rl78/rl78.c (rl78_start_function): Likewise.
13947 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13948 (rs6000_output_function_epilogue): Likewise.
13949 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13950 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13951 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13952 (sparc_asm_function_epilogue): Likewise.
13954 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13956 * tree.h (type_has_mode_precision_p): New function.
13957 * convert.c (convert_to_integer_1): Use it.
13958 * expr.c (expand_expr_real_2): Likewise.
13959 (expand_expr_real_1): Likewise.
13960 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13961 * match.pd: Likewise.
13962 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13963 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13964 * tree-tailcall.c (process_assignment): Likewise.
13965 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13966 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13967 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13968 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13969 (vectorizable_assignment): Likewise.
13970 (vectorizable_shift): Likewise.
13971 (vectorizable_operation): Likewise.
13972 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13974 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13976 * match.pd: Add pow (C, x) simplification.
13978 2017-08-21 Richard Biener <rguenther@suse.de>
13980 PR tree-optimization/81900
13981 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13982 for blocks with abnormal predecessors.
13983 (compute_antic): Do not set visited flag prematurely.
13985 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13988 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13990 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13992 * stor-layout.h (vector_type_mode): Move to...
13993 * tree.h (vector_type_mode): ...here.
13994 * stor-layout.c (vector_type_mode): Move to...
13995 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13997 2017-08-21 Richard Biener <rguenther@suse.de>
13999 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14000 register_external_die hooks.
14001 (debug_false_tree_charstarstar_uhwistar): Declare.
14002 (debug_nothing_tree_charstar_uhwi): Likewise.
14003 * debug.c (do_nothing_debug_hooks): Adjust.
14004 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14005 (debug_nothing_tree_charstar_uhwi): Likewise.
14006 * dbxout.c (dbx_debug_hooks): Adjust.
14007 (xcoff_debug_hooks): Likewise.
14008 * sdbout.c (sdb_debug_hooks): Likewise.
14009 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14010 * dwarf2out.c (macinfo_label_base): New global.
14011 (dwarf2out_register_external_die): New function for the
14012 register_external_die hook.
14013 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14014 (dwarf2_debug_hooks): Use them.
14015 (dwarf2_lineno_debug_hooks): Adjust.
14016 (struct die_struct): Add with_offset flag.
14017 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14018 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14019 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14020 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14021 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14022 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14023 defining section names for the early LTO debug variants.
14024 (reset_indirect_string): New helper.
14025 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14026 (print_dw_val): Add support for offsetted symbol references.
14027 (get_ultimate_context): Split out from is_cxx.
14028 (is_cxx): Use get_ultimate_context.
14029 (is_fortran): Add decl overload.
14030 (compute_comp_unit_symbol): Split out worker from
14031 compute_section_prefix.
14032 (compute_section_prefix): Call compute_comp_unit_symbol and
14033 set comdat_type_p here.
14034 (output_die): Skip DIE symbol output for the LTO added one.
14035 Handle DIE symbol references with offset.
14036 (output_comp_unit): Guard section name mangling properly.
14037 For LTO debug sections emit a symbol at the section beginning
14038 which we use to refer to its DIEs.
14039 (add_abstract_origin_attribute): For DIEs registered via
14040 dwarf2out_register_external_die directly refer to the early
14041 DIE rather than indirectly through the shadow one we created.
14042 Remove obsolete call to dwarf2out_abstract_function for
14043 non-function/block origins.
14044 (gen_array_type_die): When generating early LTO debug do
14045 not emit DW_AT_string_length.
14046 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14047 late when in LTO. As suggested place a gcc_unreachable for
14048 the DECL_ABSTRACT_P case.
14049 (gen_subprogram_die): Avoid another specification DIE
14050 for early built declarations/definitions for the late LTO case.
14051 (gen_variable_die): Add type references for late duplicated VLA dies
14053 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14054 we have the abstract instance already.
14055 (process_scope_var): Adjust decl DIE contexts in LTO which
14056 first puts them in limbo.
14057 (gen_decl_die): Do not generate type DIEs late apart from
14058 types for VLAs or for decls we do not yet have a DIE. Do not
14059 call dwarf2out_abstract_function late.
14060 (dwarf2out_early_global_decl): Make sure to create DIEs
14061 for abstract instances of a decl first.
14062 (dwarf2out_late_global_decl): Adjust comment.
14063 (output_macinfo_op): With multiple macro sections use
14064 macinfo_label_base to distinguish labels.
14065 (output_macinfo): Likewise. Update macinfo_label_base.
14066 Pass in the line info label.
14067 (note_variable_value_in_expr): When generating LTO resolve
14068 all variable values here by generating DIEs as needed.
14069 (init_sections_and_labels): Add early LTO debug flag parameter
14070 and generate different sections and names if set. Add generation
14071 counter for the labels so we can have multiple of them.
14072 (reset_dies): Helper to allow DIEs to be output multiple times.
14073 (dwarf2out_finish): When outputting DIEs to the fat part of an
14074 LTO object first reset DIEs.
14075 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14076 (modified_type_die): Check for decl_ultimate_origin being self
14078 (gen_type_die_with_usage): Likewise.
14079 (gen_typedef_die): Allow decl_ultimate_origin being self.
14080 (set_decl_abstract_flags): Remove.
14081 (set_block_abstract_flags): Likewise.
14082 (dwarf2out_abstract_function): Treat the early generated DIEs
14083 as the abstract copy and only add DW_AT_inline and
14084 DW_AT_artificial here and call set_decl_origin_self.
14085 If the DIE has an abstract origin don't do anything.
14086 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14087 if we have none yet (Go fails to build one, PR78628).
14088 (variably_modified_type_p): Prevent endless recursion for Ada
14089 cyclic pointer types.
14090 * lto-streamer-in.c: Include debug.h.
14091 (dref_queue): New global.
14092 (lto_read_tree_1): Stream in DIE references.
14093 (lto_input_tree): Register DIE references.
14094 (input_function): Stream DECL_DEBUG_ARGS.
14095 * lto-streamer-out.c: Include debug.h.
14096 (lto_write_tree_1): Output DIE references.
14097 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14098 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14099 (output_function): Stream DECL_DEBUG_ARGS.
14100 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14101 Stream DECL_ABSTRACT_ORIGIN.
14102 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14103 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14104 DECL_CONTEXT for file-scope decls.
14105 * lto-streamer.h (struct dref_entry): Declare.
14106 (dref_queue): Likewise.
14107 * cfgexpand.c (pass_expand::execute): Do not call the
14108 outlining_inline_function hook here.
14109 * lto-wrapper.c (debug_obj): New global.
14110 (tool_cleanup): Unlink it if required.
14111 (debug_objcopy): New function.
14112 (run_gcc): Handle early debug sections in the IL files by
14113 extracting them to separate files, partially linkin them and
14114 feeding the result back as result to the linker.
14115 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14116 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14117 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14118 sections into a separate segment.
14119 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14121 (darwin_asm_dwarf_section): Likewise.
14122 (darwin_asm_output_dwarf_offset): Likewise.
14123 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14125 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14126 Alan Hayward <alan.hayward@arm.com>
14127 David Sherwood <david.sherwood@arm.com>
14129 * read-md.h (md_reader::record_potential_iterator_use): Replace
14130 pointer argument with an rtx and an index.
14131 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14132 (apply_mode_iterator): Likewise.
14133 (apply_code_iterator): Likewise.
14134 (apply_int_iterator): Likewise.
14135 (apply_subst_iterator): Likewise.
14136 (record_iterator_use): Likewise.
14137 (record_attribute_use): Likewise.
14138 (md_reader::record_potential_iterator_use): Likewise. Update calls
14139 to record_iterator_use and apply_iterator.
14140 (iterator_use): Replace ptr with x and index.
14141 (attribute_use): Likewise.
14142 (apply_attribute_uses): Update calls to apply_iterator.
14143 (apply_iterators): Likewise. Update initialization of iterator_use.
14144 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14145 and record_potential_iterator_use.
14146 (rtx_reader::read_rtx_operand): Likewise.
14148 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14149 Alan Hayward <alan.hayward@arm.com>
14150 David Sherwood <david.sherwood@arm.com>
14152 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14155 2017-08-21 Richard Biener <rguenther@suse.de>
14157 PR middle-end/81884
14158 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14159 at struct end conservatively when comparing common bases.
14161 2017-08-21 Richard Biener <rguenther@suse.de>
14163 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14164 (mem_ref_in_stmt): Remove.
14165 (determine_max_movement): Use ref index to get at the reference.
14166 (invariantness_dom_walker::before_dom_children): Deal with
14167 lim data already initialized.
14168 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14170 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14172 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14173 (TARGET_ISA_ROUND): Ditto.
14174 (TARGET_ROUND): Ditto.
14175 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14176 * config/i386/i386.md: Ditto.
14177 * config/i386/sse.md: Ditto.
14178 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14179 with OPTION_MASK_ISA_SSE4_1.
14181 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14184 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14185 __builtin_ia32_lzcnt_u16.
14187 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14190 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14191 (rs6000_set_current_function): Rewrite function to use it.
14193 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14196 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14197 and TYPE_WARN_IF_NOT_ALIGN.
14198 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14199 (handle_warn_if_not_align): New.
14200 (place_union_field): Call handle_warn_if_not_align.
14201 (place_field): Call handle_warn_if_not_align.
14202 Copy TYPE_WARN_IF_NOT_ALIGN.
14203 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14204 (layout_type): Likewise.
14205 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14207 (tree_decl_common): Add warn_if_not_align.
14208 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14209 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14210 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14211 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14212 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14213 * doc/extend.texi: Document warn_if_not_aligned attribute.
14214 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14216 2017-08-17 Martin Liska <mliska@suse.cz>
14219 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14220 (get_data_dependence): Use it as pointer type.
14221 (distribute_loop): Likewise.
14223 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14225 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14226 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14227 (p8_vmrgow_<mode>_direct): New define_insn.
14228 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14229 handle endianness for vmrgew and vmrgow permute patterns.
14231 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14233 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14234 * config/rs6000/rs6000-cpus.def: Remove comment.
14235 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14236 (POWERPC_MASKS): Likewise.
14237 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14238 use of TARGET_VSX_TIMODE.
14239 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14240 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14241 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14242 (rs6000_option_override_internal): Remove dead code.
14243 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14244 (rs6000_legitimize_reload_address): Likewise.
14245 (rs6000_legitimate_address_p): Likewise.
14246 (rs6000_opt_masks): Delete "vsx-timode".
14247 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14248 from function comment.
14249 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14250 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14251 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14253 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14254 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14255 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14256 (VSX_TI): Likewise.
14258 (define_peephole2): Likewise.
14260 2017-08-17 Martin Sebor <msebor@redhat.com>
14263 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14264 past the end of an array.
14265 (test_pp_format): Add test cases.
14267 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14269 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14270 missing ECF_NOTHROW flags.
14272 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14275 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14276 operands residing in integer registers.
14277 (*vsx_le_perm_load_<mode>): Likewise.
14278 (*vsx_le_perm_store_<mode>): Likewise.
14279 (define_peephole2): Add peepholes to optimize the above.
14281 2017-08-17 Marek Polacek <polacek@redhat.com>
14283 PR middle-end/81814
14284 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14285 to mimic what shorten_compare did. Change the return type to bool.
14286 (fold_cond_expr_with_comparison): Update call to
14287 operand_equal_for_comparison_p.
14288 (fold_ternary_loc): Likewise.
14290 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14292 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14294 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14296 2017-08-17 Richard Biener <rguenther@suse.de>
14298 * tree-ssa-structalias.c (solve_graph): When propagating
14299 to successors update the graphs succ edges and avoid duplicate work.
14301 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14304 * config/i386/i386.c (ix86_option_override_internal): Save target
14305 specific options after ix86_stack_protector_guard_reg was changed.
14307 2017-08-17 Richard Biener <rguenther@suse.de>
14309 PR tree-optimization/81827
14310 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14311 (new_var_info): Initialize it conservatively.
14312 (get_call_vi): Mark register vars.
14313 (new_scalar_tmp_constraint_exp): Likewise.
14314 (handle_rhs_call): Likewise.
14315 (handle_const_call): Likewise.
14316 (create_function_info_for): Likewise.
14317 (solve_constraints): Sort varinfos to separate register from
14318 non-register vars to pack points-to solution bitmaps during
14321 2017-08-17 Marek Polacek <polacek@redhat.com>
14323 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14325 2017-08-17 Richard Biener <rguenther@suse.de>
14327 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14328 to true when overflow is undefined and we saturated the result.
14330 2017-08-17 Alan Modra <amodra@gmail.com>
14333 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14334 Don't use store multiple if only one reg needs saving.
14335 (interesting_frame_related_regno): New function.
14336 (rs6000_frame_related): Don't emit frame info for regs that
14338 (rs6000_emit_epilogue): Likewise.
14340 2017-08-16 Nathan Sidwell <nathan@acm.org>
14342 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14343 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14344 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14345 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14346 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14347 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14348 (verify_type): Adjust for TYPE_BINFO move.
14349 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14350 process TYPE_BINFO directly.
14351 (hash_tree): Likewise.
14352 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14354 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14357 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14359 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14361 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14364 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14365 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14366 Add ix86_binary_operator_ok to insn constraint.
14367 (*iordi_1_bts): Ditto.
14368 (*xordi_1_btc): Ditto.
14369 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14370 Update corresponding peephole2 pattern.
14374 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14376 PR tree-optimization/81832
14377 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14378 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14380 2017-08-16 Marek Polacek <polacek@redhat.com>
14383 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14384 regarding size_zero_node.
14386 2017-08-16 Martin Liska <mliska@suse.cz>
14389 * config.gcc: Respect previously set extra_objs in case
14392 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14394 PR tree-optimization/81835
14395 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14396 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14397 not depend on the phi.
14399 2017-08-16 Alan Modra <amodra@gmail.com>
14401 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14404 2017-08-16 Alan Modra <amodra@gmail.com>
14406 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14407 (save_reg_p): ..into this. Update all callers.
14408 (first_reg_to_save): Simplify.
14410 2017-08-16 Alan Modra <amodra@gmail.com>
14412 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14415 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14419 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14420 constant count if that count is less than 32.
14422 2017-08-15 Nathan Sidwell <nathan@acm.org>
14424 * gcc.c (execute): Emit friendlier message if inferior is killed
14425 by an external cause.
14427 2017-08-15 Richard Biener <rguenther@suse.de>
14429 PR tree-optimization/81790
14430 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14431 CONSTRUCTORs from simplifying and VN.
14433 2017-08-14 Martin Sebor <msebor@redhat.com>
14435 * builtin-attrs.def: Add comments.
14437 2017-08-14 Martin Sebor <msebor@redhat.com>
14440 * doc/extend.texi (attribute nonstring): Document new attribute.
14442 2017-08-14 Martin Sebor <msebor@redhat.com>
14445 * tree-diagnostic.c (default_tree_printer): Handle %G.
14446 * gimple-pretty-print.h (percent_G_format): Declare new function.
14447 * gimple-pretty-print.c (percent_G_format): Define.
14448 * tree-pretty-print.c (percent_K_format): Add argument.
14450 2017-08-14 Martin Sebor <msebor@redhat.com>
14452 PR translation/79998
14453 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14454 Remove a stray space.
14456 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14459 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14460 (*iordi_1_bts): Ditto.
14461 (*xordi_1_btc): Ditto.
14463 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14466 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14467 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14469 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14470 quoted strings, and make more translator-friendly.
14471 (darwin_rs6000_override_options): Likewise.
14472 (rs6000_option_override_internal): Likewise.
14473 (rs6000_return_in_memory): Fix overlong line.
14474 (init_cmulative_args): Use quoted strings, and make more
14475 translator-friendly.
14476 (rs6000_pass_by_reference): Fix overlong line.
14477 (def_builtin): Use quoted strings.
14478 (altivec_expand_predicate_builtin): Use quoted strings, and make
14479 more translator-friendly.
14480 (htm_expand_builtin): Use quoted strings.
14481 (cpu_expand_builtin): Use quoted strings, and make more
14482 translator-friendly.
14483 (altivec_expand_builtin): Likewise.
14484 (paired_expand_predicate_builtin): Likewise.
14485 (rs6000_invalid_builtin): Likewise.
14486 (builtin_function_type): Use quoted strings.
14487 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14488 more translator-friendly.
14489 (rs6000_trampoline_init): Likewise.
14490 (rs6000_handle_altivec_attribute): Likewise.
14491 (rs6000_inner_target_options): Use quoted strings.
14492 (rs6000_disable_incompatible_switches): Likewise.
14493 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14494 strings, and make more translator-friendly.
14495 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14497 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14499 PR tree-optimization/81799
14500 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14501 cond_expr to simple gimple operand.
14503 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14505 PR middle-end/46932
14506 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14508 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14512 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14513 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14514 TARGET_GASISR_PROLOGUES.
14515 * config/avr/avr.c (avr_option_override): Same.
14516 (avr_pass_pre_proep::execute): Same.
14518 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14521 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14522 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14524 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14526 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14527 define_insn_and_split. Split to a memory load from 0 in
14528 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14529 using PTR mode iterator.
14530 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14531 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14532 (*add_tp_<mode>): Redefine as define_insn_and_split.
14533 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14534 address space. Merge with *add_tp_x32 using PTR mode iterator.
14535 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14536 Split to an add with a memory load from 0 in
14537 DEFAULT_TLS_SEG_REG address space.
14539 2017-08-12 Andrew Pinski <apinski@cavium.com>
14541 * config/aarch64/aarch64-option-extensions.def (rdma):
14542 Fix feature string to what Linux prints out in /proc/cpuinfo.
14544 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14547 * dwarf2out.c (modified_type_die): For C typedef types that have
14548 an ultimate origin, process the ultimate origin instead of the
14550 (gen_typedef_die): Assert that input DECLs have no ultimate
14552 (gen_type_die_with_usage): For typedef variants that have an
14553 ultimate origin, just call gen_decl_die on the original DECL.
14554 (process_scope_var): Avoid creating DIEs for local typedefs and
14555 concrete static variables.
14557 2017-08-12 Alan Modra <amodra@gmail.com>
14561 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14562 match gnu-user.h startfile.
14563 (ENDFILE_LINUX_SPEC): Similarly.
14565 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14568 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14570 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14572 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14573 * config/aarch64/aarch64.md (mov<mode>): Change.
14574 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14575 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14576 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14578 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14580 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14581 for storage order barriers.
14583 2017-08-11 Martin Liska <mliska@suse.cz>
14586 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14587 variables of void type.
14589 2017-08-11 Martin Liska <mliska@suse.cz>
14591 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14592 TARGET_SUPPORTS_ALIASES.
14593 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14594 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14595 (optimize_weakref): Likewise.
14596 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14597 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14598 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14600 2017-08-11 Martin Liska <mliska@suse.cz>
14603 * config/i386/i386.c (make_resolver_func): Do complete
14604 refactoring of the function.
14606 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14609 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14610 * config/i386/i386.c (ix86_stack_protect_guard): Use
14611 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14612 * doc/invoke.texi (x86 Options): Document
14613 -mstack-protector-guard-symbol= option.
14615 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14617 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14618 * config/i386/i386.c (ix86_split_stack_guard): New function.
14619 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14620 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14621 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14622 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14623 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14624 (split_stack_space_check): Call ix86_split_stack_guard.
14626 2017-08-10 Martin Sebor <msebor@redhat.com>
14628 * print-tree.c (print_node): Print location using the established
14630 Replace spaces with colons.
14631 (debug_raw, debug): Ditto.
14633 2017-08-10 Martin Sebor <msebor@redhat.com>
14636 * pretty-print.c (pp_format): Correct the handling of %s precision.
14638 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14641 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14643 (ix86_finalize_stack_frame_flags): This. Also clear
14644 frame_pointer_needed if -fno-omit-frame-pointer is used without
14646 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14647 with ix86_finalize_stack_frame_flags.
14648 (ix86_expand_epilogue): Likewise.
14649 (ix86_expand_split_stack_prologue): Likewise.
14650 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14652 2017-08-10 Martin Liska <mliska@suse.cz>
14655 * c-attribs.c (handle_target_attribute):
14656 Report warning for an empty string argument of target attribute.
14658 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14661 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14663 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14664 or DECL_NONLOCAL labels.
14665 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14666 or DECL_NONLOCAL labels here.
14668 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14670 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14671 to indicate when early gimple folding has been disabled.
14672 (rs6000_gimple_fold_builtin): Add debug content.
14673 (rs6000_invalid_builtin): Fix whitespace.
14674 (rs6000_expand_builtin): Fix whitespace.
14675 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14677 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14680 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14681 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14683 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14685 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14686 (qdf24xx): Likewise.
14687 * config/aarch64/aarch64.md: Include falkor.md.
14688 * config/aarch64/falkor.md: New.
14690 2017-08-09 Marek Polacek <polacek@redhat.com>
14693 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14694 * diagnostic.c (emit_diagnostic): Add a comment.
14695 (emit_diagnostic_valist): New function.
14697 2017-08-09 Marek Polacek <polacek@redhat.com>
14700 * input.c (make_location): New overload.
14701 * input.h (make_location): Declare.
14703 2017-08-08 Alan Modra <amodra@gmail.com>
14704 H.J. Lu <hongjiu.lu@intel.com>
14707 * gcc.c (NO_PIE_SPEC): Delete.
14708 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14710 (LINK_PIE_SPEC): ..to here.
14711 (LINK_COMMAND_SPEC): Support -no-pie.
14712 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14713 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14715 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14716 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14717 (ENDFILE_CRTEND_SPEC): Similarly.
14719 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14722 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14723 (mstack-protector-guard-offset=): Ditto.
14724 * config/i386/i386.c (ix86_option_override): Handle
14725 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14727 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14728 ix86_stack_protect_guard_offset variables.
14729 (TARGET_STACK_PROTECT_GUARD): Always define.
14730 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14731 and -mstack-protector-guard-offset= options.
14733 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14735 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14736 boundary case for the last candidate.
14738 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14740 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14741 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14743 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14745 PR middle-end/19706
14746 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14747 * config/aarch64/aarch64-builtins.c
14748 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14749 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14750 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14752 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14753 Andrew Pinski <pinskia@gmail.com>
14755 PR middle-end/19706
14756 * internal-fn.def (XORSIGN): New.
14757 * optabs.def (xorsign_optab): New.
14758 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14759 (convert_expand_mult_copysign): New.
14760 (pass_optimize_widening_mul::execute): Call
14761 convert_expand_mult_copysign.
14763 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14765 PR tree-optimization/81354
14766 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14767 Insert on edges rather than explicitly creating landing pads.
14768 (analyze_candidates_and_replace): Commit edge inserts.
14770 2017-08-08 Richard Biener <rguenther@suse.de>
14772 PR middle-end/81719
14773 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14774 (expand_simple_operations): Also look through ADDR_EXPRs with
14775 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14777 2017-08-08 Richard Biener <rguenther@suse.de>
14779 PR tree-optimization/81723
14780 * tree-vect-slp.c (struct bst_traits): New hash traits.
14781 (bst_fail): New global.
14782 (vect_build_slp_tree_2): New worker, split out from ...
14783 (vect_build_slp_tree): ... this now wrapping it with using
14784 bst_fail set to cache SLP tree build fails. Properly handle
14786 (vect_analyze_slp_instance): Allocate and free bst_fail.
14788 2017-08-08 Martin Liska <mliska@suse.cz>
14791 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14792 LABEL_DECLs that can be from a different function.
14794 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14796 PR tree-optimization/81744
14797 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14798 loop's number of iterations.
14800 2017-08-08 Martin Liska <mliska@suse.cz>
14802 * asan.c: Include header files.
14803 * attribs.c (build_decl_attribute_variant): New function moved
14805 (build_type_attribute_qual_variant): Likewise.
14806 (cmp_attrib_identifiers): Likewise.
14807 (simple_cst_list_equal): Likewise.
14808 (omp_declare_simd_clauses_equal): Likewise.
14809 (attribute_value_equal): Likewise.
14810 (comp_type_attributes): Likewise.
14811 (build_type_attribute_variant): Likewise.
14812 (lookup_ident_attribute): Likewise.
14813 (remove_attribute): Likewise.
14814 (merge_attributes): Likewise.
14815 (merge_type_attributes): Likewise.
14816 (merge_decl_attributes): Likewise.
14817 (merge_dllimport_decl_attributes): Likewise.
14818 (handle_dll_attribute): Likewise.
14819 (attribute_list_equal): Likewise.
14820 (attribute_list_contained): Likewise.
14821 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14822 (lookup_attribute_by_prefix): Likewise.
14823 * bb-reorder.c: Include header files.
14824 * builtins.c: Likewise.
14825 * calls.c: Likewise.
14826 * cfgexpand.c: Likewise.
14827 * cgraph.c: Likewise.
14828 * cgraphunit.c: Likewise.
14829 * convert.c: Likewise.
14830 * dwarf2out.c: Likewise.
14831 * final.c: Likewise.
14832 * fold-const.c: Likewise.
14833 * function.c: Likewise.
14834 * gimple-expr.c: Likewise.
14835 * gimple-fold.c: Likewise.
14836 * gimple-pretty-print.c: Likewise.
14837 * gimple.c: Likewise.
14838 * gimplify.c: Likewise.
14839 * hsa-common.c: Likewise.
14840 * hsa-gen.c: Likewise.
14841 * internal-fn.c: Likewise.
14842 * ipa-chkp.c: Likewise.
14843 * ipa-cp.c: Likewise.
14844 * ipa-devirt.c: Likewise.
14845 * ipa-fnsummary.c: Likewise.
14846 * ipa-inline.c: Likewise.
14847 * ipa-visibility.c: Likewise.
14849 * lto-cgraph.c: Likewise.
14850 * omp-expand.c: Likewise.
14851 * omp-general.c: Likewise.
14852 * omp-low.c: Likewise.
14853 * omp-offload.c: Likewise.
14854 * omp-simd-clone.c: Likewise.
14855 * opts-global.c: Likewise.
14856 * passes.c: Likewise.
14857 * predict.c: Likewise.
14858 * sancov.c: Likewise.
14859 * sanopt.c: Likewise.
14860 * symtab.c: Likewise.
14861 * toplev.c: Likewise.
14862 * trans-mem.c: Likewise.
14863 * tree-chkp.c: Likewise.
14864 * tree-eh.c: Likewise.
14865 * tree-into-ssa.c: Likewise.
14866 * tree-object-size.c: Likewise.
14867 * tree-parloops.c: Likewise.
14868 * tree-profile.c: Likewise.
14869 * tree-ssa-ccp.c: Likewise.
14870 * tree-ssa-live.c: Likewise.
14871 * tree-ssa-loop.c: Likewise.
14872 * tree-ssa-sccvn.c: Likewise.
14873 * tree-ssa-structalias.c: Likewise.
14874 * tree-ssa.c: Likewise.
14875 * tree-streamer-in.c: Likewise.
14876 * tree-vectorizer.c: Likewise.
14877 * tree-vrp.c: Likewise.
14878 * tsan.c: Likewise.
14879 * ubsan.c: Likewise.
14880 * varasm.c: Likewise.
14881 * varpool.c: Likewise.
14882 * tree.c: Remove functions moved to attribs.[ch].
14883 * tree.h: Likewise.
14884 * config/aarch64/aarch64.c: Add attrs.h header file.
14885 * config/alpha/alpha.c: Likewise.
14886 * config/arc/arc.c: Likewise.
14887 * config/arm/arm.c: Likewise.
14888 * config/avr/avr.c: Likewise.
14889 * config/bfin/bfin.c: Likewise.
14890 * config/c6x/c6x.c: Likewise.
14891 * config/cr16/cr16.c: Likewise.
14892 * config/cris/cris.c: Likewise.
14893 * config/darwin.c: Likewise.
14894 * config/epiphany/epiphany.c: Likewise.
14895 * config/fr30/fr30.c: Likewise.
14896 * config/frv/frv.c: Likewise.
14897 * config/ft32/ft32.c: Likewise.
14898 * config/h8300/h8300.c: Likewise.
14899 * config/i386/winnt.c: Likewise.
14900 * config/ia64/ia64.c: Likewise.
14901 * config/iq2000/iq2000.c: Likewise.
14902 * config/lm32/lm32.c: Likewise.
14903 * config/m32c/m32c.c: Likewise.
14904 * config/m32r/m32r.c: Likewise.
14905 * config/m68k/m68k.c: Likewise.
14906 * config/mcore/mcore.c: Likewise.
14907 * config/microblaze/microblaze.c: Likewise.
14908 * config/mips/mips.c: Likewise.
14909 * config/mmix/mmix.c: Likewise.
14910 * config/mn10300/mn10300.c: Likewise.
14911 * config/moxie/moxie.c: Likewise.
14912 * config/msp430/msp430.c: Likewise.
14913 * config/nds32/nds32-isr.c: Likewise.
14914 * config/nds32/nds32.c: Likewise.
14915 * config/nios2/nios2.c: Likewise.
14916 * config/nvptx/nvptx.c: Likewise.
14917 * config/pa/pa.c: Likewise.
14918 * config/pdp11/pdp11.c: Likewise.
14919 * config/powerpcspe/powerpcspe.c: Likewise.
14920 * config/riscv/riscv.c: Likewise.
14921 * config/rl78/rl78.c: Likewise.
14922 * config/rx/rx.c: Likewise.
14923 * config/s390/s390.c: Likewise.
14924 * config/sh/sh.c: Likewise.
14925 * config/sol2.c: Likewise.
14926 * config/sparc/sparc.c: Likewise.
14927 * config/spu/spu.c: Likewise.
14928 * config/stormy16/stormy16.c: Likewise.
14929 * config/tilegx/tilegx.c: Likewise.
14930 * config/tilepro/tilepro.c: Likewise.
14931 * config/v850/v850.c: Likewise.
14932 * config/vax/vax.c: Likewise.
14933 * config/visium/visium.c: Likewise.
14934 * config/xtensa/xtensa.c: Likewise.
14936 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14939 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14940 constraints since the -mupper-regs-* switches have been
14942 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14943 into a vector from a double word element that was extracted from
14944 another vector, and eliminate extra XXPERMDI instructions.
14945 (vsx_concat_<mode>_2): Likewise.
14946 (vsx_concat_<mode>_3): Likewise.
14947 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14948 concat to allow optimizing inserts from previous extracts.
14950 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14952 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14953 memory reference to a SSP offset in TLS address space.
14954 (ix86_print_operand) <case '@'>: Remove.
14955 (ix86_print_operand_punct_valid_p): Remove '@' code.
14956 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14957 UNSPEC_SP_TLS_TEST.
14958 (stack_tls_protect_set_<mode>): Remove.
14959 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14960 (stack_tls_protect_test_<mode>): Remove.
14961 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14963 2017-08-07 Olivier Hainque <hainque@adacore.com>
14966 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14968 2017-08-07 Douglas Rupp <rupp@adacore.com>
14970 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14971 variable was referenced as multidir in command.
14973 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14976 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14979 2017-08-07 Martin Liska <mliska@suse.cz>
14981 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14982 * config/rl78/rl78.c: Add include of attribs.h.
14983 * config/sh/sh.c: Likewise.
14984 * config/v850/v850.c: Likewise.
14986 2017-08-07 Tom de Vries <tom@codesourcery.com>
14988 PR middle-end/78266
14989 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14991 2017-08-07 Martin Liska <mliska@suse.cz>
14993 * config/mips/mips.c: Include attribs.h.
14995 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14998 * doc/invoke.texi: Document change in behvaior for -Ofast for
15001 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15003 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15005 (aarch64_pop_regs): Likewise.
15006 (aarch64_gen_load_pair): Likewise.
15007 (aarch64_save_callee_saves): Likewise.
15008 (aarch64_restore_callee_saves): Likewise.
15010 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15012 * config/i386/i386.c: Revert the last change.
15014 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15017 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15019 (ix86_finalize_stack_frame_flags): This. Also clear
15020 frame_pointer_needed if -fno-omit-frame-pointer is used without
15022 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15023 with ix86_finalize_stack_frame_flags.
15024 (ix86_expand_epilogue): Likewise.
15025 (ix86_expand_split_stack_prologue): Likewise.
15027 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15030 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15031 to P_AES for Westmere.
15033 2017-08-07 Jonathan Yong <10walls@gmail.com>
15035 * config/i386/mingw.opt (fset-stack-executable): Removed.
15036 * config/i386/cygming.opt (fset-stack-executable): Moved
15038 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15040 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15042 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15044 2017-08-07 Marek Polacek <polacek@redhat.com>
15046 PR middle-end/81737
15047 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15049 2017-08-07 Martin Liska <mliska@suse.cz>
15051 * attribs.h (canonicalize_attr_name): New function.
15052 (cmp_attribs): Move from c-format.c and adjusted.
15053 (is_attribute_p): Moved from tree.h.
15054 * tree-inline.c: Add new includes.
15055 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15056 (private_is_attribute_p): Remove.
15057 (private_lookup_attribute): Likewise.
15058 (private_lookup_attribute_by_prefix): Simplify.
15059 (remove_attribute): Use is_attribute_p.
15060 * tree.h: Remove removed declarations.
15062 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15064 PR middle-end/81698
15065 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15066 instead of computing it in the function. Formatting fix.
15067 (expand_case): Don't rely on default_edge being the first edge,
15068 clear it if removing it, pass default_edge to
15069 emit_case_dispatch_table.
15070 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15073 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15075 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15076 insn in the function, emit NOP after the insn.
15078 2017-08-06 Tom de Vries <tom@codesourcery.com>
15080 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15083 2017-08-06 Tom de Vries <tom@codesourcery.com>
15085 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15088 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15090 PR tree-optimization/57371
15091 * match.pd: New pattern.
15093 2017-08-04 Marek Polacek <polacek@redhat.com>
15095 PR middle-end/81695
15096 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15097 perform the computation in offset_int.
15099 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15101 PR tree-optimization/81136
15102 * tree-vectorizer.h: Include tree-hash-traits.h.
15103 (vec_base_alignments): New typedef.
15104 (vec_info): Add a base_alignments field.
15105 (vect_record_base_alignments): Declare.
15106 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15108 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15109 (create_data_ref): Add an is_conditional_in_stmt argument.
15110 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15111 the is_conditional_in_stmt field.
15112 (data_ref_loc): Add an is_conditional_in_stmt field.
15113 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15114 (find_data_references_in_stmt): Update call to create_data_ref.
15115 (graphite_find_data_references_in_stmt): Likewise.
15116 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15117 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15118 (vect_record_base_alignment): New function.
15119 (vect_record_base_alignments): Likewise.
15120 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15121 for nested statements even if we fail to compute a misalignment.
15122 Use pooled base alignments for unconditional references.
15123 (vect_find_same_alignment_drs): Compare base addresses instead
15125 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15126 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15128 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15130 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15131 Add an explicit name for the enum. Use auto_vec for slp_instances
15132 and grouped_stores.
15133 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15135 (_bb_vec_info): Add a constructor and destructor.
15136 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15137 (destroy_loop_vec_info): Delete.
15138 (vect_destroy_datarefs): Likewise.
15139 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15140 (vec_info::vec_info): New function.
15141 (vec_info::~vec_info): Likewise.
15142 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15143 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15144 destroy_loop_vec_info.
15145 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15146 (_loop_vec_info::_loop_vec_info): ...this.
15147 (destroy_loop_vec_info): Replace with...
15148 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15149 the stmt_vec_infos. Leave handling of vec_info information to its
15150 destructor. Remove explicit vector releases.
15151 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15152 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15153 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15154 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15155 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15156 (destroy_bb_vec_info): Replace with...
15157 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15158 information to its destructor.
15159 (vect_slp_analyze_bb_1): Use new and delete instead of
15160 new_bb_vec_info and destroy_bb_vec_info.
15161 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15164 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15166 * tree-data-ref.h (subscript): Add access_fn field.
15167 (data_dependence_relation): Add could_be_independent_p.
15168 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15169 (same_access_functions): Move to tree-data-ref.c.
15170 * tree-data-ref.c (ref_contains_union_access_p): New function.
15171 (access_fn_component_p): Likewise.
15172 (access_fn_components_comparable_p): Likewise.
15173 (dr_analyze_indices): Add a reference to access_fn_component_p.
15174 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15176 (constant_access_functions): Likewise.
15177 (add_other_self_distances): Likewise.
15178 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15179 (initialize_data_dependence_relation): Use XCNEW and remove
15180 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15181 of access functions that have the same type. Allow the
15182 subsequence to end with different bases in some circumstances.
15183 Record the chosen access functions in SUB_ACCESS_FN.
15184 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15185 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15186 (subscript_dependence_tester_1): Likewise dra and drb.
15187 (build_classic_dist_vector): Update calls accordingly.
15188 (subscript_dependence_tester): Likewise.
15189 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15190 DDR_COULD_BE_INDEPENDENT_P.
15191 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15192 comp_alias_ddrs instead of may_alias_ddrs.
15193 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15195 (vect_analyze_data_ref_dependence): Use it if
15196 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15197 distance vectors if that fails.
15198 (dependence_distance_ge_vf): New function.
15199 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15200 LOOP_VINFO_MAY_ALIAS_DDRS.
15202 2017-08-04 Richard Biener <rguenther@suse.de>
15204 PR middle-end/81705
15205 * fold-const.c (fold_binary_loc): Properly restrict
15206 minus_var0 && minus_var1 case when associating undefined overflow
15209 2017-08-04 Tom de Vries <tom@codesourcery.com>
15211 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15213 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15215 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15216 Don't start diagnostic messages with a capital letter.
15217 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15219 (rs6000_invalid_builtin): Likewise.
15220 (rs6000_trampoline_init): Likewise.
15222 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15225 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15226 after setting changeable df flags.
15228 2017-08-03 Richard Biener <rguenther@suse.de>
15230 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15231 up if the use is in USE - X.
15233 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15235 * toplev.c (dumpfile.h): New include.
15236 (internal_error_reentered): New static function. Use it...
15237 (internal_error_function): ...here to handle reentered internal_error.
15239 2017-08-03 Richard Biener <rguenther@suse.de>
15241 PR middle-end/81148
15242 * fold-const.c (split_tree): Add minus_var and minus_con
15243 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15244 here but always use minus_*.
15245 (associate_trees): Assert we never associate with MINUS_EXPR
15246 and NULL first operand. Do not recurse for PLUS_EXPR operands
15247 when associating as MINUS_EXPR either.
15248 (fold_binary_loc): Track minus_var and minus_con.
15250 2017-08-03 Tom de Vries <tom@codesourcery.com>
15253 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15254 ACCEL_COMPILER, apply finish_options on
15255 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15257 2017-08-03 Tom de Vries <tom@codesourcery.com>
15260 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15261 function_entry_patch_area_size > 0.
15263 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15266 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15267 instead of 10??LU, perform unit multiplication in wide_int,
15268 don't change alloc_object_size_limit if the limit is larger
15271 PR tree-optimization/81655
15272 PR tree-optimization/81588
15273 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15274 the case when ranges[i].low and high are 1 for unsigned type with
15277 PR middle-end/81052
15278 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15279 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15281 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15283 * tree-vrp.h: Add include guard.
15285 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15288 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15289 (ud2): New insn pattern.
15290 * config/i386/i386.c (ix86_expand_epilogue):
15291 For naked functions, generate ud2 instead of trap insn.
15293 2017-08-02 Marek Polacek <polacek@redhat.com>
15296 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15298 2017-08-02 Tom de Vries <tom@codesourcery.com>
15299 Cesar Philippidis <cesar@codesourcery.com>
15301 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15302 Add missing edge probabilities.
15304 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15306 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15307 Correct endianness.
15309 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15311 PR middle-end/79499
15312 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15313 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15314 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15315 prologue_seq sequences - if any.
15317 2017-08-02 Richard Biener <rguenther@suse.de>
15319 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15320 via vectors if supported, integer extracts via punning if supported
15321 or otherwise vector extracts.
15323 2017-08-02 Richard Biener <rguenther@suse.de>
15325 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15327 (bitmap_insert_into_set): ... this.
15329 2017-08-02 Richard Biener <rguenther@suse.de>
15331 PR tree-optimization/81633
15333 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15335 PR tree-optimization/71752
15336 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15338 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15340 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15341 (machine_function::call_ms2sysv_pad_out): Remove field.
15342 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15343 (ix86_compute_frame_layout): Likewise.
15345 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15348 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15349 attribute with interrupt attribute.
15351 2017-08-01 Andrew Pinski <apinski@cavium.com>
15353 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15354 BIT_INSERT_EXPR's operand 1
15355 to see if the types precision matches.
15357 2017-08-01 Martin Liska <mliska@suse.cz>
15359 PR middle-end/70140
15360 * builtins.c (expand_builtin_memcpy_args): Remove.
15361 (expand_builtin_memcpy): Call newly added function
15362 expand_builtin_memory_copy_args.
15363 (expand_builtin_memcpy_with_bounds): Likewise.
15364 (expand_builtin_mempcpy): Remove last argument.
15365 (expand_builtin_mempcpy_with_bounds): Likewise.
15366 (expand_builtin_memory_copy_args): New function created from
15367 expand_builtin_mempcpy_args with small modifications.
15368 (expand_builtin_mempcpy_args): Remove.
15369 (expand_builtin_stpcpy): Remove unused argument.
15370 (expand_builtin): Likewise.
15371 (expand_builtin_with_bounds): Likewise.
15373 2017-08-01 Martin Liska <mliska@suse.cz>
15376 Make mempcpy more optimal (PR middle-end/70140).
15378 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15381 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15382 __builtin_vec_cmpne verify both arguments are compatible vectors
15383 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15384 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15385 move computation of aligned to after checking the argument types.
15389 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15390 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15393 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15394 Doug Rupp <rupp@adacore.com>
15395 Olivier Hainque <hainque@adacore.com>
15397 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15398 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15400 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15401 for TARGET_OS_CPP_BUILTIN.
15402 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15403 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15405 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15406 passing required abi options to the assembler for EABI configurations.
15407 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15408 of .text.hot and .text.unlikely sections for kernel modules when
15409 using ARM style exceptions.
15410 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15411 options. Add EXTRA_CC1_SPEC.
15412 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15414 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15416 (ARM_TARGET2_DWARF_FORMAT): Define.
15417 * config/arm/t-vxworks: Adjust multilib control to removal of the
15418 Diab command line options.
15420 2017-08-01 Martin Liska <mliska@suse.cz>
15422 PR gcov-profile/81561
15423 * gcov.c (unblock): Make unblocking safe as we need to preserve
15424 index correspondence of blocks and block_lists.
15426 2017-08-01 Richard Biener <rguenther@suse.de>
15428 PR tree-optimization/81181
15429 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15430 (compute_antic): ... end of iteration here.
15432 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15434 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15435 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15436 (ftree-slp-vectorize): Likewise.
15437 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15438 can no longer be set independent of flag_tree_loop_vectorize.
15439 * omp-general.c (emp_max_vf): Likewise.
15440 * opts.c (enable_fdo_optimizations): Remove references to
15441 flag_tree_vectorize, these are now implicit.
15442 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15443 and leave it for the options machinery.
15445 2017-08-01 Martin Liska <mliska@suse.cz>
15447 PR middle-end/70140
15448 * builtins.c (expand_builtin_memcpy_args): Remove.
15449 (expand_builtin_memcpy): Call newly added function
15450 expand_builtin_memory_copy_args.
15451 (expand_builtin_memcpy_with_bounds): Likewise.
15452 (expand_builtin_mempcpy): Remove last argument.
15453 (expand_builtin_mempcpy_with_bounds): Likewise.
15454 (expand_builtin_memory_copy_args): New function created from
15455 expand_builtin_mempcpy_args with small modifications.
15456 (expand_builtin_mempcpy_args): Remove.
15457 (expand_builtin_stpcpy): Remove unused argument.
15458 (expand_builtin): Likewise.
15459 (expand_builtin_with_bounds): Likewise.
15461 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15464 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15465 print "ds:" only for immediates in generic address space.
15467 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15470 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15471 (ix86_function_ok_for_sibcall): Return false for naked functions.
15473 2017-08-01 Richard Biener <rguenther@suse.de>
15475 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15476 (compute_antic): Seed worklist with exit block predecessors.
15477 * cfganal.c (dfs_find_deadend): For a cycle return the source
15478 of the edge closing it.
15480 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15482 * config/aarch64/aarch64.c
15483 (aarch64_can_const_movi_rtx_p): Move 0 check.
15485 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15487 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15488 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15490 * match.pd: Ditto in address comparison pattern.
15492 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15494 PR tree-optimization/81627
15495 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15496 closed ssa form for store-store chain.
15498 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15500 PR tree-optimization/81620
15501 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15502 for store-store chain.
15504 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15506 PR tree-optimization/81588
15507 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15508 ranges[i].in_p, invert comparison code ccode. For >/>=,
15509 swap rhs1 and rhs2 and comparison code unconditionally,
15510 for </<= don't do that. Don't swap rhs1/rhs2 again if
15511 ranges[i].in_p, instead invert comparison code ccode if
15512 opcode or oe->rank is BIT_IOR_EXPR.
15515 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15516 a direct optab to conversion optab.
15517 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15518 with GET_MODE_INNER as last argument instead of optab_handler.
15519 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15520 vector extraction if possible and optab is available.
15521 * expr.c (store_constructor): Use convert_optab_handler instead
15522 of optab_handler. Use vector initialization from smaller
15523 vectors if possible and optab is available.
15524 * tree-vect-stmts.c (vectorizable_load): Likewise.
15525 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15526 now have two modes.
15527 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15528 of vec_init from half-sized vectors with the same element mode.
15529 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15530 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15531 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15532 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15533 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15534 after mode in gen_vec_extract* calls.
15535 (vec_extract<mode>): Renamed to ...
15536 (vec_extract<mode><ssescalarmodelower>): ... this.
15537 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15538 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15539 element mode after mode in gen_vec_init* calls.
15540 (VEC_INIT_HALF_MODE): New mode iterator.
15541 (vec_init<mode>): Renamed to ...
15542 (vec_init<mode><ssescalarmodelower>): ... this.
15543 (vec_init<mode><ssehalfvecmodelower>): New expander.
15544 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15545 (vec_extractv2sfsf): ... this.
15546 (vec_initv2sf): Renamed to ...
15547 (vec_initv2sfsf): ... this.
15548 (vec_extractv2si): Renamed to ...
15549 (vec_extractv2sisi): ... this.
15550 (vec_initv2si): Renamed to ...
15551 (vec_initv2sisi): ... this.
15552 (vec_extractv4hi): Renamed to ...
15553 (vec_extractv4hihi): ... this.
15554 (vec_initv4hi): Renamed to ...
15555 (vec_initv4hihi): ... this.
15556 (vec_extractv8qi): Renamed to ...
15557 (vec_extractv8qiqi): ... this.
15558 (vec_initv8qi): Renamed to ...
15559 (vec_initv8qiqi): ... this.
15560 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15561 (vec_init<mode>): Renamed to ...
15562 (vec_init<mode><VEC_base_l>): ... this.
15563 (vec_extract<mode>): Renamed to ...
15564 (vec_extract<mode><VEC_base_l>): ... this.
15565 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15566 (vec_initv2sfsf): ... this.
15567 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15568 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15569 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15570 element mode after mode in gen_vec_init* calls.
15571 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15572 (vec_init<mode><Vel>): ... this.
15573 (vec_extract<mode>): Renamed to ...
15574 (vec_extract<mode><Vel>): ... this.
15575 * config/aarch64/iterators.md (Vel): New mode attribute.
15576 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15577 Add element mode after mode in gen_vec_extract* calls.
15578 * config/s390/vector.md (non_vec_l): New mode attribute.
15579 (vec_extract<mode>): Renamed to ...
15580 (vec_extract<mode><non_vec_l>): ... this.
15581 (vec_init<mode>): Renamed to ...
15582 (vec_init<mode><non_vec_l>): ... this.
15583 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15584 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15586 * config/arm/iterators.md (V_elem_l): New mode attribute.
15587 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15588 (vec_extract<mode><V_elem_l>): ... this.
15589 (vec_extractv2di): Renamed to ...
15590 (vec_extractv2didi): ... this.
15591 (vec_init<mode>): Renamed to ...
15592 (vec_init<mode><V_elem_l>): ... this.
15593 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15594 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15595 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15596 Add element mode after gen_vec_extract* calls.
15597 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15598 (vec_init<mode><unitmode>): ... this.
15599 (vec_extract<mode>): Renamed to ...
15600 (vec_extract<mode><unitmode>): ... this.
15601 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15602 (vec_init<mode><unitmode>): ... this.
15603 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15604 (vec_initv2sfsf): ... this.
15605 (vec_extractv2sf): Renamed to ...
15606 (vec_extractv2sfsf): ... this.
15607 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15608 Add element mode after gen_vec_extract* calls.
15609 * config/mips/mips.md (unitmode): New mode iterator.
15610 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15611 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15612 * config/spu/spu.md (inner_l): New mode attribute.
15613 (vec_init<mode>): Renamed to ...
15614 (vec_init<mode><inner_l>): ... this.
15615 (vec_extract<mode>): Renamed to ...
15616 (vec_extract<mode><inner_l>): ... this.
15617 * config/sparc/sparc.md (veltmode): New mode iterator.
15618 (vec_init<VMALL:mode>): Renamed to ...
15619 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15620 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15621 (vec_initv2sisi): ... this.
15622 (vec_initv2sf): Renamed to ...
15623 (vec_initv2sfsf): ... this.
15624 (vec_extractv2sf): Renamed to ...
15625 (vec_extractv2sfsf): ... this.
15626 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15627 (vec_init<mode>): Renamed to ...
15628 (vec_init<mode><VEC_base_l>): ... this.
15629 (vec_extract<mode>): Renamed to ...
15630 (vec_extract<mode><VEC_base_l>): ... this.
15631 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15632 (vec_initv2sfsf): ... this.
15633 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15634 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15635 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15636 gen_vec_init* calls.
15638 2017-08-01 Richard Biener <rguenther@suse.de>
15640 PR tree-optimization/81297
15641 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15642 TREE_OVERFLOW from INTEGER_CSTs.
15644 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15646 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15648 2017-07-31 Carl Love <cel@us.ibm.com>
15650 * config/rs6000/rs6000-c: Add support for built-in functions
15651 vector signed char vec_xl_be (signed long long, signed char *);
15652 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15653 vector signed int vec_xl_be (signed long long, signed int *);
15654 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15655 vector signed long long vec_xl_be (signed long long, signed long long *);
15656 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15657 vector signed short vec_xl_be (signed long long, signed short *);
15658 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15659 vector double vec_xl_be (signed long long, double *);
15660 vector float vec_xl_be (signed long long, float *);
15661 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15662 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15663 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15665 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15666 (altivec_expand_builtin): Add switch statement to call
15667 altivec_expand_xl_be for each builtin.
15668 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15669 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15670 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15671 __builtin_vsx_le_be_v16qi.
15672 * doc/extend.texi: Update the built-in documentation file for the
15673 new built-in functions.
15675 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15678 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15680 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15682 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15684 * config.gcc: Add z14.
15685 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15686 CPU model numbers for z13s and z14.
15687 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15689 * config/s390/s390-opts.h (enum processor_type): Rename
15690 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15691 * config/s390/s390.c (processor_table): Add field for CPU name to
15692 be passed to Binutils.
15693 (s390_asm_output_machine_for_arch): Use the new field in
15694 processor_table for Binutils.
15695 (s390_expand_builtin): Replace arch12 with z14.
15696 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15697 (s390_get_sched_attrmask): Likewise.
15698 (s390_get_unit_mask): Likewise.
15699 * config/s390/s390.opt: Add z14 to processor_type enum.
15701 2017-07-31 Martin Jambor <mjambor@suse.cz>
15704 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15705 regardless of optimization level.
15707 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15708 Martin Liska <mliska@suse.cz>
15710 * predict.def: Remove old comment and adjust probability.
15711 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15712 PREDICT statements.
15714 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15717 * config/i386/i386.c (ix86_function_naked): New function.
15718 (ix86_can_use_return_insn_p): Return false for naked functions.
15719 (ix86_expand_prologue): Skip prologue for naked functions.
15720 (ix86_expand_epilogue): Skip epilogue for naked functions
15721 and emit trap instruction.
15722 (ix86_warn_func_return): New function.
15723 (ix86_attribute_table): Add "naked" attribute specification.
15724 (TARGET_WARN_FUNC_RETURN): Define.
15725 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15727 2017-07-31 Martin Liska <mliska@suse.cz>
15729 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15730 (dump_gimple_bb_header): Always dump BB info.
15731 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15733 2017-07-31 Martin Liska <mliska@suse.cz>
15736 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15737 also with current_function_decl non-null equality.
15739 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15742 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15743 change type to the element type, instead add eltype variable and
15744 use it where we are interested in the element type.
15746 PR tree-optimization/81603
15747 * ipa-polymorphic-call.c
15748 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15749 offset arithmetic in offset_int, bail out if the resulting bit offset
15750 doesn't fit into shwi.
15752 2017-07-31 Martin Liska <mliska@suse.cz>
15754 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15755 (gimplify_save_expr): Fix comment.
15757 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15760 * config/i386/i386.c (ix86_function_arg): Update arguments for
15762 (ix86_compute_frame_layout): Set the initial stack offset to
15763 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15764 INCOMING_FRAME_SP_OFFSET.
15765 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15766 stack before exception handler returns.
15767 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15768 the 'ERROR_CODE' for exception handler.
15770 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15772 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15773 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15774 (ASM_OUTPUT_REG_POP): Ditto.
15775 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15776 instead of asm_fprintf to output pure string.
15778 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15780 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15781 to imported_module_or_decl hook.
15782 (debug_nothing_tree_tree_tree_bool): Remove.
15783 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15784 * debug.c (do_nothing_debug_hooks): Use
15785 debug_nothing_tree_tree_tree_bool_bool instead of
15786 debug_nothing_tree_tree_tree_bool.
15787 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15788 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15789 * sdbout.c (sdb_debug_hooks): Likewise.
15790 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15791 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15793 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15794 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15795 attribute, don't add anything.
15797 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15799 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15800 (fold_build2_stat_loc): Likewise.
15801 (fold_build3_stat_loc): Likewise.
15802 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15803 (fold_build1_loc): Remove macro.
15804 (fold_build2_loc): Likewise.
15805 (fold_build3_loc): Likewise.
15807 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15809 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15810 (gimple_build_debug_bind_source_stat): Likewise.
15811 * gimple.h (gimple_build_debug_bind): Remove macro.
15812 (gimple_build_debug_bind_source): Likewise.
15814 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15816 * bitmap.c (bitmap_alloc): Adjust.
15817 (bitmap_gc_alloc): Likewise.
15818 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15820 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15822 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15823 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15824 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15825 (bitmap_gc_alloc_stat): Likewise.
15826 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15828 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15830 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15831 * rtl.h (shallow_copy_rtx): Remove macro.
15833 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15835 * emit-rtl.c (gen_raw_REG): Adjust.
15836 * gengenrtl.c (gendef): Likewise.
15837 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15838 * rtl.h (rtx_alloc): Remove macro.
15840 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15842 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15843 (build_tree_list_stat): Likewise.
15844 * tree.h (build_tree_list): Remove macro.
15845 (build_tree_list_vec): Likewise.
15847 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15849 * tree.c (make_vector_stat): Remove _stat from name.
15850 (build_vector_stat): Likewise.
15851 * tree.h (make_vector_stat): Remove macro.
15852 (build_vector_stat): Likewise.
15854 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15856 * tree.h (build_var_debug_value): Remove prototype.
15858 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15860 * tree.c (tree_cons_stat): Remove _stat from name.
15861 * tree.h (tree_cons): Remove macro.
15863 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15865 * tree.c (build_vl_exp_stat): Remove _stat from name.
15866 * tree.h (build_vl_exp): Remove macro.
15868 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15870 * tree.c (build_decl_stat): Remove _stat from name.
15871 * tree.h (build_decl): Remove macro.
15873 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15875 * gimple.c (gimple_build_with_ops_stat): Adjust.
15876 (gimple_alloc_stat): Remove _stat from name.
15877 * gimple.h (gimple_alloc): Remove macro.
15879 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15881 * tree.c (make_tree_vec_stat): Remove _stat from name.
15882 (grow_tree_vec_stat): Likewise.
15883 * tree.h (make_tree_vec_stat): Adjust prototype.
15884 (grow_tree_vec_stat): Likewise.
15885 (make_tree_vec): Remove macro.
15886 (grow_tree_vec): Likewise.
15888 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15890 * fold-const.c (fold_build1_stat_loc): Adjust.
15891 (fold_build2_stat_loc): Likewise.
15892 (fold_build3_stat_loc): Likewise.
15893 * tree.c (build0_stat): Remove _stat from name.
15894 (build1_stat): Likewise.
15895 (build2_stat): Likewise.
15896 (build3_stat): Likewise.
15897 (build4_stat): Likewise.
15898 (build5_stat): Likewise.
15899 * tree.h (build1_loc): Remove macro, and rename _stat function
15901 (build2_loc): Likewise.
15902 (build3_loc): Likewise.
15903 (build4_loc): Likewise.
15904 (build5_loc): Likewise.
15906 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15908 * tree.c (make_int_cst_stat): Remove _stat from name.
15909 * tree.h (make_int_cst_stat): Adjust prototype.
15910 (make_int_cst): Remove macro.
15912 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15914 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15915 * tree.h (make_tree_binfo_stat): Adjust prototype.
15916 (make_tree_binfo): Remove.
15918 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15920 * tree.c (copy_node_stat): Rename to copy_node.
15921 (build_distinct_type_copy): Adjust.
15922 * tree.h (copy_node_stat): Adjust prototype.
15923 (copy_node): Remove macro.
15925 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15927 * tree.c (make_node_stat): rename to make_node.
15928 (build_tree_list_stat): Adjust.
15929 (build0_stat): Likewise.
15930 (build2_stat): Likewise.
15931 (build3_stat): Likewise.
15932 (build4_stat): Likewise.
15933 (build5_stat): Likewise.
15934 (build_decl_stat): Likewise.
15935 * tree.h (make_node_stat): Adjust prototype.
15936 (make_node): remove macro.
15938 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15940 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15941 (PPC_FEATURE2_SCV): Likewise.
15942 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15944 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15946 * config/aarch64/aarch64.c
15947 (aarch64_internal_mov_immediate): Add new special pattern.
15948 * config/aarch64/aarch64.md (*movdi_aarch64):
15949 Add reg/32bit const mov case.
15951 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15952 Richard Sandiford <richard.sandiford@linaro.org>
15954 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15955 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15956 Add integer and movi cases.
15957 (movi-split-hf-df-sf split, fp16): New.
15958 (enabled): Added TARGET_FP_F16INST.
15959 * config/aarch64/iterators.md (GPF_HF): New.
15960 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15962 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15964 * config/aarch64/aarch64.c
15965 (aarch64_simd_container_mode): Add prototype.
15966 (aarch64_expand_mov_immediate): Add HI support.
15967 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15968 (aarch64_can_const_movi_rtx_p): New.
15969 (aarch64_preferred_reload_class):
15970 Remove restrictions of using FP registers for certain SIMD operations.
15971 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15972 (aarch64_valid_floating_const): Add integer move validation.
15973 (aarch64_simd_imm_scalar_p): Remove.
15974 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15975 (aarch64_legitimate_constant_p): Expand list of supported cases.
15976 * config/aarch64/aarch64-protos.h
15977 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15978 (aarch64_reinterpret_float_as_int): New.
15979 (aarch64_simd_imm_scalar_p): Remove.
15980 * config/aarch64/constraints.md (Uvi): New.
15981 (Dd): Split into Ds and new Dd.
15982 * config/aarch64/aarch64.md (*movsi_aarch64):
15984 (*movdi_aarch64): Add SIMD mov case.
15986 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15988 * tree-predcom.c: (struct chain): Handle store-store chain in which
15989 stores for elimination only store loop invariant values.
15990 (execute_pred_commoning_chain): Ditto.
15991 (prepare_initializers_chain_store_elim): Ditto.
15992 (prepare_finalizers): Ditto.
15993 (is_inv_store_elimination_chain): New function.
15994 (initialize_root_vars_store_elim_1): New function.
15996 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15998 * tree-predcom.c: Revise general description of the pass.
15999 (enum chain_type): New enum type for store elimination.
16000 (struct chain): New field supporting store elimination.
16001 (struct component): Ditto.
16002 (dump_chain): Dump store-stores chain.
16003 (release_chain): Release resources.
16004 (split_data_refs_to_components): Compute and create component
16005 contains only stores for elimination.
16006 (get_chain_last_ref_at): New function.
16007 (make_invariant_chain): Initialization.
16008 (make_rooted_chain): Specify chain type in parameter and record it.
16009 (add_looparound_copies): Skip for store-stores chain.
16010 (determine_roots_comp): Compute type of chain and pass it to
16012 (initialize_root_vars_store_elim_2): New function.
16013 (finalize_eliminated_stores): New function.
16014 (remove_stmt): Handle store for elimination.
16015 (execute_pred_commoning_chain): Execute predictive commoning on
16016 store-store chains.
16017 (determine_unroll_factor): Skip unroll for store-stores chain.
16018 (prepare_initializers_chain_store_elim): New function.
16019 (prepare_initializers_chain): Hanlde store-store chain.
16020 (prepare_finalizers_chain, prepare_finalizers): New function.
16021 (tree_predictive_commoning_loop): Return integer value indicating
16022 if loop is unrolled or lcssa form is corrupted.
16023 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16025 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16027 * tree-predcom.c (initialize_root): Delete.
16028 (execute_pred_commoning_chain): Initialize root vars and replace
16029 reference of non-combined chain directly, rather than call above
16032 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16034 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16035 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16037 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16039 * tree-predcom.c (struct chain): New field init_seq.
16040 (release_chain): Release init_seq.
16041 (prepare_initializers_chain): Record intialization stmts in above
16043 (insert_init_seqs): New function.
16044 (tree_predictive_commoning_loop): Call insert_init_seqs.
16046 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16048 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16050 2017-07-28 Richard Biener <rguenther@suse.de>
16052 * match.pd: Remove superfluous :c.
16053 * genmatch.c (simplify::id): Add member.
16054 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16056 (current_id): New global.
16057 (dt_node::parent): Move from ...
16058 (dt_operand::parent): ... here. Add for_id member.
16059 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16060 (decision_tree::find_node): Relax order requirement when
16061 merging DT_TRUE nodes to ones inbetween the current simplify
16062 and the one we try to merge with. Add diagnostic whenever
16063 we need to enforce pattern order by not merging.
16064 (decision_tree::insert): Set current_id.
16065 (decision_tree::print_node): Dump parent node and for_id.
16066 (parser::last_id): Add member.
16067 (parser::push_simplify): Assign unique id.
16068 (parser::parser): Initialize last_id.
16070 2017-07-28 Martin Liska <mliska@suse.cz>
16073 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16074 gimple_build_debug_bind.
16076 2017-07-28 Richard Biener <rguenther@suse.de>
16078 PR tree-optimization/81502
16079 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16081 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16083 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16084 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16085 for BIT_FIELD_REF args.
16086 * fold-const.c (make_bit_field_ref): Likewise.
16087 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16089 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16092 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16093 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16094 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16095 Or it into SANITIZER_UNDEFINED.
16096 * ubsan.c: Include gimple-fold.h and varasm.h.
16097 (ubsan_expand_ptr_ifn): New function.
16098 (instrument_pointer_overflow): New function.
16099 (maybe_instrument_pointer_overflow): New function.
16100 (instrument_object_size): Formatting fix.
16101 (pass_ubsan::execute): Call instrument_pointer_overflow
16102 and maybe_instrument_pointer_overflow.
16103 * internal-fn.c (expand_UBSAN_PTR): New function.
16104 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16105 * sanitizer.def (__ubsan_handle_pointer_overflow,
16106 __ubsan_handle_pointer_overflow_abort): New builtins.
16107 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16108 * internal-fn.def (UBSAN_PTR): New internal function.
16109 * opts.c (sanitizer_opts): Add pointer-overflow.
16110 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16111 * fold-const.c (build_range_check): Compute pointer range check in
16112 integral type if pointer arithmetics would be needed. Formatting
16115 2017-07-28 Martin Liska <mliska@suse.cz>
16118 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16119 parameters that are of a variable-length.
16121 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16123 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16125 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16126 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16127 (CRT_CALL_STATIC_FUNCTION): Likewise.
16128 (ASM_DEFAULT_SPEC): New define.
16129 (ASM_SPEC32): Likewise.
16130 (ASM_SPEC64): Likewise.
16131 (ASM_SPEC_COMMON): Likewise.
16132 (ASM_SPEC): Likewise.
16133 (INVALID_64BIT): Likewise.
16134 (LINK_OS_DEFAULT_SPEC): Likewise.
16135 (LINK_OS_SPEC32): Likewise.
16136 (LINK_OS_SPEC64): Likewise.
16137 (POWERPC_LINUX): Likewise.
16138 (PTRDIFF_TYPE): Likewise.
16139 (RESTORE_FP_PREFIX): Likewise.
16140 (RESTORE_FP_SUFFIX): Likewise.
16141 (SAVE_FP_PREFIX): Likewise.
16142 (SAVE_FP_SUFFIX): Likewise.
16143 (SIZE_TYPE): Likewise.
16144 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16145 (TARGET_64BIT): Likewise.
16146 (TARGET_64BIT): Likewise.
16147 (TARGET_AIX): Likewise.
16148 (WCHAR_TYPE_SIZE): Likewise.
16149 (WCHAR_TYPE): Undefine.
16150 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16151 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16152 (CPP_OS_RTEMS_SPEC): Delete.
16153 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16154 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16156 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16158 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16160 PR tree-optimization/81578
16161 * tree-parloops.c (build_new_reduction): Bail out if
16162 reduction_code isn't one of the standard OpenMP reductions.
16163 Move the details printing after that decision.
16165 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16167 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16168 related to reload_in_progress.
16169 (splat_input_operand): Likewise.
16170 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16172 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16174 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16175 (TARGET_INSTANTIATE_DECLS): Likewise.
16176 (legitimate_indexed_address_p): Delete reload_in_progress code.
16177 (rs6000_debug_legitimate_address_p): Likewise.
16178 (rs6000_eliminate_indexed_memrefs): Likewise.
16179 (rs6000_emit_le_vsx_store): Likewise.
16180 (rs6000_emit_move_si_sf_subreg): Likewise.
16181 (rs6000_emit_move): Likewise.
16182 (register_to_reg_type): Likewise.
16183 (rs6000_pre_atomic_barrier): Likewise.
16184 (rs6000_machopic_legitimize_pic_address): Likewise.
16185 (rs6000_allocate_stack_temp): Likewise.
16186 (rs6000_address_for_fpconvert): Likewise.
16187 (rs6000_address_for_altivec): Likewise.
16188 (rs6000_secondary_memory_needed_rtx): Delete function.
16189 (rs6000_check_sdmode): Likewise.
16190 (rs6000_alloc_sdmode_stack_slot): Likewise.
16191 (rs6000_instantiate_decls): Likewise.
16192 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16193 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16194 Delete reload_in_progress.
16195 (*vec_reload_and_plus_<mptrsize>): Likewise.
16196 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16197 (vsx_div_v2di): Likewise.
16198 (vsx_udiv_v2di): Likewise.
16200 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16202 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16203 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16204 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16205 (rs6000_debug_reg_global): Delete print of LRA status.
16206 (rs6000_option_override_internal): Delete dead LRA related code.
16207 (rs6000_lra_p): Delete function.
16208 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16210 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16212 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16213 * config/riscv/rtems.h: New file.
16215 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16216 Sudakshina Das <sudi.das@arm.com>
16218 * config/aarch64/aarch64.md
16219 (define_split for and<mode>3nr_compare): Move
16220 non aarch64_logical_operand to a register.
16221 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16222 register immediate operand to a register.
16223 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16225 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16227 PR middle-end/81564
16228 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16230 2017-07-27 Richard Biener <rguenther@suse.de>
16232 PR tree-optimization/81573
16233 PR tree-optimization/81494
16234 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16235 multi defuse cycle case.
16237 2017-07-27 Richard Biener <rguenther@suse.de>
16239 PR tree-optimization/81571
16240 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16243 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16245 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16246 earlier and only if MASK_FPU is set. Adjust formatting.
16248 2017-07-27 Martin Liska <mliska@suse.cz>
16250 * opt-functions.awk: Add validation of value of Init.
16251 * optc-gen.awk: Pass new argument.
16253 2017-07-27 Martin Liska <mliska@suse.cz>
16255 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16256 Fix wrong condition.
16258 2017-07-27 Martin Liska <mliska@suse.cz>
16260 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16261 BBs and edges seen by autoFDO.
16263 2017-07-27 Richard Biener <rguenther@suse.de>
16265 PR tree-optimization/81502
16266 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16267 with incompatible but same sized type.
16268 (execute_update_addresses_taken): Likewise.
16270 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16272 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16273 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16275 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16278 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16279 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16280 Change s_operand to memory_operand.
16282 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16284 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16285 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16286 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16287 Emit instructions rather than returning an expression. Handle TFmode
16288 and KFmode by casting to TImode.
16289 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16290 (rs6000_emit_le_vsx_store): Likewise.
16291 * config/rs6000/vsx.md (VSX_TI): New iterator.
16292 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16293 (*vsx_le_undo_permute_<mode>): Likewise.
16294 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16295 emit the split sequence.
16296 (*vsx_le_perm_store_<mode>): Likewise.
16298 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16300 PR tree-optimization/81555
16301 PR tree-optimization/81556
16302 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16303 if true, force CHANGED for the recursive invocation.
16304 (reassociate_bb): Remember original length of ops array, pass
16305 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16307 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16308 attributes for noipa attribute. For naked attribute use
16309 lookup_attribute first before lookup_attribute_spec.
16310 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16312 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16313 for functions with noipa attribute.
16314 (cgraph_externally_visible_p): Return true for functions with noipa
16316 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16317 for functions with noipa attribute.
16318 * doc/extend.texi: Document noipa function attribute.
16319 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16320 also for functions with noipa attribute.
16321 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16323 2017-07-26 Andrew Pinski <apinski@cavium.com>
16325 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16326 vec_unalign_load_cost and vec_unalign_store_cost.
16328 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16330 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16331 -mvsx-small-integer option.
16332 (ISA_3_0_MASKS_IEEE): Likewise.
16333 (OTHER_VSX_VECTOR_MASKS): Likewise.
16334 (POWERPC_MASKS): Likewise.
16335 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16336 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16337 code, only testing for DImode being allowed in non-VSX floating
16339 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16340 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16342 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16343 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16344 TARGET_P8_VECTOR test.
16345 (rs6000_secondary_reload_simple_move): Likewise.
16346 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16347 since TARGET_P9_VECTOR was already tested.
16348 (rs6000_opt_masks): Remove -mvsx-small-integer.
16349 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16350 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16352 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16353 test for TARGET_VEXTRACTUB was used, and that uses
16355 (p9 extract splitter): Likewise.
16356 (vsx_extract_<mode>_di_p9): Likewise.
16357 (vsx_extract_<mode>_store_p9): Likewise.
16358 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16359 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16360 the elimination of TARGET_VSX_SMALL_INTEGER.
16361 (vsx_extract_<mode>_p8): Likewise.
16362 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16363 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16364 (vsx_set_<mode>_p9): Likewise.
16365 (vsx_set_v4sf_p9): Likewise.
16366 (vsx_set_v4sf_p9_zero): Likewise.
16367 (vsx_insert_extract_v4sf_p9): Likewise.
16368 (vsx_insert_extract_v4sf_p9_2): Likewise.
16369 * config/rs6000/rs6000.md (sign extend splitter): Change
16370 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16371 (floatsi<mode>2_lfiwax_mem): Likewise.
16372 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16373 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16374 since a test for TARGET_P9_VECTOR was used.
16375 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16376 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16377 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16378 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16379 TARGET_P8_VECTOR test.
16380 (fix_trunc<mode>si2_stfiwx): Likewise.
16381 (fix_trunc<mode>si2_internal): Likewise.
16382 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16383 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16385 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16386 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16387 TARGET_P8_VECTOR test.
16388 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16389 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16390 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16392 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16393 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16394 since a test for TARGET_P9_VECTOR was used.
16395 (splitter for loading small constants): Likewise.
16397 2017-07-26 Andrew Pinski <apinski@cavium.com>
16399 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16402 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16405 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16406 (fp_valid_at): Likewise.
16408 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16410 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16411 (qdf24xx_addrcost_table): Likewise.
16412 (cortexa57_tunings): Update to use generic_branch_cost.
16413 (cortexa72_tunings): Likewise.
16414 (cortexa73_tunings): Likewise.
16415 (qdf24xx_tunings): Likewise.
16417 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16419 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16420 (thunderx2t99_branch_cost): Likewise.
16421 (cortexa35_tunings): Update to use generic_branch_cost.
16422 (cortexa53_tunings): Likewise.
16423 (cortexa57_tunings): Likewise.
16424 (cortexa72_tunings): Likewise.
16425 (cortexa73_tunings): Likewise.
16426 (thunderx2t99_tunings): Likewise.
16428 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16430 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16431 (sparc_option_override): Honour MASK_FSMULD.
16432 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16433 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16434 * config/sparc/sparc.opt (mfsmuld): New option.
16435 * doc/invoke.texi (mfsmuld): Document option.
16437 2017-07-26 Marek Polacek <polacek@redhat.com>
16439 PR middle-end/70992
16440 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16442 2017-07-26 Richard Biener <rguenther@suse.de>
16444 * gimple-match-head.c (do_valueize): Return OP if valueize
16446 (get_def): New helper to get at the def stmt of a SSA name
16447 if valueize allows.
16448 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16449 do_valueize to get at the def stmt.
16450 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16452 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16454 PR middle-end/46932
16455 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16457 2017-07-26 Martin Liska <mliska@suse.cz>
16460 * function.c (expand_function_start): Make expansion of
16461 nonlocal_goto_save_area after parm_birth_insn.
16463 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16465 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16466 from all CPU target flags enable members.
16468 2017-07-26 Richard Biener <rguenther@suse.de>
16470 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16471 (decision_tree::gen): Make 'type' const.
16472 (write_predicate): Likewise.
16474 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16476 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16477 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16478 (rs6000_option_override_internal): Likewise.
16479 (rs6000_expand_vector_set): Likewise.
16480 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16481 (TARGET_UPPER_REGS_SF): Likewise.
16482 (TARGET_UPPER_REGS_DI): Likewise.
16483 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16484 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16485 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16486 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16487 (Splitters for DI constants in Altivec registers): Likewise.
16488 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16489 (vsx_set_v4sf_p9): Likewise.
16490 (vsx_set_v4sf_p9_zero): Likewise.
16491 (vsx_insert_extract_v4sf_p9): Likewise.
16492 (vsx_insert_extract_v4sf_p9_2): Likewise.
16494 2017-07-25 Carl Love <cel@us.ibm.com>
16496 * doc/extend.texi: Update the built-in documentation file for the
16497 existing built-in functions
16498 vector signed char vec_cnttz (vector signed char);
16499 vector unsigned char vec_cnttz (vector unsigned char);
16500 vector signed short vec_cnttz (vector signed short);
16501 vector unsigned short vec_cnttz (vector unsigned short);
16502 vector signed int vec_cnttz (vector signed int);
16503 vector unsigned int vec_cnttz (vector unsigned int);
16504 vector signed long long vec_cnttz (vector signed long long);
16505 vector unsigned long long vec_cnttz (vector unsigned long long);
16507 2017-07-25 Andrew Pinski <apinski@cavium.com>
16509 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16510 accesses where the use is for the first operand of a BIT_INSERT.
16512 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16515 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16516 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16518 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16520 * config/i386/gstabs.h: Delete.
16521 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16523 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16525 * config/i386/i386.c (ix86_decompose_address): Do not check for
16526 register RTX when looking at index_reg or base_reg.
16527 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16529 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16531 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16532 to update EH info here.
16534 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16536 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16538 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16540 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16542 2017-07-25 Torsten Duwe <duwe@suse.de>
16544 * common.opt: Introduce -fpatchable-function-entry
16545 command line option, and its variables function_entry_patch_area_size
16546 and function_entry_patch_area_start.
16547 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16548 including a two-value parser.
16549 * target.def (print_patchable_function_entry): New target hook.
16550 * targhooks.h (default_print_patchable_function_entry): New function.
16551 * targhooks.c (default_print_patchable_function_entry): Likewise.
16552 * toplev.c (process_options): Switch off IPA-RA if
16553 patchable function entries are being generated.
16554 * varasm.c (assemble_start_function): Look at the
16555 patchable-function-entry command line switch and current
16556 function attributes and maybe generate NOP instructions by
16557 calling the print_patchable_function_entry hook.
16558 * doc/extend.texi: Document patchable_function_entry attribute.
16559 * doc/invoke.texi: Document -fpatchable_function_entry
16560 command line option.
16561 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16563 * doc/tm.texi: Re-generate.
16565 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16568 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16569 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16571 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16573 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16575 2017-07-25 Richard Biener <rguenther@suse.de>
16577 PR tree-optimization/81455
16578 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16579 not walk in cycles when looking for guards.
16581 2017-07-25 Richard Biener <rguenther@suse.de>
16583 PR tree-optimization/81529
16584 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16585 when optimizing backedge uses.
16587 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16589 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16591 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16592 to dl_section_ref. On AIX, append an expression to subtract
16593 the size of the section length to dl_section_ref.
16595 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16597 * configure.ac: If any of the config.* scripts fail, exit 1.
16598 * configure: Regenerate.
16600 2017-07-25 Richard Biener <rguenther@suse.de>
16602 PR middle-end/81546
16603 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16604 of immediate uses, be more verbose on errors.
16606 2017-07-25 Richard Biener <rguenther@suse.de>
16608 PR tree-optimization/81510
16609 * tree-vect-loop.c (vect_is_simple_reduction): When the
16610 reduction stmt is not inside the loop bail out.
16612 2017-07-25 Richard Biener <rguenther@suse.de>
16614 PR tree-optimization/81303
16615 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16616 profitability check against LOOP_VINFO_NITERSM1.
16618 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16620 * domwalk.c (cmp_bb_postorder): Simplify.
16621 (sort_bbs_postorder): New function. Use it...
16622 (dom_walker::walk): ...here to optimize common cases.
16624 2017-07-25 Martin Liska <mliska@suse.cz>
16627 * ipa-visibility.c (function_and_variable_visibility): Make the
16628 redirection just on target that supports aliasing.
16629 Fix GNU coding style.
16631 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16634 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16636 * config/arm/rtems.h: Likewise.
16637 * config/bfin/rtems.h: Likewise.
16638 * config/i386/rtemself.h: Likewise.
16639 * config/lm32/rtems.h: Likewise.
16640 * config/m32c/rtems.h: Likewise.
16641 * config/m68k/rtemself.h: Likewise.
16642 * config/microblaze/rtems.h: Likewise.
16643 * config/mips/rtems.h: Likewise.
16644 * config/moxie/rtems.h: Likewise.
16645 * config/nios2/rtems.h: Likewise.
16646 * config/powerpcspe/rtems.h: Likewise.
16647 * config/rs6000/rtems.h: Likewise.
16648 * config/rtems.h: Likewise.
16649 * config/sh/rtems.h: Likewise.
16650 * config/sh/rtemself.h: Likewise.
16651 * config/sparc/rtemself.h: Likewise.
16653 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16656 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16657 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16658 * tree-ssa-structalias.c (alias_get_name): Same.
16660 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16663 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16664 instructions if no du chain is found.
16666 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16668 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16670 2017-07-25 Richard Biener <rguenther@suse.de>
16672 PR middle-end/81505
16673 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16676 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16678 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16679 upper-regs options.
16680 (ISA_2_7_MASKS_SERVER): Likewise.
16681 (ISA_3_0_MASKS_IEEE): Likewise.
16682 (OTHER_P8_VECTOR_MASKS): Likewise.
16683 (OTHER_VSX_VECTOR_MASKS): Likewise.
16684 (POWERPC_MASKS): Likewise.
16685 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16686 duplicate list of options.
16687 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16688 explicit -mupper-regs options.
16689 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16690 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16691 alias for -mupper-regs-df.
16692 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16693 (rs6000_init_hard_regno_mode_ok): Likewise.
16694 (rs6000_option_override_internal): Likewise.
16695 (rs6000_opt_masks): Likewise.
16696 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16697 options in terms of whether -mvsx or -mpower8-vector was used.
16698 (TARGET_UPPER_REGS_DI): Likewise.
16699 (TARGET_UPPER_REGS_SF): Likewise.
16700 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16701 -mupper-regs-* options.
16703 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16705 * passes.c (emergency_dump_function): Print some empty lines and a
16706 header before the RTL dump.
16708 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16710 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16712 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16715 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16716 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16718 2017-07-24 Carl Love <cel@us.ibm.com>
16720 * config/rs6000/rs6000-c.c: Add support for built-in functions
16721 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16722 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16723 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16724 vec_extract_fp_from_shortl): Add defines for the two builtins.
16725 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16726 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16728 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16729 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16730 * doc/extend.texi: Update the built-in documentation file for the
16731 new built-in function.
16733 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16736 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16738 * doc/generic.texi: Likewise.
16739 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16740 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16742 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16744 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16745 (aarch64_mls_elt_merge<mode>): Likewise.
16747 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16749 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16750 having __cxa_atexit.
16752 2017-07-23 Michael Collison <michael.collison@arm.com>
16754 * config/arm/arm.c (arm_option_override): Deprecate
16755 use of -mstructure-size-boundary.
16756 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16757 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16759 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16762 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16763 Reduce cost estimate for direct moves.
16765 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16768 * config/i386/i386.c (ix86_option_override_internal): Disable
16769 BMI, BMI2 and TBM instructions for -m16.
16771 2017-07-21 Carl Love <cel@us.ibm.com>
16773 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16774 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16775 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16776 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16777 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16778 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16779 VMULOSW): New enum "unspec" values.
16780 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16781 altivec_vmulosw): New patterns.
16782 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16783 VMULOSW): Add definitions.
16785 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16787 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16788 (qdf24xx): Likewise.
16789 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16790 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16791 (AARCH64_FL_V8_1): Renumber.
16792 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16793 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16794 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16795 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16796 rdma to feature modifiers list.
16798 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16800 PR middle-end/56727
16801 * ipa-visibility (function_and_variable_visibility): Convert
16802 recursive PLT call to direct call if appropriate.
16804 2017-07-21 Andrew Pinski <apinski@cavium.com>
16806 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16807 operand 1 to see if the types precision matches.
16808 * fold-const.c (operand_equal_p): Likewise.
16810 2017-07-21 Richard Biener <rguenther@suse.de>
16812 PR tree-optimization/81303
16813 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16814 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16815 (vect_peeling_hash_get_lowest_cost): Adjust.
16816 (vect_enhance_data_refs_alignment): Likewise. Use
16817 vect_get_peeling_costs_all_drs to compute the penalty for no
16818 peeling to match up costs.
16820 2017-07-21 Richard Biener <rguenther@suse.de>
16822 PR tree-optimization/81500
16823 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16824 we didn't identify a reduction path.
16826 2017-07-21 Tom de Vries <tom@codesourcery.com>
16827 Cesar Philippidis <cesar@codesourcery.com>
16829 PR gcov-profile/81442
16830 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16833 2017-07-21 Tom de Vries <tom@codesourcery.com>
16836 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16838 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16839 nvptx_override_options_after_change.
16841 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16843 * dwarf2out.c (output_file_names): Avoid double testing for
16844 dwarf_version >= 5.
16846 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16848 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16850 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16852 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16854 (try_crossjump_to_edge): Do not punt on partitioned functions.
16856 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16858 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16859 Put all BBs reachable only via paths crossing cold region to cold
16861 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16863 2016-07-21 Richard Biener <rguenther@suse.de>
16865 PR tree-optimization/81303
16866 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16867 into account prologue and epilogue iterations when raising
16868 min_profitable_iters to sth at least covering one vector iteration.
16870 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16872 * config/arm/arm.c (arm_test_cpu_arch_dat):
16875 2017-07-20 Nathan Sidwell <nathan@acm.org>
16877 Remove TYPE_METHODS.
16878 * tree.h (TYPE_METHODS): Delete.
16879 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16880 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16881 (dbxout_type_methods): Scan TYPE_FIELDS.
16882 (dbxout_type): Don't check TYPE_METHODS here.
16883 * function.c (use_register_for_decl): Always ignore register for
16884 class types when not optimizing.
16885 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16886 * tree.c (free_lang_data_in_type): Stitch out member functions and
16887 templates from TYPE_FIELDS.
16888 (build_distinct_type_copy, verify_type_variant,
16889 verify_type): Member fns are on TYPE_FIELDS.
16890 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16891 * tree-pretty-print.c (dump_generic_node): Likewise.
16893 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16896 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16897 V2TImode and V4TImode.
16898 (ix86_expand_vector_extract): Likewise.
16899 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16900 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16901 (ssescalarmode): Handle V4TImode and V2TImode.
16902 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16903 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16904 (VEXTRACTI128_MODE): New mode iterator.
16905 (splitter for *vec_extractv?ti first element): New.
16906 (VEC_INIT_MODE): New mode iterator.
16907 (vec_init<mode>): Consolidate 3 expanders into one using
16908 VEC_INIT_MODE mode iterator.
16910 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16912 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16913 non_spilled_static_chain_regno_p.
16915 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16917 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16919 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16921 * bb-reorder.c (connect_traces): Allow copying of blocks within
16924 2017-07-20 Richard Biener <rguenther@suse.de>
16926 * gimple.h (gimple_phi_result): Add gphi * overload.
16927 (gimple_phi_result_ptr): Likewise.
16928 (gimple_phi_arg): Likewise. Adjust index assert to only
16929 allow actual argument accesses rather than all slots available
16931 (gimple_phi_arg_def): Add gphi * overload.
16932 * tree-phinodes.c (make_phi_node): Initialize only actual
16934 (resize_phi_node): Clear memory not covered by old node,
16935 do not initialize excess argument slots.
16936 (reserve_phi_args_for_new_edge): Initialize new argument slot
16939 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16941 PR tree-optimization/81388
16943 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16945 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16946 by removing computation of may_be_zero.
16948 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16949 Tom de Vries <tom@codesourcery.com>
16951 PR middle-end/81030
16952 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16953 when gimple level profile disagrees with what RTL expander did.
16955 2017-07-20 Richard Biener <rguenther@suse.de>
16957 PR tree-optimization/61171
16958 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16959 (vect_analyze_stmt): Add slp instance parameter.
16960 (vectorizable_reduction): Likewise.
16961 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16962 (vect_is_simple_reduction): Deal with chains not detected
16963 as SLP reduction chain, specifically not properly associated
16964 chains containing a mix of plus/minus.
16965 (get_reduction_op): Remove.
16966 (get_initial_defs_for_reduction): Simplify, pass in whether
16967 this is a reduction chain, pass in the SLP node for the PHIs.
16968 (vect_create_epilog_for_reduction): Get the SLP instance as
16970 (vectorizable_reduction): Get the SLP instance as arg.
16971 During analysis remember the SLP node with the PHIs in the
16972 instance. Simplify getting at the vectorized reduction PHIs.
16973 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16974 through SLP instance.
16975 (vect_slp_analyze_operations): Likewise.
16976 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16977 (vect_transform_stmt): Likewise.
16979 2017-07-20 Tom de Vries <tom@codesourcery.com>
16981 PR tree-optimization/81489
16982 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16983 read of phi arg location to before loop that modifies phi.
16985 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16987 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16990 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16992 PR middle-end/81331
16993 * except.c (execute): Fix ordering issue.
16995 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16997 PR rtl-optimization/81423
16998 * combine.c (make_compound_operation_int): Don't try to optimize
16999 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17001 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17003 PR rtl-optimization/81423
17004 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17005 with a constant that is -1 in the truncated to mode.
17007 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17009 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17010 (determine_unlikely_bbs): ... here.
17011 * predict.h (propagate_unlikely_bbs_forward): Declare.
17012 * cfgexpand.c (pass_expand::execute): Use it.
17013 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17015 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17016 propagate_unlikely_bbs_forward.
17018 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17020 PR middle-end/81331
17021 * except.c (maybe_add_nop_after_section_switch): New function.
17024 2017-07-19 Tom de Vries <tom@codesourcery.com>
17026 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17028 2017-07-19 Tom de Vries <tom@codesourcery.com>
17030 * gimple.h (gimple_phi_arg): Make assert more strict.
17032 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17034 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17035 * config/rs6000/mmintrin.h: New file.
17036 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17038 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17040 PR tree-optimization/81346
17041 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17043 2017-07-19 Tom de Vries <tom@codesourcery.com>
17045 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17047 2017-07-19 Tom de Vries <tom@codesourcery.com>
17049 * config/nvptx/nvptx-modes.def: Add V2DImode.
17050 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17051 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17052 (nvptx_output_mov_insn): Handle lack of mov.b128.
17053 (nvptx_print_operand): Handle 'H' and 'L' codes.
17054 (nvptx_vector_mode_supported): Allow V2DImode.
17055 (nvptx_preferred_simd_mode): New function.
17056 (nvptx_data_alignment): New function.
17057 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17058 nvptx_preferred_simd_mode.
17059 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17061 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17063 2017-07-19 Tom de Vries <tom@codesourcery.com>
17065 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17066 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17067 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17068 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17069 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17070 (mov<VECIM>_insn): New define_insn.
17071 (define_expand "mov<VECIM>): New define_expand.
17073 2017-07-19 Tom de Vries <tom@codesourcery.com>
17075 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17077 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17079 PR tree-optimization/81346
17080 * fold-const.h (fold_div_compare, range_check_type): Declare.
17081 * fold-const.c (range_check_type): New function.
17082 (build_range_check): Use range_check_type.
17083 (fold_div_compare): No longer static, rewritten into
17084 a match.pd helper function.
17085 (fold_comparison): Don't call fold_div_compare here.
17086 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17087 as helper function.
17089 2017-07-19 Nathan Sidwell <nathan@acm.org>
17091 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17092 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17093 * tree.c (find_decls_types_r, verify_type): Use
17094 TYPE_{MIN,MAX}_VALUE_RAW.
17095 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17096 (hash_tree): Likewise.
17097 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17099 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17102 2017-07-18 Tom de Vries <tom@codesourcery.com>
17104 PR middle-end/81464
17105 * omp-expand.c (expand_omp_for_static_chunk): Handle
17106 equal-argument loop exit phi.
17108 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17111 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17112 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17113 operand 2 predicate.
17114 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17115 operand 2 predicate.
17116 (ror,rol -> rorx splitters): Use const_int_operand as
17117 operand 2 predicate.
17119 2017-06-18 Richard Biener <rguenther@suse.de>
17121 PR tree-optimization/81410
17122 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17123 the gap in the ! slp_perm SLP case after each group.
17125 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17127 PR middle-end/81463
17128 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17131 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17133 PR middle-end/81462
17134 * predict.c (set_even_probabilities): Cleanup; do not affect
17135 probabilities that are already known.
17136 (combine_predictions_for_bb): Call even when count is set.
17138 2017-07-18 Nathan Sidwell <nathan@acm.org>
17140 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17143 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17146 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17147 optimization for loop niter analysis.
17149 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17152 * config/avr/avr.c (avr_optimize_casesi): Don't use
17153 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17155 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17157 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17158 body_cost_vec from _vect_peel_extended_info.
17159 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17160 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17163 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17165 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17167 2017-07-18 Richard Biener <rguenther@suse.de>
17169 PR tree-optimization/80620
17170 PR tree-optimization/81403
17171 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17172 info when re-using a VN table entry.
17174 2017-07-18 Richard Biener <rguenther@suse.de>
17176 PR tree-optimization/81418
17177 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17178 vectype_in. Verify that with lane-reducing reduction operations
17179 we have a single def-use cycle.
17181 2017-07-17 Carl Love <cel@us.ibm.com>
17183 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17185 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17186 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17187 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17188 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17189 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17190 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17191 VMULOSW): New enum "unspec" values.
17192 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17193 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17194 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17195 altivec_vmulosw): New patterns.
17196 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17197 VMULOSW): Add definitions.
17198 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17200 * config/alpha/alpha.c: Include predict.h.
17202 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17204 * tree-vrp.c (compare_assert_loc): Fix comparison function
17205 to return predictable results.
17207 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17209 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17211 (subdi3): Likewise.
17212 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17213 * doc/invoke.texi (mexpand-adddi): Update text.
17215 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17217 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17218 that also clobbers the CC register. The old expand code is moved
17220 (*arc_clzsi2): ... here.
17221 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17222 the CC register. The old expand code is moved to ...
17223 (arc_ctzsi2): ... here.
17225 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17227 * config/arc/arc.opt (mindexed-loads): Use initial value
17228 TARGET_INDEXED_LOADS_DEFAULT.
17229 (mauto-modify-reg): Use initial value
17230 TARGET_AUTO_MODIFY_REG_DEFAULT.
17231 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17232 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17233 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17234 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17236 2017-07-17 Martin Liska <mliska@suse.cz>
17239 * opts.c (finish_options): Do not allow -fgnu-tm
17240 w/ -fsanitize={kernel-,}address. Say sorry.
17242 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17245 * tree-loop-distribution.c (classify_partition): Only assert on
17246 numer of iterations.
17247 (merge_dep_scc_partitions): Delete prameter. Update function call.
17248 (distribute_loop): Remove code handling loop with unknown niters.
17249 (pass_loop_distribution::execute): Skip loop with unknown niters.
17251 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17254 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17255 function sort_partitions_by_post_order.
17257 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17259 PR tree-optimization/81374
17260 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17261 the max index of basic blocks, rather than number of basic blocks.
17263 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17265 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17267 (arc_legitimate_pic_operand_p): Likewise.
17268 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17270 (arc_needs_pcl_p): Likewise.
17271 (arc_legitimate_pc_offset_p): Likewise.
17272 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17273 function is also used in constrains.md.
17274 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17275 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17276 PLUS. Only return true/false in known cases, otherwise assert.
17277 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17278 is already called in arc_legitimate_constant_p.
17279 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17281 (LEGITIMATE_PIC_OPERAND_P): Use
17282 arc_raw_symbolic_reference_mentioned_p function.
17283 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17288 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17289 Andrew Burgess <andrew.burgess@embecosm.com>
17291 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17292 (arc_return_address_register): New function.
17293 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17294 (arc_handle_fndecl_attribute): Add naked attribute.
17295 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17296 (TARGET_WARN_FUNC_RETURN): Likewise.
17297 (arc_allocate_stack_slots_for_args): New function.
17298 (arc_warn_func_return): Likewise.
17299 (machine_function): Change type fn_type.
17300 (arc_compute_function_type): Consider new naked function type,
17301 change function return type.
17302 (arc_must_save_register): Adapt to handle new
17303 arc_compute_function_type's return type.
17304 (arc_expand_prologue): Likewise.
17305 (arc_expand_epilogue): Likewise.
17306 (arc_return_address_regs): Delete.
17307 (arc_return_address_register): New function.
17308 (arc_epilogue_uses): Use above function.
17309 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17310 (arc_function_type): Change encoding, add naked type.
17311 (ARC_INTERRUPT_P): Change to handle the new encoding.
17312 (ARC_FAST_INTERRUPT_P): Likewise.
17313 (ARC_NORMAL_P): Define.
17314 (ARC_NAKED_P): Likewise.
17315 (arc_compute_function_type): Delete prototype.
17316 * config/arc/arc.md (in_ret_delay_slot): Use
17317 arc_return_address_register function.
17318 (simple_return): Likewise.
17319 (p_return_i): Likewise.
17321 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17323 PR tree-optimization/81428
17324 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17325 can't be built for those types.
17327 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17329 Remove stuff dead since r239246.
17331 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17332 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17333 (avr_inform_devices): Remove dead stuff.
17335 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17337 * config/arm/arm_neon.h: Fix softp typo.
17339 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17341 PR tree-optimization/81365
17342 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17343 aggregate moves onto bb predecessor edges, make sure there are no
17344 loads that could alias the lhs in between the start of bb and the
17347 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17350 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17351 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17352 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17354 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17356 PR tree-optimization/81396
17357 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17358 (init_symbolic_number): Initialize it to 1.
17359 (perform_symbolic_merge): Add n_ops from both operands into the new
17361 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17362 without base_addr as useless if they need more than one operation.
17363 (bswap_replace): Handle !bswap case for NULL base_addr.
17365 2017-07-17 Tom de Vries <tom@codesourcery.com>
17368 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17371 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17373 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17374 conditional builtin define __FIX_LEON3FT_B2BST.
17376 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17378 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17379 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17382 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17384 PR rtl-optimization/81424
17385 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17386 to remove potential trapping from operands if -fnon-call-exceptions.
17388 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17390 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17391 profile_proability for scalling.
17392 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17394 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17396 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17398 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17400 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17401 fixpoint arithmetics.
17403 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17405 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17406 fixpoint arithmetics.
17408 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17410 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17411 fixpoint arithmetics.
17413 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17415 * profile-count.h (profile_probability::from_reg_br_prob_note,
17416 profile_probability::to_reg_br_prob_note): New functions.
17417 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17418 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17419 * predict.c (probability_reliable_p): Update.
17420 (edge_probability_reliable_p): Update.
17421 (br_prob_note_reliable_p): Update.
17422 (invert_br_probabilities): Update.
17423 (add_reg_br_prob_note): New function.
17424 (combine_predictions_for_insn): Update.
17425 * asan.c (asan_clear_shadow): Update.
17426 * cfgbuild.c (compute_outgoing_frequencies): Update.
17427 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17428 (update_br_prob_note): Update.
17429 (rtl_verify_edges): Update.
17430 (purge_dead_edges): Update.
17431 (fixup_reorder_chain): Update.
17432 * emit-rtl.c (try_split): Update.
17433 * ifcvt.c (cond_exec_process_insns): Update.
17434 (cond_exec_process_if_block): Update.
17435 (dead_or_predicable): Update.
17436 * internal-fn.c (expand_addsub_overflow): Update.
17437 (expand_neg_overflow): Update.
17438 (expand_mul_overflow): Update.
17439 * loop-doloop.c (doloop_modify): Update.
17440 * loop-unroll.c (compare_and_jump_seq): Update.
17441 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17442 * predict.h: Update.
17443 * reorg.c (mostly_true_jump): Update.
17445 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17446 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17447 * config/arc/arc.c: (emit_unlikely_jump): Update.
17448 * config/arm/arm.c: (emit_unlikely_jump): Update.
17449 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17450 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17451 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17452 (ix86_print_operand): Update.
17453 (ix86_split_fp_branch): Update.
17454 (predict_jump): Update.
17455 * config/ia64/ia64.c (ia64_print_operand): Update.
17456 * config/mmix/mmix.c (mmix_print_operand): Update.
17457 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17458 (rs6000_expand_split_stack_prologue): Update.
17459 * config/rs6000/rs6000.c: Update.
17460 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17461 (s390_expand_vec_movstr): Update.
17462 (s390_expand_cs_tdsi): Update.
17463 (s390_expand_split_stack_prologue): Update.
17464 * config/sh/sh.c (sh_print_operand): Update.
17465 (expand_cbranchsi4): Update.
17466 (expand_cbranchdi4): Update.
17467 * config/sparc/sparc.c (output_v9branch): Update.
17468 * config/spu/spu.c (get_branch_target): Update.
17469 (ea_load_store_inline): Update.
17470 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17471 * config/tilepro/tilepro.c: Update.
17473 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17475 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17476 (gimplify_save_expr): Likewise.
17478 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17480 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17482 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17484 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17486 * timevar.def (TV_IPA_FNSUMMARY): Define.
17488 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17490 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17491 to back store errata sensitive sequence from being generated.
17492 (sqrtdf2_fix): Likewise.
17494 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17496 * tree-ssa-threadupdate.c (compute_path_counts,
17497 update_joiner_offpath_counts): Use profile_probability.
17499 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17502 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17504 * config/arm/arm-c.c (arm_cpu_builtins): Define
17505 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17507 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17509 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17510 array entries to represent __ieee128 versions of the
17511 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17512 scalar_extract_sig, and scalar_insert_exp built-in functions.
17513 (altivec_resolve_overloaded_builtin): Add special case handling
17514 for the __builtin_scalar_insert_exp function, as represented by
17515 the P9V_BUILTIN_VEC_VSIEDP constant.
17516 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17517 exponent support for __ieee128 argument.
17518 (VSESQP): Add scalar extract signature support for __ieee128
17520 (VSTDCNQP): Add scalar test negative support for __ieee128
17522 (VSIEQP): Add scalar insert exponent support for __int128 argument
17523 with __ieee128 result.
17524 (VSIEQPF): Add scalar insert exponent support for __ieee128
17525 argument with __ieee128 result.
17526 (VSTDCQP): Add scalar test data class support for __ieee128
17528 (VSTDCNQP): Add overload support for scalar test negative with
17529 __ieee128 argument.
17530 (VSTDCQP): Add overload support for scalar test data class
17531 __ieee128 argument.
17532 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17533 UNSPEC_VSX_SXSIGDP.
17534 (UNSPEC_VSX_SIEXPQP): New constant.
17535 (xsxexpqp): New insn for VSX scalar extract exponent quad
17537 (xsxsigqp): New insn for VSX scalar extract significand quad
17539 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17540 precision with floating point argument.
17541 (xststdcqp): New expand for VSX scalar test data class quad
17543 (xststdcnegqp): New expand for VSX scalar test negative quad
17545 (xststdcqp): New insn to match expansions for VSX scalar test data
17546 class quad precision and VSX scalar test negative quad precision.
17547 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17548 special case operand checking to enforce that second operand of
17549 VSX scalar test data class with quad precision argument is a 7-bit
17551 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17552 prototypes and descriptions of __ieee128 versions of
17553 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17554 scalar_test_data_class, and scalar_test_neg built-in functions.
17556 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17558 PR tree-optimization/81162
17559 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17560 replace a negate with an add.
17562 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17564 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17566 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17568 * config/arm/arm-c.c (arm_cpu_builtins): Define
17569 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17571 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17573 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17574 (armv8-r): Set ARM Cortex-R52 as default CPU.
17575 * config/arm/arm-tables.opt: Regenerate.
17576 * config/arm/arm-tune.md: Regenerate.
17577 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17579 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17580 extension for -mcpu=cortex-r52.
17582 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17584 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17585 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17586 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17587 (fp-armv8): Define it as FP_ARMv8 only.
17588 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17589 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17591 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17592 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17593 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17594 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17595 than TARGET_FPU_ARMV8.
17596 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17597 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17598 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17600 * config/arm/neon.md (neon_vrint): Likewise.
17601 (neon_vcvt): Likewise.
17602 (neon_<fmaxmin_op><mode>): Likewise.
17603 (<fmaxmin><mode>3): Likewise.
17604 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17605 * config/arm/predicates.md (arm_cond_move_operator): Check against
17606 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17608 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17610 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17611 to top of function.
17613 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17615 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17616 loop in comment with memset.
17618 2017-07-14 Martin Liska <mliska@suse.cz>
17620 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17621 * dwarf2out.c (is_java): Remove the function.
17622 (output_pubname): Remove usage of the function.
17623 (lower_bound_default): Remove usage of DW_LANG_Java.
17624 (gen_compile_unit_die): Likewise.
17625 * gcc.c: Remove compiler defaults for .java and .zip files.
17626 * gimple-expr.c (remove_suffix): Change as there's no longer
17627 extension than 4-letter one.
17628 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17629 (gimplify_save_expr): Likewise.
17630 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17631 as it's possible even for other languages than Java.
17632 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17633 * lto-opts.c (lto_write_options): Remove reference to Java.
17634 * opts.c (strip_off_ending): Update file extension handling.
17635 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17636 * tree-eh.c (lower_resx): Likewise.
17637 * tree.c (free_lang_data_in_type): Remove dead code.
17638 (find_decls_types_r): Likewise.
17639 (build_common_builtin_nodes): Remove Java from a comment.
17640 (verify_type): Remove dead code.
17641 * varasm.c (assemble_external): Remove Java from a comment.
17643 2017-07-14 Martin Liska <mliska@suse.cz>
17645 * opts.c (finish_options): Add quotes.
17646 (common_handle_option): Likewise.
17648 2017-07-14 Martin Liska <mliska@suse.cz>
17650 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17651 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17652 Remove N_SO_PASCAL.
17653 * dwarf2out.c (lower_bound_default): Do not handle
17655 (gen_compile_unit_die): Likewise.
17656 * gcc.c: Remove default extension binding for GNU Pascal.
17657 * stmt.c: Remove Pascal language from a comment.
17658 * xcoffout.c: Likewise.
17660 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17663 * diagnostic-show-locus.c (fixit_cmp): New function.
17664 (layout::layout): Sort m_fixit_hints.
17665 (column_range::column_range): Assert that the values are valid.
17666 (struct char_span): New struct.
17667 (correction::overwrite): New method.
17668 (struct source_line): New struct.
17669 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17670 calls in terms of classes source_line and char_span, and
17671 correction::overwrite.
17672 (selftest::test_overlapped_fixit_printing_2): New function.
17673 (selftest::diagnostic_show_locus_c_tests): Call it.
17675 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17677 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17678 early if there is no lhs.
17680 2017-07-13 Martin Liska <mliska@suse.cz>
17682 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17683 (gen_reference_type_die): Likewise.
17684 * stor-layout.c: Remove Pascal-related comment.
17686 2017-07-13 Martin Liska <mliska@suse.cz>
17688 * opts.c (finish_options): Add quotes to error messages.
17689 (parse_sanitizer_options): Likewise.
17691 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17693 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17695 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17697 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17699 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17701 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17703 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17705 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17708 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17709 provides the hardware capability bits, define the macro
17710 __BUILTIN_CPU_SUPPORTS__.
17711 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17712 if GLIBC does not provide the hardware capability bits. Add a
17713 gcc_unreachable call if the built-in cpu function is neither
17714 __builtin_cpu_is nor __builtin_cpu_supports.
17715 (rs6000_get_function_versions_dispatcher): Change the warning
17716 that an old GLIBC is used which does not export the capability
17717 bits to be an error.
17718 * doc/extend.texi (target_clones attribute): Document the
17719 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17720 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17721 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17722 the macros defined by GCC if the newer GLIBC is available.
17724 2017-07-12 Jeff Law <law@redhat.com>
17726 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17727 remaining includes slightly.
17728 * config/riscv/riscv-builtins.c: Include profile-count.h.
17730 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17733 * config/avr/avr.c (avr_set_current_function): In diagnostic
17734 messages: Quote keywords and (parts of) identifiers.
17735 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17738 2017-07-12 Carl Love <cel@us.ibm.com>
17740 * config/rs6000/rs6000-c.c: Add support for built-in functions
17741 vector bool char vec_revb (vector bool char);
17742 vector bool short vec_revb (vector short char);
17743 vector bool int vec_revb (vector bool int);
17744 vector bool long long vec_revb (vector bool long long);
17745 * doc/extend.texi: Update the built-in documentation file for the
17746 new built-in functions.
17748 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17750 * config/s390/s390.md: Remove movcc splitter.
17752 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17754 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17755 load/store on condition.
17757 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17760 * config/avr/avr.c (avr_encode_section_info)
17761 [progmem && !TREE_READONLY]: Error if progmem object needs
17764 2017-07-11 Michael Collison <michael.collison@arm.com>
17766 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17769 2017-07-11 Carl Love <cel@us.ibm.com>
17771 * config/rs6000/rs6000-c.c: Add support for builtins
17772 vector unsigned int vec_parity_lsbb (vector signed int);
17773 vector unsigned int vec_parity_lsbb (vector unsigned int);
17774 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17775 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17776 vector unsigned long long vec_parity_lsbb (vector signed long long);
17777 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17778 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17779 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17780 * doc/extend.texi: Update the built-in documentation file for the
17781 new built-in functions.
17783 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17785 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17786 (layout::m_primary_loc): New field.
17787 (layout::layout): Initialize new field. Move location filtering
17788 logic from here to...
17789 (layout::maybe_add_location_range): ...this new method. Add
17790 support for filtering to just the lines already specified by other
17792 (layout::will_show_line_p): New method.
17793 (gcc_rich_location::add_location_if_nearby): New method.
17794 (selftest::test_add_location_if_nearby): New test function.
17795 (selftest::diagnostic_show_locus_c_tests): Call it.
17796 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17799 2017-07-11 Tom de Vries <tom@codesourcery.com>
17801 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17802 (bb_first_real_insn): New function.
17803 (nvptx_single): Add extra initialization of broadcasted condition
17806 2017-07-11 Nathan Sidwell <nathan@acm.org>
17808 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17810 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17812 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17813 Binutils doc as TEXI will mess them up.
17814 * doc/invoke.texi (AVR Options): Same here.
17816 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17818 * config/sparc/sparc.opt (mfix-ut700): New option.
17819 (mfix-gr712rc): Likewise.
17820 (sparc_fix_b2bst): New variable.
17821 * doc/invoke.texi (SPARC options): Document them.
17822 (ARM options): Fix warnings.
17823 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17824 instructions to prevent sequences that can trigger the store-store
17825 errata for certain LEON3FT processors.
17826 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17827 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17828 * config/sparc/sparc.md (fix_b2bst): New attribute.
17829 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17831 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17834 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17836 (*rsqrtsf2_sse): Ditto.
17838 (div<mode>3): Macroize insn from divdf3 and divsf3
17839 using MODEF mode iterator.
17841 2017-07-10 Martin Sebor <msebor@redhat.com>
17843 PR tree-optimization/80397
17844 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17845 instead of testing for equality to INTEGER_TYPE.
17847 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17849 * config.gcc: Remove uclibc from arc target spec.
17851 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17853 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17855 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17858 * lto-wrapper.c (remove_option): New function.
17859 (merge_and_complain): Merge PIC/PIE options more realistically.
17861 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17863 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17867 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17868 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17869 * config.in: Regenerate.
17870 * configure: Regenerate.
17871 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17872 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17873 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17874 (TARGET_GASISR_PROLOGUES): ...target mask.
17875 * common/config/avr/avr-common.c
17876 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17877 Set -mgas-isr-prologues.
17878 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17879 INSERT_PASS_BEFORE for it.
17880 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17881 * config/avr/avr.c (avr_option_override)
17882 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17883 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17884 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17885 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17886 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17888 (make_avr_pass_pre_proep): New function.
17889 (emit_push_sfr) <treg>: Add argument to function and use it
17890 instead of TMP_REG.
17891 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17892 and set machine->gasisr.yes.
17893 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17894 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17895 __gcc_isr.n_pushed to .L__stack_usage.
17896 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17897 (avr_asm_final_postscan_insn): ...this new static function.
17898 * config/avr/avr.h (machine_function)
17899 <is_no_gccisr, use_L__stack_usage>: New fields.
17900 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17901 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17902 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17903 (gasisr, *gasisr): New expander and insn.
17904 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17905 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17906 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17908 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17910 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17913 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17915 Move jump-tables out of .text again.
17918 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17919 (ASM_OUTPUT_ADDR_VEC): New function.
17920 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17921 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17922 INSN_ADDRESSes as asm comment.
17923 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17924 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17925 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17926 * config/avr/avr.md (*tablejump): Adjust comment.
17927 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17928 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17930 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17931 (avr_output_addr_vec): New proto.
17932 (avr_log_t) <insn_addresses>: New field.
17934 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17937 * config/i386/i386.c (ix86_function_arg_advance): Set
17938 outgoing_args_on_stack to true if there are outgoing arguments
17940 (ix86_function_arg): Likewise.
17941 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17942 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17943 * config/i386/i386.h (machine_function): Add
17944 outgoing_args_on_stack.
17946 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17948 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17949 supporting pthreds.
17950 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17952 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17954 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17955 (REAL_H): Remove $(MACHMODE_H).
17956 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17958 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17959 $(MACHMODE_H) and double-int.h.
17960 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17962 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17965 2017-07-07 Andrew Pinski <apinski@cavium.com>
17967 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17968 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17970 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17972 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17973 Add warning if GCC was not configured to link against a GLIBC that
17974 exports the hardware capability bits.
17975 (make_resolver_func): Make resolver function private and not a
17976 COMDAT function. Create the name with clone_function_name instead
17977 of make_unique_name.
17980 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17981 correct operand in doing the split.
17983 2017-07-07 Carl Love <cel@us.ibm.com>
17985 * config/rs6000/rs6000-c: Add support for built-in function
17986 vector unsigned short vec_pack_to_short_fp32 (vector float,
17988 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17989 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17990 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17991 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17992 (convert_4f32_8i16): Add define_expand.
17993 * doc/extend.texi: Update the built-in documentation file for the
17994 new built-in function.
17996 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17998 * config/sparc/m8.md: New file.
17999 * config/sparc/sparc.md: Include m8.md.
18001 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18003 * config/sparc/sparc.opt: New option -mvis4b.
18004 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18005 (sparc_option_override): Handle VIS4B.
18006 (enum sparc_builtins): Define
18007 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18008 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18009 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18010 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18011 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18012 (check_constant_argument): New function.
18013 (sparc_vis_init_builtins): Define builtins
18014 __builtin_vis_dictunpack{8,16,32},
18015 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18016 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18017 __builtin_vis_fpcmpde{8,16,32}shl and
18018 __builtin_vis_fpcmpur{8,16,32}shl.
18019 (sparc_expand_builtin): Check that the constant operands to
18020 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18021 constant and in range.
18022 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18024 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18025 (SPARC_IMM5_P): Likewise.
18026 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18027 (enabled): Handle vis4b.
18028 (UNSPEC_DICTUNPACK): New unspec.
18029 (UNSPEC_FPCMPSHL): Likewise.
18030 (UNSPEC_FPUCMPSHL): Likewise.
18031 (UNSPEC_FPCMPDESHL): Likewise.
18032 (UNSPEC_FPCMPURSHL): Likewise.
18033 (cpu_feature): New CPU feature `vis4b'.
18034 (dictunpack{8,16,32}): New insns.
18035 (FPCSMODE): New mode iterator.
18036 (fpcscond): New code iterator.
18037 (fpcsucond): Likewise.
18038 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18039 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18040 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18041 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18042 * config/sparc/constraints.md: Define constraints `q' for unsigned
18043 2-bit integer constants and `t' for unsigned 5-bit integer
18045 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18047 (imm5_operand_dictunpack16): Likewise.
18048 (imm5_operand_dictunpack32): Likewise.
18049 (imm2_operand): Likewise.
18050 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18051 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18052 ditunpack* and fpcmp*shl builtins.
18054 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18056 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18057 * config.in: Add HAVE_AS_SPARC6 define.
18058 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18060 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18062 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18063 (CPP_CPU_SPEC): Handle m8.
18064 (ASM_CPU_SPEC): Likewise.
18065 * config/sparc/sparc-opts.h (enum processor_type): Add
18067 * config/sparc/sparc.c (m8_costs): New struct.
18068 (sparc_option_override): Handle TARGET_CPU_m8.
18069 (sparc32_initialize_trampoline): Likewise.
18070 (sparc64_initialize_trampoline): Likewise.
18071 (sparc_issue_rate): Likewise.
18072 (sparc_register_move_cost): Likewise.
18073 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18074 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18075 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18076 (CPP_CPU_SPEC): Handle M8.
18077 (ASM_CPU_SPEC): Likewise.
18078 (AS_M8_FLAG): Define.
18079 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18080 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18081 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18083 * configure: Regenerate.
18084 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18087 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18089 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18091 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18092 ("*movdi_insn_sp32"): Do not set v3pipe.
18093 ("*movsi_insn"): Likewise.
18094 ("*movdi_insn_sp64"): Likewise.
18095 ("*movsf_insn"): Likewise.
18096 ("*movdf_insn_sp32"): Likewise.
18097 ("*movdf_insn_sp64"): Likewise.
18098 ("*zero_extendsidi2_insn_sp64"): Likewise.
18099 ("*sign_extendsidi2_insn"): Likewise.
18100 ("*mov<VM32:mode>_insn"): Likewise.
18101 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18102 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18103 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18104 ("<vlop:code><VL:mode>3"): Likewise.
18105 ("*not_<vlop:code><VL:mode>3"): Likewise.
18106 ("*nand<VL:mode>_vis"): Likewise.
18107 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18108 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18109 ("one_cmpl<VL:mode>2"): Likewise.
18110 ("faligndata<VM64:mode>_vis"): Likewise.
18111 ("alignaddrsi_vis"): Likewise.
18112 ("alignaddrdi_vis"): Likweise.
18113 ("alignaddrlsi_vis"): Likewise.
18114 ("alignaddrldi_vis"): Likewise.
18115 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18116 ("bmaskdi_vis"): Likewise.
18117 ("bmasksi_vis"): Likewise.
18118 ("bshuffle<VM64:mode>_vis"): Likewise.
18119 ("cmask8<P:mode>_vis"): Likewise.
18120 ("cmask16<P:mode>_vis"): Likewise.
18121 ("cmask32<P:mode>_vis"): Likewise.
18122 ("pdistn<P:mode>_vis"): Likewise.
18123 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18125 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18127 * config/sparc/sparc.md ("subtype"): New insn attribute.
18128 ("*wrgsr_sp64"): Set insn subtype.
18129 ("*rdgsr_sp64"): Likewise.
18130 ("alignaddrsi_vis"): Likewise.
18131 ("alignaddrdi_vis"): Likewise.
18132 ("alignaddrlsi_vis"): Likewise.
18133 ("alignaddrldi_vis"): Likewise.
18134 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18135 ("fexpand_vis"): Likewise.
18136 ("fpmerge_vis"): Likewise.
18137 ("faligndata<VM64:mode>_vis"): Likewise.
18138 ("bshuffle<VM64:mode>_vis"): Likewise.
18139 ("cmask8<P:mode>_vis"): Likewise.
18140 ("cmask16<P:mode>_vis"): Likewise.
18141 ("cmask32<P:mode>_vis"): Likewise.
18142 ("fchksm16_vis"): Likewise.
18143 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18144 ("fmean16_vis"): Likewise.
18145 ("fp<plusminus_insn>64_vis"): Likewise.
18146 ("<plusminus_insn>v8qi3"): Likewise.
18147 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18148 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18149 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18150 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18151 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18152 ("*movqi_insn"): Likewise.
18153 ("*movhi_insn"): Likewise.
18154 ("*movsi_insn"): Likewise.
18155 ("movsi_pic_gotdata_op"): Likewise.
18156 ("*movdi_insn_sp32"): Likewise.
18157 ("*movdi_insn_sp64"): Likewise.
18158 ("movdi_pic_gotdata_op"): Likewise.
18159 ("*movsf_insn"): Likewise.
18160 ("*movdf_insn_sp32"): Likewise.
18161 ("*movdf_insn_sp64"): Likewise.
18162 ("*zero_extendhisi2_insn"): Likewise.
18163 ("*zero_extendqihi2_insn"): Likewise.
18164 ("*zero_extendqisi2_insn"): Likewise.
18165 ("*zero_extendqidi2_insn"): Likewise.
18166 ("*zero_extendhidi2_insn"): Likewise.
18167 ("*zero_extendsidi2_insn_sp64"): Likewise.
18168 ("ldfsr"): Likewise.
18169 ("prefetch_64"): Likewise.
18170 ("prefetch_32"): Likewise.
18171 ("tie_ld32"): Likewise.
18172 ("tie_ld64"): Likewise.
18173 ("*tldo_ldub_sp32"): Likewise.
18174 ("*tldo_ldub1_sp32"): Likewise.
18175 ("*tldo_ldub2_sp32"): Likewise.
18176 ("*tldo_ldub_sp64"): Likewise.
18177 ("*tldo_ldub1_sp64"): Likewise.
18178 ("*tldo_ldub2_sp64"): Likewise.
18179 ("*tldo_ldub3_sp64"): Likewise.
18180 ("*tldo_lduh_sp32"): Likewise.
18181 ("*tldo_lduh1_sp32"): Likewise.
18182 ("*tldo_lduh_sp64"): Likewise.
18183 ("*tldo_lduh1_sp64"): Likewise.
18184 ("*tldo_lduh2_sp64"): Likewise.
18185 ("*tldo_lduw_sp32"): Likewise.
18186 ("*tldo_lduw_sp64"): Likewise.
18187 ("*tldo_lduw1_sp64"): Likewise.
18188 ("*tldo_ldx_sp64"): Likewise.
18189 ("*mov<VM32:mode>_insn"): Likewise.
18190 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18191 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18193 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18195 * config/sparc/sparc.md ("type"): New insn type viscmp.
18196 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18198 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18199 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18200 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18201 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18203 ("n7_vis_logical_11cycle"): Likewise.
18204 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18205 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18206 * config/sparc/niagara.md ("niag_vis"): Likewise.
18207 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18208 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18210 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18212 * config/sparc/sparc.md: New instruction type `bmask'.
18213 (bmaskdi_vis): Use the `bmask' type.
18214 (bmasksi_vis): Likewise.
18215 * config/sparc/ultra3.md (us3_array): Likewise.
18216 * config/sparc/niagara7.md (n7_array): Likewise.
18217 * config/sparc/niagara4.md (n4_array): Likewise.
18218 * config/sparc/niagara2.md (niag2_vis): Likewise.
18219 (niag3_vis): Likewise.
18220 * config/sparc/niagara.md (niag_vis): Likewise.
18222 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18224 * ipa-comdats.c: Remove optimize check from gate.
18225 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18226 for functions not optimized.
18227 (ipa_fn_summary_read): Skip optimize check.
18228 (ipa_fn_summary_write): Likewise.
18229 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18231 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18233 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18234 (check_callers): Give up on caller which is not optimized.
18235 (inline_small_functions): Likewise.
18236 (ipa_inline): Do not give up when not optimizing.
18237 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18238 away unoptimizes cdtors.
18239 (whole_program_function_and_variable_visibility): Do
18240 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18241 * ipa.c (process_references): Do not check optimize.
18242 (symbol_table::remove_unreachable_nodes): Update optimize check.
18243 (set_writeonly_bit): Update optimize check.
18244 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18245 (pass_ipa_single_use::gate): Remove.
18247 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18249 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18250 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18251 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18252 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18253 permute_load, permute_store, adjust_extract, adjust_splat,
18254 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18255 replace_swap_with_copy, dump_swap_insn_table,
18256 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18257 recombine_lvx_pattern, recombine_stvx_pattern,
18258 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18259 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18260 to file rs6000-p8swap.c.
18261 * config/rs6000/rs6000-p8swap.c: New file.
18262 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18263 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18264 and rs6000*-*-* targets.
18266 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18268 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18270 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18272 * lto-wrapper.c (merge_and_complain): Do not merge
18273 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18274 fsigned_zeros, ftrapping_math, fwrapv.
18275 (append_compiler_options): Do not track these options.
18276 (append_linker_options): Likewie
18278 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18280 * cgraphunit.c (cgraph_node::finalize_function): When
18281 !flag_toplevel_reorde set no_reorder flag.
18282 (varpool_node::finalize_decl): Likewise.
18283 (symbol_table::compile): Drop no toplevel reorder path.
18285 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18287 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18288 edges; zero probability is not better than uninitialized.
18290 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18292 * asan.h (asan_sanitize_allocas_p): Declare.
18293 * asan.c (asan_sanitize_allocas_p): New function.
18294 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18295 (handle_builtin_alloca): Likewise.
18296 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18297 if !asan_sanitize_allocas_p.
18298 * params.def (asan-instrument-allocas): Add new option.
18299 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18300 * opts.c (common_handle_option): Disable allocas sanitization for
18303 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18305 * asan.c: Include gimple-fold.h.
18306 (get_last_alloca_addr): New function.
18307 (handle_builtin_stackrestore): Likewise.
18308 (handle_builtin_alloca): Likewise.
18309 (asan_emit_allocas_unpoison): Likewise.
18310 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18311 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18312 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18313 (instrument_builtin_call): Pass gimple iterator to
18314 get_mem_refs_of_builtin_call.
18315 (last_alloca_addr): New global.
18316 * asan.h (asan_emit_allocas_unpoison): Declare.
18317 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18318 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18319 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18320 if function calls alloca.
18321 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18322 * gimple-fold.h (replace_call_with_value): Declare.
18323 * internal-fn.c: Include asan.h.
18324 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18325 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18327 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18329 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18330 (C_SELFTEST_FLAGS): New.
18331 (CPP_SELFTEST_FLAGS): New.
18332 (SELFTEST_DEPS): New, from deps of s-selftest.
18333 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18334 (CPP_SELFTEST_DEPS): New.
18335 (selftest): Add dependency on s-selftest-c++.
18336 (s-selftest): Rename to...
18337 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18338 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18339 than SELFTEST_FLAGS.
18340 (selftest-gdb): Rename to...
18341 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18343 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18344 (selftest-valgrind): Rename to...
18345 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18347 (selftest-valgrind): Reintroduce as an alias for
18348 selftest-c-valgrind.
18349 (s-selftest-c++): New.
18350 (selftest-c++-gdb): New.
18351 (selftest-c++-valgrind): New.
18353 2017-07-06 Olivier Hainque <hainque@adacore.com>
18355 * gcc.c (process_command): When deciding if undefined variables
18356 should be ignored when processing specs, accept "gcc -v" as well.
18358 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18360 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18361 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18363 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18365 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18366 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18367 * config/arm/arm-tables.opt: Regenerate.
18368 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18370 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18372 2017-07-06 Carl Love <cel@us.ibm.com>
18374 * ChangeLog: Clean up from mid air collision
18376 2017-07-06 Carl Love <cel@us.ibm.com>
18378 * config/rs6000/rs6000-c.c: Add support for built-in functions
18379 vector signed int vec_subc (vector signed int, vector signed int);
18380 vector signed __int128 vec_subc (vector signed __int128,
18381 vector signed __int128);
18382 vector unsigned __int128 vec_subc (vector unsigned __int128,
18383 vector unsigned __int128);
18384 vector signed int vec_sube (vector signed int, vector signed int,
18385 vector signed int);
18386 vector unsigned int vec_sube (vector unsigned int,
18387 vector unsigned int,
18388 vector unsigned int);
18389 vector signed __int128 vec_sube (vector signed __int128,
18390 vector signed __int128,
18391 vector signed__int128);
18392 vector unsigned __int128 vec_sube (vector unsigned __int128,
18393 vector unsigned __int128,
18394 vector unsigned __int128);
18395 vector signed int vec_subec (vector signed int, vector signed int,
18396 vector signed int);
18397 vector unsigned int vec_subec (vector unsigned int,
18398 vector unsigned int,
18399 vector unsigned int);
18400 vector signed __int128 vec_subec (vector signed __int128,
18401 vector signed __int128,
18402 vector signed__int128);
18403 vector unsigned __int128 vec_subec (vector unsigned __int128,
18404 vector unsigned __int128,
18405 vector unsigned __int128);
18406 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18407 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18408 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18409 BU_ALTIVEC_OVERLOAD_X definitions.
18410 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18411 * doc/extend.texi: Update the built-in documentation file for the new
18412 built-in functions.
18414 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18417 * diagnostic-show-locus.c (layout::layout): Use start and finish
18418 spelling location for the start and finish of each range.
18419 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18420 Add unused aspect param.
18421 * input.c (expand_location_1): Add "aspect" param, and use it
18422 to access the correct part of the location.
18423 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18425 (expand_location_to_spelling_point): Likewise.
18426 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18427 param, and pass it to expand_location_1.
18429 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18431 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18432 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18433 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18434 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18435 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18436 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18437 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18438 _mm_maskz_getmant_ss): New intrinsics.
18439 (__builtin_ia32_getexpss128_mask): Changed to ...
18440 __builtin_ia32_getexpss128_round ... this.
18441 (__builtin_ia32_getexpsd128_mask): Changed to ...
18442 __builtin_ia32_getexpsd128_round ... this.
18443 * config/i386/i386-builtin-types.def
18444 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18445 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18446 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18447 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18448 __builtin_ia32_getmantss_mask_round): New builtins.
18449 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18450 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18451 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18452 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18453 * config/i386/sse.md
18454 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18455 avx512f_sgetexp<mode><mask_scalar_name>
18456 <round_saeonly_scalar_name> ... this.
18457 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18458 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18459 vgetexp<ssescalarmodesuffix>
18460 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18461 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18462 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18463 avx512f_vgetmant<mode><mask_scalar_name>
18464 <round_saeonly_scalar_name> ... this.
18465 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18466 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18467 vgetmant<ssescalarmodesuffix>
18468 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18469 %0<mask_scalar_operand4>, %1, %2
18470 <round_saeonly_scalar_mask_op4>, %3} ... this.
18471 * config/i386/subst.md (mask_scalar_operand4,
18472 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18473 round_saeonly_scalar_nimm_predicate): New subst attributes.
18475 2017-07-06 Julia Koval <julia.koval@intel.com>
18477 * config/i386/i386.c (ix86_erase_embedded_rounding):
18478 Remove code for old rounding pattern.
18480 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18482 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18484 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18486 * doc/sourcebuild.texi (Test Directives, Variants of
18487 dg-require-support): Add documentation for dg-require-stack-check.
18489 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18491 * config/i386/subst.md (mask_scalar, round_scalar,
18492 round_saeonly_scalar): New meta-templates.
18493 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18494 round_scalar_mask_operand3, round_scalar_mask_op3,
18495 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18496 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18497 round_saeonly_scalar_constraint,
18498 round_saeonly_scalar_prefix): New subst attribute.
18499 * config/i386/sse.md
18500 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18501 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18502 <round_scalar_name> ... this.
18503 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18504 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18505 <round_scalar_name> ... this.
18506 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18507 <sse>_vm<code><mode>3<mask_scalar_name>
18508 <round_saeonly_scalar_name> ... this.
18509 (v<plusminus_mnemonic><ssescalarmodesuffix>
18510 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18511 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18512 v<plusminus_mnemonic><ssescalarmodesuffix>
18513 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18514 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18515 (v<multdiv_mnemonic><ssescalarmodesuffix>
18516 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18517 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18518 v<multdiv_mnemonic><ssescalarmodesuffix>
18519 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18520 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18521 (v<maxmin_float><ssescalarmodesuffix>
18522 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18523 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18524 v<maxmin_float><ssescalarmodesuffix>
18525 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18526 %0<mask_scalar_operand3>, %1, %<iptr>2
18527 <round_saeonly_scalar_mask_op3>} ... this.
18529 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18531 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18532 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18534 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18535 Alan Hayward <alan.hayward@arm.com>
18536 David Sherwood <david.sherwood@arm.com>
18538 * combine.c (simplify_if_then_else): Remove "enum" before
18540 * compare-elim.c (can_eliminate_compare): Likewise.
18541 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18543 (aarch64_lookup_simd_builtin_type): Likewise.
18544 (aarch64_simd_builtin_type): Likewise.
18545 (aarch64_init_simd_builtin_types): Likewise.
18546 (aarch64_simd_expand_args): Likewise.
18547 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18549 (aarch64_reverse_mask): Likewise.
18550 (aarch64_simd_emit_reg_reg_move): Likewise.
18551 (aarch64_gen_adjusted_ldpstp): Likewise.
18552 (aarch64_ccmp_mode_to_code): Likewise.
18553 (aarch64_operands_ok_for_ldpstp): Likewise.
18554 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18555 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18557 (aarch64_min_divisions_for_recip_mul): Likewise.
18558 (aarch64_reassociation_width): Likewise.
18559 (aarch64_get_condition_code_1): Likewise.
18560 (aarch64_simd_emit_reg_reg_move): Likewise.
18561 (aarch64_simd_attr_length_rglist): Likewise.
18562 (aarch64_reverse_mask): Likewise.
18563 (aarch64_operands_ok_for_ldpstp): Likewise.
18564 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18565 (aarch64_gen_adjusted_ldpstp): Likewise.
18566 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18568 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18569 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18570 (arm_lookup_simd_builtin_type): Likewise.
18571 (arm_simd_builtin_type): Likewise.
18572 (arm_init_simd_builtin_types): Likewise.
18573 (arm_expand_builtin_args): Likewise.
18574 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18575 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18576 (ft32_setup_incoming_varargs): Likewise.
18577 (ft32_function_arg): Likewise.
18578 (ft32_function_arg_advance): Likewise.
18579 (ft32_pass_by_reference): Likewise.
18580 (ft32_arg_partial_bytes): Likewise.
18581 (ft32_valid_pointer_mode): Likewise.
18582 (ft32_addr_space_pointer_mode): Likewise.
18583 (ft32_addr_space_legitimate_address_p): Likewise.
18584 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18586 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18587 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18588 (iamcu_alignment): Likewise.
18589 (canonicalize_vector_int_perm): Likewise.
18590 (ix86_noce_conversion_profitable_p): Likewise.
18591 (ix86_mpx_bound_mode): Likewise.
18592 (ix86_operands_ok_for_move_multiple): Likewise.
18593 * config/microblaze/microblaze-protos.h
18594 (microblaze_expand_conditional_branch_reg): Likewise.
18595 * config/microblaze/microblaze.c
18596 (microblaze_expand_conditional_branch_reg): Likewise.
18597 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18599 (rs6000_reassociation_width): Likewise.
18600 (rs6000_invalid_binary_op): Likewise.
18601 (fusion_p9_p): Likewise.
18602 (emit_fusion_p9_load): Likewise.
18603 (emit_fusion_p9_store): Likewise.
18604 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18606 (riscv_hard_regno_mode_ok_p): Likewise.
18607 (riscv_address_insns): Likewise.
18608 (riscv_split_symbol): Likewise.
18609 (riscv_legitimize_move): Likewise.
18610 (riscv_function_value): Likewise.
18611 (riscv_hard_regno_nregs): Likewise.
18612 (riscv_expand_builtin): Likewise.
18613 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18614 (riscv_build_integer): Likewise.
18615 (riscv_split_integer): Likewise.
18616 (riscv_legitimate_constant_p): Likewise.
18617 (riscv_cannot_force_const_mem): Likewise.
18618 (riscv_regno_mode_ok_for_base_p): Likewise.
18619 (riscv_valid_base_register_p): Likewise.
18620 (riscv_valid_offset_p): Likewise.
18621 (riscv_valid_lo_sum_p): Likewise.
18622 (riscv_classify_address): Likewise.
18623 (riscv_legitimate_address_p): Likewise.
18624 (riscv_address_insns): Likewise.
18625 (riscv_load_store_insns): Likewise.
18626 (riscv_force_binary): Likewise.
18627 (riscv_split_symbol): Likewise.
18628 (riscv_force_address): Likewise.
18629 (riscv_legitimize_address): Likewise.
18630 (riscv_move_integer): Likewise.
18631 (riscv_legitimize_const_move): Likewise.
18632 (riscv_legitimize_move): Likewise.
18633 (riscv_address_cost): Likewise.
18634 (riscv_subword): Likewise.
18635 (riscv_output_move): Likewise.
18636 (riscv_canonicalize_int_order_test): Likewise.
18637 (riscv_emit_int_order_test): Likewise.
18638 (riscv_function_arg_boundary): Likewise.
18639 (riscv_pass_mode_in_fpr_p): Likewise.
18640 (riscv_pass_fpr_single): Likewise.
18641 (riscv_pass_fpr_pair): Likewise.
18642 (riscv_get_arg_info): Likewise.
18643 (riscv_function_arg): Likewise.
18644 (riscv_function_arg_advance): Likewise.
18645 (riscv_arg_partial_bytes): Likewise.
18646 (riscv_function_value): Likewise.
18647 (riscv_pass_by_reference): Likewise.
18648 (riscv_setup_incoming_varargs): Likewise.
18649 (riscv_print_operand): Likewise.
18650 (riscv_elf_select_rtx_section): Likewise.
18651 (riscv_save_restore_reg): Likewise.
18652 (riscv_for_each_saved_reg): Likewise.
18653 (riscv_register_move_cost): Likewise.
18654 (riscv_hard_regno_mode_ok_p): Likewise.
18655 (riscv_hard_regno_nregs): Likewise.
18656 (riscv_class_max_nregs): Likewise.
18657 (riscv_memory_move_cost): Likewise.
18658 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18659 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18660 (rl78_addr_space_address_mode): Likewise.
18661 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18663 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18664 (rs6000_reassociation_width): Likewise.
18665 (rs6000_invalid_binary_op): Likewise.
18666 (fusion_p9_p): Likewise.
18667 (emit_fusion_p9_load): Likewise.
18668 (emit_fusion_p9_store): Likewise.
18669 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18670 (ok_for_simple_move_operands): Likewise.
18671 (ok_for_simple_move_strict_operands): Likewise.
18672 (ok_for_simple_arith_logic_operands): Likewise.
18673 (visium_legitimize_reload_address): Likewise.
18674 (visium_select_cc_mode): Likewise.
18675 (output_cbranch): Likewise.
18676 (visium_split_double_move): Likewise.
18677 (visium_expand_copysign): Likewise.
18678 (visium_expand_int_cstore): Likewise.
18679 (visium_expand_fp_cstore): Likewise.
18680 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18681 (visium_function_arg): Likewise.
18682 (visium_function_arg_advance): Likewise.
18683 (visium_libcall_value): Likewise.
18684 (visium_setup_incoming_varargs): Likewise.
18685 (visium_legitimate_constant_p): Likewise.
18686 (visium_legitimate_address_p): Likewise.
18687 (visium_legitimize_address): Likewise.
18688 (visium_secondary_reload): Likewise.
18689 (visium_register_move_cost): Likewise.
18690 (visium_memory_move_cost): Likewise.
18691 (prepare_move_operands): Likewise.
18692 (ok_for_simple_move_operands): Likewise.
18693 (ok_for_simple_move_strict_operands): Likewise.
18694 (ok_for_simple_arith_logic_operands): Likewise.
18695 (visium_function_value_1): Likewise.
18696 (rtx_ok_for_offset_p): Likewise.
18697 (visium_legitimize_reload_address): Likewise.
18698 (visium_split_double_move): Likewise.
18699 (visium_expand_copysign): Likewise.
18700 (visium_expand_int_cstore): Likewise.
18701 (visium_expand_fp_cstore): Likewise.
18702 (visium_split_cstore): Likewise.
18703 (visium_select_cc_mode): Likewise.
18704 (visium_split_cbranch): Likewise.
18705 (output_cbranch): Likewise.
18706 (visium_print_operand_address): Likewise.
18707 * expmed.c (flip_storage_order): Likewise.
18708 * expmed.h (emit_cstore): Likewise.
18709 (flip_storage_order): Likewise.
18710 * genrecog.c (validate_pattern): Likewise.
18711 * hsa-gen.c (gen_hsa_addr): Likewise.
18712 * internal-fn.c (expand_arith_overflow): Likewise.
18713 * ira-color.c (allocno_copy_cost_saving): Likewise.
18714 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18715 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18716 (process_invariant_for_inheritance): Likewise.
18717 * lra-eliminations.c (move_plus_up): Likewise.
18718 * omp-low.c (lower_oacc_reductions): Likewise.
18719 * simplify-rtx.c (simplify_subreg): Likewise.
18720 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18721 (TARGET_CHKP_BOUND_MODE): Likewise..
18722 * targhooks.c (default_chkp_bound_mode): Likewise.
18723 (default_setup_incoming_vararg_bounds): Likewise.
18724 * targhooks.h (default_chkp_bound_mode): Likewise.
18725 (default_setup_incoming_vararg_bounds): Likewise.
18726 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18727 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18728 (have_whole_vector_shift): Likewise.
18729 * tree-vect-stmts.c (vectorizable_load): Likewise.
18730 * doc/tm.texi: Regenerate.
18732 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18734 Graceful degrade if Binutils PR21472 is not available.
18737 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18738 .rodata in flash test fails.
18739 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18740 * confgure: Regenerate.
18741 * config.in: Regenerate.
18742 * config/avr/avr.c (avr_asm_named_section)
18743 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18744 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18745 (avr_asm_init_sections): Same.
18747 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18749 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18750 (fma<VH:mode>4_intrinsic): Likewise.
18751 (*fmsub<VCVTF:mode>4): Likewise.
18752 (*fmsub<VH:mode>4_intrinsic): Likewise.
18754 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18757 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18758 Don't depend on "optimize > 0".
18759 (out_movhi_r_mr, out_movqi_mr_r): Same.
18760 (out_movhi_mr_r, out_movqi_r_mr): Same.
18761 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18762 io_address_operand on "optimize > 0".
18764 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18766 * tree-loop-distribution.c: Add general explanantion on the pass.
18767 (generate_loops_for_partition): Mark distributed loop.
18768 (pg_add_dependence_edges): New parameter. Handle alias data
18769 dependence specially and record it in the parameter if asked.
18770 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18771 (init_partition_graph_vertices, add_partition_graph_edge): New.
18772 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18773 (free_partition_graph_vdata, build_partition_graph): New.
18774 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18775 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18776 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18777 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18778 (version_for_distribution_p, finalize_partitions): New.
18779 (distribute_loop): Handle alias data dependence specially. Factor
18780 out loop fusion code as functions and call these functions.
18782 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18784 * tree-loop-distribution.c (classify_partition): New parameter and
18785 better handle reduction statement.
18786 (rdg_build_partitions): Revise comment.
18787 (distribute_loop): Compute statements in all partitions and pass it
18788 to classify_partition.
18790 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18792 * tree-loop-distribution.c (enum partition_type): New.
18793 (struct partition): New field type.
18794 (partition_merge_into): Add parameter. Update partition type.
18795 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18796 (build_rdg_partition_for_vertex): Compute partition type.
18797 (rdg_build_partitions): Dump partition type.
18798 (distribute_loop): Update calls to partition_merge_into.
18800 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18802 * tree-loop-distribution.c (struct ddr_hasher): New.
18803 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18805 (classify_partition): Call get_data_dependence.
18806 (pg_add_dependence_edges): Ditto.
18807 (distribute_loop): Release data dependence hash table.
18809 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18811 * tree-loop-distribution.c (ref_base_address): Delete.
18812 (similar_memory_accesses): Rename ...
18813 (share_memory_accesses): ... to this. Check if partitions access
18814 the same memory reference.
18815 (distribute_loop): Call share_memory_accesses.
18817 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18819 * tree-loop-distribution.c (struct partition): New field recording
18820 its data reference.
18821 (partition_alloc, partition_free): Init and release data refs.
18822 (partition_merge_into): Merge data refs.
18823 (build_rdg_partition_for_vertex): Collect data refs for partition.
18824 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18826 (distribute_loop): Remve data refs from vertice data of partition
18829 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18831 * tree-loop-distribution.c (params.h): Include header file.
18832 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18833 (datarefs_vec): New global var.
18834 (create_rdg_vertices): Use datarefs_vec directly.
18835 (free_rdg): Don't free data references.
18836 (build_rdg): Update use. Don't free data references.
18837 (distribute_loop): Compute global variable for data references.
18838 Bail out if there are too many data references.
18840 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18842 * tree-loop-distribution.c (loop_nest): New global var.
18843 (build_rdg): Use loop directly, rather than loop nest.
18844 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18846 (distribute_loop): Compute global variable loop nest. Update use.
18848 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18850 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18851 (partition_merge_into): New parameter. Dump reason for fusion.
18852 (distribute_loop): Update use of partition_merge_into.
18854 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18856 * tree-loop-distribution.c (bb_top_order_index): New.
18857 (bb_top_order_index_size, bb_top_order_cmp): New.
18858 (stmts_from_loop): Use topological order.
18859 (pass_loop_distribution::execute): Compute and release topological
18860 order for basic blocks.
18862 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18864 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18867 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18869 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18870 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18871 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18872 * internal-fn.def (LOOP_DIST_ALIAS): New.
18873 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18874 (fold_loop_internal_call): ... this.
18875 (vect_loop_dist_alias_call): New function.
18876 (set_uid_loop_bbs): Call fold_loop_internal_call.
18877 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18880 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18883 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18884 Require dead FLAGS_REG at the beginning of a peephole.
18886 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18889 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18890 arguments in the call to __builtin_ia32_sbb_u32.
18891 (_subborrow_u64): Swap _X and _Y arguments in the call to
18892 __builtin_ia32_sbb_u64.
18894 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18897 * tree-vrp.c (compare_assert_loc): Turn into a function template
18898 with stable template parameter. Only test if a->e is NULL,
18899 !a->e == !b->e has been verified already. Use e == NULL or
18900 e != NULL instead of e or ! e tests. If stable is true, don't use
18901 iterative_hash_expr, on the other side allow a or b or both NULL
18902 and sort the NULLs last.
18903 (process_assert_insertions): Sort using compare_assert_loc<false>
18904 instead of compare_assert_loc, later sort using
18905 compare_assert_loc<true> before calling process_assert_insertions_for
18906 in a loop. Use break instead of continue once seen NULL pointer.
18908 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18910 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18911 Cortex-R7 and Cortex-R8 processors.
18913 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18915 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18916 uninitialized while src is not.
18918 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18920 * common/config/arm/arm-common.c: Adjust include path for
18922 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18923 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18924 (arm-cpu-data.h): Likewise.
18925 (arm-cpu-cdata.h): Likewise.
18926 * config/arm/arm-cpu.h: Delete.
18927 * config/arm/arm-cpu-cdata.h: Delete.
18928 * config/arm/arm-cpu-data.h: Delete.
18930 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18932 * config/arm/arm-cpus.in (cortex-a55): New.
18933 (cortex-a75): Likewise.
18934 (cortex-a75.cortex-a55): Likewise.
18935 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18937 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18938 * config/arm/arm-cpu-cdata.h: Regenerate.
18939 * config/arm/arm-cpu-data.h: Regenerate.
18940 * config/arm/arm-cpu.h: Regenerate.
18941 * config/arm/arm-tables.opt: Regenerate.
18942 * config/arm/arm-tune.md: Regenerate.
18944 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18946 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18948 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18950 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18953 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18955 PR tree-optimization/81292
18956 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18957 full_string_p, also call adjust_related_strinfos if the adjustment
18958 is simple, otherwise invalidate related strinfos.
18960 2017-07-04 Martin Liska <mliska@suse.cz>
18963 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18964 newly created variable as DECL_IGNORED_P.
18966 2017-07-04 Martin Liska <mliska@suse.cz>
18969 * ipa-inline.c (inline_small_functions):
18970 Use xstrdup_for_dump.
18972 2017-07-04 Tom de Vries <tom@codesourcery.com>
18974 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18976 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18979 * config/darwin.c (darwin_function_switched_text_sections):
18982 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18984 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18986 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18988 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18990 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18992 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18993 min_profitable_iters, and th as inclusive lower bounds.
18994 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18995 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18996 for min_profitable_iters and min_profitable_estimate.
18997 (vect_transform_loop): Treat th as an inclusive lower bound.
18998 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19000 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19003 * config/darwin.c (darwin_function_switched_text_sections):
19004 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19005 in two pieces, and suppress the use of buf.
19007 2017-07-03 Nathan Sidwell <nathan@acm.org>
19009 * hash-table.h (hash_table_mod1): Fix indentation.
19011 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19013 PR middle-end/81290
19014 * predict.c (force_edge_cold): Be more careful about propagation
19016 * profile-count.h (profile_probability::guessed,
19017 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19019 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19021 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19023 * doc/invoke.texi (rcpc architecture extension): Document it.
19025 2017-07-03 Richard Biener <rguenther@suse.de>
19027 PR tree-optimization/60510
19028 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19029 the scalar reduction PHI and use it.
19030 (vectorizable_reduction): Properly guard the single_defuse_cycle
19031 path for non-SLP reduction chains where we cannot use it.
19032 Rework reduc_def/index and vector type deduction. Rework
19033 vector operand gathering during reduction op code-gen.
19034 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19035 chains dissolve the chain and leave it to non-SLP reduction
19038 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19040 * tree-data-ref.h (dr_alignment): Declare.
19041 * tree-data-ref.c (dr_alignment): New function.
19042 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19043 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19045 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19047 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19049 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19050 and base_misalignment fields.
19051 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19052 * tree-data-ref.c: Include builtins.h.
19053 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19054 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19055 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19056 * tree-vect-data-refs.c: Include tree-cfg.h.
19057 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19058 fields instead of calculating an alignment here.
19059 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19060 innermost_loop_behavior fields.
19062 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19064 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19066 (DR_STEP_ALIGNMENT): New macro.
19067 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19068 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19069 (create_data_ref): Print it.
19070 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19071 to tell whether the step preserves vector (mis)alignment.
19072 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19073 Move the check for an integer step and generalise to all INTEGER_CST.
19074 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19075 Print the outer step alignment.
19077 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19079 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19080 with offset_alignment.
19081 (DR_ALIGNED_TO): Delete.
19082 (DR_OFFSET_ALIGNMENT): New macro.
19083 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19084 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19085 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19086 (create_data_ref): Likewise.
19087 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19088 (vect_analyze_data_refs): Likewise.
19089 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19090 creating dummy innermost behavior.
19092 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19094 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19095 with a "innermost_loop_behavior *" and refeence tree.
19096 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19097 (create_data_ref): Update call accordingly.
19098 * tree-predcom.c (find_looparound_phi): Likewise.
19100 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19102 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19103 fields with dr_wrt_vec_loop.
19104 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19105 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19106 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19107 (vect_dr_behavior): New function.
19108 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19109 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19110 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19111 track whether the step preserves the misalignment.
19112 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19113 Use vect_dr_behavior.
19114 (vect_setup_realignment): Update call accordingly.
19115 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19116 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19117 call to vect_create_addr_base_for_vector_ref.
19118 (vect_create_cond_for_align_checks): Likewise.
19119 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19120 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19121 (vect_recog_mask_conversion_pattern): Likewise.
19122 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19123 (new_stmt_vec_info): Remove redundant zeroing.
19125 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19127 * common/config/arm/arm-common.c (arm_be8_option): New function.
19128 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19129 (ISA_ARMv6): Add isa_bit_be8.
19130 * config/arm/arm.h (arm_be8_option): Add prototype.
19131 (BE8_SPEC_FUNCTION): New define.
19132 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19133 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19134 (mlittle-endian): Similarly.
19135 (mbe8, mbe32): New options.
19136 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19137 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19139 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19141 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19143 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19145 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19146 (cleanup_tree_cfg_bb): Use it.
19147 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19149 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19151 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19154 * loop-doloop.c (add_test): Update profile.
19156 2017-07-03 Martin Liska <mliska@suse.cz>
19159 * sanopt.c (rewrite_usage_of_param): New function.
19160 (sanitize_rewrite_addressable_params): Likewise.
19161 (pass_sanopt::execute): Call rewrite_usage_of_param.
19163 2017-07-03 Richard Biener <rguenther@suse.de>
19165 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19166 back to using VIEW_CONVERT_EXPR.
19168 2017-07-03 Martin Liska <mliska@suse.cz>
19171 * doc/extend.texi: Document when a resolver function is
19172 generated for target_clones.
19174 2017-07-03 Martin Liska <mliska@suse.cz>
19176 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19177 and shadow memory of auto variables which are subject of
19178 use-after-scope sanitization.
19179 (asan_expand_mark_ifn): Add do set only when is_poison.
19181 2016-07-03 Richard Biener <rguenther@suse.de>
19183 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19185 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19186 (vectorizable_reduction): Perform reduction PHI creation when
19187 visiting a reduction PHI and adjust and simplify code generation
19188 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19189 (vect_transform_loop): Visit reduction PHIs.
19190 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19191 defs into the SLP tree.
19192 (vect_build_slp_tree): Reduction defs terminate the recursion.
19193 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19195 (vect_get_vec_defs_for_stmt_copy): Export.
19196 (vect_get_vec_defs): Likewise.
19197 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19199 (vect_get_vec_defs_for_stmt_copy): Declare.
19200 (vect_get_vec_defs): Likewise.
19202 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19204 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19205 parameter with a "loop" parameter and use it instead of the
19206 loop containing DR_STMT. Don't check simple_iv when doing
19207 BB analysis. Describe the two analysis modes in the comment.
19209 2017-07-03 Tom de Vries <tom@codesourcery.com>
19211 PR tree-optimization/69468
19212 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19213 (find_same_succ_bb): Handle ignore_edge_flags.
19215 2017-07-03 Tom de Vries <tom@codesourcery.com>
19217 PR tree-optimization/81192
19218 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19220 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19222 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19224 2017-07-03 Tom de Vries <tom@codesourcery.com>
19226 PR tree-optimization/81192
19227 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19228 BB_SAME_SUCC (bb) == NULL.
19230 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19232 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19235 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19237 * dumpfile.c: Include profile-count.h
19238 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19240 (insert_cond_bb): Update profile.
19241 * tree-cfg.h (insert_cond_bb): Update prototype.
19242 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19243 * tree-dump.c: Do not include tree-cfg.
19245 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19247 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19249 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19251 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19254 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19256 * tree-complex.c (expand_complex_div_wide): update profile.
19258 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19259 Alan Hayward <alan.hayward@arm.com>
19260 David Sherwood <david.sherwood@arm.com>
19262 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19263 (CORETYPES_H): New define.
19264 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19265 (insn-modes-inline.h, s-modes-inline-h): New rules.
19266 (generated_files): Add insn-modes-inline.h.
19267 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19268 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19269 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19270 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19271 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19272 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19273 (build/gencodes.o, build/genconditions.o): Likewise.
19274 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19275 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19276 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19277 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19278 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19279 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19280 * coretypes.h: Include everything up to real.h for generators.
19281 Include insn-modes.h first. Include wide-int-print.h after
19282 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19283 * machmode.h: Don't include insn-modes.h here.
19284 * function-tests.c: Remove includes of signop.h, machmode.h,
19285 double-int.h and wide-int.h.
19287 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19289 * optc-save-gen.awk: Likewise.
19290 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19291 * godump.c: Remove include of wide-int-print.h.
19292 * pretty-print.h: Likewise.
19293 * wide-int-print.cc: Likewise.
19294 * wide-int.cc: Likewise.
19295 * hash-map-tests.c: Remove include of signop.h.
19296 * hash-set-tests.c: Likewise.
19297 * rtl-tests.c: Likewise.
19298 * mkconfig.sh: Remove include of machmode.h.
19299 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19301 (emit_insn_modes_inline_h): ...this new function. Emit the code
19302 into an insn-modes-inline.h header file, adding appropriate
19303 include guards and end comments.
19304 (emit_insn_modes_c_header): Remove include of machmode.h.
19305 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19307 (main): Handle -i flag and call emit_insn_modes_inline_h when
19310 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19312 * tree-ssa-strlen.c (strinfo): Rename the length field to
19313 nonzero_chars. Add a full_string_p field.
19314 (compare_nonzero_chars, zero_length_string_p): New functions.
19315 (get_addr_stridx): Add an offset_out parameter.
19316 Use compare_nonzero_chars.
19317 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19318 (new_strinfo): Update after above changes to strinfo.
19319 (set_endptr_and_length): Set full_string_p.
19320 (get_string_length): Update after above changes to strinfo.
19321 (unshare_strinfo): Update call to new_strinfo.
19322 (maybe_invalidate): Likewise.
19323 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19324 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19325 as a uhwi instead of an shwi. Update after above changes to
19326 strinfo and new_strinfo.
19327 (zero_length_string): Assert that chainsi contains full strings.
19328 Use zero_length_string_p. Update call to new_strinfo.
19329 (adjust_related_strinfos): Update after above changes to strinfo.
19330 Copy full_string_p from origsi.
19331 (adjust_last_stmt): Use zero_length_string_p.
19332 (handle_builtin_strlen): Update after above changes to strinfo and
19333 new_strinfo. Install the lhs as the string length if the previous
19334 entry didn't describe a full string.
19335 (handle_builtin_strchr): Update after above changes to strinfo
19337 (handle_builtin_strcpy): Likewise.
19338 (handle_builtin_strcat): Likewise.
19339 (handle_builtin_malloc): Likewise.
19340 (handle_pointer_plus): Likewise.
19341 (handle_builtin_memcpy): Likewise. Track nonzero characters
19342 that aren't necessarily followed by a nul terminator.
19343 (handle_char_store): Likewise.
19345 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19347 PR tree-optimization/80769
19348 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19349 for malloc and calloc. Document the new invariant that all related
19350 strinfos have delayed lengths or none do.
19351 (verify_related_strinfos): Move earlier in file.
19352 (set_endptr_and_length): New function, split out from...
19353 (get_string_length): ...here. Also set the lengths of related
19355 (zero_length_string): Assert that chainsi has known (rather than
19357 (adjust_related_strinfos): Likewise.
19359 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19361 PR tree-optimization/81136
19362 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19363 assert that two references with the same misalignment have the same
19364 compile-time misalignment if those compile-time misalignments
19367 2017-07-01 Andi Kleen <ak@linux.intel.com>
19369 * print-tree.c (print_node): Print all attributes.
19371 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19373 * cfg.c (scale_bbs_frequencies): New function.
19374 * cfg.h (scale_bbs_frequencies): Declare it.
19375 * cfgloopanal.c (single_likely_exit): Cleanup.
19376 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19378 (scale_loop_profile): Likewise.
19379 (loop_version): Likewise.
19380 (create_empty_loop_on_edge): Update.
19381 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19382 scale_loop_frequencies, scale_loop_profile, loopify,
19383 loop_version): Update prototypes.
19384 * modulo-sched.c (sms_schedule): Update.
19385 * predict.c (unlikely_executed_edge_p): Also check probability.
19386 (probably_never_executed_edge_p): Fix typo.
19387 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19388 * tree-parloops.c (gen_parallel_loop): Update.
19389 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19390 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19391 * tree-ssa-loop-split.c (split_loop): Update.
19392 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19393 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19394 (vect_loop_versioning): Update.
19395 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19397 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19399 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19401 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19403 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19404 to keep profile consistent.
19406 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19408 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19409 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19410 * profile-count.h (max_safe_multiplier): Make unsigned.
19411 (profile_count::guessed_zero): New.
19413 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19415 * bb-reorder.c (fix_up_crossing_landing_pad,
19416 fix_crossing_conditional_branches): Use make_single_succ_edge
19417 to keep profile consistent.
19419 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19421 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19424 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19427 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19428 the right scopes, make sure cond_jump isn't preserved between multiple
19429 iterations. Search for fallthru edge whenever there are 3+ edges and
19430 use find_fallthru_edge for it.
19432 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19434 Patch by Alexander Monakov <amonakov@ispras.ru>
19435 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19436 probabilities consistently.
19438 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19440 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19441 emit_cmp_and_jump_insns.
19443 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19446 * tree-inline.c (expand_call_inline): Combine profile statuses.
19448 2017-06-30 Andrew Pinski <apinski@cavium.com>
19450 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19451 fold_stmt returned true.
19453 2017-06-30 Nathan Sidwell <nathan@acm.org>
19455 * ggc.h (empty_string): Delete.
19456 * cfgexpand.c (expand_asm_stmt): Use plain "".
19457 * optabs.c (expand_asm_memory_barrier): Likewise.
19458 * stringpool.c (empty_string): Delete.
19459 (digit_vector, digit_string): Delete.
19460 (ggc_alloc_string): Use plain "", don't optimize single digit
19461 strings. Use ggc_alloc_atomic.
19463 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19465 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19466 comparison set and one other set, use the cost of the non-comparison
19469 2017-06-30 Nathan Sidwell <nathan@acm.org>
19471 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19474 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19476 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19477 loops. Remove now unneeded calls to gimple_switch_set_label() that
19478 just set removed labels to NULL_TREE.
19480 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19482 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19483 (set_range_info): ...here. Only call set_range_info_raw if domain
19485 (set_nonzero_bits): Call set_range_info_raw.
19486 * tree-ssanames.h (set_range_info_raw): New.
19488 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19491 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19492 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19493 of nonimmediate_operand and <store_mask_constraint> instead of m
19494 for the input operand. For V8FI iterator, always split if input
19495 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19496 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19497 <store_mask_predicate> instead of register_operand and
19498 <store_mask_constraint> instead of v for the input operand. Make
19499 sure both operands aren't MEMs for if not <mask_applied>.
19501 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19503 * lto-wrapper.c (copy_file) Close both file descriptors before
19506 2017-06-30 Martin Liska <mliska@suse.cz>
19509 * multiple_target.c (create_dispatcher_calls): Make ifunc
19510 also for function that don't have calls or are not referenced.
19512 2017-06-30 Richard Biener <rguenther@suse.de>
19514 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19515 analyze the first scalar stmt. Move vector type computation
19516 for the BB case here from ...
19517 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19518 live operation processing in the SLP case properly.
19520 2017-06-30 Richard Biener <rguenther@suse.de>
19522 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19524 2017-06-30 Martin Liska <mliska@suse.cz>
19527 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19528 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19530 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19532 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19535 2017-06-30 Richard Biener <rguenther@suse.de>
19537 PR tree-optimization/81249
19538 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19539 condition reduction result to original scalar type.
19541 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19543 * profile-count.h (enum profile_quality): Fix typos and whitespace
19546 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19548 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19549 type for branch probabilities.
19551 2017-06-29 Julian Brown <julian@codesourcery.com>
19552 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19554 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19555 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19556 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19557 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19559 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19561 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19562 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19563 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19564 CC usage from generic code to here.
19565 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19566 CC usage into the target macros.
19568 2017-06-29 Maya Rashish <coypu@sdf.org>
19570 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19573 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19575 * arm/arm-builtins.c: Include profile-count.h
19576 * except.c (sjlj_emit_function_enter): Use
19577 profile_probability::unlikely.
19579 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19581 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19582 and tocrel_offset be pointer args rather than implicitly using
19584 (legitimate_constant_pool_address_p, rs6000_emit_move,
19585 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19586 tocrel_offset and use in toc_relative_expr_p call.
19587 (print_operand, print_operand_address): Use static tocrel_base_oac
19588 and tocrel_offset_oac.
19589 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19592 2017-06-29 Maya Rashish <coypu@sdf.org>
19594 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19596 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19598 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19599 objects, take into account only the alignment of 'op0' and 'mode1' if
19602 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19604 * ccmp.c (ccmp_tree_comparison_p): New function.
19605 (ccmp_candidate_p): Update to use above function.
19606 (get_compare_parts): New function.
19607 (expand_ccmp_next): Update to use new functions.
19608 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19610 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19611 take mode as argument.
19612 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19613 * expr.c (expand_expr_real_1): Pass mode as argument.
19615 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19617 * combine.c (combine_instructions): Print insns to dump_file, together
19620 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19622 * asan.c (asan_emit_stack_protection): Update.
19623 (create_cond_insert_point): Update.
19624 * auto-profile.c (afdo_propagate_circuit): Update.
19625 * basic-block.h (struct edge_def): Turn probability to
19626 profile_probability.
19627 (EDGE_FREQUENCY): Update.
19628 * bb-reorder.c (find_traces_1_round): Update.
19629 (better_edge_p): Update.
19630 (sanitize_hot_paths): Update.
19631 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19632 (make_single_succ_edge): Update.
19633 (check_bb_profile): Update.
19634 (dump_edge_info): Update.
19635 (update_bb_profile_for_threading): Update.
19636 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19637 probabilitycount to 0.
19638 * cfgbuild.c (compute_outgoing_frequencies): Update.
19639 * cfgcleanup.c (try_forward_edges): Update.
19640 (outgoing_edges_match): Update.
19641 (try_crossjump_to_edge): Update.
19642 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19643 (expand_gimple_tailcall): Update.
19644 (construct_init_block): Use make_single_succ_edge.
19645 (construct_exit_block): Use make_single_succ_edge.
19646 * cfghooks.c (verify_flow_info): Update.
19647 (redirect_edge_succ_nodup): Update.
19648 (split_edge): Update.
19649 (account_profile_record): Update.
19650 * cfgloopanal.c (single_likely_exit): Update.
19651 * cfgloopmanip.c (scale_loop_profile): Update.
19652 (set_zero_probability): Remove.
19653 (duplicate_loop_to_header_edge): Update.
19654 * cfgloopmanip.h (loop_version): Update prototype.
19655 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19656 (force_nonfallthru_and_redirect): Update.
19657 (update_br_prob_note): Update.
19658 (rtl_verify_edges): Update.
19659 (purge_dead_edges): Update.
19660 (rtl_lv_add_condition_to_bb): Update.
19661 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19662 * cgraphunit.c (init_lowered_empty_function): Update.
19663 (cgraph_node::expand_thunk): Update.
19664 * cilk-common.c: Include profile-count.h
19665 * dojump.c (inv): Remove.
19666 (jumpifnot): Update.
19667 (jumpifnot_1): Update.
19668 (do_jump_1): Update.
19670 (do_jump_by_parts_greater_rtx): Update.
19671 (do_compare_rtx_and_jump): Update.
19672 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19673 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19674 * dwarf2cfi.c: Include profile-count.h
19675 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19676 (sjlj_emit_dispatch_table): Likewise.
19677 * explow.c: Include profile-count.h
19678 * expmed.c (emit_store_flag_force): Update.
19679 (do_cmp_and_jump): Update.
19680 * expr.c (compare_by_pieces_d::generate): Update.
19681 (compare_by_pieces_d::finish_mode): Update.
19682 (emit_block_move_via_loop): Update.
19683 (store_expr_with_bounds): Update.
19684 (store_constructor): Update.
19685 (expand_expr_real_2): Update.
19686 (expand_expr_real_1): Update.
19687 * expr.h (try_casesi, try_tablejump): Update prototypes.
19688 * gimple-pretty-print.c (dump_probability): Update.
19689 (dump_profile): New.
19690 (dump_gimple_label): Update.
19691 (dump_gimple_bb_header): Update.
19692 * graph.c (draw_cfg_node_succ_edges): Update.
19693 * hsa-gen.c (convert_switch_statements): Update.
19694 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19695 (find_if_case_1): Update.
19696 (find_if_case_2): Update.
19697 * internal-fn.c (expand_arith_overflow_result_store): Update.
19698 (expand_addsub_overflow): Update.
19699 (expand_neg_overflow): Update.
19700 (expand_mul_overflow): Update.
19701 (expand_vector_ubsan_overflow): Update.
19702 * ipa-cp.c (good_cloning_opportunity_p): Update.
19703 * ipa-split.c (split_function): Use make_single_succ_edge.
19704 * ipa-utils.c (ipa_merge_profiles): Update.
19705 * loop-doloop.c (add_test): Update.
19706 (doloop_modify): Update.
19707 * loop-unroll.c (compare_and_jump_seq): Update.
19708 (unroll_loop_runtime_iterations): Update.
19709 * lra-constraints.c (lra_inheritance): Update.
19710 * lto-streamer-in.c (input_cfg): Update.
19711 * lto-streamer-out.c (output_cfg): Update.
19712 * mcf.c (adjust_cfg_counts): Update.
19713 * modulo-sched.c (sms_schedule): Update.
19714 * omp-expand.c (expand_omp_for_init_counts): Update.
19715 (extract_omp_for_update_vars): Update.
19716 (expand_omp_ordered_sink): Update.
19717 (expand_omp_for_ordered_loops): Update.
19718 (expand_omp_for_generic): Update.
19719 (expand_omp_for_static_nochunk): Update.
19720 (expand_omp_for_static_chunk): Update.
19721 (expand_cilk_for): Update.
19722 (expand_omp_simd): Update.
19723 (expand_omp_taskloop_for_outer): Update.
19724 (expand_omp_taskloop_for_inner): Update.
19725 * omp-simd-clone.c (simd_clone_adjust): Update.
19726 * optabs.c (expand_doubleword_shift): Update.
19727 (expand_abs): Update.
19728 (emit_cmp_and_jump_insn_1): Update.
19729 (expand_compare_and_swap_loop): Update.
19730 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19731 * predict.c (predictable_edge_p): Update.
19732 (edge_probability_reliable_p): Update.
19733 (set_even_probabilities): Update.
19734 (combine_predictions_for_insn): Update.
19735 (combine_predictions_for_bb): Update.
19736 (propagate_freq): Update.
19737 (estimate_bb_frequencies): Update.
19738 (force_edge_cold): Update.
19739 * profile-count.c (profile_count::dump): Add missing space into dump.
19740 (profile_count::debug): Add newline.
19741 (profile_count::differs_from_p): Explicitly convert to unsigned.
19742 (profile_count::stream_in): Update.
19743 (profile_probability::dump): New member function.
19744 (profile_probability::debug): New member function.
19745 (profile_probability::differs_from_p): New member function.
19746 (profile_probability::differs_lot_from_p): New member function.
19747 (profile_probability::stream_in): New member function.
19748 (profile_probability::stream_out): New member function.
19749 * profile-count.h (profile_count_quality): Rename to ...
19750 (profile_quality): ... this one.
19751 (profile_probability): New.
19752 (profile_count): Update.
19753 * profile.c (compute_branch_probabilities): Update.
19754 * recog.c (peep2_attempt): Update.
19755 * sched-ebb.c (schedule_ebbs): Update.
19756 * sched-rgn.c (find_single_block_region): Update.
19757 (compute_dom_prob_ps): Update.
19758 (schedule_region): Update.
19759 * sel-sched-ir.c (compute_succs_info): Update.
19760 * stmt.c (struct case_node): Update.
19761 (do_jump_if_equal): Update.
19762 (get_outgoing_edge_probs): Update.
19763 (conditional_probability): Update.
19764 (emit_case_dispatch_table): Update.
19765 (expand_case): Update.
19766 (expand_sjlj_dispatch_table): Update.
19767 (emit_case_nodes): Update.
19768 * targhooks.c: Update.
19769 * tracer.c (better_p): Update.
19770 (find_best_successor): Update.
19771 * trans-mem.c (expand_transaction): Update.
19772 * tree-call-cdce.c: Update.
19773 * tree-cfg.c (gimple_split_edge): Upate.
19774 (move_sese_region_to_fn): Upate.
19775 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19776 * tree-eh.c (lower_resx): Upate.
19777 (cleanup_empty_eh_move_lp): Upate.
19778 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19779 * tree-inline.c (copy_edges_for_bb): Update.
19780 (copy_cfg_body): Update.
19781 * tree-parloops.c (gen_parallel_loop): Update.
19782 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19783 (gimple_gen_time_profiler): Update.
19784 * tree-ssa-dce.c (remove_dead_stmt): Update.
19785 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19786 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19787 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19788 (unloop_loops): Update.
19789 (try_peel_loop): Update.
19790 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19791 * tree-ssa-loop-split.c (connect_loops): Update.
19792 (split_loop): Update.
19793 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19794 (hoist_guard): Update.
19795 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19796 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19797 (value_replacement): Update.
19798 * tree-ssa-reassoc.c (branch_fixup): Update.
19799 * tree-ssa-tail-merge.c (replace_block_by): Update.
19800 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19801 (create_edge_and_update_destination_phis): Update.
19802 (compute_path_counts): Update.
19803 (recompute_probabilities): Update.
19804 (update_joiner_offpath_counts): Update.
19805 (freqs_to_counts_path): Update.
19806 (duplicate_thread_path): Update.
19807 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19808 (struct switch_conv_info): Update.
19809 (gen_inbound_check): Update.
19810 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19811 (vect_do_peeling): Update.
19812 (vect_loop_versioning): Update.
19813 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19814 (optimize_mask_stores): Update.
19815 * ubsan.c (ubsan_expand_null_ifn): Update.
19816 * value-prof.c (gimple_divmod_fixed_value): Update.
19817 (gimple_divmod_fixed_value_transform): Update.
19818 (gimple_mod_pow2): Update.
19819 (gimple_mod_pow2_value_transform): Update.
19820 (gimple_mod_subtract): Update.
19821 (gimple_mod_subtract_transform): Update.
19822 (gimple_ic): Update.
19823 (gimple_stringop_fixed_value): Update.
19824 (gimple_stringops_transform): Update.
19825 * value-prof.h: Update.
19827 2017-06-29 Carl Love <cel@us.ibm.com>
19829 * config/rs6000/rs6000-c.c: Add support for built-in functions
19830 vector signed int vec_signed (vector float);
19831 vector signed long long vec_signed (vector double);
19832 vector signed int vec_signed2 (vector double, vector double);
19833 vector signed int vec_signede (vector double);
19834 vector signed int vec_signedo (vector double);
19835 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19836 instruction generator.
19837 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19838 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19839 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19841 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19842 vunsignede_v2df): Add define_expands.
19843 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19844 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19845 VEC_UNSIGNEDO): Add definitions.
19846 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19847 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19848 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19849 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19850 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19851 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19852 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19853 vec_unsignede, vec_unsignedo): Add builtin defines.
19854 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19856 * doc/extend.texi: Update the built-in documentation file for the
19857 new built-in functions.
19859 2017-06-29 Richard Biener <rguenther@suse.de>
19861 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19862 reduction chains to LOOP_VINFO_REDUCTIONS.
19863 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19864 SLP reductions after processing reduction chains.
19866 2017-06-29 Nathan Sidwell <nathan@acm.org>
19868 * builtins.c (fold_builtin_FUNCTION): Use
19869 lang_hooks.decl_printable_name.
19871 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19873 PR middle-end/81194
19874 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19875 with only one label.
19876 * stmt.c (expand_case): Assert NCASES is greater than one.
19878 2017-06-29 Richard Biener <rguenther@suse.de>
19880 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19882 (group_case_labels): Likewise.
19883 (find_taken_edge): Push sanity checking on val to workers...
19884 (find_taken_edge_cond_expr): ... here
19885 (find_taken_edge_switch_expr): ... and here, handle cases
19886 with just a default label.
19887 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19888 (group_case_labels): Likewise.
19889 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19890 group_case_labels does anything cleanup the CFG again.
19892 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19894 PR tree-optimization/81196
19895 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19896 exit condition comparing two IVs.
19898 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19900 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19901 profile to the dummy entry at the end of the list of architectures.
19902 * config/arm/arm-cpu-cdata.h: Regenerated.
19904 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19905 Michael Collison <michael.collison@arm.com>
19908 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19910 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19911 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19912 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19913 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19914 of shift when the shift amount is masked with constant equal to
19915 the size of the mode.
19916 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19919 2017-06-29 Martin Liska <mliska@suse.cz>
19921 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19923 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19926 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19929 (edge_growth_cache_entry::edge_growth_cache_entry): New
19931 (reset_edge_growth_cache): Update to use constructor.
19933 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19935 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19936 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19937 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19939 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19941 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19942 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19944 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19946 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19947 (*-linux-uclibc*): Add t-uclibc tmake_file.
19948 * config/t-musl: New.
19949 * config/t-uclibc: New.
19951 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19953 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19955 (gen_comm_data): Emit architectural setting of arch_prof.
19956 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19958 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19959 (armv8-m.base, armv8-m.main): Likewise.
19960 * arm-protos.h (arm_build_target): Add profile field.
19961 (arch_option): Likewise.
19962 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19964 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19965 arm_active_target.profile.
19967 2017-06-28 Richard Biener <rguenther@suse.de>
19969 PR middle-end/81227
19970 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19971 TYPE_OVERFLOW_WRAPS.
19972 * match.pd (negate_expr_p): Likewise.
19973 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19974 fold_build2, not fold_binary.
19976 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19978 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19979 Convert memory address to Pmode.
19980 (aarch64_print_operand): Assert MEM operands are always Pmode.
19982 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19985 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19986 Remove redundant if.
19987 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19988 * config/arm/aarch-common-protos.h
19989 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19990 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19992 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19995 * multiple_target.c (create_dispatcher_calls): Set the default
19996 clone to be static, not public.
19998 2017-06-28 Richard Biener <rguenther@suse.de>
20000 * tree-vect-loop.c (vectorizable_reduction): Move special
20001 cond reduction IV var creation ...
20002 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20003 parameter. Use STMT_VINFO_VECTYPE.
20004 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20007 2017-06-28 Martin Liska <mliska@suse.cz>
20010 * ipa-visibility.c (non_local_p): Handle visibility.
20012 2017-06-28 Martin Liska <mliska@suse.cz>
20015 * common.opt: Add IntegerRange to various options.
20016 * opt-functions.awk (integer_range_info): New function.
20017 * optc-gen.awk: Add integer_range_info to cl_options struct.
20018 * opts-common.c (decode_cmdline_option): Handle
20019 CL_ERR_INT_RANGE_ARG.
20020 (cmdline_handle_error): Likewise.
20021 * opts.c (print_filtered_help): Show valid interval in
20022 when --help is provided.
20023 * opts.h (struct cl_option): Add range_min and range_max fields.
20024 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20026 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20028 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20029 (x * C EQ/NE y * C): New transformation.
20031 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20033 * genmultilib (combination_space): Accept '+' in option names.
20035 2017-06-28 Martin Liska <mliska@suse.cz>
20038 * asan.c (instrument_derefs): Bail out inner references
20039 that are hard register variables.
20041 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20044 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20045 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20047 2017-06-28 Richard Biener <rguenther@suse.de>
20049 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20050 (vect_get_slp_defs): Adjust.
20051 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20053 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20055 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20056 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20057 (vectorizable_reduction): Adjust.
20058 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20060 (vect_get_slp_defs): Likewise.
20061 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20062 (vectorizable_bswap): Adjust.
20063 (vectorizable_call): Likewise.
20064 (vectorizable_conversion): Likewise.
20065 (vectorizable_assignment): Likewise.
20066 (vectorizable_shift): Likewise.
20067 (vectorizable_operation): Likewise.
20068 (vectorizable_store): Likewise.
20069 (vectorizable_condition): Likewise.
20070 (vectorizable_comparison): Likewise.
20072 2017-06-28 Michael Collison <michael.collison@arm.com>
20075 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20077 (arm_gen_movmemqi): Removed unused variable 'i'.
20078 Convert 'for' loop into 'while' loop.
20079 (arm_expand_prologue): Remove last unnecessary set of insn.
20080 (thumb_pop): Remove unused variable 'pushed_words'.
20081 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20083 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20085 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20086 * config/s390/s390-protos.h: Add prototype of
20087 s390_rel_address_ok_p.
20088 * config/s390/s390.c (s390_got_symbol): New function.
20089 (s390_rel_address_ok_p): New function.
20090 (legitimize_pic_address): Use s390_rel_address_ok_p.
20091 (s390_load_got): Use s390_got_symbol.
20092 (s390_option_override): Issue error if
20093 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20094 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20096 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20098 2017-06-27 Andrew Pinski <apinski@cavium.com>
20100 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20101 (X * copysign (1.0, X)): New pattern.
20102 (X * copysign (1.0, -X)): New pattern.
20103 (copysign (-1.0, CST)): New pattern.
20105 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20107 * genmultilib (combination_space): Remove variable.
20108 Validate reuse rules against regular expression for any sequence
20109 of multilib options in any order.
20111 2017-06-27 Michael Collison <michael.collison@arm.com>
20113 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20114 call aarch64_split_simd_combine.
20115 * (aarch64_combine_internal<mode>): Delete pattern.
20116 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20117 Allow register and subreg operands.
20119 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20121 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20122 specific need, just fallback on defaults.
20123 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20125 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20126 Olivier Hainque <hainque@adacore.com>
20128 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20130 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20131 targets. Pick a default if no particular attempt applied.
20132 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20135 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20137 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20138 (x86_64-wrs-vxworks7): Likewise.
20140 2017-06-27 Marek Polacek <polacek@redhat.com>
20143 * ubsan.c (instrument_null): Check get_base_address's result for null.
20145 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20147 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20149 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20151 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20152 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20153 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20154 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20155 New function types.
20156 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20157 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20158 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20159 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20160 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20161 BUILT_IN_FEUPDATEENV): New builtins.
20162 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20163 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20164 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20165 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20167 (builtin_structptr_types): Adjust size.
20168 * tree.c (builtin_structptr_types): Add four entries.
20170 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20171 Olivier Hainque <hainque@adacore.com>
20173 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20174 (TLS_SYM): New local macro, forcing reference to __tls__ on
20175 link command lines for VxWorks 7 RTPs, triggering initialization
20177 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20178 OS features TLS support, true for RTPs on VxWorks 7.
20179 * config/vxworks.c (vxworks_override_options): Setup emutls
20182 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20184 * predict.c (test_prediction_value_range): Use -1U instead of -1
20185 to avoid narrowing conversion warning.
20186 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20187 to avoid narrowing conversion warning.
20188 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20190 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20192 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20194 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20195 64bit configurations.
20196 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20197 (SIZE_TYPE): Likewise.
20198 * config/vxworks.c (vxworks_emutls_var_fields): Use
20199 long_unsigned_type_node instead of unsigned_type_node as the offset
20200 field type, which is "pointer" mode in emutls.c.
20202 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20205 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20207 PR middle-end/81207
20208 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20209 gimple_vuse copying separately from gimple_vdef copying.
20211 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20213 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20216 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20217 Olivier Hainque <hainque@adacore.com>
20219 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20220 for all vxworks7 targets.
20221 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20222 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20223 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20224 variations for VX6/VX7 and 32/64bits later on in ...
20225 (VXWORKS_LIB_SPEC): Leverage new macros.
20226 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20227 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20229 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20231 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20232 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20234 2017-06-26 Carl Love <cel@us.ibm.com>
20236 * config/rs6000/rs6000-c.c: Add support for built-in functions
20237 vector bool char vec_reve (vector bool char);
20238 vector signed char vec_reve (vector signed char);
20239 vector unsigned char vec_reve (vector unsigned char);
20240 vector bool int vec_reve (vector bool int);
20241 vector signed int vec_reve (vector signed int);
20242 vector unsigned int vec_reve (vector unsigned int);
20243 vector bool long long vec_reve (vector bool long long);
20244 vector signed long long vec_reve (vector signed long long);
20245 vector unsigned long long vec_reve (vector unsigned long long);
20246 vector bool short vec_reve (vector bool short);
20247 vector signed short vec_reve (vector signed short);
20248 vector double vec_reve (vector double);
20249 vector float vec_reve (vector float);
20250 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20251 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20252 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20253 (altivec_vreve): New pattern.
20254 * config/rs6000/altivec.h (vec_reve): New define.
20255 * doc/extend.texi (vec_rev): Update the built-in documentation file
20256 for the new built-in functions.
20258 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20260 PR tree-optimization/71815
20261 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20263 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20265 (slsr_process_phi): Likewise.
20266 (replace_uncond_cands_and_profitable_phis): Don't replace a
20267 multiply candidate with a stride of 1 (copy or cast).
20268 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20270 (lowest_cost_path): Likewise.
20271 (total_savings): Likewise.
20273 2017-06-26 Richard Biener <rguenther@suse.de>
20276 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20277 Use def_builtin_pure for all gather builtins.
20279 2017-06-26 Richard Biener <rguenther@suse.de>
20281 PR tree-optimization/81203
20282 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20283 non-dominating BBs.
20285 2017-06-26 Marek Polacek <polacek@redhat.com>
20288 * doc/invoke.texi: Document -Wmultistatement-macros.
20290 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20292 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20293 arm_neon_ok_no_float_abi effective target.
20295 2017-06-26 Richard Biener <rguenther@suse.de>
20297 PR tree-optimization/80928
20298 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20299 (copy_bbs): Set BB_DUPLICATED flag early.
20300 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20302 (execute_on_shrinking_pred): Likewise.
20303 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20304 BB_DUPLICATED blocks.
20305 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20306 iterate over all PHIs considering removal of *gsi.
20308 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20310 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20313 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20315 * config/rs6000/rs6000-string.c: (expand_block_clear,
20316 do_load_for_compare, select_block_compare_mode,
20317 compute_current_alignment, expand_block_compare,
20318 expand_strncmp_align_check, expand_strn_compare,
20319 expand_block_move, rs6000_output_load_multiple)
20320 Move functions related to string/block move/compare
20321 to a separate file.
20322 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20323 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20324 for this function which is now used in two files.
20325 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20326 * config.gcc: Add rs6000-string.o to extra_objs for
20327 targets powerpc*-*-* and rs6000*-*-*.
20329 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20332 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20333 32-bit, since indexed is not valid for DImode.
20334 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20335 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20336 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20337 (define_peephole2 for Altivec d-form store): Likewise.
20340 * multiple_target.c (create_dispatcher_calls): Only create the
20341 dispatcher call if the function is the default clone of a
20342 versioned function.
20344 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20346 PR middle-end/80902
20347 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20348 a call, force the call to not be a tail call.
20350 2017-06-23 Jeff Law <law@redhat.com>
20352 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20355 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20357 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20358 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20359 handling for early expansion of vector shifts (sl,sr,sra,rl).
20360 (builtin_function_type): Add vector shift right instructions
20361 to the unsigned argument list.
20363 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20365 rtl-optimizatoin/79286
20366 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20367 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20368 trap. PIC register plus a const unspec without offset can never trap.
20370 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20372 * tree.h (builtin_structptr_type): New type.
20373 (builtin_structptr_types): Declare new array.
20374 * tree.c (builtin_structptr_types): New array.
20375 (free_lang_data, build_common_tree_nodes): Use it.
20377 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20380 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20383 2017-06-22 Matt Turner <mattst88@gmail.com>
20385 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20386 Lake models to skylake case. Assume skylake for unknown
20387 models with clflushopt.
20389 2017-06-22 Jeff Law <law@redhat.com>
20391 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20392 frame sizes that do not satisfy aarch64_uimm12_shift.
20394 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20396 * profile-count.h (apply_probability,
20397 apply_scale, probability_in): Fix checks for zero.
20399 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20401 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20402 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20404 2016-06-22 Richard Biener <rguenther@suse.de>
20406 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20407 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20408 REDUC_MAX_EXPR support.
20409 (vectorizable_reduction): Likewise.
20410 (vect_create_epilog_for_reduction): Likewise.
20412 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20414 * match.pd (A / (1 << B) -> A >> B): New.
20415 * generic-match-head.c: Include optabs-tree.h.
20416 * gimple-match-head.c: Likewise.
20417 * optabs-tree.h (target_supports_op_p): New.
20418 * optabs-tree.c (target_supports_op_p): New.
20420 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20422 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20423 $gcc_cv_ld --help output.
20424 (gcc_cv_ld_demangle): Likewise.
20425 (gcc_cv_ld_eh_frame_hdr): Likewise.
20426 (gcc_cv_ld_pie): Likewise.
20427 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20428 (gcc_cv_ld_buildid): Likewise.
20429 (gcc_cv_ld_sysroot): Likewise.
20430 (ld_bndplt_support): Likewise.
20431 (ld_pushpopstate_support): Likewise.
20432 * configure: Regenerate.
20433 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20435 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20438 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20439 operands indexes to avoid gap between operands and match_dups.
20441 2017-06-21 Andrew Pinski <apinski@cavium.com>
20443 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20444 Increment Arith_shift and Arith_shift_reg by 1.
20445 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20447 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20448 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20449 (aarch64_strip_extend): Add new argument and test for it.
20450 (aarch64_cheap_mult_shift_p): New function.
20451 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20452 add a cost if it is true.
20453 Update calls to aarch64_strip_extend.
20454 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20456 2017-06-21 Andrew Pinski <apinski@cavium.com>
20458 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20460 (thunderxt88): Likewise.
20461 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20462 (thunderx_prefetch_tune): New variable.
20463 (thunderx2t99_prefetch_tune): Update for the correct values.
20464 (thunderxt88_tunings): New variable.
20465 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20466 generic_prefetch_tune.
20467 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20469 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20471 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20472 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20473 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20474 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20475 (aarch64_atomic_cas<mode>, GPI): Likewise.
20477 2017-06-21 Martin Liska <mliska@suse.cz>
20479 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20480 statements on cold and hot labels.
20481 * predict.c (tree_estimate_probability_bb): Remove the
20482 prediction from this place.
20484 2017-06-21 Martin Liska <mliska@suse.cz>
20486 PR tree-optimization/79489
20487 * gimplify.c (maybe_add_early_return_predict_stmt): New
20489 (gimplify_return_expr): Call the function.
20490 * predict.c (tree_estimate_probability_bb): Remove handling
20492 * predict.def: Update comment about early return predictor.
20493 * gimple-predict.h (is_gimple_predict): New function.
20494 * predict.def: Change default value of early return to 66.
20495 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20497 * passes.def: Put pass_strip_predict_hints to the beginning of
20500 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20502 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20503 FUNCTION_DECL declarations.
20504 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20506 (dwaf2out_decl): Likewise.
20507 * godump.c (go_early_global_decl): Skip call to the real debug hook
20508 for FUNCTION_DECL declarations.
20509 * passes.c (rest_of_decl_compilation): Skip call to the
20510 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20511 -fdump-go-spec is passed.
20513 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20515 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20516 Reorder for compactness.
20517 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20518 (def_builtin_pure, def_builtin_pure2): New functions.
20519 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20521 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20523 * match.pd (nop_convert): New predicate.
20524 ((A +- CST1) +- CST2): Allow some NOP conversions.
20526 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20529 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20530 with ctors/dtors if GOVD_SHARED is set.
20532 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20534 * config/aarch64/aarch64.md (movti_aarch64):
20535 Emit mov rather than orr.
20536 (movtf_aarch64): Likewise.
20537 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20538 Emit mov rather than orr.
20540 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20542 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20543 Swap alternatives, make integer dup more expensive.
20545 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20547 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20548 Return true for non-tls symbols.
20550 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20552 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20553 (cortex-a75): Likewise.
20554 (cortex-a75.cortex-a55): Likewise.
20555 * config/aarch64/aarch64-tune.md: Regenerate.
20556 * doc/invoke.texi (-mtune): Document new values for -mtune.
20558 2017-06-21 Tom de Vries <tom@codesourcery.com>
20560 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20561 stack_size feature.
20562 (Effective-Target Keywords, Other attributes): Suggest using
20563 dg-add-options stack_size feature to get stack limit in stack_size
20564 effective target documentation.
20566 2017-06-21 Julian Brown <julian@codesourcery.com>
20567 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20569 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20570 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20571 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20573 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20574 attribute type list for neon_multiply.
20575 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20576 attribute type list for neon_multiply.
20577 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20578 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20579 attribute type list for neon_multiply.
20580 * config/arm/types.md (crypto_pmull): Add.
20581 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20582 attribute type list.
20584 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20586 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20589 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20591 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20592 to make sure not to dereference a NULL cost_classes_ptr pointer.
20594 2017-06-20 Carl Love <cel@us.ibm.com>
20596 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20597 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20598 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20599 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20600 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20601 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20602 VMULOSW): New enum "unspec" values.
20603 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20604 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20605 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20606 altivec_vmulosw): New patterns.
20607 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20608 VMULOSW): Add definitions.
20610 2017-06-20 Julia Koval <julia.koval@intel.com>
20612 * config/i386/i386.c: Fix rounding expand for new pattern.
20613 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20615 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20617 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20618 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20620 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20622 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20625 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20627 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20628 architecture, then by implementer ID.
20629 * config/aarch64/aarch64-tune.md: Regenerate.
20631 2017-06-20 Richard Biener <rguenther@suse.de>
20633 PR middle-end/81097
20634 * fold-const.c (split_tree): Fold to type before negating.
20636 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20638 * diagnostic-show-locus.c
20639 (selftest::test_fixit_deletion_affecting_newline): New function.
20640 (selftest::diagnostic_show_locus_c_tests): Call it.
20642 2017-06-20 Andreas Schwab <schwab@suse.de>
20645 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20648 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20650 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20651 __ARM_FEATURE_COPROC according to support.
20653 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20655 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20656 Rewritten to avoid overflow for > 32-bit pointers.
20659 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20660 by removing enum keyword.
20661 (ubsan_type_descriptor): Likewise. Formatting fix.
20664 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20665 splitter): Require TARGET_SSE2 in the condition.
20667 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20670 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20671 for doing vector set of SFmode on ISA 3.0.
20672 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20673 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20675 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20676 SFmode value into a V4SF variable that was extracted from another
20677 V4SF variable without converting the element to double precision
20678 and back to single precision vector format.
20679 (vsx_insert_extract_v4sf_p9_2): Likewise.
20681 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20683 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20684 in UWHI to avoid undefined overflow.
20687 * ubsan.h (enum ubsan_encode_value_phase): New.
20688 (ubsan_encode_value): Change second argument to
20689 enum ubsan_encode_value_phase with default value of
20690 UBSAN_ENCODE_VALUE_GENERIC.
20691 * ubsan.c (ubsan_encode_value): Change second argument to
20692 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20693 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20694 create_tmp_var_raw instead of create_tmp_var and use a
20696 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20697 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20698 ubsan_encode_value callers.
20701 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20702 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20703 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20705 2017-06-19 Richard Biener <rguenther@suse.de>
20707 PR middle-end/81118
20708 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20709 estimates if we changed anything.
20711 2017-06-19 Richard Biener <rguenther@suse.de>
20713 PR tree-optimization/80887
20714 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20715 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20716 simplified lookups, then reset mprts_hook.
20717 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20719 (try_to_simplify): Likewise.
20721 2017-06-19 Martin Liska <mliska@suse.cz>
20724 * gimplify.c (gimplify_switch_expr):
20725 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20727 2017-06-19 Martin Liska <mliska@suse.cz>
20729 * doc/install.texi: Document that PGO runs in 4 stages.
20731 2017-06-19 Martin Liska <mliska@suse.cz>
20734 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20735 to dispatcher function name.
20736 * multiple_target.c (replace_function_decl): New function.
20737 (create_dispatcher_calls): Redirect both edges and references.
20739 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20741 * profile-count.c (profile_count::dump): Dump quality.
20742 (profile_count::differs_from_p): Update for unsigned val.
20743 * profile-count.h (profile_count_quality): New enum.
20744 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20746 2017-06-19 Richard Biener <rguenther@suse.de>
20748 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20749 struct function as arg.
20750 (estimate_numbers_of_iterations): Export overload with loop arg.
20751 (free_numbers_of_iterations_estimates_loop): Use an overload of
20752 free_numbers_of_iterations_estimates instead.
20753 * tree-cfg.c (remove_bb): Adjust.
20754 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20755 * tree-parloops.c (gen_parallel_loop): Likewise.
20756 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20758 (tree_unroll_loops_completely): Likewise.
20759 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20760 Use an overload instead and export.
20761 (estimated_loop_iterations): Adjust.
20762 (max_loop_iterations): Likewise.
20763 (likely_max_loop_iterations): Likewise.
20764 (estimate_numbers_of_iterations): Take struct function as arg
20766 (loop_exits_before_overflow): Adjust.
20767 (free_numbers_of_iterations_estimates_loop): Use an overload.
20768 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20769 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20771 2017-06-19 Richard Biener <rguenther@suse.de>
20774 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20775 RANGE_EXPR conservatively.
20777 2017-06-16 Carl Love <cel@us.ibm.com>
20779 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20780 definitions for vec_float, vec_float2, vec_floato,
20781 vec_floate built-ins.
20782 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20783 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20785 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20786 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20787 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20788 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20789 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20790 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20791 vec_floato): Add builtin defines.
20792 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20793 Update the built-in documentation file for the new built-in
20796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20798 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20799 (mthumb): Mark as the negative of -marm.
20801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20803 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20805 (ARM Options, -mtune): Document that this accepts the same
20806 extension options as -mcpu.
20807 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20811 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20812 permitted extensions.
20814 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20816 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20817 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20818 (armv8-m.main): Add option +nodsp.
20819 * config/arm/arm-cpu-cdata.h: Regenerated.
20821 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20823 * config/arm/t-fuchsia: New file.
20824 * config.gcc (arm*-*-fuchsia*): Use it.
20826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20828 * config/arm/t-symbian: Rewrite for new option infrastructure.
20830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20832 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20833 (MULTILIB_REQUIRED): Likewise.
20835 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20837 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20838 (MULTILIB_RESUE): Likewise.
20839 (MULTILIB_MATCHES): Likewise.
20840 (MULTLIB_REQUIRED): Likewise.
20842 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20844 * config/arm/t-rtems: Rewrite for new option framework.
20846 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20848 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20849 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20850 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20851 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20852 * config/arm/t-multilib: ... here.
20853 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20854 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20855 armv7-a and armv8*-a when A-profile libraries have not been built.
20856 * config/arm/t-rmprofile: Rewrite.
20858 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20860 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20861 with a backslash. Remove the backslash after substituting unescaped
20863 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20865 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20867 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20868 the driver to pass through the default setting of -mfloat-abi.
20869 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20871 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20872 (all_feat_combs): New rule.
20873 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20875 * config/arm/t-aprofile: Rewrite.
20877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20879 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20880 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20881 fpu is not specified by the user/command-line.
20882 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20883 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20884 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20885 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20886 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20887 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20891 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20892 * config/arm/t-arm-elf: Rewritten.
20894 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20896 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20897 have some floating-point instructions.
20898 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20899 (TARGET_MAYBE_HARD_FLOAT): New macro.
20900 * config/arm/arm-builtins.c (arm_init_builtins): Use
20901 TARGET_MAYBE_HARD_FLOAT.
20902 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20906 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20907 (configargs.h): Include it.
20908 (arm_print_hint_for_fpu_option): New function.
20909 (arm_parse_fpu_option): New function.
20910 (candidate_extension): New class.
20911 (arm_canon_for_multilib): New function.
20912 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20913 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20914 (ARCH_CANONICAL_SPECS): New macro.
20915 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20917 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20919 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20920 are set after handling multilib fragments. Set target_cpu_default2
20923 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20925 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20927 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20928 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20930 (check_arch): Likewise.
20931 * config/arm/arm.c (arm_configure_build_target): Handle
20932 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20933 options in the default.
20935 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20937 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20938 when an option is an alias of another.
20939 * config/arm/parsecpu.awk (optalias): New parser token.
20940 (gen_comm_data): Mark non-alias options as such. Emit entries
20941 for extension aliases.
20942 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20943 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20944 (armv6kz, armv6zk, armv6t2): Likewise.
20945 (armv7): Make vfpv3-d16 an alias.
20946 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20948 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20949 Sort in canonical order.
20950 (armv8-a): Sort in canonical order.
20951 (armv8.1-a, armv8.2-a): Likewise.
20952 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20954 (cortex-a9): Sort in canonical order.
20955 * config/arm/arm.c (selftests.h): Include it.
20956 (arm_test_cpu_arch_data): New function.
20957 (arm_run_self_tests): New function.
20958 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20959 (targetm): Move declaration to the end of the file.
20960 * arm-cpu-cdata.h: Regenerated.
20962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20964 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20965 call to target_mode_check describing the type of option passed.
20966 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20967 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20968 arm_parse_cpu_option_name to match parameters against list of
20970 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20971 arm_arch_core_flags data structure.
20972 * config/arm/arm-cpu_cdata.h: Regenerated.
20974 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20976 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20978 (arm_print_hint_for_cpu_option): Likewise.
20979 (arm_print_hint_for_arch_option): Likewise.
20980 (arm_parse_cpu_option_name): Likewise.
20981 (arm_parse_arch_option_name): Likewise.
20982 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20983 of entries in the all_fpus list.
20984 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20985 (arm_parse_cpu_option_name): Declare.
20986 (arm_parse_arch_option_name): Declare.
20987 (arm_parse_option_features): Declare.
20988 (arm_intialize_isa): Declare.
20989 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20991 (gen_comm_data): ... here. Make definitions non-static.
20992 * config/arm/arm-cpu-data.h: Regenerated.
20993 * config/arm/arm-cpu-cdata.h: Regenerated.
20995 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20997 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20998 (cpu_arch_extension): New structure.
20999 (cpu_arch_option, arch_option, cpu_option): New structures.
21000 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21001 architecture types.
21002 (gen_data): Generate new format data tables.
21003 * config/arm/arm.c (cpu_tune): New structure.
21004 (cpu_option, processors): Delete.
21005 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21006 (arm_print_hint_for_cpu_option): ... this and ...
21007 (arm_print_hint_for_arch_option): ... this.
21008 (arm_parse_arch_cpu_name): Delete. Replace with ...
21009 (arm_parse_cpu_option_name): ... this and ...
21010 (arm_parse_arch_option_name): ... this.
21011 (arm_unrecognized_feature): Change type of target parameter to
21013 (arm_parse_arch_cpu_features): Delete. Replace with ...
21014 (arm_parse_option_features): ... this.
21015 (arm_configure_build_target): Rework to use new configuration data
21017 (arm_print_tune_info): Rework for new configuration data tables.
21018 * config/arm/arm-cpu-data.h: Regenerated.
21019 * config/arm/arm-cpu.h: Regenerated.
21021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21023 * Makefile.in (OBJS): Move sbitmap.o from here ...
21024 (OBJS-libcommon): ... to here.
21026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21028 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21029 (ISA_ALL_CRYPTO): New macro.
21030 (ISA_ALL_SIMD): New macro
21031 (ISA_ALL_FP): New macro.
21032 * config/arm/arm.c (fpu_bitlist): Update initializer.
21033 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21035 (arm9e): Add fpu. Add option for nofp
21036 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21037 (arm926ej-s, arm1026ej-s): Likewise.
21038 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21039 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21040 neon-fp16, neon-vfpv4, nofp and nosimd.
21041 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21042 (cortex-a8): Add fpu. Add option for nofp.
21043 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21044 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21045 (cortex-r4f): Add fpu.
21046 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21047 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21049 (cortex-r8): Likewise.
21050 (cortex-m4): Add fpu. Add option for nofp.
21051 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21052 (cortex-a17.cortex-a7): Likewise.
21053 (cortex-a32): Add fpu. Add options for crypto and nofp.
21054 (cortex-a35, cortex-a53): Likewise.
21055 (cortex-a57): Add fpu. Add option for crypto.
21056 (cortex-a72, cortex-a73): Likewise.
21057 (exynos-m1): Likewise.
21058 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21059 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21060 (cortex-m33): Add fpu. Add option for nofp.
21061 * config/arm/arm-cpu-cdata.h: Regenerated
21062 * config/arm/arm-cpu-data.h: Regenerated.
21064 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21066 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21067 (armv5te, armv5tej): Likewise.
21068 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21069 (armv7): Add options fp and vfpv3-d16.
21070 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21071 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21073 (armv7ve): Likewise.
21074 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21075 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21076 (armv8-a): Add nocrypto option.
21077 (armv8.1-a, armv8.2-a): Likewise.
21078 (armv8-m.main): add options fp, fp.dp and nofp.
21080 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21082 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21084 (armv8-a+crc): Delete.
21085 (armv8.1-a): Add options simd, crypto and nofp.
21086 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21087 (armv8.2-a+fp16): Delete.
21088 (armv8-m.main): Add option dsp.
21089 (armv8-m.main+dsp): Delete.
21090 (cortex-a8): Add fpu. Add nofp option.
21091 (cortex-a9): Add fpu. Add nofp and nosimd options.
21092 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21093 link to main cpu and architecture data structures.
21094 (gen_comm_data): Only put isa attributes from the main architecture
21096 (option): New statement for architecture and CPU entries.
21097 * arm.c (struct cpu_option): New structure.
21098 (struct processors): Add entry for options.
21099 (arm_unrecognized_feature): New function.
21100 (arm_parse_arch_cpu_name): Ignore any characters after the first
21102 (arm_parse_arch_cpu_feature): New function.
21103 (arm_configure_build_target): Separate out any CPU and architecture
21104 features and parse separately. Don't error out if -mfpu=auto is
21105 used with only an architecture string.
21106 (arm_print_asm_arch_directives): New function.
21107 (arm_file_start): Call it.
21108 * config/arm/arm-cpu-cdata.h: Regenerated.
21109 * config/arm/arm-cpu-data.h: Likewise.
21110 * config/arm/arm-tables.opt: Likewise.
21112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21114 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21115 assembler when it is not -mfpu=auto.
21117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21119 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21120 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21121 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21122 (ASM_CPU_SPEC): Rewrite.
21123 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21124 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21125 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21126 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21127 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21128 copied string is NUL-terminated. Also strip any characters prefixed
21130 (arm_rewrite_selected_arch): New function.
21131 (arm_rewrite_march): New function.
21133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21135 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21136 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21137 (march, mcpu, mtune): Convert to string-based options.
21138 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21139 (arm_parse_arch_cpu_name): New function.
21140 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21141 identify selected architecture or CPU.
21142 (arm_option_save): New function.
21143 (TARGET_OPTION_SAVE): Redefine.
21144 (arm_option_restore): Restore string options.
21145 (arm_option_print): Print string options.
21147 2017-06-16 Martin Sebor <msebor@redhat.com>
21149 PR tree-optimization/80933
21150 PR tree-optimization/80934
21151 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21152 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21153 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21154 (gimple_fold_builtin): Call them.
21156 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21158 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21159 as unlikely; update profile.
21161 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21163 * predict.c (force_edge_cold): Handle declaring edges impossible
21166 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21168 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21170 (try_unroll_loop_completely): Fix reporting.
21172 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21174 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21176 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21179 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21180 if given a non-constant argument for an intrinsic which requires a
21183 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21185 * profile.c (compare_freqs): New function.
21186 (branch_prob): Sort edge list.
21187 (find_spanning_tree): Assume that the list is priority sorted.
21189 2017-06-16 Richard Biener <rguenther@suse.de>
21191 PR tree-optimization/81090
21192 * passes.def (pass_record_bounds): Remove.
21193 * tree-pass.h (make_pass_record_bounds): Likewise.
21194 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21195 make_pass_record_bounds): Likewise.
21196 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21197 not free niter estimates at the beginning but at the end.
21198 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21200 2017-06-16 Richard Biener <rguenther@suse.de>
21202 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21203 initializer to workaround ICE in host GCC 4.8.
21205 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21207 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21209 (clone_inlined_nodes): Update.
21211 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21213 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21214 prefetch settings, and enable prefetching by default at -O3.
21216 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21218 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21219 Set flag_prefetch_loop_arrays according to tuning data.
21221 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21223 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21224 New tune structure.
21225 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21226 [Unrelated to main purpose of the patch] Place the pointer field last
21227 to enable type checking errors when tune structure are wrongly merged.
21228 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21229 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21230 (thunderx2t99_prefetch_tune): New tune constants.
21231 (tune_params *_tunings): Update all tunings (no functional change).
21232 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21233 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21234 from tunings structures.
21236 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21239 * ubsan.c (instrument_null): Add T argument, use it instead
21240 of computing it based on IS_LHS.
21241 (instrument_object_size): Likewise.
21242 (pass_ubsan::execute): Adjust instrument_null and
21243 instrument_object_size callers to pass gimple_get_lhs or
21244 gimple_assign_rhs1 result to it. Use instrument_null instead of
21245 calling get_base_address and instrument_mem_ref. Handle
21246 aggregate call arguments for object-size sanitization.
21248 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21250 PR tree-optimization/81089
21251 * tree-vrp.c (is_masked_range_test): Validate operands of
21254 2017-06-15 Martin Sebor <msebor@redhat.com>
21257 * dumpfile.c (dump_register): Avoid calling memset to initialize
21258 a class with a default ctor.
21259 * gcc.c (struct compiler): Remove const qualification.
21260 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21261 * hash-table.h: Ditto.
21262 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21264 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21265 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21267 * params.h (struct param_info): Make struct members non-const.
21268 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21269 with default initialization.
21270 * vec.h (vec_copy_construct, vec_default_construct): New helper
21272 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21273 with vec_copy_construct.
21274 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21275 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21276 * doc/invoke.texi (-Wclass-memaccess): Document.
21278 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21280 * emit-rtl.h (is_leaf): Update comment about local
21281 register allocator.
21283 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21286 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21287 for a variable to have a section before checking if the section has a
21289 Set section to.persistent if persistent attribute is set.
21290 Warn if .persistent attribute is used on an automatic variable.
21292 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21294 PR rtl-optimization/80474
21295 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21297 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21299 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21302 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21304 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21305 (casesi): Emit bounds checking as RTL.
21306 (casesi_internal_mips16_<mode>): Remove bounds checking.
21308 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21310 * config/xtensa/xtensa.c (xtensa_option_override): Append
21311 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21312 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21313 xtensa_doloop_hooks): Define unconditionally.
21314 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21316 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21317 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21318 for it in xtensa_option_override.
21319 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21320 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21322 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21324 * doc/cppopts.texi: Document '-' special value to -MF.
21326 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21328 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21329 (cortex_a53_fconst): Likewise.
21330 (cortex_a53_fpmul): Likewise.
21331 (cortex_a53_f_load_64): Likewise.
21332 (cortex_a53_f_load_many): Likewise.
21333 (cortex_a53_advsimd_alu): Likewise.
21334 (cortex_a53_advsimd_alu_q): Likewise.
21335 (cortex_a53_advsimd_mul): Likewise.
21336 (cortex_a53_advsimd_mul_q): Likewise.
21337 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21338 Add missing fmul, r2f_cvt and fconst cases.
21340 2017-06-14 Richard Biener <rguenther@suse.de>
21342 PR middle-end/81088
21343 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21345 (fold_binary_loc): When associating do not treat pre-existing
21346 TREE_OVERFLOW on literal constants as a reason to allow
21347 TREE_OVERFLOW on associated literal constants.
21349 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21351 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21352 (MASK_FEATURES): New macro.
21353 * config/sparc/sparc.c (sparc_option_override): Remove the special
21354 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21356 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21358 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21359 a division of 0 if non-call exceptions are enabled.
21361 2017-06-14 Andrew Pinski <apinski@cavium.com>
21362 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21365 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21366 Improve vector initialization code gen for only variable case.
21368 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21370 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21372 2017-06-14 Richard Biener <rguenther@suse.de>
21374 PR tree-optimization/81083
21375 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21378 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21380 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21381 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21382 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21383 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21384 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21385 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21387 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21389 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21390 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21392 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21394 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21396 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21398 * config/rs6000/t-rtems: Don't handle SPE.
21400 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21402 * config/rs6000/t-linux: Don't handle SPE.
21404 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21406 * config/rs6000/eabispe.h: Delete file.
21408 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21410 * config/rs6000/t-spe: Delete file.
21412 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21414 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21415 (rs6000_legitimate_offset_address_p): Return false for anything in
21416 V2SImode or V2SFmode.
21418 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21420 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21421 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21422 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21424 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21425 (rs6000_legitimate_offset_address_p): Ditto.
21426 (rs6000_emit_move): Ditto.
21427 (rs6000_init_builtins): Remove V4HI_type_node.
21429 2017-06-13 Martin Liska <mliska@suse.cz>
21432 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21433 (gate_asan): Likewise.
21434 * asan.h (asan_no_sanitize_address_p): Remove the function.
21435 (sanitize_flags_p): New function.
21436 * builtins.def: Fix coding style.
21437 * common.opt: Use renamed enum value.
21438 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21439 * doc/extend.texi: Document no_sanitize attribute.
21440 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21441 to SANITIZE_UNDEFINED_NONDEFAULT.
21442 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21443 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21444 Use sanitize_flags_p.
21445 * gimplify.c (gimplify_function_tree): Likewise.
21446 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21447 * opts.c (parse_no_sanitize_attribute): New function.
21448 (common_handle_option): Use renamed enum value.
21449 * opts.h (parse_no_sanitize_attribute): Declare.
21450 * tree.c (sanitize_flags_p): New function.
21451 * tree.h: Declared here.
21452 * tsan.c: Use sanitize_flags_p.
21453 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21454 (instrument_mem_ref): Likewise.
21455 (instrument_bool_enum_load): Likewise.
21456 (do_ubsan_in_current_function): Remove the function.
21457 (pass_ubsan::execute): Use sanitize_flags_p.
21458 * ubsan.h: Remove do_ubsan_in_current_function
21459 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21460 (dump_function_to_file): Use it here.
21462 2017-06-13 Martin Jambor <mjambor@suse.cz>
21464 PR tree-optimization/80803
21465 PR tree-optimization/81063
21466 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21467 (propagate_subaccesses_across_link): Enqueue subtree whenever
21468 necessary instead of relying on the caller.
21470 2017-06-13 Martin Jambor <mjambor@suse.cz>
21472 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21473 that have a first_link.
21474 (sort_and_splice_var_accesses): Do not check first_link before
21476 (subtree_mark_written_and_enqueue): Likewise.
21477 (propagate_all_subaccesses): Likewise and do not stop at first
21478 parent with a first_link.
21480 2017-06-13 Martin Jambor <mjambor@suse.cz>
21482 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21485 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21487 * match.pd: New pattern.
21489 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21491 * tree-vrp.c (is_masked_range_test): New function.
21492 (register_edge_assert_for): Determine ranges for
21495 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21497 PR tree-optimization/67328
21498 * fold-const.c (maskable_range_p): New function.
21499 (build_range_check): Generate bittests if possible.
21501 2017-06-13 Martin Liska <mliska@suse.cz>
21503 * gimple-pretty-print.c (dump_probability): Add new argument.
21504 (dump_edge_probability): Dump both probability and count.
21505 (dump_gimple_label): Likewise.
21506 (dump_gimple_bb_header): Likewise.
21508 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21511 * config/avr/avr-devices.c: Fix indentation.
21512 * config/avr/gen-avr-mmcu-specs.c: Dito.
21514 2017-06-13 Richard Biener <rguenther@suse.de>
21516 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21517 instead get vector type from stmt_info.
21518 (vectorizable_reduction): Adjust. Remove dead code.
21520 2017-06-13 Richard Biener <rguenther@suse.de>
21522 PR middle-end/81065
21523 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21524 case of C * (x * C2 + C3).
21525 (fold_addr_of_array_ref_difference): Properly fold index difference.
21527 2017-06-12 David S. Miller <davem@davemloft.net>
21530 * config/sparc/sparc.md (return expander): Emit frame blockage if
21531 function uses alloca.
21533 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21535 * combine.c (make_field_assignment): Check len rather than the mode
21536 precision when calling force_to_mode.
21538 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21540 Support multilibs and devices that see flash in RAM address range.
21543 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21544 (avr_mcu_t) <flash_pm_offset>: New field.
21545 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21546 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21547 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21548 (AVR_TINY_PM_OFFSET): Remove macro.
21549 * config/avr/avr.opt (-mshort-calls): New option.
21550 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21551 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21552 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21553 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21554 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21555 instead of avr_arch->have_jmp_call.
21556 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21557 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21558 avr_arch->flash_pm_offset to define.
21559 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21560 new field flash_pm_offset. Add entry for avrxmega3.
21561 (avr_texinfo): Add entry for avrxmega3.
21562 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21563 attiny212, attiny214,
21564 attiny412, attiny414, attiny416, attiny417,
21565 attiny814, attiny816, attiny817,
21566 attiny1614, attiny1616, attiny1617,
21567 attiny3214, attiny3216, attiny3217.
21568 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21569 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21570 (avr_print_operand_address) [AVR_TINY]: Same.
21571 (avr_asm_init_sections) <readonly_data_section>: Only patch
21572 callback if avr_arch->flash_pm_offset = 0.
21573 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21574 for rodata if avr_arch->flash_pm_offset != 0.
21575 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21576 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21577 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21578 (m_options): Append opt_rcall.
21579 (m_dirnames): Append dir_rcall.
21580 * config/avr/t-multilib: Regenerate.
21582 * configure.ac [target=avr]: Check whether avrxmega3 default
21583 linker description file works as needed.
21584 * configure: Regenerate.
21585 * doc/avr-mmcu.texi: Regenerate.
21586 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21587 <__AVR_ARCH__>: Document avrxmega3 and 103.
21588 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21589 <__AVR_SHORT_CALLS__>: Document it.
21590 <__AVR_PM_BASE_ADDRESS__>: Document it.
21591 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21592 (AVR Variable Attributes) <progmem>: Document this is
21593 not needed for avrxmega3.
21594 (AVR Named Address Spaces) <__flash>: Dito.
21596 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21598 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21600 2017-06-12 Doug Rupp <rupp@adacore.com>
21602 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21603 Append vxworks-stdint.h to the tm_file list.
21604 * config/vxworks-stdint.h: New file.
21606 2017-06-12 Martin Liska <mliska@suse.cz>
21608 PR tree-optimization/81041
21609 * tree-profile.c (gimple_gen_ic_func_profiler):
21610 Create an extra BB in profile-generate
21611 (gimple_gen_time_profiler): Likewise.
21613 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21615 PR tree-optimization/81003
21616 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21617 (update_range_test): Use it instead of force_gimple_operand_gsi.
21619 2017-06-12 Richard Biener <rguenther@suse.de>
21621 PR tree-optimization/81053
21622 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21623 with backedge value not defined in loop. Simplify def stmt
21626 2017-06-11 Tom de Vries <tom@codesourcery.com>
21629 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21631 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21632 nvptx_cannot_force_const_mem.
21634 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21636 * opts.c (finish_options): Move test for flag_split_stack after
21637 it has been initialized.
21639 2017-06-11 Jason Merrill <jason@redhat.com>
21641 * tree.h (id_equal): New.
21642 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21643 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21644 instead of strcmp of IDENTIFIER_POINTER.
21646 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21648 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21649 (mark_all_inlined_calls_cdtor): Fix formating.
21650 (inline_transform): Rescale profile before inlining.
21652 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21654 * cgraph.h (cgraph_edge::clone): Update prototype.
21655 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21656 (cgraph_node::create_clone): Update.
21657 (cgraph_node::create_version_clone): Update.
21658 * tree-inline.c (copy_bb): Update.
21659 (expand_call_inline): Update.
21661 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21663 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21664 factored out from ...
21665 (rs6000_emit_prologue): ... here.
21667 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21669 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21670 factored out from ...
21671 (rs6000_emit_prologue): ... here.
21673 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21675 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21677 (handle_missing_profiles): Fix computation of tp_first_run.
21678 (counts_to_freqs): Do not touch freqs when count is 0.
21680 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21682 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21685 2017-06-10 Tom de Vries <tom@codesourcery.com>
21687 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21688 attributes): Document signal effective target.
21690 2017-06-10 Tom de Vries <tom@codesourcery.com>
21692 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21693 Document effective target stack_size.
21695 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21697 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21698 to the edit_context if they can be auto-applied.
21700 2017-06-9 Ian Lance Taylor <iant@golang.org>
21702 * opts.c (finish_options): If -fsplit-stack, disable implicit
21703 -forder-blocks-and-partition.
21704 * doc/invoke.texi (Optimize Options): Document that when using
21705 -fsplit-stack -forder-blocks-and-partition is not implicitly
21708 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21710 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21711 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21712 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21713 * builtins.def (abort, trap, unreachable): Declare cold.
21714 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21715 * tree-core.h (ECF_COLD): New.
21716 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21717 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21719 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21721 * predict.c (unlikely_executed_stmt_p): Cleanup.
21723 2017-06-09 Richard Biener <rguenther@suse.de>
21725 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21726 model if the ref is always written to.
21728 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21730 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21732 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21734 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21737 2017-06-09 Tom de Vries <tom@codesourcery.com>
21740 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21741 "target cannot support label values" when encountering LABEL_REF.
21743 2017-06-09 Martin Liska <mliska@suse.cz>
21745 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21746 (gimple_gen_ic_func_profiler): Emit direct comparison
21747 of __gcov_indirect_call_callee with NULL.
21748 (gimple_gen_time_profiler): Change probability from
21749 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21751 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21753 * profile.c (edge_gcov_counts): Turn to pointer.
21754 (compute_branch_probabilities, compute_branch_probabilities): Update.
21755 (branch_prob): Do not clear edge_gcov_count.
21756 * profile.h (edge_gcov_counts): Turn to pointer.
21757 (edge_gcov_count): Update.
21759 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21761 * gimple.h (gimple_check_failed): Mark cold.
21763 2017-06-09 Richard Biener <rguenther@suse.de>
21765 PR tree-optimization/66623
21766 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21767 refactor check_reduction into two parts, properly computing
21768 whether we have to check reduction validity for outer loop
21771 2017-06-09 Richard Biener <rguenther@suse.de>
21773 PR tree-optimization/79483
21774 * graphite-scop-detection.c (order): New global.
21775 (get_order): Compute bb to order mapping that satisfies code
21776 generation constraints.
21777 (cmp_pbbs): New helper.
21778 (build_scops): Start domwalk at entry block, sort generated
21781 2017-06-09 Richard Biener <rguenther@suse.de>
21783 PR middle-end/81007
21784 * ipa-polymorphic-call.c
21785 (ipa_polymorphic_call_context::restrict_to_inner_class):
21786 Skip FIELD_DECLs with error_mark_node type.
21787 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21790 2017-06-09 Martin Liska <mliska@suse.cz>
21792 * predict.c (struct branch_predictor): New struct.
21793 (test_prediction_value_range): New test.
21794 (predict_c_tests): New function.
21795 * selftest-run-tests.c (selftest::run_tests): Run the function.
21796 * selftest.h: Declare new tests.
21798 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21801 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21802 gen_add3_insn did not fail.
21803 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21804 r0, construct that number in a temporary reg and add that reg to r0.
21805 If asked to put the result in r0 as well, fail.
21807 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21809 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21810 for early expansion of vec_eqv.
21812 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21814 PR middle-end/81005
21815 * ubsan.c (instrument_null): Avoid pointless code temporary.
21816 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21818 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21822 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21825 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21828 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21830 * predict.c (unlikely_executed_edge_p): Move ahead.
21831 (probably_never_executed_edge_p): Use it.
21833 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21835 PR middle-end/79988
21836 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21837 gimple_call_builtin_p call.
21839 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21841 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21842 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21843 rtl_check_failed_type2, rtl_check_failed_code1,
21844 rtl_check_failed_code2, rtl_check_failed_code_mode,
21845 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21846 rtvec_check_failed_bounds, rtl_check_failed_flag,
21847 _fatal_insn_not_found, _fatal_insn): Likewise.
21848 * tree.h (tree_contains_struct_check_failed,
21849 tree_check_failed, tree_not_check_failed,
21850 tree_class_check_failed, tree_range_check_failed,
21851 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21852 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21853 tree_operand_check_failed, omp_clause_check_failed,
21854 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21857 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21859 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21860 flag_branch_probabilities.
21861 * ipa-inline.c (edge_badness): Likewise.
21862 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21863 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21864 * predict.c (maybe_hot_frequency_p): Likewise.
21865 (probably_never_executed): Likewise.
21866 * sched-ebb.c (schedule_ebbs): Likewise.
21867 * sched-rgn.c (find_single_block_region): Likewise.
21868 * tracer.c (tail_duplicate): Likewise.
21870 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21872 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21873 longer requires x_flag_profile_use.
21875 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21877 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21878 instead of flag_reorder_blocks_and_partition.
21879 * dbxout.c (dbxout_function_end): Likewise.
21880 * dwarf2out.c (gen_subprogram_die): Likewise.
21881 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21882 * hw-doloop.c (reorg_loops): Likewise.
21883 * varasm.c (assemble_start_function,
21884 assemble_end_function): Likewise.
21885 (decide_function_section): Do not check for
21886 flag_reorder_blocks_and_partition.
21888 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21890 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21892 (chkp_get_hard_register_fake_addr_expr): Ditto.
21893 (chkp_build_addr_expr): Add check for hard reg case.
21894 (chkp_parse_array_and_component_ref): Ditto.
21895 (chkp_find_bounds_1): Ditto.
21896 (chkp_process_stmt): Don't generate bounds store for
21899 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21901 * predict.c (maybe_hot_bb_p): Do not check profile status.
21902 (maybe_hot_edge_p): Likewise.
21903 (probably_never_executed): Check for zero counts even if profile
21905 (unlikely_executed_edge_p): New function.
21906 (unlikely_executed_stmt_p): New function.
21907 (unlikely_executed_bb_p): New function.
21908 (set_even_probabilities): Use unlikely predicates.
21909 (combine_predictions_for_bb): Likewise.
21910 (predict_paths_for_bb): Likewise.
21911 (predict_paths_leading_to_edge): Likewise.
21912 (determine_unlikely_bbs): New function.
21913 (estimate_bb_frequencies): Use it.
21914 (compute_function_frequency): Use zero counts even if profile is
21916 * profile-count.h: Fix typo.
21918 2017-08-08 Julia Koval <julia.koval@intel.com>
21920 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21921 _mm512_mask_cvtsepi16_storeu_epi8,
21922 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21923 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21924 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21925 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21926 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21927 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21928 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21929 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21930 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21931 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21932 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21933 __builtin_ia32_pmovuswb256mem_mask,
21934 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21935 __builtin_ia32_pmovwb512mem_mask): New builtins.
21937 2017-08-08 Julia Koval <julia.koval@intel.com>
21939 PR target/73350,80862
21940 * config/i386/subst.md (round): Fix round pattern.
21941 * config/i386/i386.c (ix86_erase_embedded_rounding):
21942 Fix erasing rounding for the fixed pattern.
21944 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21946 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21948 2017-06-08 Martin Liska <mliska@suse.cz>
21950 PR gcov-profile/80911
21951 * gcov.c (block_info::block_info): New constructor.
21953 2017-06-07 Carl Love <cel@us.ibm.com>
21955 * config/rs6000/rs6000-c: The return type of the following
21956 built-in functions was implemented as int not long long. Fix sign
21957 of return value for the unsigned version of vec_mulo and vec_mule.
21958 vector unsigned long long vec_bperm (vector unsigned long long,
21959 vector unsigned char)
21960 vector signed long long vec_mule (vector signed int,
21962 vector unsigned long long vec_mule (vector unsigned int,
21963 vector unsigned int)
21964 vector signed long long vec_mulo (vector signed int,
21966 vector unsigned long long vec_mulo (vector unsigned int,
21967 vector unsigned int)
21968 * doc/extend.texi: Fix the documentation for the built-in
21971 2017-06-07 Carl Love <cel@us.ibm.com>
21974 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21977 2017-06-07 Carl Love <cel@us.ibm.com>
21979 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21980 support, Generate doublehv for signed int/float for BE case only.
21982 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21984 * doc/invoke.texi (mcx16): Rewrite.
21986 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21988 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21989 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21990 *mov<mode>_softfloat, and an anonymous splitter): Use
21991 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21993 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21995 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21997 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21998 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21999 (rs6000_debug_reg_global): Adjust.
22000 (rs6000_init_hard_regno_mode_ok): Adjust.
22001 (rs6000_dbx_register_number): Adjust.
22002 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22003 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22004 Remove SPE_ACC and SPEFSCR.
22005 (REG_ALLOC_ORDER): Ditto.
22006 (FRAME_POINTER_REGNUM): Change to 111.
22007 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22008 (REG_CLASS_NAMES): Ditto.
22009 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22010 (REGISTER_NAMES): Ditto.
22011 (ADDITIONAL_REG_NAMES): Ditto.
22012 (rs6000_reg_names): Ditto.
22013 * config/rs6000/rs6000.md: Renumber some register number
22016 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22018 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22020 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22021 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22023 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22024 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22025 Delete the SPE high registers.
22026 (REG_ALLOC_ORDER): Ditto.
22027 (enum reg_class): Remove SPE_HIGH_REGS.
22028 (REG_CLASS_NAMES): Ditto.
22029 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22030 (REGISTER_NAMES): Ditto.
22031 (rs6000_reg_names): Ditto.
22032 * doc/tm.texi.in: Remove SPE as example.
22033 * doc/tm.texi: Regenerate.
22035 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22037 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22038 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22039 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22040 * config/rs6000/rs6000.md (type): Remove "brinc".
22042 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22044 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22045 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22046 * config/rs6000/linuxspe.h: Delete file.
22047 * config/rs6000/rs6000.md: Don't include spe.md.
22048 * config/rs6000/spe.h: Delete file.
22049 * config/rs6000/spe.md: Delete file.
22050 * config/rs6000/t-rs6000: Remove spe.md.
22052 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22054 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22055 (reg_or_none500mem_operand): Delete.
22056 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22057 instead of reg_or_none500mem_operand.
22059 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22061 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22062 handling of SPE flags.
22063 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22065 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22067 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22069 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22070 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22071 paired_divv2sf3): Similar.
22072 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22073 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22074 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22076 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22077 Rename the paired_* instruction patterns.
22078 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22080 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22081 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22082 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22083 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22084 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22085 PAIRED_VECTOR_MODE.
22086 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22087 (spe_func_has_64bit_regs_p): Delete.
22088 (spe_expand_predicate_builtin): Delete.
22089 (spe_expand_evsel_builtin): Delete.
22090 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22091 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22092 (invalid_e500_subreg): Delete.
22093 (rs6000_legitimize_address): Always force_reg op2 as well, for
22094 paired single memory accesses.
22095 (rs6000_member_type_forces_blk): Delete.
22096 (rs6000_spe_function_arg): Delete.
22097 (rs6000_expand_unop_builtin): Delete SPE handling.
22098 (rs6000_expand_binop_builtin): Ditto.
22099 (spe_expand_stv_builtin): Delete.
22100 (bdesc_2arg_spe): Delete.
22101 (spe_expand_builtin): Delete.
22102 (spe_expand_predicate_builtin): Delete.
22103 (spe_expand_evsel_builtin): Delete.
22104 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22105 (spe_init_builtins): Delete.
22106 (spe_func_has_64bit_regs_p): Delete.
22107 (savres_routine_name): Delete "info" parameter. Adjust callers.
22108 (rs6000_emit_stack_reset): Ditto.
22109 (rs6000_dwarf_register_span): Delete.
22110 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22111 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22112 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22114 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22116 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22117 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22118 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22119 mulv2sf3, divv2sf3): Delete expanders.
22121 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22123 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22125 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22127 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22128 * config/rs6000/rs6000.c: Ditto.
22130 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22132 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22133 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22134 comparison_operator.
22136 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22138 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22139 * config/rs6000/rs6000.opt: Ditto.
22140 * config/rs6000/t-rtems: Ditto.
22142 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22144 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22145 TARGET_E500_SINGLE by 0, simplify.
22146 * config/rs6000/rs6000.c: Ditto.
22147 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22148 (spe_build_register_parallel): Delete.
22149 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22150 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22151 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22152 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22153 (E500_CONVERT): Delete.
22154 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22156 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22158 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22159 * config/rs6000/dfp.md: Ditto.
22160 (negdd2, *negdd2_fpr): Merge.
22161 (absdd2, *absdd2_fpr): Merge.
22162 (negtd2, *negtd2_fpr): Merge.
22163 (abstd2, *abstd2_fpr): Merge.
22164 * config/rs6000/e500.h: Delete file.
22165 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22166 TARGET_FPRS by 1 and simplify.
22167 * config/rs6000/rs6000-c.c: Ditto.
22168 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22169 TARGET_DF_SPE by 0.
22170 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22172 * config/rs6000/rs6000.md: Ditto.
22173 (floatdidf2, *floatdidf2_fpr): Merge.
22174 (move_from_CR_gt_bit): Delete.
22175 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22176 (E500_CR_IOR_COMPARE): Delete.
22177 (All patterns that require !TARGET_FPRS): Delete.
22178 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22180 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22182 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22184 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22186 * graphds.c (add_edge): Intitialize edge's attached data.
22187 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22188 pointer parameter. Call pointed function on each edge during
22189 graph traversing. Skip traversing the edge when the function
22191 (graphds_dfs, graphds_scc): Ditto.
22192 (for_each_edge): New parameter. Pass the new parameter to callback
22194 * graphds.h (skip_edge_callback): New function pointer type.
22195 (graphds_dfs, graphds_scc): New function pointer parameter.
22196 (graphds_edge_callback, for_each_edge): New parameter.
22198 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22200 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22201 out code checking if runtime alias check is possible to below ...
22202 Call the new function.
22203 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22204 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22206 2017-06-07 Marek Polacek <polacek@redhat.com>
22209 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22210 TYPE_OVERFLOW_WRAPS check.
22212 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22214 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22215 if versioning is required.
22216 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22217 peeling with the check for versioning.
22219 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22221 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22222 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22223 Set true to new parameter if new ssa variable is defined.
22224 (vect_gen_vector_loop_niters): Refactor. Set range information
22225 for the new vector loop bound variable.
22226 (vect_do_peeling): Ditto.
22228 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22230 * tree-affine.c (ssa.h): Include header file.
22231 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22232 has wrapping overflow behavior.
22234 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22236 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22238 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22240 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22241 (tree_to_aff_combination): ... here.
22243 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22245 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22246 reg_pressure model function.
22247 (ivopts_global_cost_for_size): Delete.
22248 (determine_set_costs, iv_ca_recount_cost): Call new model function
22249 ivopts_estimate_reg_pressure.
22251 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22253 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22254 expensive than udiv. Remove floating point cases from mod.
22256 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22258 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22259 Increase idiv cost.
22261 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22263 * config/aarch64/aarch64.md
22264 (copysignsf3): Fix mask generation.
22266 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22268 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22270 (class dump_manager): Add register_dumps method.
22271 * dumpfile.c: Include langhooks.h.
22272 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22273 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22274 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22275 (dump_manager::dump_register): Start with 512 entries instead of 32.
22276 (dump_manager::register_dumps): New method.
22277 * toplev.c (general_init): Instead of invoking register_dumps
22278 langhook, invoke register_dumps method on the dump manager.
22279 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22282 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22284 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22285 Say that # requires an associated define_split to exist, and that
22286 the define_split must be suitable for use after register allocation.
22288 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22290 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22291 (compute_outgoing_frequencies): Also initialize zero counts.
22292 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22293 around loops; preserve more of profile when nothing changes.
22295 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22297 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22299 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22300 * config/arm/arm-cpu-cdata.h: Regenerate.
22301 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22302 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22303 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22304 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22305 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22307 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22309 * config/arm/t-rmprofile: Likewise.
22310 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22312 2017-06-06 David S. Miller <davem@davemloft.net>
22315 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22316 blockage if function uses alloca.
22318 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22320 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22321 New "uid" fields to hold pretty-print IDs of group and ref.
22322 Memory references are now identified as <group_id>:<ref_id>
22323 instead of using [random] addresses.
22324 (dump_mem_details): Simplify, no functional change.
22325 (dump_mem_ref): Simplify and make output more concise.
22326 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22327 (find_or_create_group): Initialize group uid.
22328 (record_ref): Initialize ref uid. Improve debug output.
22329 (prune_group_by_reuse, should_issue_prefetch_p,)
22330 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22331 (mark_nontemporal_store, determine_loop_nest_reuse):
22332 Improve debug output.
22334 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22336 * dbgcnt.def (prefetch): New debug counter.
22337 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22338 (schedule_prefetches): Stop issueing prefetches if debug counter
22341 2017-06-06 Tom de Vries <tom@codesourcery.com>
22343 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22344 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22346 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22348 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22349 Use aarch64_reg_or_zero predicate for operand 4.
22350 (aarch64_compare_and_swap<mode> define_insn_and_split):
22351 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22352 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22354 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22356 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22357 (arm_compute_save_core_reg_mask): This.
22358 (thumb1_compute_save_reg_mask): Rename into ...
22359 (thumb1_compute_save_core_reg_mask): This.
22360 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22361 (arm_compute_frame_layout): Likewise.
22363 2017-06-06 Richard Biener <rguenther@suse.de>
22365 PR tree-optimization/80974
22366 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22367 keep or clear leaders SSA info.
22369 2017-06-06 Tom de Vries <tom@codesourcery.com>
22371 * config/nvptx/nvptx.c (split_mode_p): New function.
22372 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22374 2017-06-06 Tom de Vries <tom@codesourcery.com>
22376 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22378 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22381 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22384 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22386 * shrink-wrap.c (handle_simple_exit): Update profile.
22387 (try_shrink_wrapping): Upate profile.
22389 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22391 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22392 (tree_guess_outgoing_edge_probabilities): New.
22393 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22394 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22396 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22398 * ipa-split.c (split_function): Initialize return bb profile.
22400 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22402 * profile.c (compute_branch_probabilities): Also initialize
22403 EXIT_BLOCK profile.
22405 2017-06-06 Richard Biener <rguenther@suse.de>
22407 PR tree-optimization/80928
22408 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22409 (vect_analyze_loop_operations): Properly guard analysis for
22411 (vect_transform_loop): Likewise.
22412 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22413 (vect_model_induction_cost): Do not cost for pure SLP.
22414 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22415 of induction in inner loop vectorization.
22416 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22417 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22418 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22420 (vect_analyze_slp_cost_1): Cost induction.
22421 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22422 (vect_get_slp_vect_defs): Likewise.
22423 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22424 (vect_transform_stmt): Handle SLP reductions.
22425 * tree-vectorizer.h (vectorizable_induction): Adjust.
22427 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22429 * config/rs6000/rs6000.c (make_resolver_func): Update
22430 init_lowered_empty_function call.
22432 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22434 * doc/invoke.texi: Document the -fprofile-abs-path option.
22435 * common.opt (fprofile-abs-path): New option.
22436 * gcov-io.h (gcov_write_filename): Declare.
22437 * gcov-io.c (gcov_write_filename): New function.
22438 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22439 * profile.c (output_location): Likewise.
22441 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22443 * shring-wrap.c: Revert accidental commit.
22445 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22447 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22449 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22451 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22453 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22454 profile in callgraph edge.
22455 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22456 (apply_scale): Likewise.
22457 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22458 Also scale profile when inlining function with zero profile.
22459 (initialize_cfun): Update exit block profile even when it is zero.
22460 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22461 when profile is read.
22463 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22465 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22466 (CLONE_*): New constants to define the processors we can generate
22467 code for with the target_clone attribute.
22468 (rs6000_clone_map): New array to identify which clone processors
22469 the current program is running on.
22470 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22471 target_clone attribute.
22472 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22473 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22474 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22475 (cpu_expand_builtin): Add support for target_clone attribute.
22476 (rs6000_valid_attribute_p): Allow "default" attribute.
22477 (get_decl_name): New debug function to simplify printing the
22478 current function name in debugging statements.
22479 (rs6000_clone_priority): New functions to support the target_clone
22480 attribute, and be able to generate code to switch between ISA 2.05
22481 through ISA 3.0 (power6 through power9).
22482 (rs6000_compare_version_priority): Likewise.
22483 (rs6000_get_function_versions_dispatcher): Likewise.
22484 (make_resolver_func): Likewise.
22485 (add_condition_to_bb): Likewise.
22486 (dispatch_function_versions): Likewise.
22487 (rs6000_generate_version_dispatcher_body): Likewise.
22488 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22489 (fusion_gpr_load_p): Fix a spacing issue.
22490 * doc/extend.texi (Common Function Attributes): Document that the
22491 PowerPC supports the target_clone attribute.
22493 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22495 * config/arm/arm.h: explain F symbol found in description of ARM
22496 register allocation in its legend.
22498 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22500 * config/mips/frame-header-opt.c: Include profile-count.h.
22501 * config/riscv/riscv.c: Include profile-count.h
22503 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22505 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22507 (sm_set_flag_if_changed): Add bbs field.
22508 (execute_sm_if_changed_flag_set): Pass BBS.
22509 (execute_sm): Update.
22511 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22513 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22516 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22518 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22519 (peephole2): New peephole2 to emit the above.
22520 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22522 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22524 * config/aarch64/aarch64.c (define_peephole2 above
22525 *sub_<shift>_<mode>): New peephole.
22527 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22529 * config/i386/i386.c (make_resolver_func): Update.
22530 * Makefile.in: Add profile-count.h and profile-count.o
22531 * auto-profile.c (afdo_indirect_call): Update to new API.
22532 (afdo_set_bb_count): Update.
22533 (afdo_propagate_edge): Update.
22534 (afdo_propagate_circuit): Update.
22535 (afdo_calculate_branch_prob): Update.
22536 (afdo_annotate_cfg): Update.
22537 * basic-block.h: Include profile-count.h
22538 (struct edge_def): Turn count to profile_count.
22539 (struct basic_block_def): Likewie.
22540 (REG_BR_PROB_BASE): Move to profile-count.h
22541 (RDIV): Move to profile-count.h
22542 * bb-reorder.c (max_entry_count): Turn to profile_count.
22543 (find_traces): Update.
22544 (rotate_loop):Update.
22545 (connect_traces):Update.
22546 (sanitize_hot_paths):Update.
22547 * bt-load.c (migrate_btr_defs): Update.
22548 * cfg.c (RDIV): Remove.
22549 (init_flow): Use alloc_block.
22550 (alloc_block): Uninitialize count.
22551 (unchecked_make_edge): Uninitialize count.
22552 (check_bb_profile): Update.
22553 (dump_edge_info): Update.
22554 (dump_bb_info): Update.
22555 (update_bb_profile_for_threading): Update.
22556 (scale_bbs_frequencies_int): Update.
22557 (scale_bbs_frequencies_gcov_type): Update.
22558 (scale_bbs_frequencies_profile_count): New.
22559 * cfg.h (update_bb_profile_for_threading): Update.
22560 (scale_bbs_frequencies_profile_count): Declare.
22561 * cfgbuild.c (compute_outgoing_frequencies): Update.
22562 (find_many_sub_basic_blocks): Update.
22563 * cfgcleanup.c (try_forward_edges): Update.
22564 (try_crossjump_to_edge): Update.
22565 * cfgexpand.c (expand_gimple_tailcall): Update.
22566 (construct_exit_block): Update.
22567 * cfghooks.c (verify_flow_info): Update.
22568 (dump_bb_for_graph): Update.
22569 (split_edge): Update.
22570 (make_forwarder_block): Update.
22571 (duplicate_block): Update.
22572 (account_profile_record): Update.
22573 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22574 (get_estimated_loop_iterations): Update.
22575 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22576 (single_likely_exit): Update.
22577 * cfgloopmanip.c (scale_loop_profile): Update.
22579 (set_zero_probability): Update.
22580 (lv_adjust_loop_entry_edge): Update.
22581 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22582 (purge_dead_edges): Update.
22583 (rtl_account_profile_record): Update.
22584 * cgraph.c (cgraph_node::create): Uninitialize count.
22585 (symbol_table::create_edge): Uninitialize count.
22586 (cgraph_update_edges_for_call_stmt_node): Update.
22587 (cgraph_edge::dump_edge_flags): Update.
22588 (cgraph_node::dump): Update.
22589 (cgraph_edge::maybe_hot_p): Update.
22590 * cgraph.h: Include profile-count.h
22591 (create_clone), create_edge, create_indirect_edge): Update.
22592 (cgraph_node): Turn count to profile_count.
22593 (cgraph_edge0: Likewise.
22594 (make_speculative, clone): Update.
22595 (create_edge): Update.
22596 (init_lowered_empty_function): Update.
22597 * cgraphclones.c (cgraph_edge::clone): Update.
22598 (duplicate_thunk_for_node): Update.
22599 (cgraph_node::create_clone): Update.
22600 * cgraphunit.c (cgraph_node::analyze): Update.
22601 (cgraph_node::expand_thunk): Update.
22602 * final.c (dump_basic_block_info): Update.
22603 * gimple-streamer-in.c (input_bb): Update.
22604 * gimple-streamer-out.c (output_bb): Update.
22605 * graphite.c (print_global_statistics): Update.
22606 (print_graphite_scop_statistics): Update.
22607 * hsa-brig.c: Include basic-block.h.
22608 * hsa-dump.c: Include basic-block.h.
22609 * hsa-gen.c (T sum_slice): Update.
22610 (convert_switch_statements):Update.
22611 * hsa-regalloc.c: Include basic-block.h.
22612 * ipa-chkp.c (chkp_produce_thunks): Update.
22613 * ipa-cp.c (struct caller_statistics): Update.
22614 (init_caller_stats): Update.
22615 (gather_caller_stats): Update.
22616 (ipcp_cloning_candidate_p): Update.
22617 (good_cloning_opportunity_p): Update.
22618 (get_info_about_necessary_edges): Update.
22619 (dump_profile_updates): Update.
22620 (update_profiling_info): Update.
22621 (update_specialized_profile): Update.
22622 (perhaps_add_new_callers): Update.
22623 (decide_about_value): Update.
22624 (ipa_cp_c_finalize): Update.
22625 * ipa-devirt.c (struct odr_type_warn_count): Update.
22626 (struct decl_warn_count): Update.
22627 (struct final_warning_record): Update.
22628 (possible_polymorphic_call_targets): Update.
22629 (ipa_devirt): Update.
22630 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22631 * ipa-icf.c (sem_function::merge): Update.
22632 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22633 * ipa-inline.c (compute_uninlined_call_time): Update.
22634 (compute_inlined_call_time): Update.
22635 (want_inline_small_function_p): Update.
22636 (want_inline_self_recursive_call_p): Update.
22637 (edge_badness): Update.
22638 (lookup_recursive_calls): Update.
22639 (recursive_inlining): Update.
22640 (inline_small_functions): Update.
22641 (dump_overall_stats): Update.
22642 (dump_inline_stats): Update.
22643 * ipa-profile.c (ipa_profile_generate_summary): Update.
22644 (ipa_propagate_frequency): Update.
22645 (ipa_profile): Update.
22646 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22647 * ipa-utils.c (ipa_merge_profiles): Update.
22648 * loop-doloop.c (doloop_modify): Update.
22649 * loop-unroll.c (report_unroll): Update.
22650 (unroll_loop_runtime_iterations): Update.
22651 * lto-cgraph.c (lto_output_edge): Update.
22652 (lto_output_node): Update.
22653 (input_node): Update.
22654 (input_edge): Update.
22655 (merge_profile_summaries): Update.
22656 * lto-streamer-in.c (input_cfg): Update.
22657 * lto-streamer-out.c (output_cfg): Update.
22658 * mcf.c (create_fixup_graph): Update.
22659 (adjust_cfg_counts): Update.
22660 (sum_edge_counts): Update.
22661 * modulo-sched.c (sms_schedule): Update.
22662 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22663 * predict.c (maybe_hot_count_p): Update.
22664 (probably_never_executed): Update.
22665 (dump_prediction): Update.
22666 (combine_predictions_for_bb): Update.
22667 (propagate_freq): Update.
22668 (handle_missing_profiles): Update.
22669 (counts_to_freqs): Update.
22670 (rebuild_frequencies): Update.
22671 (force_edge_cold): Update.
22672 * predict.h: Include profile-count.h
22673 (maybe_hot_count_p, counts_to_freqs): UPdate.
22674 * print-rtl-function.c: Do not include cfg.h
22675 * print-rtl.c: Include basic-block.h
22676 * profile-count.c: New file.
22677 * profile-count.h: New file.
22678 * profile.c (is_edge_inconsistent): Update.
22679 (correct_negative_edge_counts): Update.
22680 (is_inconsistent): Update.
22681 (set_bb_counts): Update.
22682 (read_profile_edge_counts): Update.
22683 (compute_frequency_overlap): Update.
22684 (compute_branch_probabilities): Update; Initialize and deinitialize
22686 (branch_prob): Update.
22687 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22688 (edge_gcov_count): New.
22689 (bb_gcov_count): New.
22690 * shrink-wrap.c (try_shrink_wrapping): Update.
22691 * tracer.c (better_p): Update.
22692 * trans-mem.c (expand_transaction): Update.
22693 (ipa_tm_insert_irr_call): Update.
22694 (ipa_tm_insert_gettmclone_call): Update.
22695 * tree-call-cdce.c: Update.
22696 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22697 (gimple_duplicate_sese_tail): Update.
22698 (gimple_account_profile_record): Update.
22699 (execute_fixup_cfg): Update.
22700 * tree-inline.c (copy_bb): Update.
22701 (copy_edges_for_bb): Update.
22702 (initialize_cfun): Update.
22703 (freqs_to_counts): Update.
22704 (copy_cfg_body): Update.
22705 (expand_call_inline): Update.
22706 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22707 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22708 (try_unroll_loop_completely): Update.
22709 (try_peel_loop): Update.
22710 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22711 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22712 * tree-ssa-loop-split.c (connect_loops): Update.
22713 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22714 * tree-ssa-reassoc.c (branch_fixup): Update.
22715 * tree-ssa-tail-merge.c (replace_block_by): Update.
22716 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22717 (compute_path_counts): Update.
22718 (update_profile): Update.
22719 (recompute_probabilities): Update.
22720 (update_joiner_offpath_counts): Update.
22721 (estimated_freqs_path): Update.
22722 (freqs_to_counts_path): Update.
22723 (clear_counts_path): Update.
22724 (ssa_fix_duplicate_block_edges): Update.
22725 (duplicate_thread_path): Update.
22726 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22727 (struct switch_conv_info): Update.
22728 * tree-tailcall.c (decrease_profile): Update.
22729 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22730 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22731 * value-prof.c (check_counter): Update.
22732 (gimple_divmod_fixed_value): Update.
22733 (gimple_mod_pow2): Update.
22734 (gimple_mod_subtract): Update.
22735 (gimple_ic_transform): Update.
22736 (gimple_stringop_fixed_value): Update.
22737 * value-prof.h (gimple_ic): Update.
22739 2017-06-02 Carl Love <cel@us.ibm.com>
22741 * config/rs6000/rs6000-c: Add support for built-in functions
22742 vector double vec_doublee (vector signed int);
22743 vector double vec_doublee (vector unsigned int);
22744 vector double vec_doublee (vector float);
22745 vector double vec_doubleh (vector signed int);
22746 vector double vec_doubleh (vector unsigned int);
22747 vector double vec_doubleh (vector float);
22748 vector double vec_doublel (vector signed int);
22749 vector double vec_doublel (vector unsigned int);
22750 vector double vec_doublel (vector float);
22751 vector double vec_doubleo (vector signed int);
22752 vector double vec_doubleo (vector unsigned int);
22753 vector double vec_doubleo (vector float);.
22754 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22755 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22757 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22758 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22759 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22761 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22762 vec_doublel, vec_doubleh.
22763 * doc/extend.texi: Update the built-in documentation file for the
22764 new built-in functions.
22766 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22769 * ipa-inline-analysis.c (free_growth_caches): Set
22770 edge_removal_hook_holder to NULL after removing it.
22772 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22774 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22775 comparision with zero.
22777 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22778 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22779 for early expansion of vec_min and vec_max builtins.
22780 (builtin_function_type): Add min/max unsigned variants to those
22781 identified as having unsigned arguments.
22783 2017-06-02 Olivier Hainque <hainque@adacore.com>
22785 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22787 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22789 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22790 Use VALL_F16 iterator rather than VALL.
22792 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22794 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22795 Emit CBNZ inside loop when doing a strong exchange and comparing
22796 against zero. Generate the CC flags after the loop.
22798 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22800 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22801 (dl_section_ref): New.
22802 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22803 On AIX, append an expression to subtract the size of the
22804 section length to dl_section_ref.
22806 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22808 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22809 for early expansion of vector absolute builtins.
22811 2017-06-02 Richard Biener <rguenther@suse.de>
22813 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22814 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22816 2017-06-02 Richard Biener <rguenther@suse.de>
22818 PR tree-optimization/80948
22819 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22820 stmt order as well.
22822 2017-06-02 Richard Biener <rguenther@suse.de>
22824 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22826 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22827 for inductions as relevant.
22829 2017-06-02 Richard Biener <rguenther@suse.de>
22831 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22832 (vectorizable_induction): ... this. Remove dead code.
22834 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22836 * builtins. (expand_builtin_alloca): Remove second parameter and
22837 infer its value from the first parameter instead.
22838 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22840 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22842 PR rtl-optimization/80903
22843 * loop-doloop.c (add_test): Unshare sequence.
22845 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22847 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22849 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22851 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22853 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22854 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22855 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22856 (xlogue_layout::get_stub_rtx): Make static.
22857 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22858 (xlogue_layout::compute_stub_managed_regs): Rename to...
22859 (xlogue_layout::count_stub_managed_regs): ...this.
22860 (xlogue_layout::is_stub_managed_reg): New function.
22861 (xlogue_layout::m_stub_names): Rename to...
22862 (xlogue_layout::s_stub_names): ...this, make static.
22863 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22864 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22865 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22866 xlogue_layout::s_stub_names): Instantiate statics.
22867 (stub_managed_regs): Remove.
22868 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22869 (disable_call_ms2sysv_xlogues): Rename to...
22870 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22871 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22873 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22874 change after reload_completed.
22875 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22877 (ix86_expand_prologue): Likewise.
22878 (ix86_expand_epilogue): Likewise.
22879 (ix86_expand_split_stack_prologue): Likewise.
22880 (ix86_compute_frame_layout): Remove frame parameter ...
22881 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22882 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22884 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22885 (ix86_frame): Move from here ...
22886 * config/i386/i386.h (ix86_frame): ... to here.
22887 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22888 complete ix86_frame data structure instead. Remove some_ld_name.
22890 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22892 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22893 symbols that hold a DECL_VALUE_EXPR.
22895 2017-06-01 Martin Jambor <mjambor@suse.cz>
22897 PR tree-optimization/80898
22898 * tree-sra.c (process_subtree_disqualification): Removed.
22899 (disqualify_candidate): Do not acll
22900 process_subtree_disqualification.
22901 (subtree_mark_written_and_enqueue): New function.
22902 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22903 RHS has been disqualified and re-queue LHS if necessary. Apart
22904 from that, ignore disqualified RHS.
22906 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22908 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22909 address fetch for z10 or later.
22911 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22913 * config/arc/arc.md (tst_movb): Add guard when splitting.
22915 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22917 * config/arc/arc.c (arc_can_eliminate): Test against
22918 arc_frame_pointer_needed.
22920 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22922 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22923 to prevent store reordering.
22924 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22925 (type): Add block type.
22926 (stack_tie): Define special instruction to be used in
22929 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22931 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22932 constraint. It is not valid for the pattern.
22933 (noncommutative_binary_comparison): Likewise.
22935 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22937 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22940 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22942 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22943 be used by the reg-alloc.
22945 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22947 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22948 reg-alloc when having mul64 or mul32x16 instructions.
22949 (mulsidi3): Likewise.
22950 (umulsidi3): Likewise.
22951 (mulsi32x16): New pattern.
22952 (mulsi64): Likewise.
22953 (mulsidi64): Likewise.
22954 (umulsidi64): Likewise.
22955 (MUL32x16_REG): Define.
22956 (mul64_600): Use MUL32x16_REG.
22957 (mac64_600): Likewise.
22958 (umul64_600): Likewise.
22959 (umac64_600): Likewise.
22961 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22963 * config/arc/arc.md (mulsi3_700): Make it commutative.
22965 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22967 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22969 (*sign_extendsidi2_insn): Likewise for movstosw.
22971 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22973 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22974 the type of the input discriminant value. Convert the
22975 discriminant value of signedness vary.
22977 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22979 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22980 Add to -Wall section.
22982 2017-06-01 Richard Biener <rguenther@suse.de>
22984 PR middle-end/66313
22985 * fold-const.c (fold_plusminus_mult_expr): If the factored
22986 factor may be zero use a wrapping type for the inner operation.
22987 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22988 and handle moved defs.
22989 (process_assignment): Properly guard the unary op case. Return a
22990 tri-state indicating that moving the stmt before the call may allow
22991 to continue. Pass through to_move.
22992 (find_tail_calls): Handle moving unrelated defs before
22995 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22998 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22999 splitter result in the canonical way.
23001 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23003 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23004 also for 32bit target. Update insn attributes.
23005 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23007 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23009 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23010 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23011 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23012 (_mm_maskz_min_ss): New intrinsics.
23014 2017-05-31 Martin Liska <mliska@suse.cz>
23016 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23017 Change comment style to one we normally use.
23018 (vectorizable_reduction): Likewise.
23019 (vectorizable_induction): Likewise.
23020 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23021 (vectorizable_call): Likewise.
23022 (vectorizable_simd_clone_call): Likewise.
23023 (vectorizable_conversion): Likewise.
23024 (vectorizable_assignment): Likewise.
23025 (vectorizable_shift): Likewise.
23026 (vectorizable_operation): Likewise.
23027 (vectorizable_store): Likewise.
23028 (vectorizable_load): Likewise.
23029 * tree-vectorizer.h: Likewise.
23031 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23033 * passes.c (emergency_dump_function): New.
23034 * tree-pass.h (emergency_dump_function): Declare.
23035 * plugin.c (plugins_internal_error_function): Remove.
23036 * plugin.h (plugins_internal_error_function): Remove declaration.
23037 * toplev.c (internal_error_function): New static function. Use it...
23038 (general_init): ...here.
23040 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23042 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23043 (arc_rtx_costs): Add costs for new patterns.
23044 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23045 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23047 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23049 * tree-ssa-strlen.c (get_next_strinfo): New function.
23050 (get_stridx_plus_constant): Use it.
23051 (zero_length_string): Likewise.
23052 (adjust_related_strinfos): Likewise.
23053 (adjust_last_stmt): Likewise.
23055 2017-05-31 Richard Biener <rguenther@suse.de>
23058 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23059 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23061 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23063 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23064 loop_vinfo argument and use of dependence distance vectors.
23065 Check instead whether the two references differ only in their
23066 initial value and assume that they have the same alignment if the
23067 difference is a multiple of the vector alignment.
23068 (vect_analyze_data_refs_alignment): Update call accordingly.
23070 2017-05-31 Martin Liska <mliska@suse.cz>
23073 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23075 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23077 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23078 (create_intersect_range_checks): Move from ...
23079 * tree-data-ref.c (create_intersect_range_checks_index)
23080 (create_intersect_range_checks): ... to here.
23081 (create_runtime_alias_checks): New function factored from ...
23082 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23083 here. Call above function.
23084 * tree-data-ref.h (create_runtime_alias_checks): New function.
23086 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23088 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23089 segment length for dr_b and compute it in wide_int.
23091 2017-05-31 Richard Biener <rguenther@suse.de>
23093 PR tree-optimization/80906
23094 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23095 and pass through iv_map.
23096 (copy_bb_and_scalar_dependences): Adjust.
23097 (translate_pending_phi_nodes): Likewise.
23098 (copy_loop_close_phi_args): Handle code-generating IVs instead
23101 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23103 * diagnostic-color.c (color_dict): Add "type-diff".
23104 (parse_gcc_colors): Update comment.
23105 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23106 -fdiagnostics-show-template-tree and -fno-elide-type.
23107 (GCC_COLORS): Add type-diff to example.
23109 (-fdiagnostics-show-template-tree): New.
23110 (-fno-elide-type): New.
23111 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23112 the pp_format_decoder callback. Call any m_format_postprocessor's
23114 (pretty_printer::pretty_printer): Initialize
23115 m_format_postprocessor.
23116 (pretty_printer::~pretty_printer): Delete any
23117 m_format_postprocessor.
23118 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23119 (class format_postprocessor): New class.
23120 (struct pretty_printer::format_decoder): Document the new parameters.
23121 (struct pretty_printer::m_format_postprocessor): New field.
23122 * tree-diagnostic.c (default_tree_printer): Update for new
23123 bool and const char ** params.
23124 * tree-diagnostic.h (default_tree_printer): Likewise.
23126 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23128 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23129 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23130 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23131 rs6000_gen_cell_microcode code.
23132 (rs6000_final_prescan_insn): Delete.
23133 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23134 "warn-cell-microcode" entries.
23135 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23136 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23137 throughout. Change cc_reg_not_micro_cr0_operand to
23138 cc_reg_not_cr0_operand throughout.
23139 (*extendhi<mode>2_noload): Delete.
23140 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23141 (mwarn-cell-microcode): Delete.
23142 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23143 -mgen-cell-microcode and -mwarn-cell-microcode.
23145 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23148 * config/i386/constraints.md (Yd): New constraint.
23150 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23151 and (?Yd, r) alternatives. Update insn attributes.
23152 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23153 and (?*Yd, r) alternatives. Update insn attributes.
23154 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23156 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23158 * gimplify.c (gimplify_modify_expr): Don't create a
23159 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23162 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23164 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23166 2017-05-30 Richard Biener <rguenther@suse.de>
23168 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23169 and reduc_def fields.
23170 (STMT_VINFO_REDUC_TYPE): New define.
23171 (STMT_VINFO_REDUC_DEF): Likewise.
23172 (vect_force_simple_reduction): Adjust prototype.
23173 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23174 (vect_is_simple_reduction): Remove check_reduction argument.
23175 (vect_force_simple_reduction): Adjust and set
23176 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23177 (vectorizable_reduction): Do not re-do reduction analysis
23178 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23179 * tree-parloops.c (gather_scalar_reductions): Adjust.
23181 2017-05-30 Richard Biener <rguenther@suse.de>
23183 PR middle-end/80901
23184 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23187 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23189 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23190 Introduce unknown_misalignment parameter and remove vf.
23191 (vect_peeling_hash_get_lowest_cost):
23192 Pass unknown_misalignment parameter.
23193 (vect_enhance_data_refs_alignment):
23194 Fix unsupportable data ref treatment.
23196 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23198 * tree-vect-data-refs.c (vect_get_data_access_cost):
23199 Workaround for SLP handling.
23200 (vect_enhance_data_refs_alignment):
23201 Compute costs for doing no peeling at all, compare to the best
23202 peeling costs so far and avoid peeling if cheaper.
23204 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23206 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23207 Return peeling info and set costs to zero for unlimited cost
23209 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23210 with unknown misalignment. Compute and costs for unknown
23211 misalignment, compare them to the costs for known misalignment
23212 and choose the cheapest for peeling.
23214 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23216 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23217 (vect_get_peeling_costs_all_drs): Create function.
23218 (vect_peeling_hash_get_lowest_cost):
23219 Use vect_get_peeling_costs_all_drs.
23220 (vect_peeling_supportable): Create function.
23221 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23223 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23225 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23226 DR_HAS_NEGATIVE_STEP.
23227 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23228 (vect_enhance_data_refs_alignment): Use.
23229 (vect_duplicate_ssa_name_ptr_info): Use.
23230 * tree-vectorizer.h (dr_misalignment): Use.
23231 (known_alignment_for_access_p): Use.
23233 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23236 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23238 (has_section_name): New function.
23240 2017-05-30 Martin Liska <mliska@suse.cz>
23243 * auto-profile.c (get_function_decl_from_block): Fix
23246 2017-05-30 Richard Biener <rguenther@suse.de>
23248 PR middle-end/80876
23249 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23251 2017-05-30 Martin Liska <mliska@suse.cz>
23253 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23254 * dumpfile.h (struct dump_file_info): Remove ctors.
23256 2017-05-30 Martin Liska <mliska@suse.cz>
23258 * predict.def: Fix GNU coding style.
23260 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23262 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23263 Mark 'to' argument with ATTRIBUTE_UNUSED.
23265 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23267 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23268 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23269 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23272 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23274 * doc/install.texi (Options specification): Restore entry of
23275 --enable-sjlj-exceptions.
23277 2017-05-27 Michael Eager <eager@eagercon.com>
23280 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23282 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23284 * config/microblaze/microblaze.h
23285 (FIXED_REGISTERS): Update in macro.
23286 (CALL_USED_REGISTERS): Update in macro.
23288 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23290 * doc/install.texi: Add links to macOS binary distributions.
23292 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23296 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23298 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23300 2017-05-26 Martin Liska <mliska@suse.cz>
23302 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23304 2017-05-26 Martin Liska <mliska@suse.cz>
23306 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23307 always leading ';; '.
23308 (dump_bb_info): Likewise.
23309 (brief_dump_cfg): Likewise.
23310 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23311 * dumpfile.c: Remove usage of TDF_VERBOSE.
23312 * dumpfile.h (enum dump_kind): Likewise.
23313 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23314 * print-tree.c (debug_verbose): Remove.
23315 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23316 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23317 * tree-diagnostic.c (default_tree_printer): Replace
23318 TDF_DIAGNOSTIC with TDF_SLIM.
23320 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23322 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23323 in parameter loop, rather than loop_vinfo.
23324 (create_intersect_range_checks): Ditto.
23325 (vect_create_cond_for_alias_checks): Update call to above functions.
23327 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23329 PR tree-optimization/80815
23330 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23331 for merging runtime alias checks. Handle negative DR_STEPs.
23333 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23335 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23337 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23338 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23339 out code pruning runtime alias checks.
23340 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23341 factored out from above.
23342 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23344 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23346 (prune_runtime_alias_test_list): New decalaration.
23348 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23350 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23351 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23352 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23353 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23354 (operator==, comp_dr_with_seg_len_pair): Ditto.
23355 (vect_prune_runtime_alias_test_list): Ditto.
23357 2017-05-26 Martin Liska <mliska@suse.cz>
23360 * params.def: Bound partial-inlining-entry-probability param.
23362 2017-05-26 Marek Polacek <polacek@redhat.com>
23365 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23368 2017-05-26 Richard Biener <rguenther@suse.de>
23370 PR tree-optimization/80842
23371 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23374 2017-05-26 Richard Biener <rguenther@suse.de>
23376 PR tree-optimization/80844
23377 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23379 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23381 * doc/md.texi (Machine Constraints): Update x86 family
23382 machine constraints section to match 'config/i386/constraints.md'.
23384 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23386 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23388 2017-05-25 Nathan Sidwell <nathan@acm.org>
23390 * doc/invoke.texi (--enable-languages): Update documentation.
23392 2017-05-25 Martin Liska <mliska@suse.cz>
23394 * dumpfile.c: Add TDF_FOLDING.
23395 * dumpfile.h (enum dump_kind): Likewise.
23396 * genmatch.c (dt_simplify::gen_1): Use it.
23398 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23400 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23402 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23404 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23405 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23407 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23409 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23410 * match.pd (X == C): Rewrite it here.
23411 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23412 with_certain_nonzero_bits2): New predicates.
23413 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23415 2017-05-24 Nathan Sidwell <nathan@acm.org>
23417 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23420 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23423 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23425 * config/powerpcspe: New port. Files are copied from the rs6000
23426 port, with "rs6000" in filenames replaced by "powerpcspe".
23428 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23430 PR rtl-optimization/80754
23431 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23433 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23435 * config/sparc/sparc.md (length): Return the correct value for -mflat
23436 sibcalls to match output_sibcall.
23438 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23442 * config/rs6000/rs6000.c (struct machine_function): Add new field
23444 (rs6000_get_separate_components): Init that field, use it.
23445 (rs6000_components_for_bb): Use the field.
23447 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23449 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23451 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23453 PR middle-end/80823
23454 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23456 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23459 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23460 address against address_operand predicate.
23461 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23463 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23465 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23466 parameters passed indirectly.
23468 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23470 * config/i386/i386.md (*movdi_internal): Remove SSE4
23471 alternative 18 (?r, *v). Update insn attributes.
23472 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23473 Update insn attributes.
23474 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23475 Update insn attributes.
23476 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23477 alternative 1 (r, v). Remove isa attribute.
23478 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23479 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23480 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23482 2017-05-23 Tom de Vries <tom@codesourcery.com>
23484 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23487 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23489 * cgraphunit.c (symbol_table::process_new_functions): Update.
23490 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23491 (inline_generate_summary): Rename to ...
23492 (ipa_fn_summary_generate): ... this one.
23493 (inline_read_summary): Rename to ...
23494 (ipa_fn_summary_read): ... this one.
23495 (inline_write_summary): Rename to ...
23496 (ipa_fn_summary_write): ... this one.
23497 (inline_free_summary): Rename to ...
23498 (ipa_free_fn_summary): ... this one.
23499 (pass_data_local_fn_summary, pass_local_fn_summary,
23500 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23501 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23502 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23503 make_pass_ipa_fn_summary): New.
23504 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23505 inline_write_summary, inline_free_summary): Remove.
23506 (ipa_free_fn_summary) : New.
23507 * ipa-inline.c (ipa_inline): Update.
23508 (pass_ipa_inline): Do not generate summaries.
23509 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23511 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23512 and add pass_ipa_fn_summary.
23513 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23515 (make_pass_inline_parameters): Remove.
23517 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23519 * omp-low.c (struct omp_context): Remove "default_kind" member.
23522 * omp-offload.c (execute_oacc_device_lower): Remove the
23523 parallelism dimensions function attributes for unparallelized
23524 OpenACC kernels constructs.
23526 2017-05-23 Martin Liska <mliska@suse.cz>
23528 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23530 (cgraph_edge::make_speculative): Likewise.
23531 (cgraph_edge::resolve_speculation): Likewise.
23532 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23533 (cgraph_node::dump): Likewise.
23534 * cgraph.h: Likewise.
23535 * cgraphunit.c (analyze_functions): Likewise.
23536 (symbol_table::compile): Likewise.
23537 * ipa-cp.c (print_all_lattices): Likewise.
23538 (determine_versionability): Likewise.
23539 (initialize_node_lattices): Likewise.
23540 (ipcp_verify_propagated_values): Likewise.
23541 (estimate_local_effects): Likewise.
23542 (update_profiling_info): Likewise.
23543 (create_specialized_node): Likewise.
23544 (perhaps_add_new_callers): Likewise.
23545 (decide_about_value): Likewise.
23546 (decide_whether_version_node): Likewise.
23547 (identify_dead_nodes): Likewise.
23548 (ipcp_store_bits_results): Likewise.
23549 * ipa-devirt.c (dump_targets): Likewise.
23550 (ipa_devirt): Likewise.
23551 * ipa-icf.c (sem_item::dump): Likewise.
23552 (sem_function::equals): Likewise.
23553 (sem_variable::equals): Likewise.
23554 (sem_item_optimizer::read_section): Likewise.
23555 (sem_item_optimizer::execute): Likewise.
23556 (congruence_class::dump): Likewise.
23557 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23558 (dump_inline_summary): Likewise.
23559 (estimate_node_size_and_time): Likewise.
23560 (inline_analyze_function): Likewise.
23561 * ipa-inline-transform.c (inline_call): Likewise.
23562 * ipa-inline.c (report_inline_failed_reason): Likewise.
23563 (want_early_inline_function_p): Likewise.
23564 (edge_badness): Likewise.
23565 (update_edge_key): Likewise.
23566 (inline_small_functions): Likewise.
23567 * ipa-profile.c (ipa_profile): Likewise.
23568 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23569 (ipa_make_edge_direct_to_target): Likewise.
23570 (remove_described_reference): Likewise.
23571 (ipa_impossible_devirt_target): Likewise.
23572 (propagate_controlled_uses): Likewise.
23573 (ipa_print_node_params): Likewise.
23574 (ipcp_transform_function): Likewise.
23575 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23576 (propagate_pure_const): Likewise.
23577 * ipa-reference.c (generate_summary): Likewise.
23578 (read_write_all_from_decl): Likewise.
23579 (propagate): Likewise.
23580 (ipa_reference_read_optimization_summary): Likewise.
23581 * ipa-utils.c (ipa_merge_profiles): Likewise.
23582 * ipa.c (walk_polymorphic_call_targets): Likewise.
23583 (symbol_table::remove_unreachable_nodes): Likewise.
23584 (ipa_single_use): Likewise.
23585 * passes.c (execute_todo): Likewise.
23586 * predict.c (drop_profile): Likewise.
23587 * symtab.c (symtab_node::get_dump_name): New function.
23588 (symtab_node::dump_name): Likewise.
23589 (symtab_node::dump_asm_name): Likewise.
23590 (symtab_node::dump_references): Likewise.
23591 (symtab_node::dump_referring): Likewise.
23592 (symtab_node::dump_base): Likewise.
23593 (symtab_node::debug_symtab): Likewise.
23594 * tree-sra.c (convert_callers_for_node): Likewise.
23595 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23596 * value-prof.c (init_node_map): Likewise.
23598 2017-05-23 Martin Liska <mliska@suse.cz>
23600 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23601 and symtab_node::debug_symtab to symbol_table::debug.
23602 * cgraphunit.c (analyze_functions): Use the renamed function.
23603 (symbol_table::compile): Likewise.
23604 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23605 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23606 * passes.c (execute_todo): Likewise.
23607 * symtab.c (symbol_table::dump): New function.
23608 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23610 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23612 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23613 that nonconst implies exec.
23615 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23617 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23618 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23619 (inline_edge_summary_vec): Turn into ...
23620 (ipa_call_summaries): ... this one.
23621 (redirect_to_unreachable, edge_set_predicate,
23622 evaluate_properties_for_edge, inline_summary_alloc,
23623 reset_ipa_call_summary, reset_inline_summary,
23624 inline_summary_t::duplicate): Update.
23625 (inline_edge_duplication_hook): Turn to ...
23626 (ipa_call_summary_t::duplicate): ... this one.
23627 (inline_edge_removal_hook): Turn to ...
23628 (ipa_call_summary_t::remove): ... this one.
23629 (dump_inline_edge_summary): Turn to ...
23630 (dump_ipa_call_summary): ... this one.
23631 (estimate_function_body_sizes): Update.
23632 (inline_update_callee_summaries): Update.
23633 (remap_edge_change_prob): Update.
23634 (remap_edge_summaries): Update.
23635 (inline_merge_summary): Update.
23636 (do_estimate_edge_time): Update.
23637 (inline_generate_summary): Update.
23638 (inline_read_section): Update.
23639 (inline_read_summary): Update.
23640 (inline_free_summary): Update.
23641 * ipa-inline.c (can_inline_edge_p): Update.
23642 (compute_inlined_call_time): Update.
23643 (want_inline_small_function_p): Update.
23644 (edge_badness): Update.
23645 (early_inliner): Update.
23646 * ipa-inline.h (inline_edge_summary): Turn to ...
23647 (ipa_call_summary): ... this one.
23648 (ipa_call_summary_t): New class.
23649 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23650 (ipa_call_summaries): New.
23651 (inline_edge_summary): Remove.
23652 (estimate_edge_growth): Update.
23653 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23654 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23655 * ipa-split.c (execute_split_functions): Update.
23656 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23658 2017-05-23 Tom de Vries <tom@codesourcery.com>
23660 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23661 attributes): Document rdrand effective target.
23663 2017-05-23 Tom de Vries <tom@codesourcery.com>
23665 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23666 attributes): Sort alphabetically.
23668 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23670 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23672 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23675 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23676 V2DF/V2DI splat into two separate patterns, one that handles
23677 registers, and the other that only handles memory. Drop support
23678 for splatting from a GPR on ISA 2.07 and then splitting the
23679 splat into direct move and splat.
23680 (vsx_splat_<mode>_reg): Likewise.
23681 (vsx_splat_<mode>_mem): Likewise.
23683 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23685 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23687 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23689 PR middle-end/80809
23690 * omp-low.c (finish_taskreg_remap): New function.
23691 (finish_taskreg_scan): If unit size of ctx->record_type
23692 is non-constant, unshare the size expression and replace
23693 decls in it with possible outer var refs.
23695 PR middle-end/80809
23696 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23697 GOVD_SHARED rather than GOVD_PRIVATE with it.
23698 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23699 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23701 PR middle-end/80853
23702 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23703 as last argument to build_outer_var_ref for pointer bases of array
23704 section reductions.
23706 2017-05-19 Martin Sebor <msebor@redhat.com>
23708 * print-tree.c (print_node): Print DECL_READ_P flag.
23710 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23712 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23713 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23714 * cgraph.c: Likewise.
23715 * cgraphunit.c: Likewise.
23716 * gengtype.c: Likewise.
23717 * ipa-cp.c: Likewise.
23718 * ipa-devirt.c: Likewise.
23719 * ipa-icf.c: Likewise.
23720 * ipa-predicate.c: Likewise.
23721 * ipa-profile.c: Likewise.
23722 * ipa-prop.c: Likewise.
23723 * ipa-split.c: Likewise.
23725 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23726 edge_predicate_pool, dump_inline_hints,
23727 inline_summary::account_size_time, redirect_to_unreachable,
23728 edge_set_predicate, set_hint_predicate,
23729 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23730 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23731 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23732 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23733 ipa_call_summary_t::remove, initialize_growth_caches,
23734 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23735 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23736 mark_modified, unmodified_parm_1, unmodified_parm,
23737 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23738 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23739 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23740 will_be_nonconstant_predicate, record_modified_bb_info,
23741 get_minimal_bb, record_modified, param_change_prob,
23742 phi_result_unknown_predicate, predicate_for_phi_result,
23743 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23744 estimate_function_body_sizes, compute_inline_parameters,
23745 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23746 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23747 inline_update_callee_summaries, remap_edge_change_prob,
23748 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23749 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23750 inline_analyze_function, inline_summary_t::insert,
23751 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23752 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23753 inline_free_summary): Move to ipa-fnsummary.h
23754 (predicate_t): Remove.
23755 * ipa-fnsummary.c: New file.
23756 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23757 (enum inline_hints_vals, inline_hints, agg_position_info,
23758 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23759 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23760 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23761 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23762 inline_read_summary, inline_write_summary, inline_free_summary,
23763 inline_analyze_function, initialize_inline_failed,
23764 inline_merge_summary, inline_update_overall_summary,
23765 compute_inline_parameters): Move to ipa-fnsummary.h
23766 * ipa-fnsummary.h: New file.
23767 * ipa-inline-transform.h: Include ipa-inline.h.
23768 * ipa-inline.c: LIkewise.
23770 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23772 * ipa-inline.c (edge_badness): Use inlined_time instead of
23773 inline_summaries->get.
23775 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23777 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23779 2017-05-22 Nathan Sidwell <nathan@acm.org>
23781 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23782 (fdump-lang): Document 'raw' option.
23783 * dumpfile.h (TDI_tu): Delete.
23784 * dumpfile.c (dump_files): Remove translation-unit.
23785 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23787 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23789 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23790 command option from $(AWK) call.
23791 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23792 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23793 [FORMAT]: Remove handling of variable.
23794 * config/avr/t-multilib: Regenerate.
23796 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23798 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23800 (dump_inline_summary): Do not print self_time.
23801 (estimate_function_body_sizes): Do not set self_time.
23802 (compute_inline_parameters): Likewise.
23803 (inline_read_section, inline_write_summary): Do not stream self_time.
23804 * ipa-inline.h (inline_summary): Drop self_time.
23806 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23808 * ipa-inline-analysis.c (account_size_time): Rename to ...
23809 (inline_summary::account_size_time): ... this one.
23810 (reset_ipa_call_summary): Turn to ...
23811 (ipa_call_summary::reset): ... this one.
23812 (reset_inline_summary): Turn to ...
23813 (inline_summary::reset): ... this one.
23814 (inline_summary_t::remove): Update.
23815 (inline_summary_t::duplicate): Update.
23816 (ipa_call_summary_t::remove): Update.
23817 (dump_inline_summary): Update.
23818 (estimate_function_body_sizes): Update.
23819 (compute_inline_parameters): Update.
23820 (estimate_node_size_and_time): Update.
23821 (inline_merge_summary): Update.
23822 (inline_update_overall_summary): Update.
23823 (inline_read_section): Update.
23824 (inline_write_summary): Update.
23825 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23826 add account_size_time and reset member functions.
23827 (ipa_call_summary): Add reset function.
23828 * ipa-predicate.h (predicate::operator &): Constify.
23830 2017-05-22 Richard Biener <rguenther@suse.de>
23832 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23834 2017-05-19 Jason Merrill <jason@redhat.com>
23836 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23838 2017-05-19 Marek Polacek <polacek@redhat.com>
23841 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23842 TYPE_OVERFLOW_WRAPS checks.
23844 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23846 * tree-core.h (enum omp_clause_default_kind): Add
23847 "OMP_CLAUSE_DEFAULT_PRESENT".
23848 * tree-pretty-print.c (dump_omp_clause): Handle it.
23849 * gimplify.c (enum gimplify_omp_var_data): Add
23850 "GOVD_MAP_FORCE_PRESENT".
23851 (gimplify_adjust_omp_clauses_1): Map it to
23852 "GOMP_MAP_FORCE_PRESENT".
23853 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23855 * gimplify.c (oacc_default_clause): Clarify.
23857 2017-05-19 Nathan Sidwell <nathan@acm.org>
23859 LANG_HOOK_REGISTER_DUMPS
23860 * toplev.c (general_init): Call register dump lang hook.
23861 * doc/invoke.texi: Document -fdump-lang option family.
23862 * dumpfile.c (dump_files): Remove class dump here.
23863 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23864 * dumpfile.h (tree_dump_index): Remove TDI_class.
23865 * langhooks-def.h (lhd_register_dumps): Declare.
23866 (LANG_HOOKS_REGISTER_DUMPS): Define.
23867 (LANG_HOOKS_INITIALIZER): Add it.
23868 * langhooks.c (lhd_register_dumps): Define.
23869 * langhooks.h (struct lang_hooks): Add register_dumps.
23871 2017-05-19 Nathan Sidwell <nathan@acm.org>
23873 * context.h (context::set_passes): New.
23874 * context.c (context::context): Do not create pass manager.
23875 * toplev.c (general_init): Create pass manager here.
23877 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23879 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23880 use this splitter if two add or or instructions would also work for
23881 the constant we want to generate.
23883 2017-05-19 Richard Biener <rguenther@suse.de>
23886 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23887 predicate evaluation.
23889 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23891 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23893 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23896 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23898 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23899 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23900 (inline_edge_summary_vec): Turn into ...
23901 (ipa_call_summaries): ... this one.
23902 (redirect_to_unreachable, edge_set_predicate,
23903 evaluate_properties_for_edge, inline_summary_alloc,
23904 reset_ipa_call_summary, reset_inline_summary,
23905 inline_summary_t::duplicate): Update.
23906 (inline_edge_duplication_hook): Turn to ...
23907 (ipa_call_summary_t::duplicate): ... this one.
23908 (inline_edge_removal_hook): Turn to ...
23909 (ipa_call_summary_t::remove): ... this one.
23910 (dump_inline_edge_summary): Turn to ...
23911 (dump_ipa_call_summary): ... this one.
23912 (estimate_function_body_sizes): Update.
23913 (inline_update_callee_summaries): Update.
23914 (remap_edge_change_prob): Update.
23915 (remap_edge_summaries): Update.
23916 (inline_merge_summary): Update.
23917 (do_estimate_edge_time): Update.
23918 (inline_generate_summary): Update.
23919 (inline_read_section): Update.
23920 (inline_read_summary): Update.
23921 (inline_free_summary): Update.
23922 * ipa-inline.c (can_inline_edge_p): Update.
23923 (compute_inlined_call_time): Update.
23924 (want_inline_small_function_p): Update.
23925 (edge_badness): Update.
23926 (early_inliner): Update.
23927 * ipa-inline.h (inline_edge_summary): Turn to ...
23928 (ipa_call_summary): ... this one.
23929 (ipa_call_summary_t): New class.
23930 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23931 (ipa_call_summaries): New.
23932 (inline_edge_summary): Remove.
23933 (estimate_edge_growth): Update.
23934 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23935 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23936 * ipa-split.c (execute_split_functions): Update.
23937 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23939 2017-05-19 Richard Biener <rguenther@suse.de>
23941 PR middle-end/80764
23942 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23944 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23946 * config/rs6000/rs6000.c (struct machine_function): Add field
23947 fpr_is_wrapped_separately.
23948 (rs6000_get_separate_components): Use 64 components. Handle the
23949 new FPR components.
23950 (rs6000_components_for_bb): Handle the FPR components.
23951 (rs6000_emit_prologue_components): Handle the FPR components.
23952 (rs6000_emit_epilogue_components): Handle the FPR components.
23953 (rs6000_set_handled_components): Handle the FPR components.
23954 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23955 that are already separately shrink-wrapped.
23956 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23957 that are already separately shrink-wrapped.
23959 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23962 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23965 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23966 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23967 cases where the register allocator uses a move and an offsettable
23968 memory operation to/from a FPR register on ISA 2.06/2.07.
23969 (define_peephole2 for Altivec d-form store): Likewise.
23971 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23974 * config/i386/mmx.md (*mov<mode>_internal): Enable
23975 alternatives 11, 12, 13 and 14 also for 32bit targets.
23976 Remove alternatives 15, 16, 17 and 18.
23977 * config/i386/sse.md (vec_concatv2di): Change
23978 alternative (!x, *y) to (x, ?!*Yn).
23980 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23982 * dumpfile.h (enum dump_kind): Remove stray comma.
23984 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23986 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23987 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23988 predicate::num_conditions
23989 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23990 (CHANGED): turn into predicate::changed.
23991 (agg_position_info): Move to ipa-predicate.h
23992 (add_condition, predicate::add_clause, predicate::operator &=,
23993 predicate::or_with, predicate::evaluate, predicate::probability,
23994 dump_condition, dump_clause, predicate::dump,
23995 predicate::remap_after_duplication, predicate::remap_after_inlining,
23996 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23997 (evaluate_conditions_for_known_args): Update.
23998 (set_cond_stmt_execution_predicate): Update.
23999 * ipa-inline.h: Include ipa-predicate.h
24000 (condition, inline_param_summary, conditions, agg_position_info,
24001 predicate): Move to ipa-predicate.h
24002 * ipa-predicate.c: New file.
24003 * ipa-predicate.h: New file.
24005 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24007 * final.c (leaf_function_p): Check we are not in a sequence.
24009 2017-05-18 Martin Liska <mliska@suse.cz>
24011 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24012 * dumpfile.c (dump_register): Use new enum dump_kind.
24013 (get_dump_file_name): Likewise.
24014 (dump_enable_all): Likewise.
24015 (dump_switch_p_1): Likewise.
24016 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24017 * dumpfile.h (enum dump_kind): New enum type.
24018 (struct dump_file_info): Create constructor and
24019 format fields and comments.
24020 * passes.c (pass_manager::register_one_dump_file):
24022 * statistics.c (statistics_early_init): Likewise.
24023 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24024 TDF_TREE with TDF_SLIM.
24025 (gather_memory_references_ref): Likewise.
24027 2017-05-18 Martin Liska <mliska@suse.cz>
24029 * vec.h (struct vnull): Use it.
24031 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24033 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24034 (true_predicate, false_predicate, true_predicate_p,
24035 false_predicate_p): Remove.
24036 (single_cond_predicate, not_inlined_predicate): Turn to member function
24038 (add_condition): Update.
24039 (add_clause): Turn to...
24040 (predicate::add_clause): ... this one; update; allow passing NULL
24042 (and_predicates): Turn to ...
24043 (predicate::operator &=): ... this one.
24044 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24045 (or_predicates): Turn to ...
24046 (predicate::or_with): ... this one.
24047 (evaluate_predicate): Turn to ...
24048 (predicate::evaluate): ... this one.
24049 (predicate_probability): Turn to ...
24050 (predicate::probability): ... this one.
24051 (dump_condition): Update.
24052 (dump_predicate): Turn to ...
24053 (predicate::dump): ... this one.
24054 (account_size_time): Update.
24055 (edge_set_predicate): Update.
24056 (set_hint_predicate): UPdate.
24057 (evaluate_conditions_for_known_args): Update.
24058 (evaluate_properties_for_edge): Update.
24059 (remap_predicate_after_duplication): Turn to...
24060 (predicate::remap_after_duplication): ... this one.
24061 (remap_hint_predicate_after_duplication): Update.
24062 (inline_summary_t::duplicate): UPdate.
24063 (dump_inline_edge_summary): Update.
24064 (dump_inline_summary): Update.
24065 (set_cond_stmt_execution_predicate): Update.
24066 (set_switch_stmt_execution_predicate): Update.
24067 (compute_bb_predicates): Update.
24068 (will_be_nonconstant_expr_predicate): Update.
24069 (will_be_nonconstant_predicate): Update.
24070 (phi_result_unknown_predicate): Update.
24071 (predicate_for_phi_result): Update.
24072 (array_index_predicate): Update.
24073 (estimate_function_body_sizes): Update.
24074 (estimate_node_size_and_time): Update.
24075 (estimate_ipcp_clone_size_and_time): Update.
24076 (remap_predicate): Rename to ...
24077 (predicate::remap_after_inlining): ... this one.
24078 (remap_hint_predicate): Update.
24079 (inline_merge_summary): Update.
24080 (inline_update_overall_summary): Update.
24081 (estimate_size_after_inlining): Update.
24082 (read_predicate): Rename to ...
24083 (predicate::stream_in): ... this one.
24084 (read_inline_edge_summary): Update.
24085 (write_predicate): Rename to ...
24086 (predicate::stream_out): ... this one.
24087 (write_inline_edge_summary): Update.
24088 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24089 (clause_t): Turn to uint32_t
24090 (predicate): Turn to class; implement constructor and operators
24092 (size_time_entry): Update.
24093 (inline_summary): Update.
24094 (inline_edge_summary): Update.
24096 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24098 * fold-const.c (fold_binary_loc): Move transformation...
24099 * match.pd (C - X CMP X): ... here.
24101 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24103 * config/sparc/sparc.c (sparc_option_override): Set function
24104 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24105 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24107 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24109 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24111 2017-05-18 Marek Polacek <polacek@redhat.com>
24114 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24115 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24116 gimple_assign_load_p.
24118 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24120 PR middle-end/80692
24121 * real.c (do_compare): Give decimal_do_compare preference over
24122 comparing just the signs.
24124 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24126 * doc/md.texi (Canonicalization of Instructions): Describe the
24127 canonical form of instructions that inherently set a condition
24130 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24132 PR middle-end/80775
24133 * tree-cfg.c: Move deletion of unreachable case statements to after
24134 the merging of consecutive case labels.
24136 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24138 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24139 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24140 restoring of callee-saved registers.
24142 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24144 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24145 * config/visium/visium.c (single_set_and_flags): Likewise.
24146 * config/visium/visium.md (Substitutions): Likewise.
24148 2017-05-17 Martin Liska <mliska@suse.cz>
24150 * cfg.c: Introduce dump_flags_t type and
24151 use it instead of int type.
24153 * cfghooks.c: Likewise.
24154 * cfghooks.h (struct cfg_hooks): Likewise.
24155 * cfgrtl.c: Likewise.
24156 * cfgrtl.h: Likewise.
24157 * cgraph.c (cgraph_node::get_body): Likewise.
24158 * coretypes.h: Likewise.
24159 * domwalk.c: Likewise.
24160 * domwalk.h: Likewise.
24161 * dumpfile.c (struct dump_option_value_info): Likewise.
24162 (dump_enable_all): Likewise.
24163 (dump_switch_p_1): Likewise.
24164 (opt_info_switch_p): Likewise.
24165 * dumpfile.h (enum tree_dump_index): Likewise.
24166 (struct dump_file_info): Likewise.
24167 * genemit.c: Likewise.
24168 * generic-match-head.c: Likewise.
24169 * gengtype.c (open_base_files): Likewise.
24170 * gimple-pretty-print.c: Likewise.
24171 * gimple-pretty-print.h: Likewise.
24172 * graph.c (print_graph_cfg): Likewise.
24173 * graphite-scop-detection.c (dot_all_sese): Likewise.
24174 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24175 * loop-unroll.c (report_unroll): Likewise.
24176 * passes.c (pass_manager::register_one_dump_file): Likewise.
24177 * print-tree.c: Likewise.
24178 * statistics.c: Likewise.
24179 * tree-cfg.c: Likewise.
24180 * tree-cfg.h: Likewise.
24181 * tree-dfa.c: Likewise.
24182 * tree-dfa.h: Likewise.
24183 * tree-dump.c (dump_function): Likewise.
24184 * tree-dump.h (struct dump_info): Likewise.
24185 * tree-pretty-print.c: Likewise.
24186 * tree-pretty-print.h: Likewise.
24187 * tree-ssa-live.c: Likewise.
24188 * tree-ssa-live.h: Likewise.
24189 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24190 * tree-vect-loop.c: Likewise.
24191 * tree-vect-slp.c: Likewise.
24193 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24194 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24196 PR tree-optimization/80457
24197 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24198 of all arguments to a statement as scalar_to_vec operations.
24199 (vectorizable_call): Adjust call to vect_model_simple_cost for
24201 (vectorizable_conversion): Likewise.
24202 (vectorizable_assignment): Likewise.
24203 (vectorizable_shift): Likewise.
24204 (vectorizable_operation): Likewise.
24205 (vectorizable_comparison): Likewise.
24206 (vect_is_simple_cond): Record the def types for operands.
24207 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24208 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24209 for statement argument count.
24211 2017-05-16 Carl Love <cel@us.ibm.com>
24213 * config/rs6000/rs6000-c: Add support for built-in functions
24214 vector unsigned long long vec_bperm (vector unsigned long long,
24215 vector unsigned char)
24216 vector signed long long vec_mule (vector signed int,
24218 vector unsigned long long vec_mule (vector unsigned int,
24219 vector unsigned int)
24220 vector signed long long vec_mulo (vector signed int,
24222 vector unsigned long long vec_mulo (vector unsigned int,
24223 vector unsigned int)
24224 vector signed char vec_sldw (vector signed char,
24225 vector signed char,
24227 vector unsigned char vec_sldw (vector unsigned char,
24228 vector unsigned char,
24230 vector signed short vec_sldw (vector signed short,
24231 vector signed short,
24233 vector unsigned short vec_sldw (vector unsigned short,
24234 vector unsigned short,
24236 vector signed int vec_sldw (vector signed int,
24239 vector unsigned int vec_sldw (vector unsigned int,
24240 vector unsigned int,
24242 vector signed long long vec_sldw (vector signed long long,
24243 vector signed long long,
24245 vector unsigned long long vec_sldw (vector unsigned long long,
24246 vector unsigned long long,
24248 * config/rs6000/rs6000-c: Add support for built-in functions
24249 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24250 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24251 * doc/extend.texi: Update the built-in documentation for the
24252 new built-in functions.
24254 2017-05-16 Marek Polacek <polacek@redhat.com>
24258 * tree.c (save_expr): Don't fold the expression.
24260 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24262 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24263 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24264 and (?*y,m). Update insn attributes.
24266 2017-05-16 Martin Liska <mliska@suse.cz>
24268 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24269 flags argument of print_gimple_stmt, print_gimple_expr,
24270 print_generic_stmt and print_generic_expr.
24271 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24272 * coretypes.h: Likewise.
24273 * except.c (dump_eh_tree): Likewise.
24274 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24275 * gimple-pretty-print.h: Likewise.
24276 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24277 (backprop::push_to_worklist): Likewise.
24278 (backprop::pop_from_worklist): Likewise.
24279 (backprop::process_use): Likewise.
24280 (backprop::intersect_uses): Likewise.
24281 (note_replacement): Likewise.
24282 * gimple-ssa-store-merging.c
24283 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24284 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24285 (pass_store_merging::execute): Likewise.
24286 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24287 (ssa_base_cand_dump_callback): Likewise.
24288 (dump_incr_vec): Likewise.
24289 (replace_refs): Likewise.
24290 (replace_mult_candidate): Likewise.
24291 (create_add_on_incoming_edge): Likewise.
24292 (create_phi_basis): Likewise.
24293 (insert_initializers): Likewise.
24294 (all_phi_incrs_profitable): Likewise.
24295 (introduce_cast_before_cand): Likewise.
24296 (replace_one_candidate): Likewise.
24297 * gimplify.c (gimplify_expr): Likewise.
24298 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24299 (set_rename): Likewise.
24300 (rename_uses): Likewise.
24301 (copy_loop_phi_nodes): Likewise.
24302 (add_close_phis_to_merge_points): Likewise.
24303 (copy_loop_close_phi_args): Likewise.
24304 (copy_cond_phi_args): Likewise.
24305 (graphite_copy_stmts_from_block): Likewise.
24306 (translate_pending_phi_nodes): Likewise.
24307 * graphite-poly.c (print_pdr): Likewise.
24308 (dump_gbb_cases): Likewise.
24309 (dump_gbb_conditions): Likewise.
24310 (print_scop_params): Likewise.
24311 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24312 (build_cross_bb_scalars_use): Likewise.
24313 (gather_bbs::before_dom_children): Likewise.
24314 * hsa-dump.c (dump_hsa_immed): Likewise.
24315 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24316 (get_replacement_map): Likewise.
24317 * ipa-inline-analysis.c (dump_condition): Likewise.
24318 (estimate_function_body_sizes): Likewise.
24319 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24320 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24321 * ipa-prop.c (ipa_dump_param): Likewise.
24322 (ipa_print_node_jump_functions_for_edge): Likewise.
24323 (ipa_modify_call_arguments): Likewise.
24324 (ipa_modify_expr): Likewise.
24325 (ipa_dump_param_adjustments): Likewise.
24326 (ipa_dump_agg_replacement_values): Likewise.
24327 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24328 * ipa-pure-const.c (check_stmt): Likewise.
24329 (pass_nothrow::execute): Likewise.
24330 * ipa-split.c (execute_split_functions): Likewise.
24331 * omp-offload.c (dump_oacc_loop_part): Likewise.
24332 (dump_oacc_loop): Likewise.
24333 * trans-mem.c (tm_log_emit): Likewise.
24334 (tm_memopt_accumulate_memops): Likewise.
24335 (dump_tm_memopt_set): Likewise.
24336 (dump_tm_memopt_transform): Likewise.
24337 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24338 (print_loop): Likewise.
24339 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24340 (chkp_gather_checks_info): Likewise.
24341 (chkp_get_check_result): Likewise.
24342 (chkp_remove_check_if_pass): Likewise.
24343 (chkp_use_outer_bounds_if_possible): Likewise.
24344 (chkp_reduce_bounds_lifetime): Likewise.
24345 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24346 (chkp_mark_completed_bounds): Likewise.
24347 (chkp_register_incomplete_bounds): Likewise.
24348 (chkp_mark_invalid_bounds): Likewise.
24349 (chkp_maybe_copy_and_register_bounds): Likewise.
24350 (chkp_build_returned_bound): Likewise.
24351 (chkp_get_bound_for_parm): Likewise.
24352 (chkp_build_bndldx): Likewise.
24353 (chkp_get_bounds_by_definition): Likewise.
24354 (chkp_generate_extern_var_bounds): Likewise.
24355 (chkp_get_bounds_for_decl_addr): Likewise.
24356 * tree-chrec.c (chrec_apply): Likewise.
24357 * tree-data-ref.c (dump_data_reference): Likewise.
24358 (dump_subscript): Likewise.
24359 (dump_data_dependence_relation): Likewise.
24360 (analyze_overlapping_iterations): Likewise.
24361 * tree-inline.c (expand_call_inline): Likewise.
24362 (tree_function_versioning): Likewise.
24363 * tree-into-ssa.c (dump_defs_stack): Likewise.
24364 (dump_currdefs): Likewise.
24365 (dump_names_replaced_by): Likewise.
24366 (dump_update_ssa): Likewise.
24367 (update_ssa): Likewise.
24368 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24369 * tree-parloops.c (build_new_reduction): Likewise.
24370 (try_create_reduction_list): Likewise.
24371 (ref_conflicts_with_region): Likewise.
24372 (oacc_entry_exit_ok_1): Likewise.
24373 (oacc_entry_exit_single_gang): Likewise.
24374 * tree-pretty-print.h: Likewise.
24375 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24376 (get_scalar_evolution): Likewise.
24377 (add_to_evolution): Likewise.
24378 (get_loop_exit_condition): Likewise.
24379 (analyze_evolution_in_loop): Likewise.
24380 (analyze_initial_condition): Likewise.
24381 (analyze_scalar_evolution): Likewise.
24382 (instantiate_scev): Likewise.
24383 (number_of_latch_executions): Likewise.
24384 (gather_chrec_stats): Likewise.
24385 (final_value_replacement_loop): Likewise.
24386 (scev_const_prop): Likewise.
24387 * tree-sra.c (dump_access): Likewise.
24388 (disqualify_candidate): Likewise.
24389 (create_access): Likewise.
24390 (reject): Likewise.
24391 (maybe_add_sra_candidate): Likewise.
24392 (create_access_replacement): Likewise.
24393 (analyze_access_subtree): Likewise.
24394 (analyze_all_variable_accesses): Likewise.
24395 (sra_modify_assign): Likewise.
24396 (initialize_constant_pool_replacements): Likewise.
24397 (find_param_candidates): Likewise.
24398 (decide_one_param_reduction): Likewise.
24399 (replace_removed_params_ssa_names): Likewise.
24400 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24401 * tree-ssa-copy.c (dump_copy_of): Likewise.
24402 (copy_prop_visit_cond_stmt): Likewise.
24403 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24404 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24405 (record_equivalences_from_stmt): Likewise.
24406 * tree-ssa-dse.c (compute_trims): Likewise.
24407 (delete_dead_call): Likewise.
24408 (delete_dead_assignment): Likewise.
24409 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24410 (forward_propagate_into_cond): Likewise.
24411 (pass_forwprop::execute): Likewise.
24412 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24413 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24415 (move_computations_worker): Likewise.
24416 (execute_sm): Likewise.
24417 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24418 (remove_exits_and_undefined_stmts): Likewise.
24419 (remove_redundant_iv_tests): Likewise.
24420 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24421 (adjust_iv_update_pos): Likewise.
24422 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24423 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24424 (value_replacement): Likewise.
24425 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24426 * tree-ssa-pre.c (print_pre_expr): Likewise.
24427 (get_representative_for): Likewise.
24428 (create_expression_by_pieces): Likewise.
24429 (insert_into_preds_of_block): Likewise.
24430 (eliminate_insert): Likewise.
24431 (eliminate_dom_walker::before_dom_children): Likewise.
24432 (eliminate): Likewise.
24433 (remove_dead_inserted_code): Likewise.
24434 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24435 * tree-ssa-reassoc.c (get_rank): Likewise.
24436 (eliminate_duplicate_pair): Likewise.
24437 (eliminate_plus_minus_pair): Likewise.
24438 (eliminate_not_pairs): Likewise.
24439 (undistribute_ops_list): Likewise.
24440 (eliminate_redundant_comparison): Likewise.
24441 (update_range_test): Likewise.
24442 (optimize_range_tests_var_bound): Likewise.
24443 (optimize_vec_cond_expr): Likewise.
24444 (rewrite_expr_tree): Likewise.
24445 (rewrite_expr_tree_parallel): Likewise.
24446 (linearize_expr): Likewise.
24447 (break_up_subtract): Likewise.
24448 (linearize_expr_tree): Likewise.
24449 (attempt_builtin_powi): Likewise.
24450 (attempt_builtin_copysign): Likewise.
24451 (transform_stmt_to_copy): Likewise.
24452 (transform_stmt_to_multiply): Likewise.
24453 (dump_ops_vector): Likewise.
24454 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24455 (print_scc): Likewise.
24456 (set_ssa_val_to): Likewise.
24457 (visit_reference_op_store): Likewise.
24458 (visit_use): Likewise.
24459 (sccvn_dom_walker::before_dom_children): Likewise.
24460 (run_scc_vn): Likewise.
24461 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24463 (expr_hash_elt::print): Likewise.
24464 (const_and_copies::pop_to_marker): Likewise.
24465 (const_and_copies::record_const_or_copy_raw): Likewise.
24466 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24467 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24468 (dump_predicates): Likewise.
24469 (find_uninit_use): Likewise.
24470 (warn_uninitialized_phi): Likewise.
24471 (pass_late_warn_uninitialized::execute): Likewise.
24472 * tree-ssa.c (verify_vssa): Likewise.
24473 (verify_ssa): Likewise.
24474 (maybe_optimize_var): Likewise.
24475 * tree-vrp.c (dump_value_range): Likewise.
24476 (dump_all_value_ranges): Likewise.
24477 (dump_asserts_for): Likewise.
24478 (register_edge_assert_for_2): Likewise.
24479 (vrp_visit_cond_stmt): Likewise.
24480 (vrp_visit_switch_stmt): Likewise.
24481 (vrp_visit_stmt): Likewise.
24482 (vrp_visit_phi_node): Likewise.
24483 (simplify_cond_using_ranges_1): Likewise.
24484 (fold_predicate_in): Likewise.
24485 (evrp_dom_walker::before_dom_children): Likewise.
24486 (evrp_dom_walker::push_value_range): Likewise.
24487 (evrp_dom_walker::pop_value_range): Likewise.
24488 (execute_early_vrp): Likewise.
24490 2017-05-16 Richard Biener <rguenther@suse.de>
24492 * dwarf2out.c (loc_list_from_tree_1): Do not create
24493 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24495 2017-05-16 Richard Biener <rguenther@suse.de>
24497 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24499 (note_variable_value_in_expr): If we resolved the decl ref
24500 do not push to the stack.
24502 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24504 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24505 operations in fast-math mode.
24506 (vaddq_f16): Likewise.
24507 (vmul_f16): Likewise.
24508 (vmulq_f16): Likewise.
24509 (vsub_f16): Likewise.
24510 (vsubq_f16): Likewise.
24511 * config/arm/neon.md (add<mode>3): New.
24513 (fma:<VH:mode>3): New. Also remove outdated comment.
24516 2017-05-16 Martin Liska <mliska@suse.cz>
24520 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24521 (odr_types_equivalent_p): Likewise.
24523 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24525 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24527 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24530 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24531 non-interunit SSE move alternatives with '?'.
24532 (zero-extendsidi peephole2): New peephole to skip intermediate
24533 general register in SSE zero-extend sequence.
24535 2017-05-15 Jeff Law <law@redhat.com>
24537 * reorg.c (relax_delay_slots): Create a new variable to hold
24538 the temporary target rather than clobbering TARGET_LABEL.
24540 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24541 missing argument to extract_bit_field call.
24542 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24544 2017-05-15 Martin Liska <mliska@suse.cz>
24547 * gcc.c (process_command): Do not allow empty argument of -o option.
24549 2017-05-15 Renlin Li <renlin.li@arm.com>
24551 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24552 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24553 * config/aarch64/constraints.md (Usf): Add long call check.
24554 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24555 (call_value): Likewise.
24556 (sibcall): Likewise.
24557 (sibcall_value): Likewise.
24559 (call_value_insn): New.
24560 (sibcall_insn): Update rtx pattern.
24561 (sibcall_value_insn): Likewise.
24562 (call_internal): Remove.
24563 (call_value_internal): Likewise.
24564 (sibcall_internal): Likewise.
24565 (sibcall_value_internal): Likewise.
24566 (call_reg): Likewise.
24567 (call_symbol): Likewise.
24568 (call_value_reg): Likewise.
24569 (call_value_symbol): Likewise.
24571 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24574 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24576 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24578 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24579 compatible with CCGOCmode and with CCZmode.
24581 2017-05-14 Martin Sebor <msebor@redhat.com>
24583 PR middle-end/77671
24584 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24585 (gimple_fold_builtin_snprintf): Same.
24586 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24587 (gimple_fold_builtin_snprintf): Same.
24588 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24589 of character types.
24590 (is_call_safe): New function.
24591 (try_substitute_return_value): Call it.
24592 (try_simplify_call): New function.
24593 (pass_sprintf_length::handle_gimple_call): Call it.
24595 2017-05-14 Martin Sebor <msebor@redhat.com>
24597 PR middle-end/80669
24598 * builtins.c (expand_builtin_stpncpy): Simplify.
24600 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24602 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24603 * config/i386/i386.h
24604 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24605 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24606 (struct machine_function): Add new members call_ms2sysv,
24607 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24608 (struct machine_frame_state): New fields sp_realigned and
24609 sp_realigned_offset.
24610 * config/i386/i386.c
24611 (enum xlogue_stub): New enum.
24612 (enum xlogue_stub_sets): New enum.
24613 (class xlogue_layout): New class.
24614 (struct ix86_frame): New fields stack_realign_allocate_offset,
24615 stack_realign_offset and outlined_save_offset. Modify comments to
24616 detail stack layout when using out-of-line stubs.
24617 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24618 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24619 -mcall-ms2sysv-xlogues.
24620 (stub_managed_regs): New static variable.
24621 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24622 registers managed by out-of-line stub.
24623 (disable_call_ms2sysv_xlogues): New function.
24624 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24625 m->call_ms2sysv when appropriate and compute frame layout for
24627 (sp_valid_at, fp_valid_at): New inline functions.
24628 (choose_basereg): New function.
24629 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24631 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24632 Use align parameter of choose_baseaddr to generated aligned SSE movs
24634 (pro_epilogue_adjust_stack): Modify to track
24635 machine_frame_state::sp_realigned.
24636 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24637 (ix86_nsaved_sseregs): Likewise.
24638 (ix86_emit_save_regs): Likewise.
24639 (ix86_emit_save_regs_using_mov): Likewise.
24640 (ix86_emit_save_sse_regs_using_mov): Likewise.
24641 (get_scratch_register_on_entry): Likewise.
24642 (gen_frame_set): New function.
24643 (gen_frame_load): Likewise.
24644 (gen_frame_store): Likewise.
24645 (emit_outlined_ms2sysv_save): Likewise.
24646 (emit_outlined_ms2sysv_restore): Likewise.
24647 (ix86_expand_prologue): Modify stack re-alignment code and call
24648 emit_outlined_ms2sysv_save when appropriate.
24649 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24650 parameter rtx_insn *insn, which allows the function to be used to only
24651 generate the notes.
24652 (ix86_expand_epilogue): Modify validity checks of frame and stack
24653 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24654 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24655 * config/i386/predicates.md
24656 (save_multiple): New predicate.
24657 (restore_multiple): Likewise.
24658 * config/i386/sse.md
24659 (save_multiple<mode>): New pattern.
24660 (save_multiple_realign<mode>): Likewise.
24661 (restore_multiple<mode>): Likewise.
24662 (restore_multiple_and_return<mode>): Likewise.
24663 (restore_multiple_leave_return<mode>): Likewise.
24664 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24666 2017-05-14 Julia Koval <julia.koval@intel.com>
24668 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24669 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24670 (__builtin_ia32_xsetbv): New builtins.
24671 * config/i386/i386.c (ix86_expand_special_args_builtin):
24673 (ix86_expand_builtin): Special expand for new intrinsics.
24674 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24675 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24676 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24678 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24680 * cfganal.c (inverted_post_order_compute): Change argument type
24682 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24683 * df-core.c (rest_of_handle_df_initialize): Adjust.
24684 (rest_of_handle_df_finish): Likewise.
24685 (df_analyze_1): Likewise.
24686 (df_analyze): Likewise.
24687 (loop_inverted_post_order_compute): Change argument to be a vec *.
24688 (df_analyze_loop): Adjust.
24689 (df_get_n_blocks): Likewise.
24690 (df_get_postorder): Likewise.
24691 * df.h (struct df_d): Change field to be a vec.
24692 * lcm.c (compute_laterin): Adjust.
24693 (compute_available): Likewise.
24694 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24695 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24696 * tree-ssa-pre.c (compute_antic): Likewise.
24698 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24700 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24701 (depth_first_search::depth_first_search): Change structure init
24702 function to this constructor.
24703 (depth_first_search::add_bb): Rename function to this member.
24704 (depth_first_search::execute): Likewise.
24705 (flow_dfs_compute_reverse_finish): Adjust.
24707 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24709 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24710 (longest_simple_path): Likewise.
24711 * shrink-wrap.c (spread_components): Likewise.
24712 (disqualify_problematic_components): Likewise.
24713 (emit_common_heads_for_components): Likewise.
24714 (emit_common_tails_for_components): Likewise.
24715 (insert_prologue_epilogue_for_components): Likewise.
24717 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24719 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24722 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24724 * df-core.c (df_set_blocks): Start using auto_bitmap.
24725 (df_compact_blocks): Likewise.
24726 * df-problems.c (df_rd_confluence_n): Likewise.
24727 * df-scan.c (df_insn_rescan_all): Likewise.
24728 (df_process_deferred_rescans): Likewise.
24729 (df_update_entry_block_defs): Likewise.
24730 (df_update_exit_block_uses): Likewise.
24731 (df_entry_block_bitmap_verify): Likewise.
24732 (df_exit_block_bitmap_verify): Likewise.
24733 (df_scan_verify): Likewise.
24734 * lra-constraints.c (lra_constraints): Likewise.
24735 (undo_optional_reloads): Likewise.
24736 (lra_undo_inheritance): Likewise.
24737 * lra-remat.c (calculate_gen_cands): Likewise.
24738 (do_remat): Likewise.
24739 * lra-spills.c (assign_spill_hard_regs): Likewise.
24740 (spill_pseudos): Likewise.
24741 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24742 (bitmap_set_subtract_values): Likewise.
24744 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24746 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24747 management with auto_bitmap.
24748 (fix_inter_tick): Likewise.
24749 (fix_recovery_deps): Likewise.
24750 * ira.c (add_store_equivs): Likewise.
24751 (find_moveable_pseudos): Likewise.
24752 (split_live_ranges_for_shrink_wrap): Likewise.
24753 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24754 (rtx_reuse_manager::seen_def_p): Likewise.
24755 (rtx_reuse_manager::set_seen_def): Likewise.
24756 * print-rtl.h (class rtx_reuse_manager): Likewise.
24758 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24760 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24762 (migrate_btr_def): Likewise.
24763 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24764 * df-core.c (loop_post_order_compute): Likewise.
24765 (loop_inverted_post_order_compute): Likewise.
24766 * hsa-common.h: Likewise.
24767 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24768 * init-regs.c (initialize_uninitialized_regs): Likewise.
24769 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24770 (inline_small_functions): Likewise.
24771 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24772 * ira.c (combine_and_move_insns): Likewise.
24773 (build_insn_chain): Likewise.
24774 * loop-invariant.c (find_invariants): Likewise.
24775 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24776 * predict.c (tree_predict_by_opcode): Likewise.
24777 (predict_paths_leading_to): Likewise.
24778 (predict_paths_leading_to_edge): Likewise.
24779 (estimate_loops_at_level): Likewise.
24780 (estimate_loops): Likewise.
24781 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24782 (spread_components): Likewise.
24783 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24784 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24785 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24786 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24787 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24788 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24789 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24790 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24791 (mark_threaded_blocks): Likewise.
24792 (thread_through_all_blocks): Likewise.
24793 * tree-ssa.c (verify_ssa): Likewise.
24794 (execute_update_addresses_taken): Likewise.
24795 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24797 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24799 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24801 (post_order_compute): Likewise.
24802 (inverted_post_order_compute): Likewise.
24803 (pre_and_rev_post_order_compute_fn): Likewise.
24805 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24807 * genrecog.c (int_set::int_set): Explicitly construct our
24808 auto_vec base class.
24809 * vec.h (auto_vec::auto_vec): New constructor.
24811 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24813 * bitmap.h (class auto_bitmap): New constructor taking
24814 bitmap_obstack * argument.
24816 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24818 * bitmap.h (class auto_bitmap): Change type of m_bits to
24819 bitmap_head, and adjust ctor / dtor and member operators.
24821 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24823 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24824 when returned register mode doesn't match original mode.
24826 2017-05-12 Jeff Law <law@redhat.com>
24827 Jakub Jelinek <jakub@redhat.com>
24829 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24830 we look for cc setter after the compare-elim changes.
24831 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24832 within the vector to match what compare-elim now expects.
24833 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24834 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24836 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24837 after the compare-elim changes.
24838 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24839 the vector to match what compare-elim now expects.
24840 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24841 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24842 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24843 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24844 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24846 * config/visium/visium.c (single_set_and_flags): Fix where
24847 we look for cc setter after the compare-elim changes.
24848 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24849 with the vector to match what compare-elim now expects.
24850 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24851 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24852 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24853 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24854 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24855 (neg<mode>2_insn_set_overflow): Likewise.
24857 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24859 PR middle-end/79794
24860 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24861 maybe_expand_insn call, set ops[0].target. If still set after call,
24862 set alt_rtl. Add extra arg to recursive calls.
24863 (extract_bit_field): Add alt_rtl argument. Pass to
24865 * expmed.h (extract_bit_field): Fix prototype.
24866 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24867 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24868 to extract_bit_field_calls.
24869 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24870 Pass alt_rtl to extract_bit_field calls.
24871 * calls.c (store_unaligned_arguments_into_psuedos)
24872 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24873 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24875 * optabs.h (struct expand_operand): Add target bitfield.
24877 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24879 * compare-elim.c (try_eliminate_compare): Canonicalize
24880 operation with embedded compare to
24881 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24882 (set (reg) (operation)].
24884 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24886 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24889 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24890 cost of adding a carry flag for ADC instruction.
24891 [case MINUS]: Ignore the cost of subtracting a carry flag
24892 for SBB instruction.
24894 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24896 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24898 * config/rs6000/bmiintrin.h: New file.
24899 * config/rs6000/bmi2intrin.h: New file.
24900 * config/rs6000/x86intrin.h: New file.
24902 2017-05-12 Jeff Law <law@redhat.com>
24904 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24907 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24909 PR middle-end/80707
24910 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24912 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24914 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24915 early expansion of vector divide builtins.
24916 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24917 builtins identified as having unsigned arguments.
24919 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24921 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24922 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24923 expansion of vector logical operations (and, andc, or, xor,
24926 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24928 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24929 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24931 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24933 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24934 early GIMPLE expansion of vector multiplies.
24936 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24938 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24939 TARGET_HAVE_MOVT conditional.
24940 (movt splitter): Likewise.
24942 2017-05-12 Richard Biener <rguenther@suse.de>
24944 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24945 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24946 Fold all stmts not inplace.
24948 2017-05-12 Richard Biener <rguenther@suse.de>
24950 PR tree-optimization/80713
24951 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24952 inserted_exprs bit for not removed stmts.
24954 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24956 PR middle-end/69921
24957 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24958 parallelized" attribute for parallelized OpenACC kernels.
24959 * omp-offload.c (execute_oacc_device_lower): Use it.
24961 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24962 Set "oacc kernels" attribute.
24963 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24964 parameter. Adjust all users.
24965 (oacc_fn_attrib_kernels_p): Remove function.
24966 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24967 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24968 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24969 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24970 assert "oacc kernels" attribute is set.
24972 2017-05-11 Carl Love <cel@us.ibm.com>
24974 * config/rs6000/rs6000-c: Add support for built-in functions
24975 vector unsigned char vec_popcnt (vector signed char)
24976 vector unsigned char vec_popcnt (vector unsigned char)
24977 vector unsigned short vec_popcnt (vector signed short)
24978 vector unsigned short vec_popcnt (vector unsigned short)
24979 vector unsigned int vec_popcnt (vector signed int)
24980 vector unsigned int vec_popcnt (vector unsigned int)
24981 vector unsigned long long vec_popcnt (vector signed long long)
24982 vector unsigned long long vec_popcnt (vector unsigned long long)
24983 vector signed long long vec_slo (vector signed long long,
24984 vector signed char)
24985 vector signed long long vec_slo (vector signed long long,
24986 vector unsigned char)
24987 vector unsigned long long vec_slo (vector unsigned long long,
24988 vector signed char)
24989 vector unsigned long long vec_slo (vector unsigned long long,
24990 vector unsigned char)
24991 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24992 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24993 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24994 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24995 * doc/extend.texi: Update the built-in documentation file for the
24996 new built-in functions.
24998 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25000 * attribs.h (sorted_attr_string): Move machine independent
25001 functions for target clone support from the i386 port to common
25002 code. Rename ix86_function_versions to common_function_versions.
25003 Rename make_name to make_unique_name.
25004 (common_function_versions): Likewise.
25005 (make_unique_name): Likewise.
25006 (make_dispatcher_decl): Likewise.
25007 (is_function_default_version): Likewise.
25008 * attribs.c (attr_strcmp): Likewise.
25009 (sorted_attr_string): Likewise.
25010 (common_function_versions): Likewise.
25011 (make_unique_name): Likewise.
25012 (make_dispatcher_decl): Likewise.
25013 (is_function_default_version): Likewise.
25014 * config/i386/i386.c (attr_strcmp): Likewise.
25015 (sorted_attr_string): Likewise.
25016 (ix86_function_versions): Likewise.
25017 (make_name): Likewise.
25018 (make_dispatcher_decl): Likewise.
25019 (is_function_default_version): Likewise.
25020 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25022 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25025 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25026 Account for direct move costs for vec_construct of integer
25029 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25032 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25033 (UNSPEC_STX_ATOMIC): Ditto.
25034 (loaddi_via_sse): New insn.
25035 (storedi_via_sse): Ditto.
25036 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25037 Update corresponding peephole2 patterns.
25038 (atomic_storedi_fpu): Ditto.
25040 2017-05-11 Julia Koval <julia.koval@intel.com>
25042 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25043 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25045 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25046 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25047 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25049 2017-05-11 Nathan Sidwell <nathan@acm.org>
25051 * graphite-poly.c: Include dumpfile.h.
25053 * dumpfle.h (dump_function): Declare here ...
25054 * tree-dump.h (dump_function): ... not here.
25055 * dumpfile.c: #include tree-cfg.h.
25056 (dump_function): Move here from ...
25057 * tree-dump.c (dump_function): ... here.
25058 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25059 * graphite-poly.c: Don't include tree-dump.h.
25060 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25061 * print-tree.c: Likewise.
25062 * stor-layout.c: Likewise.
25063 * tree-nested.c: Likewise.
25065 * dumpfile.c (dump_start): Use TDF_FLAGS.
25066 (dump_enable_all): Fix TDF_KIND check thinko.
25068 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25070 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25071 array entries to represent two legal parameterizations of the
25072 overloaded __builtin_cmpb function, as represented by the
25073 P6_OV_BUILTIN_CMPB constant.
25074 (altivec_resolve_overloaded_builtin): Add special case handling
25075 for the __builtin_cmpb function, as represented by the
25076 P6_OV_BUILTIN_CMPB constant.
25077 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25078 (BU_P6_64BIT_2): New macro.
25079 (BU_P6_OVERLOAD_2): New macro
25080 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25081 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25082 (CMPB): Add overload support to represent both 32-bit and 64-bit
25083 compare-bytes function.
25084 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25085 support for TARGET_CMPB.
25086 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25087 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25088 documentation of the __builtin_cmpb overloaded built-in function.
25090 2017-05-11 Richard Biener <rguenther@suse.de>
25092 PR tree-optimization/80705
25093 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25094 bases are not vectorizable.
25096 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25098 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25099 when counting register pressure.
25101 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25103 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25104 (iv_ca_more_deps): Renamed to ...
25105 (iv_ca_compare_deps): ... this.
25106 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25108 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25110 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25112 (determine_group_iv_costs): ... here.
25113 (find_inv_vars_cb): Record inv var if it's not recorded before.
25115 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25117 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25118 (get_shiftadd_cost): Ditto.
25120 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25122 * tree-ssa-address.c: Include header file.
25123 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25125 (add_to_parts): Refactor.
25126 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25127 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25130 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25132 PR tree-optimization/53090
25133 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25135 (extract_cond_operands): Detect condition with IV on both sides
25136 and return COMP_IV_EXPR_2.
25137 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25138 (rewrite_use_compare): Simplify by removing call to function
25139 extract_cond_operands.
25141 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25143 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25144 (extract_cond_operands): Detect condition comparing against non-
25145 invariant bound and return appropriate enum value.
25146 (find_interesting_uses_cond): Update use of extract_cond_operands.
25147 Handle its return value accordingly.
25148 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25150 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25152 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25153 nonlinear iv_use computation in loop invariant sensitive way.
25155 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25157 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25158 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25160 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25162 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25163 (dump_cand): Support iv_cand.inv_exprs.
25164 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25166 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25169 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25171 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25173 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25174 as local function. Include necessary header files.
25175 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25177 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25179 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25181 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25183 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25184 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25185 RSHIFT_EXPR and BIT_NOT_EXPR.
25187 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25189 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25190 (adjust_setup_cost): New parameter supporting round up adjustment.
25191 (struct address_cost_data): Delete.
25192 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25193 (split_address_cost, ptr_difference_cost): Delete.
25194 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25195 (struct ainc_cost_data): New struct.
25196 (get_address_cost_ainc): New function.
25197 (get_address_cost, get_computation_cost): Reimplement.
25198 (determine_group_iv_cost_address): Record inv_expr for all uses of
25200 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25201 (iv_ca_has_deps): Reimplemented to ...
25202 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25204 (iv_ca_extend): Call iv_ca_more_deps.
25206 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25208 * tree-ssa-address.c (struct mem_address): Move to header file.
25209 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25210 * tree-ssa-address.h (struct mem_address): Move from C file.
25211 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25213 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25215 * tree-affine.h (aff_combination_type): New interface.
25216 (aff_combination_zero_p): Remove static.
25217 (aff_combination_const_p): New interface.
25218 (aff_combination_singleton_var_p): New interfaces.
25220 2017-05-11 Richard Biener <rguenther@suse.de>
25222 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25223 Skip unreachable blocks and destinations.
25224 (eliminate): Move stmt removal and fixup ...
25225 (fini_eliminate): ... here. Skip inserted exprs.
25226 (pass_pre::execute): Move fini_pre after fini_eliminate.
25227 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25228 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25229 PRE to get rid of dead code that has invalid SSA form and
25230 split critical edges again.
25232 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25234 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25236 2017-05-11 Richard Biener <rguenther@suse.de>
25238 * passes.c (execute_function_todo): Verify loops if they are
25239 said to be up-to-date.
25240 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25241 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25243 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25246 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25247 handle calling assemble_external ourself.
25250 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25251 modes with zero size. Enhance comment.
25253 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25255 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25256 built-ins for vec_xl and vec_xst with short and char pointer
25259 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25261 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25262 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25263 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25264 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25265 (_mm_maskz_min_round_ss): New intrinsics.
25266 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25267 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25268 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25269 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25270 (__builtin_ia32_minss_mask_round): New builtins.
25271 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25272 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25273 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25275 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25276 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25278 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25281 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25283 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25284 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25285 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25286 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25287 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25288 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25289 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25290 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25291 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25292 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25293 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25294 (__builtin_ia32_mulss_mask_round): New builtins.
25295 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25296 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25297 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25299 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25300 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25302 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25305 2017-05-10 Julia Koval <julia.koval@intel.com>
25307 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25308 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25309 (_mm256_setr_m128i): New intrinsics.
25311 2017-05-10 Julia Koval <julia.koval@intel.com>
25313 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25314 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25315 (_mm_maskz_rcp14_ss): New intrinsics.
25316 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25317 (__builtin_ia32_rcp14ss_mask): New builtins.
25318 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25320 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25322 PR tree-optimization/51513
25323 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25324 (assert_unreachable_fallthru_edge_p): Use it.
25325 (group_case_labels_stmt): Likewise.
25326 * tree-cfg.h: Prototype it.
25327 * stmt.c: Include cfghooks.h and tree-cfg.h.
25328 (emit_case_dispatch_table) <gap_label>: New local variable.
25329 Use it to fill dispatch table gaps.
25330 Test for default_label before updating probabilities.
25331 (expand_case) <default_label>: Remove unneeded initialization.
25332 Test for unreachable default case statement and remove its edge.
25333 Set default_label accordingly.
25334 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25336 2017-05-10 Carl Love <cel@us.ibm.com>
25338 * config/rs6000/rs6000-c: Add support for built-in functions
25339 vector signed char vec_neg (vector signed char)
25340 vector signed short int vec_neg (vector short int)
25341 vector signed int vec_neg (vector signed int)
25342 vector signed long long vec_neg (vector signed long long)
25343 vector float vec_neg (vector float)
25344 vector double vec_neg (vector double)
25345 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25347 * config/rs6000/altivec.h: Add define for vec_neg
25348 * doc/extend.texi: Update the built-in documentation for the
25349 new built-in functions.
25351 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25353 PR tree-optimization/77644
25354 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25356 2017-05-10 Nathan Sidwell <nathan@acm.org>
25358 * dumpfile.h (TDI_lang_all): New.
25359 (TDF_KIND): New. Renumber others
25360 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25362 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25364 (get_dump_file_name): Adjust suffix generation.
25365 (dump_enable_all): Use TDF_KIND.
25366 * doc/invoke.texi (-fdump-lang-all): Document.
25368 * dumpfile.h: Tabify.
25370 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25373 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25374 Move member access before delete.
25376 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25378 * tree-inline.c (expand_call_inline): Split block at stmt
25381 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25384 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25385 are now unused after splitting mov{sf,sd}_hardfloat.
25386 (f32_lr2): Likewise.
25387 (f32_lm): Likewise.
25388 (f32_lm2): Likewise.
25389 (f32_li): Likewise.
25390 (f32_li2): Likewise.
25391 (f32_lv): Likewise.
25392 (f32_sr): Likewise.
25393 (f32_sr2): Likewise.
25394 (f32_sm): Likewise.
25395 (f32_sm2): Likewise.
25396 (f32_si): Likewise.
25397 (f32_si2): Likewise.
25398 (f32_sv): Likewise.
25399 (f32_dm): Likewise.
25400 (f32_vsx): Likewise.
25401 (f32_av): Likewise.
25402 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25403 For movsf, order stores so the VSX stores occur before the GPR
25404 store which encourages the register allocator to use a traditional
25405 FPR instead of a GPR. For movsd, order the stores so that the GPR
25406 store comes before the VSX stores to allow the power6 to work.
25407 This is due to the power6 not having a 32-bit integer store
25408 instruction from a FPR.
25409 (movsf_hardfloat): Likewise.
25410 (movsd_hardfloat): Likewise.
25412 2017-05-09 Martin Sebor <msebor@redhat.com>
25414 PR translation/80280
25415 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25418 PR translation/80280
25419 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25420 data member added in r247778.
25421 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25423 2017-05-09 Nathan Sidwell <nathan@acm.org>
25425 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25427 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25430 2017-05-09 Marek Polacek <polacek@redhat.com>
25432 * doc/invoke.texi: Fix typo.
25434 2017-05-09 Richard Biener <rguenther@suse.de>
25436 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25437 (vrp_val_is_min): Likewise.
25438 (set_value_range_to_value): Likewise.
25439 (set_value_range_to_nonnegative): Likewise.
25440 (gimple_assign_nonzero_p): Likewise.
25441 (gimple_stmt_nonzero_p): Likewise.
25442 (vrp_int_const_binop): Likewise. Remove unreachable case.
25443 (adjust_range_with_scev): Adjust comments.
25444 (compare_range_with_value): Likewise.
25445 (extract_range_from_phi_node): Likewise.
25446 (test_for_singularity): Likewise.
25448 2017-05-09 Richard Biener <rguenther@suse.de>
25450 * tree-vrp.c (get_single_symbol): Add assert that we don't
25451 get overflowed constants as invariant part.
25452 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25453 checks. Use wi::cmp instead of recursing for integer constants.
25454 (compare_values): Just ignore whether we assumed undefined
25455 overflow instead of failing the compare.
25456 (extract_range_for_var_from_comparison_expr): Add comment before the
25457 TREE_NO_WARNING sets.
25458 (test_for_singularity): Likewise.
25459 (extract_range_from_comparison): Do not disable optimization
25460 when we assumed undefined overflow.
25461 (extract_range_basic): Remove init of unused var.
25463 2017-05-09 Richard Biener <rguenther@suse.de>
25465 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25466 (extract_range_from_multiplicative_op_1): Adjust.
25467 (extract_range_from_binary_expr_1): Use int_const_binop.
25469 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25472 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25473 rs6000_store_data_bypass_p in seven define_bypass directives and
25474 in several comments.
25475 * config/rs6000/rs6000-protos.h: Add prototype for
25476 rs6000_store_data_bypass_p function.
25477 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25478 function implements slightly different (rs6000-specific) semantics
25479 than store_data_bypass_p, returning false rather than aborting
25480 with assertion error when arguments do not satisfy the
25481 requirements of store data bypass.
25482 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25483 rs6000_store_data_bypass_p.
25485 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25487 * config/xtensa/xtensa-protos.h
25488 (xtensa_initial_elimination_offset): New declaration.
25489 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25490 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25491 macro definition, add case for FRAME_POINTER_REGNUM when
25492 FRAME_GROWS_DOWNWARD.
25493 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25494 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25495 xtensa_initial_elimination_offset.
25497 2017-05-08 Nathan Sidwell <nathan@acm.org>
25499 * doc/invoke.texi: Alphabetize -fdump options.
25501 2017-05-08 Martin Sebor <msebor@redhat.com>
25503 PR translation/80280
25504 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25506 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25508 * target.def (compute_frame_layout): New optional target hook.
25509 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25510 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25511 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25513 * reload1.c (verify_initial_elim_offsets): Likewise.
25514 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25515 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25516 (arm_get_frame_offsets): Split up into this ...
25517 (arm_compute_frame_layout): ... and this function.
25519 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25521 * config/aarch64/constraints.md (Usa): New constraint.
25522 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25524 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25526 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25527 with_multilib_list after it has been checked.
25529 2017-05-08 Richard Biener <rguenther@suse.de>
25531 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25532 (bitmap_set_subtract_values): Likewise.
25534 2017-05-08 Richard Biener <rguenther@suse.de>
25536 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25537 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25539 (gimple_stmt_nonzero_warnv_p): Rename to ...
25540 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25542 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25543 (extract_range_basic): Adjust, do not disable propagation on
25544 strict overflow sensitive simplification.
25545 (vrp_visit_cond_stmt): Likewise.
25547 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25549 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25550 body size unconditionally.
25552 2017-05-07 Jeff Law <law@redhat.com>
25555 2017-05-06 Jeff Law <law@redhat.com>
25556 PR tree-optimization/78496
25557 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25560 PR tree-optimization/78496
25561 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25562 (simplify_stmt_using_ranges): Call it.
25563 (vrp_dom_walker::before_dom_children): Extract equivalences
25564 from an ASSERT_EXPR with an equality comparison against a
25567 2017-05-06 Jeff Law <law@redhat.com>
25569 PR tree-optimization/78496
25570 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25573 PR tree-optimization/78496
25574 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25575 (simplify_stmt_using_ranges): Call it.
25576 (vrp_dom_walker::before_dom_children): Extract equivalences
25577 from an ASSERT_EXPR with an equality comparison against a
25580 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25582 * lra-constraints.c (lra_copy_reg_equiv): New function.
25583 (split_reg): Use it to copy equivalence information from the
25584 original register to the spill register.
25586 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25588 PR rtl-optimization/75964
25589 * simplify-rtx.c (simplify_const_relational_operation): Remove
25590 invalid handling of comparisons of integer ABS.
25592 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25594 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25595 initialize to zero.
25596 (init_regs): Remove declaration.
25597 (function_arg_advance_32): Initialize error_p as boolean variable.
25599 2017-05-05 Nathan Sidwell <nathan@acm.org>
25601 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25602 lines. Use for (;;).
25604 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25606 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25607 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25608 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25609 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25610 VF=2 that require versioning.
25612 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25614 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25617 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25619 * diagnostic.h (diagnostic_override_option_index): Convert from
25620 macro to inline function.
25622 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25624 * diagnostic.c (last_module_changed_p): New function.
25625 (set_last_module): New function.
25626 (diagnostic_report_current_module): Convert macro usage to
25627 the above functions.
25628 * diagnostic.h (diagnostic_context::last_module): Strengthen
25629 from const line_map * to const line_map_ordinary *.
25630 (diagnostic_last_module_changed): Delete macro.
25631 (diagnostic_set_last_module): Delete macro.
25633 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25635 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25636 with diagnostic_report_diagnostic.
25637 (diagnostic_n_impl_richloc): Likewise.
25638 * diagnostic.h (report_diagnostic): Delete macro.
25639 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25640 with diagnostic_report_diagnostic.
25641 * substring-locations.c (format_warning_va): Likewise.
25643 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25645 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25646 save/restor of format_spec. Move option-printing code to...
25647 (print_option_information): ...this new function, and
25648 reimplement by simply printing to the pretty_printer,
25649 rather than appending to the format string.
25651 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25653 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25654 handling logic into...
25655 (update_effective_level_from_pragmas): ...this new function.
25657 2017-05-04 Andrew Waterman <andrew@sifive.com>
25659 * config/riscv/riscv.opt (mstrict-align): New option.
25660 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25661 (SLOW_UNALIGNED_ACCESS): Define.
25662 (riscv_slow_unaligned_access): Declare.
25663 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25665 (riscv_slow_unaligned_access): New variable.
25666 (rocket_tune_info): Set slow_unaligned_access to true.
25667 (optimize_size_tune_info): Set slow_unaligned_access to false.
25668 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25669 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25670 (riscv_option_override): Set riscv_slow_unaligned_access.
25671 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25673 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25675 * config/riscv/riscv.md: Unify indentation.
25677 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25682 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25684 (extendsi<mode>2): Add support for doing sign extension via
25685 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25686 don't have ISA 3.0 instructions.
25687 (extendsi<mode>2 splitter): Likewise.
25688 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25689 generate the normal insns since SImode can now go in vector
25690 registers. Disallow the special UNSPECs needed for previous
25691 machines to hide SImode being used. Add new insns
25692 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25693 (fix_trunc<mode>si2_stfiwx): Likewise.
25694 (fix_trunc<mode>si2_internal): Likewise.
25695 (fixuns_trunc<mode>si2): Likewise.
25696 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25697 (fctiw<u>z_<mode>_smallint): Likewise.
25698 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25699 of floating point to 32-bit integer from doing a direct move to
25700 the GPR registers to do a store.
25701 (fctiwz_<mode>): Break long line.
25703 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25705 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25706 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25707 (addr_list, addr_offset_valid_p): New.
25708 (split_address_groups): Check offset validity with above function.
25709 (gt-tree-ssa-loop-ivopts.h): Include header file.
25711 2017-05-05 Nathan Sidwell <nathan@acm.org>
25713 * config.gcc (arm*-*-*): Add missing 'fi'.
25715 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25717 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25718 included in -fopt-info does not matter.
25719 * doc/optinfo.texi (-fopt-info): Fix description of default
25720 behavour. Explicitly say order of options included in -fopt-info
25723 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25725 * config.gcc: Allow combinations of aprofile and rmprofile values for
25726 --with-multilib-list.
25727 * config/arm/t-multilib: New file.
25728 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25729 variables. Remove setting of ISA and floating-point ABI in
25730 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25731 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25732 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25733 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25735 * config/arm/t-rmprofile: Likewise except for the matches changes.
25736 * doc/install.texi (--with-multilib-list): Document the combination of
25737 aprofile and rmprofile values and warn about pitfalls in doing that.
25739 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25741 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25742 (movdi_aarch64): Likewise.
25744 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25746 PR tree-optimization/80632
25747 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25749 (build_arrays): Initialize it for virtual phis.
25750 (fix_phi_nodes): Use it for virtual phis.
25752 PR tree-optimization/80558
25753 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25754 [x, y] op z into [x op, y op z] for op & or | if conditions
25757 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25758 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25761 * config/arm/arm.md (use_literal_pool): Remove.
25762 (64-bit immediate split): No longer takes cost into consideration
25763 if arm_disable_literal_pool is enabled.
25764 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25765 used when arm_disable_literal_pool is enabled.
25766 (arm_max_const_double_inline_cost): Remove use of
25767 arm_disable_literal_pool.
25768 (push_minipool_fix): Add assert.
25769 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25770 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25771 (no_literal_pool_sf_immediate): New.
25773 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25775 PR tree-optimization/80613
25776 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25777 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25779 2017-05-05 Richard Biener <rguenther@suse.de>
25781 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25783 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25785 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25786 of this flag from insn conditions due to removal from r247495.
25788 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25790 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25792 (arm_early_store_addr_dep_ptr): Likewise.
25793 * config/arm/aarch-common-protos.h
25794 (arm_early_load_addr_dep_ptr): Add prototype.
25795 (arm_early_store_addr_dep_ptr): Likewise.
25796 * config/arm/cortex-a53.md: Add new bypasses.
25798 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25800 * tree.c (next_type_uid): Change type to unsigned.
25801 (type_hash_canon): Decrement back next_type_uid if
25802 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25803 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25806 2017-05-04 Martin Sebor <msebor@redhat.com>
25808 * builtins.c: Fix a trivial typo in a comment.
25810 PR middle-end/79234
25811 * builtins.c (check_sizes): Adjust to handle reading past the end.
25812 Avoid printing excessive upper bound of ranges. Use %E to print
25813 tree nodes instead of converting them to %wu.
25814 (expand_builtin_memchr): New function.
25815 (compute_dest_size): Rename...
25816 (compute_objsize): ...to this.
25817 (expand_builtin_memcpy): Adjust.
25818 (expand_builtin_mempcpy): Adjust.
25819 (expand_builtin_strcat): Adjust.
25820 (expand_builtin_strcpy): Adjust.
25821 (check_strncat_sizes): Adjust.
25822 (expand_builtin_strncat): Adjust.
25823 (expand_builtin_strncpy): Adjust and simplify.
25824 (expand_builtin_memset): Adjust.
25825 (expand_builtin_bzero): Adjust.
25826 (expand_builtin_memcmp): Adjust.
25827 (expand_builtin): Handle memcmp.
25828 (maybe_emit_chk_warning): Check strncat just once.
25830 2017-05-04 Martin Sebor <msebor@redhat.com>
25832 PR preprocessor/79214
25833 PR middle-end/79222
25834 PR middle-end/79223
25835 * builtins.c (check_sizes): Add inlining context and issue
25836 warnings even when -Wno-system-headers is set.
25837 (check_strncat_sizes): Same.
25838 (expand_builtin_strncat): Same.
25839 (expand_builtin_memmove): New function.
25840 (expand_builtin_stpncpy): Same.
25841 (expand_builtin): Handle memmove and stpncpy.
25843 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25845 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25846 which is not used any more.
25848 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25850 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25852 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25854 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25855 (cortexa53_tunings): Likewise.
25856 (cortexa57_tunings): Likewise.
25857 (cortexa72_tunings): Likewise.
25858 (cortexa73_tunings): Likewise.
25860 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25862 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25863 Set loop alignment to 8.
25865 2017-05-04 Martin Sebor <msebor@redhat.com>
25867 PR translation/80280
25868 * builtins.c (expand_builtin_object_size): Add missing quoting to
25869 %D and like directives.
25870 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25871 (hsa_type_for_tree_type): Same.
25872 (verify_function_arguments): Same.
25873 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25874 * varasm.c (get_section): Same.
25877 2017-05-04 Martin Sebor <msebor@redhat.com>
25879 PR translation/80280
25880 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25882 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25884 * config/aarch64/aarch64.c (generic_addrcost_table):
25885 Change HI/TI mode setting.
25887 2017-05-04 Martin Jambor <mjambor@suse.cz>
25889 PR tree-optimization/80622
25890 * tree-sra.c (comes_initialized_p): New function.
25891 (build_accesses_from_assign): Only set write lazily when
25892 comes_initialized_p is false.
25893 (analyze_access_subtree): Use comes_initialized_p.
25894 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25895 instead of testing for PARM_DECL.
25897 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25899 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25900 constraint on operand 0 to allow more general addressing modes.
25901 Adjust output template.
25902 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25904 * config/aarch64/aarch64-protos.h
25905 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25906 * config/aarch64/constraints.md (Dp): New address constraint.
25907 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25910 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25912 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25913 update use of estimate_ipcp_clone_size_and_time.
25914 (estimate_local_effects): Update use of
25915 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25916 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25917 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25918 Return nonspecialized time.
25920 2017-05-04 Richard Biener <rguenther@suse.de>
25922 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25923 for the last VUSE which def dominates the PHI. Directly call
25925 (get_continuation_for_phi_1): Remove.
25927 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25929 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25930 to explain the use of truncating division. Cap the number of
25931 iterations to the maximum given by nb_iterations_upper_bound,
25934 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25936 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25937 * configure: Regenerate.
25938 * config.in: Regenerate.
25939 * config/i386/driver-mingw32.c: new file.
25940 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25941 * config.host: Link driver-mingw32.o on MinGW host.
25942 * doc/install.texi: Document new --enable-mingw-wildcard configure
25945 2017-05-04 Marek Polacek <polacek@redhat.com>
25947 PR tree-optimization/80612
25948 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25950 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25951 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25953 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25954 (movt splitter): Likewise.
25955 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25956 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25957 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25958 block for Thumb-1 with MOVT.
25959 (thumb2_legitimate_address_p): Move code block ...
25960 (can_avoid_literal_pool_for_label_p): ... into this new function.
25961 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25963 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25964 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25965 "M-profile targets with the MOVT instruction".
25967 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25969 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25970 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25971 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25973 2017-05-04 Martin Liska <mliska@suse.cz>
25975 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25976 variable cond_code.
25978 2017-05-04 Richard Biener <rguenther@suse.de>
25980 * tree.c (array_at_struct_end_p): Handle arrays at struct
25981 end with flexarrays more conservatively. Refactor and treat
25982 arrays of arrays or aggregates more strict. Fix
25983 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25984 * tree.c (array_at_struct_end_p): Adjust prototype.
25985 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25986 * gimple-fold.c (get_range_strlen): Likewise.
25987 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25989 2017-05-04 Richard Biener <rguenther@suse.de>
25991 PR tree-optimization/31130
25992 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25994 (supports_overflow_infinity): Likewise.
25995 (is_negative_overflow_infinity): Likewise.
25996 (is_positive_overflow_infinity): Likewise.
25997 (is_overflow_infinity): Likewise.
25998 (stmt_overflow_infinity): Likewise.
25999 (overflow_infinity_range_p): Likewise.
26000 (usable_range_p): Remove as always returning true.
26001 (make_overflow_infinity): Remove.
26002 (negative_overflow_infinity): Likewise.
26003 (positive_overflow_infinity): Likewise.
26004 (avoid_overflow_infinity): Likewise.
26005 (set_value_range): Adjust accordingly.
26006 (set_value_range_to_nonnegative): Likewise, remove now unused
26007 overflow_infinity arg.
26008 (vrp_operand_equal_p): Adjust.
26009 (update_value_range): Likewise.
26010 (range_int_cst_singleton_p): Likewise.
26011 (operand_less_p): Likewise.
26012 (compare_values_warnv): Likewise.
26013 (extract_range_for_var_from_comparison_expr): Likewise.
26014 (vrp_int_const_binop): Likewise.
26015 (zero_nonzero_bits_from_vr): Likewise.
26016 (extract_range_from_multiplicative_op_1): Likewise.
26017 (extract_range_from_binary_expr_1): Likewise.
26018 (extract_range_from_unary_expr): Likewise.
26019 (extract_range_from_comparison): Likewise.
26020 (extract_range_basic): Likewise.
26021 (adjust_range_with_scev): Likewise.
26022 (compare_ranges): Likewise.
26023 (compare_range_with_value): Likewise.
26024 (dump_value_range): Likewise.
26025 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26027 (simplify_cond_using_ranges): Adjust.
26029 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26031 * brig-builtins.def: Added a builtin for class_f64.
26032 * builtin-types.def: Added a builtin type needed by class_f64.
26034 2017-05-03 Jason Merrill <jason@redhat.com>
26036 * timevar.def: Add TV_CONSTEXPR.
26038 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26040 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26042 2017-05-03 Martin Jambor <mjambor@suse.cz>
26044 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26045 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26046 * ipa-cp.c (ipcp_propagate_stage): Do not call
26047 ipa_update_after_lto_read.
26048 * ipa-inline.c (ipa_inline): Likewise.
26050 2017-05-03 Martin Jambor <mjambor@suse.cz>
26052 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26053 tag. Added a default constructor and a destructor.
26054 (ipa_edge_args_sum_t): New class;
26055 (ipa_edge_args_sum): Declare.
26056 (ipa_edge_args_vector): Remove declaration.
26057 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26058 (ipa_free_edge_args_substructures): Remove declaration.
26059 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26060 (ipa_edge_args_info_available_for_edge_p): Likewise.
26061 * ipa-prop.c (ipa_edge_args_vector): Removed.
26062 (edge_removal_hook_holder): Likewise.
26063 (edge_duplication_hook_holder): Likewise.
26064 (ipa_edge_args_sum): New variable.
26065 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26066 ipa_edge_args_vector.
26067 (ipa_free_edge_args_substructures): Likewise.
26068 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26069 ipa_edge_args_vector.
26070 (ipa_edge_removal_hook): Turned into method
26071 ipa_edge_args_sum_t::remove.
26072 (ipa_edge_duplication_hook): Turned into method
26073 ipa_edge_args_sum_t::duplicate.
26074 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26075 registering edge hooks.
26076 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26077 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26078 ipa_edge_args_sum instead of ipa_edge_args_vector.
26079 * ipa-profile.c (ipa_profile): Likewise.
26081 2017-05-03 Martin Jambor <mjambor@suse.cz>
26083 * symbol-summary.h (function_summary): New method exists.
26084 (function_summary::symtab_removal): Deallocate through release.
26085 (call_summary): New class.
26086 (gt_ggc_mx): New overload.
26087 (gt_pch_nx): Likewise.
26088 (gt_pch_nx): Likewise.
26090 2017-05-03 Jeff Law <law@redhat.com>
26092 PR tree-optimization/78496
26093 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26094 from simplify_cond_using_ranges. Split off code to walk
26095 backwards through casts into ...
26096 (simplify_cond_using_ranges_2): New function.
26097 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26098 (execute_vrp): After identifying jump threads, call
26099 simplify_cond_using_ranges_2.
26101 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26104 * ipa-inline.h (inline_summary): Add ctor.
26105 (create_ggc): Do not use ggc_cleared_alloc.
26107 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26108 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26110 * gcc.c (handle_braces): Support escaping in switch matching
26112 * doc/invoke.texi (Spec Files): Document it.
26113 Remove superfluous @code markup in items.
26115 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26117 * diagnostic-show-locus.c (struct column_range): New struct.
26118 (get_affected_columns): New function.
26119 (get_printed_columns): New function.
26120 (struct correction): New struct.
26121 (correction::ensure_capacity): New function.
26122 (correction::ensure_terminated): New function.
26123 (struct line_corrections): New struct.
26124 (line_corrections::~line_corrections): New dtor.
26125 (line_corrections::add_hint): New function.
26126 (layout::print_trailing_fixits): Reimplement in terms of the new
26128 (selftest::test_overlapped_fixit_printing): New function.
26129 (selftest::diagnostic_show_locus_c_tests): Call it.
26131 2017-05-03 Nathan Sidwell <nathan@acm.org>
26133 Canonicalize canonical type hashing
26134 * tree.h (type_hash_canon_hash): Declare.
26135 * tree.c (type_hash_list, attribute_hash_list): Move into
26136 type_hash_canon_hash.
26137 (build_type_attribute_qual_variant): Break out hash code calc into
26138 type_hash_canon_hash.
26139 (type_hash_canon_hash): New. Generic type hash computation.
26140 (build_range_type_1, build_array_type_1, build_function_type,
26141 build_method_type_directly, build_offset_type, build_complex_type,
26142 make_vector_type): Call it.
26144 2017-05-03 Richard Biener <rguenther@suse.de>
26146 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26147 When all DRs have unknown misaligned do not always peel
26148 when there is a store but apply the same costing model as if
26149 there were only loads.
26151 2017-05-03 Richard Biener <rguenther@suse.de>
26154 PR tree-optimization/80492
26155 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26156 compare_base_decls returning dont-know properly.
26158 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26160 * config/arm/iterators.md (CCSI): New mode iterator.
26161 (arch): New mode attribute.
26162 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26163 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26164 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26165 code iterator for success result mode.
26166 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26167 the corresponding new insn generators.
26169 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26172 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26173 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26175 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26177 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26178 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26179 (DDR_A): Wrap DDR argument in brackets.
26180 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26181 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26182 (DDR_REVERSED_P): Likewise.
26184 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26186 PR tree-optimization/79472
26187 * tree-switch-conversion.c (struct switch_conv_info): Add
26188 contiguous_range and default_case_nonstandard fields.
26189 (collect_switch_conv_info): Compute contiguous_range and
26190 default_case_nonstandard fields, don't clear final_bb if
26191 contiguous_range and only the default case doesn't have the required
26193 (check_all_empty_except_final): Set default_case_nonstandard instead
26194 of failing if contiguous_range and the default case doesn't have empty
26196 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26197 and only the default case doesn't have the required constants. Skip
26199 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26200 if default_case_nonstandard.
26201 (build_constructors): Build constant 1 just once. Assert that default
26202 values aren't inserted in between cases if contiguous_range. Skip
26204 (build_arrays): Skip virtual phis.
26205 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26206 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26207 Handle virtual phis.
26208 (gen_inbound_check): Handle default_case_nonstandard case.
26209 (process_switch): Adjust check_final_bb caller. Call
26210 gather_default_values with the first non-default case instead of
26211 default case if default_case_nonstandard.
26213 2017-05-02 Nathan Sidwell <nathan@acm.org>
26215 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26216 check. Fix formatting.
26218 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26220 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26221 errors when comparing specialized and unspecialized times.
26223 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26225 * diagnostic-show-locus.c
26226 (layout::should_print_annotation_line_p): Make private.
26227 (layout::print_annotation_line): Make private.
26228 (layout::annotation_line_showed_range_p): Make private.
26229 (layout::show_ruler): Make private.
26230 (layout::print_source_line): Make private. Pass in line and
26231 line_width, rather than calling location_get_source_line. Drop
26233 (layout::print_leading_fixits): New method.
26234 (layout::print_any_fixits): Rename to...
26235 (layout::print_trailing_fixits): ...this, and make private.
26236 Don't print newline fixits.
26237 (diagnostic_show_locus): Move logic for printing one row into...
26238 (layout::print_line): ...this new function. Move the
26239 location_get_source_line call and error-handling from
26240 print_source_line to here. Call print_leading_fixits, and rename
26241 print_any_fixits to print_trailing_fixits.
26242 (selftest::test_fixit_insert_containing_newline): Update now that
26243 newlines are partially supported.
26244 (selftest::test_fixit_insert_containing_newline_2): New test.
26245 (selftest::test_fixit_replace_containing_newline): Update comments.
26246 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26247 * edit-context.c (class added_line): New class.
26248 (class edited_line): Describe newline handling in comment.
26249 (edited_line::actually_edited_p): New method.
26250 (edited_line::print_content): Delete redundant decl.
26251 (edited_line::m_predecessors): New field.
26252 (edited_file::print_content): Call edited_line::print_content.
26253 (edited_file::print_diff): Update to support newlines.
26254 (edited_file::print_diff_hunk): Likewise.
26255 (edited_file::print_run_of_changed_lines): New function.
26256 (edited_file::print_diff_line): Convert to...
26257 (print_diff_line): ...this.
26258 (edited_file::get_effective_line_count): New function.
26259 (edited_line::edited_line): Initialize new field m_predecessors.
26260 (edited_line::~edited_line): Clean up m_predecessors.
26261 (edited_line::apply_fixit): Handle newlines.
26262 (edited_line::get_effective_line_count): New function.
26263 (edited_line::print_content): New function.
26264 (edited_line::print_diff_lines): New function.
26265 (selftest::test_applying_fixits_insert_containing_newline): New
26267 (selftest::test_applying_fixits_replace_containing_newline): New
26269 (selftest::insert_line): New function.
26270 (selftest::test_applying_fixits_multiple_lines): Add example of
26272 (selftest::edit_context_c_tests): Call the new tests.
26274 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26276 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26277 parameter cand. Update dump information.
26278 (get_computation_cost): Update uses.
26280 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26282 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26283 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26284 (get_computation_at, rewrite_use_address): Update use of
26285 get_computation_aff.
26287 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26289 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26290 (get_computation): Delete.
26291 (get_computation_cost): Implement like get_computation_cost_at.
26292 Use get_computation_at.
26293 (get_computation_cost_at): Delete.
26294 (rewrite_use_nonlinear_expr): Use get_computation_at.
26295 (rewrite_use_compare, remove_unused_ivs): Ditto.
26297 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26299 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26301 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26303 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26304 (ivopts_global_cost_for_size): Rename parameter and update uses.
26305 (iv_ca_recount_cost): Update uses.
26306 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26307 candidates seperately in n_invs and n_cands.
26308 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26310 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26312 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26313 (find_inv_vars_cb): New.
26314 (find_depends): Renamed to ...
26315 (find_inv_vars): ... this.
26316 (add_candidate_1, force_var_cost): Call find_inv_vars.
26317 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26319 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26321 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26322 inv_vars. Add inv_exprs.
26323 (struct iv_cand): Rename depends_on to inv_vars.
26324 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26325 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26326 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26327 (dump_cand): Dump inv_vars.
26328 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26329 (record_invariant, find_depends, add_candidate_1): Ditto.
26330 (set_group_iv_cost, force_var_cost): Ditto.
26331 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26332 (get_computation_cost_at, get_computation_cost): Ditto.
26333 (determine_group_iv_cost_generic): Ditto.
26334 (determine_group_iv_cost_address): Ditto.
26335 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26336 (determine_group_iv_costs): Ditto.
26337 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26338 (iv_ca_set_remove_invariants): Renamed to ...
26339 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26340 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26341 (iv_ca_set_add_invariants): Renamed to ...
26342 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26343 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26344 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26345 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26346 (create_new_ivs): Remove useless dump.
26348 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26350 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26352 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26353 (iv_ca_set_no_cp, create_new_iv): Ditto.
26355 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26357 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26359 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26361 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26362 function tree_check2.
26364 2017-05-02 Martin Liska <mliska@suse.cz>
26366 * doc/gcov.texi: Add missing preposition.
26367 * gcov.c (function_info::function_info): Properly fill up
26368 all member variables.
26370 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26372 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26374 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26376 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26378 2017-05-02 Martin Liska <mliska@suse.cz>
26381 * lto-streamer-in.c (lto_read_tree_1): Remove
26382 LTO_STREAMER_DEBUG.
26383 * lto-streamer.c (struct tree_hash_entry): Likewise.
26384 (struct tree_entry_hasher): Likewise.
26385 (tree_entry_hasher::hash): Likewise.
26386 (tree_entry_hasher::equal): Likewise.
26387 (lto_streamer_init): Likewise.
26388 (lto_orig_address_map): Likewise.
26389 (lto_orig_address_get): Likewise.
26390 (lto_orig_address_remove): Likewise.
26391 * lto-streamer.h: Likewise.
26392 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26393 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26395 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26397 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26398 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26399 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26400 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26401 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26402 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26403 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26404 (mm_maskz_sub_ss): New intrinsics.
26405 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26406 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26407 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26408 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26409 (__builtin_ia32_subss_mask_round): New builtins.
26410 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26411 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26412 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26414 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26415 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26417 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26420 2017-05-02 Martin Jambor <mjambor@suse.cz>
26422 PR tree-optimization/78687
26423 * tree-sra.c (access): New field parent.
26424 (process_subtree_disqualification): New function.
26425 (disqualify_candidate): Call it.
26426 (build_accesses_from_assign): Reset write flag if creating an
26428 (build_access_subtree): Fill in parent field and also prpagate
26429 down grp_write flag.
26430 (create_artificial_child_access): New parameter set_grp_write, set
26431 grp_write to its value.
26432 (propagate_subaccesses_across_link): Also propagate grp_write flag
26434 (propagate_all_subaccesses): Push the closest parent back to work
26435 queue if add_access_to_work_queue returned true.
26437 2017-05-02 Richard Biener <rguenther@suse.de>
26439 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26440 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26441 -fstrict-overflow documentation.
26442 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26443 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26444 flag_strict_overflow.
26445 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26446 * lto-opts.c (lto_write_options): Do not stream it.
26447 * lto-wrapper.c (merge_and_complain): Do not handle it.
26448 * opts.c (default_options_table): Do not set -fstrict-overflow.
26449 (finish_options): Likewise do not clear it when sanitizing.
26450 * simplify-rtx.c (simplify_const_relational_operation): Do not
26451 test flag_strict_overflow.
26453 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26455 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26456 using enabled attribute.
26457 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26458 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26459 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26460 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26461 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26462 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26463 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26464 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26465 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26466 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26468 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26470 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26472 2017-05-02 Richard Biener <rguenther@suse.de>
26474 PR tree-optimization/80591
26476 2017-04-10 Richard Biener <rguenther@suse.de>
26478 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26481 2017-05-02 Richard Biener <rguenther@suse.de>
26483 PR tree-optimization/80549
26484 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26485 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26486 headers if they do not have a preheader.
26488 2017-05-02 Martin Liska <mliska@suse.cz>
26491 * common.opt: Fix typo.
26492 * doc/invoke.texi: Likewise.
26494 2017-05-01 Jan Beulich <jbeulich@suse.com>
26496 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26497 swapping, add (x,x,m,x,n) alternative.
26499 2017-05-01 Nathan Sidwell <nathan@acm.org>
26501 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26502 unnecessary unadjusted_alignment check.
26504 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26507 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26509 * gimplify.c (gimplify_cilk_detach): New function.
26510 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26511 * tree-core.h: Document EXPR_CILK_SPAWN.
26512 * tree.h (EXPR_CILK_SPAWN): Define.
26514 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26516 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26517 to use new fixit_hint representation, using the "replace" logic.
26518 (get_line_span_for_fixit_hint): Likewise.
26519 (layout::print_any_fixits): Likewise.
26520 (selftest::test_one_liner_many_fixits): Rename to...
26521 (selftest::test_one_liner_many_fixits_1): ...this, and update
26522 comment and expected output to reflect that the multiple fix-it
26523 hints are now consolidated into one insertion.
26524 (selftest::test_one_liner_many_fixits_2): New test.
26525 (selftest::test_diagnostic_show_locus_one_liner): Update for
26527 (selftest::test_fixit_consolidation): Update for fix-it API
26529 * diagnostic.c (print_parseable_fixits): Likewise.
26530 * edit-context.c (edited_line::m_line_events): Convert from
26531 auto_vec <line_event *> to auto_vec <line_event>.
26532 (class line_event): Convert from abstract base class to a concrete
26533 class, taking over the role of replace_event.
26534 (class insert_event): Delete.
26535 (class replace_event): Rename to class line_event. Convert to
26537 (edit_context::add_fixits): Reimplement.
26538 (edit_context::apply_insert): Delete.
26539 (edit_context::apply_replace): Rename to...
26540 (edit_context::apply_fixit): ...this. Convert to half-open range.
26541 (edited_file::apply_insert): Delete.
26542 (edited_file::apply_replace): Rename to...
26543 (edited_file::apply_fixit): ...this.
26544 (edited_line::~edited_line): Drop deletion of events.
26545 (edited_line::apply_insert): Delete.
26546 (edited_line::apply_replace): Rename to...
26547 (edited_line::apply_fixit): ...this. Convert to half-open range.
26548 Update for change to type of m_line_events.
26549 * edit-context.h (edit_context::apply_insert): Delete.
26550 (edit_context::apply_replace): Rename to...
26551 (edit_context::apply_fixit): ...this.
26553 2017-05-01 Martin Sebor <msebor@redhat.com>
26555 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26558 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26561 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26562 __get_cpuid_max returns 0.
26563 (__get_cpuid_count): Ditto.
26565 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26567 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26568 replacement expression is another instance of one of its arguments.
26570 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26573 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26574 check for stack push/pop autoinc.
26575 * config/i386/i386.c (ix86_agi_dependent): Return false
26576 if the only reason why modified_in_p returned true is that
26577 addr is SP based and set_insn is a push or pop.
26579 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26581 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26584 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26587 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26588 (account_size_time): Use two predicates - exec_pred and
26590 (evaluate_conditions_for_known_args): Compute both clause and
26592 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26593 (inline_summary_t::duplicate): Update.
26594 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26596 (compute_inline_parameters): Likewise.
26597 (estimate_edge_size_and_time): Update caluclation of time.
26598 (estimate_node_size_and_time): Compute both time and nonspecialized
26600 (estimate_ipcp_clone_size_and_time): Update.
26601 (inline_merge_summary): Update.
26602 (do_estimate_edge_time): Update.
26603 (do_estimate_edge_size): Update.
26604 (do_estimate_edge_hints): Update.
26605 (inline_read_section, inline_write_summary): Stream both new predicates.
26606 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26608 (compute_inlined_call_time): Cleanup.
26609 (big_speedup_p): Update.
26610 (edge_badness): Update.
26611 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26612 (size_time_entry): Replace predicate by exec_predicate and
26613 nonconst_predicate.
26614 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26615 (estimate_edge_time): Return also nonspec_time.
26616 (reset_edge_growth_cache): Update.
26618 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26620 PR rtl-optimization/80491
26621 * ifcvt.c (noce_process_if_block): When looking for x setter
26622 with missing else_bb, don't check only the insn right before
26623 cond_earliest, but look for the last insn that x is modified in
26624 within the same bb.
26626 PR rtl-optimization/80491
26627 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26629 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26631 PR tree-optimization/80487
26632 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26634 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26636 PR tree-optimization/79697
26637 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26638 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26639 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26641 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26642 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26644 2017-04-28 Martin Sebor <msebor@redhat.com>
26646 PR tree-optimization/80523
26647 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26648 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26650 (maybe_warn, format_directive, parse_directive): Use new functions.
26651 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26653 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26655 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26657 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26659 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26660 target_header_dir): Set correctly.
26661 * configure: Regenerated.
26662 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26663 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26664 instead of SYSTEM_HEADER_DIR.
26666 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26668 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26669 (estimate_local_effects): Likewise.
26670 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26671 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26672 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26673 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26674 do_estimate_edge_time, estimate_edge_time): Likewise.
26675 * ipa-inline-analysis.c (estimate_node_size_and_time,
26676 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26677 (estimate_time_after_inlining): Remove.
26679 2017-04-28 Martin Liska <mliska@suse.cz>
26681 * doc/gcov.texi: Enhance documentation of gcov.
26683 2017-04-28 Martin Liska <mliska@suse.cz>
26685 * doc/gcov.texi: Sort options in alphabetic order.
26686 * doc/gcov-dump.texi: Likewise.
26687 * doc/gcov-tool.texi: Likewise.
26688 * gcov.c (print_usage): Likewise.
26689 * gcov-dump.c (print_usage): Likewise.
26690 * gcov-tool.c (print_merge_usage_message): Likewise.
26691 (print_rewrite_usage_message): Likewise.
26692 (print_overlap_usage_message): Likewise.
26694 2017-04-28 Martin Liska <mliska@suse.cz>
26696 PR gcov-profile/53915
26697 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26699 2017-04-28 Martin Liska <mliska@suse.cz>
26701 PR gcov-profile/79891
26702 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26703 is marked by compiler as living on a line.
26704 (get_cycles_count): Remove usage of the union.
26705 (output_intermediate_file): Likewise.
26706 (find_source): Fix GNU coding style.
26707 (accumulate_line_counts): Remove old non-all block mode.
26708 (output_lines): Remove usage of the union.
26709 * profile.c (output_location): Include all BBs, even if
26710 belonging to a same line (and file) as a previous BB.
26712 2017-04-28 Martin Liska <mliska@suse.cz>
26714 * gcov.c (process_args): Handle new argument 'w'.
26715 (read_graph_file): Assign ID to BBs.
26716 (output_branch_count): Display BB # if verbose flag is set.
26717 (output_lines): Likewise for arcs.
26718 (print_usage): Add '--verbose' option help.
26719 * doc/gcov.texi: Document --verbose (-w) option.
26721 2017-04-28 Martin Liska <mliska@suse.cz>
26723 * gcov.c (struct block_location_info): New struct.
26724 (process_file): Fill up the new structure.
26725 (read_graph_file): Replace usage of encoding by the newly added
26727 (add_line_counts): Likewise.
26728 (accumulate_line_counts): Remove usage of the union.
26729 (function_info::function_info): New function.
26730 (function_info::~function_info): Likewise.
26731 (process_file): Call delete instead of release_function.
26732 (release_function): Release the function.
26733 (release_structures): Call delete instead of release_function.
26734 (solve_flow_graph): Replace usage of num_blocks.
26735 (find_exception_blocks): Likewise.
26736 (output_lines): Fix GNU coding style.
26738 2017-04-28 Martin Liska <mliska@suse.cz>
26741 * coverage.c (coverage_remove_note_file): New function.
26742 * coverage.h: Declare the function.
26743 * toplev.c (finalize): Clean if an error has been seen.
26745 2017-04-28 Martin Liska <mliska@suse.cz>
26747 PR gcov-profile/80031
26748 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26749 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26750 * gcov.c (read_graph_file): Read just number of blocks.
26751 * profile.c (branch_prob): Do not stream 0 flags per a basic
26754 2017-04-28 Martin Liska <mliska@suse.cz>
26756 * gcov-dump.c (tag_*): Add new argument to declarations.
26757 (dump_gcov_file): Likewise.
26758 (tag_blocks): Add and use new argument depth.
26759 (tag_arcs): Likewise.
26760 (tag_lines): Likewise.
26761 (tag_counters): Likewise.
26762 (tag_summary): Likewise.
26763 (dump_working_sets): Use depth to do a proper indentation.
26765 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26768 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26769 * symtab.c (symtab_node::debug_symtab): Move definition here.
26771 2017-04-28 Richard Biener <rguenther@suse.de>
26773 * lto-streamer.h (LTO_major_version): Bump to 7.
26775 2017-04-28 Richard Biener <rguenther@suse.de>
26777 * tree-vrp.c (assert_info): New struct.
26778 (add_assert_info): New helper.
26779 (register_edge_assert_for_2): Refactor to add asserts to a vector
26781 (register_edge_assert_for_1): Likewise.
26782 (register_edge_assert_for): Likewise.
26783 (finish_register_edge_assert_for): New helper actually registering
26784 asserts where live on edge.
26785 (find_conditional_asserts): Adjust.
26786 (find_switch_asserts): Likewise.
26787 (evrp_dom_walker::try_find_new_range): Generalize.
26788 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26790 2017-04-27 Marek Polacek <polacek@redhat.com>
26793 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26794 arg10 and arg11 to itype.
26796 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26798 * doc/extend.texi (Object Size Checking): Improve grammar.
26800 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26803 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26804 that the logic for permitting reciprocal estimates matches that
26807 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26810 * tree.c (type_cache_hasher::equal): Only compare
26811 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26812 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26813 non-aggregate element types.
26814 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26815 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26817 2017-04-27 Richard Biener <rguenther@suse.de>
26819 PR middle-end/80533
26820 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26821 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26822 keeping a reference to a trailing array.
26824 2017-04-27 Richard Biener <rguenther@suse.de>
26826 PR middle-end/80539
26827 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26828 being in loop-closed SSA form conservatively.
26829 (chrec_fold_multiply_poly_poly): Likewise.
26831 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26833 PR middle-end/79665
26834 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26835 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26837 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26840 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26841 (aarch64_function_arg_alignment): Return unsigned int again, but still
26842 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26843 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26844 Don't emit -Wpsabi note.
26845 (aarch64_function_arg_boundary): Likewise.
26846 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26849 2017-04-26 Nathan Sidwell <nathan@acm.org>
26851 * tree.h (crc32_unsigned_n): Declare.
26852 (crc32_unsigned, crc32_unsigned): Make inline.
26853 * tree.c (crc32_unsigned_bits): Replace with ...
26854 (crc32_unsigned_n): ... this.
26855 (crc32_unsigned, crc32_byte): Remove.
26856 (crc32_string): Remove unnecessary braces.
26858 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26860 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26861 * ipa-inline-analysis.c (MAX_TIME): Remove.
26862 (account_size_time): Use sreal for time.
26863 (dump_inline_summary): Update.
26864 (estimate_function_body_sizes): Update.
26865 (estimate_edge_size_and_time): Update.
26866 (estimate_calls_size_and_time): Update.
26867 (estimate_node_size_and_time): Update.
26868 (inline_merge_summary): Update.
26869 (inline_update_overall_summary): Update.
26870 (estimate_time_after_inlining): Update.
26871 (inline_read_section): Update.
26872 (inline_write_summary): Update.
26873 * ipa-inline.c (compute_uninlined_call_time): Update.
26874 (compute_inlined_call_time): Update.
26875 (recursive_inlining): Update.
26876 (inline_small_functions): Update.
26877 (dump_overall_stats): Update.
26878 * ipa-inline.h: Include sreal.h.
26879 (size_time_entry): Turn time to sreal.
26880 (inline_summary): Turn self_time nad time to sreal.
26882 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26884 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26886 (sreal::stream_out, sreal::stream_in): New.
26887 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26889 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26891 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26894 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26897 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26899 (dimode_scalar_chain::compute_convert_gain): Ditto.
26900 (dimode_scalar_chain::make_vector_copies): Ditto.
26901 (dimode_scalar_chain::convert_reg): Ditto.
26902 (dimode_scalar_chain::convert_insn): Ditto.
26903 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26904 (VI248_AVX512BW_1): New mode iterator.
26905 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26906 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26909 2017-04-25 Martin Sebor <msebor@redhat.com>
26911 PR tree-optimization/80497
26912 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26913 constants are representable in HOST_WIDE_INT.
26914 (parse_directive): Ditto.
26916 2017-04-25 Martin Sebor <msebor@redhat.com>
26919 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26920 (new_zero_array): Adjust signature.
26921 (dom_info::dom_init): Used unsigned rather that size_t.
26922 (dom_info::dom_info): Same.
26924 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26925 Jakub Jelinek <jakub@redhat.com>
26928 * config/arm/arm.c: Include gimple.h.
26929 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26930 returns negative, increment ncrn only if it returned positive.
26931 (arm_needs_doubleword_align): Return int instead of bool,
26932 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26933 members, but if there is any such non-FIELD_DECL
26934 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26935 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26936 returns negative, increment nregs only if it returned positive.
26937 (arm_setup_incoming_varargs): Likewise.
26938 (arm_function_arg_boundary): Emit -Wpsabi note if
26939 arm_needs_doubleword_align returns negative, return
26940 DOUBLEWORD_ALIGNMENT only if it returned positive.
26942 2017-04-25 Marek Polacek <polacek@redhat.com>
26945 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26946 first argument to type.
26948 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26951 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26952 type checks to test for compatibility instead of equality.
26954 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26955 Jakub Jelinek <jakub@redhat.com>
26958 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26960 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26961 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26962 the alignment computation, but return their maximum in warn_alignment.
26963 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26964 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26966 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26967 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26970 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26972 * config/arc/simdext.md (dmpyh): Fix typo.
26974 2017-04-25 Richard Biener <rguenther@suse.de>
26976 PR tree-optimization/80492
26977 * alias.c (compare_base_decls): Handle registers with asm
26978 specification conservatively.
26979 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26980 compare_base_decls returning dont-know properly.
26982 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26984 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26985 (legitimate_offset_address_p): New function.
26986 (arc_legitimate_address_p): Use above function.
26988 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26990 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26992 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26994 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26995 ACCH registers whenever they are available.
26997 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26999 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27000 double regs fix when not used.
27002 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27004 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27006 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27007 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27009 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27011 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27012 when emitting short ADD instructions.
27014 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27016 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27018 (cmpsi_cc_c_insn): Likewise.
27019 (cbranchsi4_scratch): Compute proper instruction length using
27020 compact_hreg_operand.
27021 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27023 2017-04-25 Richard Biener <rguenther@suse.de>
27025 PR middle-end/80509
27026 * passes.c (pass_manager::pass_manager): Initialize
27027 m_name_to_pass_map.
27029 2017-04-25 Richard Biener <rguenther@suse.de>
27031 PR tree-optimization/79201
27032 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27034 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27037 * config/s390/vector.md: Split MEM->GPR vector moves for
27038 non-s_operand addresses.
27040 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27043 * config/s390/predicates.md (reload_const_wide_int_operand): New
27045 * config/s390/s390.md ("movti"): Remove d/P alternative.
27046 ("movti_bigconst"): New pattern definition.
27048 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27051 * s390-protos.h (s390_expand_cs_hqi): Removed.
27052 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27053 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27054 modes as well as CCZ1mode and CCZmode.
27055 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27056 signature of s390_emit_compare_and_swap.
27057 (s390_expand_cs_hqi): Likewise, make static.
27058 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27059 compare-and-swap, in some cases.
27060 (s390_expand_cs): Wrapper function.
27061 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27063 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27064 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27065 patterns for small and large integers. Forbid symref memory operands.
27066 Move expander to s390.c. Require cc register.
27067 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27068 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27069 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27070 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27071 symref memory operands. Remove CC mode and call s390_match_ccmode
27073 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27075 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27077 * config/s390/s390.md (define_peephole2): New peephole to help
27078 combining the load-and-test pattern with volatile memory.
27080 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27082 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27083 with CCZmode for TARGET_Z196.
27085 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27087 PR rtl-optimization/80501
27088 * combine.c (make_compound_operation_int): Set subreg_code to SET
27089 even for AND with mask of the sign bit of mode.
27091 PR rtl-optimization/80500
27092 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27093 sum's initial value.
27095 2017-04-25 Julian Brown <julian@codesourcery.com>
27096 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27098 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27100 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27102 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27104 2017-04-25 Julian Brown <julian@codesourcery.com>
27105 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27107 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27108 (thunderx2t99_sha): New Reservation.
27110 2017-04-25 Julian Brown <julian@codesourcery.com>
27111 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27113 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27114 type for 1-element load.
27116 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27118 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27120 2017-04-24 Martin Jambor <mjambor@suse.cz>
27122 PR tree-optimization/80293
27123 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27124 char arrays not totally scalarizable if it is false.
27125 (analyze_all_variable_accesses): Pass correct value in the new
27126 parameter. Add a statistics counter.
27128 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27130 PR middle-end/79931
27131 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27133 2017-04-24 Richard Biener <rguenther@suse.de>
27135 PR tree-optimization/80494
27136 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27137 out for complex types.
27139 2017-04-24 Richard Biener <rguenther@suse.de>
27141 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27142 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27143 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27144 (DFS): Adjust and never fail.
27145 (sccvn_dom_walker::fail): Remove.
27146 (sccvn_dom_walker::before_dom_children): Adjust.
27147 (run_scc_vn): Likewise and never fail.
27148 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27149 (pass_fre::execute): Likewise.
27151 2017-04-24 Richard Biener <rguenther@suse.de>
27153 PR tree-optimization/79725
27154 * tree-ssa-sink.c (statement_sink_location): Return whether
27155 failure reason was zero uses. Move that check later.
27156 (sink_code_in_bb): Deal with zero uses by removing the stmt
27159 2017-04-24 Richard Biener <rguenther@suse.de>
27162 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27163 pointer-based references.
27165 2017-04-24 Richard Biener <rguenther@suse.de>
27168 * pass_manager.h (pass_manager::operator new): Remove.
27169 (pass_manager::operator delete): Likewise.
27170 * passes.c (pass_manager::operator new): Remove.
27171 (pass_manager::operator delete): Likewise.
27172 (pass_manager::pass_manager): Zero individual pass members.
27174 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27177 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27178 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27179 Check "XEXP (src, 1)" operand here.
27180 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27181 Check "XEXP (src, 1)" operand here.
27182 (dimode_scalar_chain::make_vector_copies): Detect count register
27183 of a shift instruction. Zero extend count register from QImode
27184 to DImode to satisfy vector shift pattern count operand predicate.
27185 Substitute vector shift count operand with a DImode copy.
27186 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27189 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27191 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27192 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27193 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27194 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27195 (UNSPEC_NOREX_MEM): Remove definition.
27197 2017-04-21 Richard Biener <rguenther@suse.de>
27199 PR tree-optimization/79547
27200 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27201 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27202 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27203 without any constraints.
27205 2017-04-21 Richard Biener <rguenther@suse.de>
27207 PR tree-optimization/78847
27208 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27210 2017-04-21 Richard Biener <rguenther@suse.de>
27212 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27213 (build_distinct_type_copy): Likewise.
27214 (build_variant_type_copy): Likewise.
27215 * tree.c (build_qualified_type): Pass down mem-stat info.
27216 (build_distinct_type_copy): Likewise.
27217 (build_variant_type_copy): Likewise.
27219 2017-04-21 Richard Biener <rguenther@suse.de>
27221 PR tree-optimization/80237
27222 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27224 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27225 for a simplified result.
27227 2016-04-21 Richard Biener <rguenther@suse.de>
27229 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27230 sth as strict as a simple_iv but a chrec without symbols and an
27231 operand defined in the loop we are peeling (and not some subloop).
27232 (propagate_constants_for_unrolling): Propagate all constants.
27234 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27237 * config/i386/i386.c (print_reg): Remove assert for disalowed
27238 regno values, call output_operand_lossage instead.
27240 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27243 * config/i386/constraints.md (Yc): New register constraint.
27244 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27245 Use Yc constraint for alternative 2 of operand 0. Remove
27246 preferred_for_speed attribute.
27248 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27250 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27251 lastprivate clauses in SIMT case.
27253 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27255 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27257 2017-04-20 Richard Biener <rguenther@suse.de>
27259 PR tree-optimization/57796
27260 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27261 as N scalar stores.
27262 (vect_model_load_cost): Cost gathers as N scalar loads.
27264 2017-04-20 Richard Biener <rguenther@suse.de>
27266 * ggc-page.c (ggc_allocated_p): Rename to ...
27267 (safe_lookup_page_table_entry): ... this and return the lookup
27269 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27271 2017-04-20 Richard Biener <rguenther@suse.de>
27273 PR tree-optimization/80453
27274 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27275 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27276 from the conditions.
27277 (vn_phi_eq): Pass them down.
27278 (vn_phi_lookup): Record them.
27279 (vn_phi_insert): Likewise.
27281 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27283 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27284 uninitialized variable warning to avoid buffer overrun.
27286 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27289 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27290 is suppressed for '{ 0 }' in C.
27292 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27294 * BASE-VER: Set to 8.0.0.
27296 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27298 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27299 priority .init_array and .fini_array section with SECTION_NOTYPE
27302 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27304 PR middle-end/80423
27305 * tree.h (build_array_type): Add typeless_storage default argument.
27306 * tree.c (type_cache_hasher::equal): Also compare
27307 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27308 (build_array_type): Add typeless_storage argument, set
27309 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27311 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27312 (build_array_type): Likewise. Add typeless_storage argument.
27314 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27315 Jakub Jelinek <jakub@redhat.com>
27317 PR tree-optimization/80426
27318 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27319 operation on symbolic operands, also compute the overflow for the
27320 invariant part when the operation degenerates into a negation.
27322 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27325 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27326 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27329 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27331 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27334 * config/avr/avr.c (tree.h): Include it.
27335 (cgraph.h): Include it.
27336 (avr_encode_section_info): Don't warn for uninitialized progmem
27337 variable if it's just an alias.
27339 2017-04-19 Richard Biener <rguenther@suse.de>
27342 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27343 when needed by AutoPGO.
27345 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27348 * doc/lto.texi: Remove an extra 'that'.
27350 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27352 PR rtl-optimization/80429
27353 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27354 are only used in debug insns.
27356 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27357 Vladimir Makarov <vmakarov@redhat.com>
27359 * config/sparc/predicates.md (input_operand): Add comment. Return
27360 true for any memory operand when LRA is in progress.
27361 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27363 2017-04-18 Jeff Law <law@redhat.com>
27366 * mips.md ({return,simple_return}_internal): Do not overwrite
27369 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27371 PR tree-optimization/80443
27372 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27373 instead of adding 1, subtract -1 and similarly instead of subtracting
27376 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27378 PR rtl-optimization/80357
27379 * haifa-sched.c (tmp_bitmap): New variable.
27380 (model_recompute): Handle duplicate use records.
27381 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27382 (free_global_sched_pressure_data): Free it.
27384 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27387 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27388 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27389 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27390 instead of SYSTEM_HEADER_DIR.
27392 2017-04-18 Jeff Law <law@redhat.com>
27394 PR middle-end/80422
27395 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27396 predecessors after walking up the insn chain.
27398 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27401 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27402 sizetype type into debug info.
27404 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27407 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27408 unneeded test for TARGET_UPPER_REGS_SF.
27409 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27411 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27414 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27415 instead of gsi_after_labels.
27417 2017-04-18 Jeff Law <law@redhat.com>
27419 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27423 2017-04-13 Jeff Law <law@redhat.com>
27424 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27425 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27427 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27430 * config/avr/avr.c (intl.h): Include it.
27431 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27433 2017-04-18 Martin Liska <mliska@suse.cz>
27435 PR gcov-profile/78783
27436 * gcov-tool.c (gcov_output_files): Validate that destination
27437 file is either removed by the tool or by a user.
27439 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27440 Guy Benyei <guybe@mellanox.com>
27442 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27443 block, and do not negate it, the stored id is already negative.
27445 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27447 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27449 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27452 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27453 masks of options that should be turned off if the VSX vector
27454 options are turned off.
27455 (OTHER_P8_VECTOR_MASKS): Likewise.
27456 (OTHER_VSX_VECTOR_MASKS): Likewise.
27457 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27458 rs6000_disable_incompatible_switches to validate no type switches
27460 (rs6000_incompatible_switch): New function to disallow turning on
27461 other vector options if -mno-vsx, -mno-power8-vector, or
27462 -mno-power9-vector are specified.
27464 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27466 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27468 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27470 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27471 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27472 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27473 (ARG_POINTER_CFA_OFFSET): Likewise.
27475 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27477 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27478 conditions to take advantage of various optimizations.
27480 2017-04-13 Jeff Law <law@redhat.com>
27482 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27483 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27484 (zero_extendsidi2_dext): Likewise.
27486 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27489 * fold-const.c (fold_ternary_loc): Revert
27490 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27493 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27495 PR rtl-optimization/80343
27496 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27497 new scratch pseudo.
27499 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27502 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27503 to ubsan_encode_value.
27505 2017-04-13 Jeff Law <law@redhat.com>
27507 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27508 appearing in DEBUG_INSNs.
27510 2017-04-13 Martin Liska <mliska@suse.cz>
27512 PR gcov-profile/80413
27513 * gcov-io.c (gcov_write_string): Copy to buffer just when
27514 allocated size is greater than zero.
27516 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27519 * dwarf2out.c (decls_for_scope): Ignore declarations of
27520 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27522 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27525 * ipa-visibility.c (non_local_p): Fix typos.
27526 (localize_node): When localizing symbol in same comdat group,
27527 dissolve the group only when we know external symbols are going
27529 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27531 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27533 PR tree-optimization/79390
27534 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27535 order does not result in usable sequence, retry with reversed operand
27541 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27542 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27543 op0 instead of fold_convert_loc (loc, type, arg0).
27545 2017-04-12 Jeff Law <law@redhat.com>
27547 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27548 has a delay slot in the generated code.
27550 * config/cris/cris.md (cris_preferred_reload_class): Return
27551 GENNONACR_REGS rather than GENERAL_REGS.
27553 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27556 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27557 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27558 signedness of the result type.
27560 2017-04-12 Richard Biener <rguenther@suse.de>
27561 Jeff Law <law@redhat.com>
27563 PR tree-optimization/80359
27564 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27565 trim stores to TARGET_MEM_REFs.
27567 2017-04-12 Richard Biener <rguenther@suse.de>
27569 PR tree-optimization/79390
27570 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27571 threading case even more.
27573 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27576 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27577 for quad_address_p for TImode, instead of just not indexed_address.
27579 2017-04-12 Richard Biener <rguenther@suse.de>
27580 Bernd Edlinger <bernd.edlinger@hotmail.de>
27582 PR middle-end/79671
27583 * alias.c (component_uses_parent_alias_set_from): Handle
27584 TYPE_TYPELESS_STORAGE.
27585 (get_alias_set): Likewise.
27586 * tree-core.h (tree_type_common): Add typeless_storage flag.
27587 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27588 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27589 for types containing members with TYPE_TYPELESS_STORAGE.
27590 (place_field): Likewise.
27591 (layout_type): Likewise for ARRAY_TYPE.
27592 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27593 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27594 TYPE_TYPELESS_STORAGE.
27595 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27597 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27600 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27601 first argument to type.
27603 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27607 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27608 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27609 (rs6000_expand_binop_builtin): Likewise.
27610 (rs6000_expand_ternop_builtin): Likewise; also add missing
27611 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27612 vshasigma built-ins.
27613 * doc/extend.texi: Document that vec_xxpermdi's third argument
27614 must be a constant.
27616 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27618 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27619 Use shift_const cost parameter when calculating gain of STV shifts.
27621 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27623 PR rtl-optimization/70478
27624 * lra-constraints.c (process_alt_operands): Check memory for
27625 disfavoring memory insn operand.
27627 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27629 PR middle-end/80100
27630 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27631 left shift in unsigned HOST_WIDE_INT type.
27633 PR rtl-optimization/80385
27634 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27635 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27638 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27639 if they have any depend clauses.
27641 2017-04-11 Martin Liska <mliska@suse.cz>
27644 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27645 * ipa-split.c (split_function): Create a local comdat symbol
27646 if caller is in a comdat group.
27648 2017-04-11 Martin Liska <mliska@suse.cz>
27651 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27654 2017-04-11 Martin Sebor <msebor@redhat.com>
27656 PR middle-end/80364
27657 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27658 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27660 (directive::set_width, directive::set_precision, format_character):
27662 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27665 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27668 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27669 conflict, set target->arch_name instead of target->cpu_name.
27671 2017-04-11 Richard Biener <rguenther@suse.de>
27673 PR tree-optimization/80374
27674 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27675 build_zero_cst, remove fold_convertible_p check again.
27677 2017-04-11 Martin Liska <mliska@suse.cz>
27680 * ubsan.c (instrument_object_size): Do not instrument register
27683 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27686 * config/i386/i386-builtin-types.def
27687 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27688 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27689 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27690 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27691 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27692 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27693 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27694 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27695 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27696 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27697 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27698 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27699 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27700 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27701 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27702 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27703 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27704 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27705 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27706 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27707 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27708 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27709 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27710 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27711 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27712 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27713 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27714 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27715 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27716 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27717 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27718 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27719 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27720 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27721 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27722 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27723 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27724 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27725 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27726 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27727 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27728 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27729 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27730 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27731 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27732 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27734 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27735 flag to second_arg_count, handle 4 argument function type _COUNT
27736 aliases, handle second_arg_count on second argument rather than last.
27738 2017-04-10 Jeff Law <law@redhat.com>
27740 PR tree-optimization/80374
27741 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27742 record anything if we can not convert integer_zero_node to the
27745 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27748 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27749 Enhance special handling given to the TARGET_P9_MINMAX option in
27750 relation to certain other options.
27752 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27754 PR tree-optimization/80153
27755 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27756 remove POINTER_PLUS_EXPR's base part directly, rather than through
27759 2017-04-10 Richard Biener <rguenther@suse.de>
27760 Bin Cheng <bin.cheng@arm.com>
27762 PR tree-optimization/80153
27763 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27764 the first element of pointer type aff_tree. Build result expr in
27766 (add_elt_to_tree): Convert to type unconditionally. Remove other
27767 fold_convert calls.
27768 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27769 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27771 2017-04-10 Richard Biener <rguenther@suse.de>
27773 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27776 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27778 PR rtl-optimization/70478
27779 * lra-constraints.c (curr_small_class_check): New.
27780 (update_and_check_small_class_inputs): New.
27781 (process_alt_operands): Update curr_small_class_check. Disfavor
27782 alternative insn memory operands. Check available regs for small
27785 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27788 * config/mips/mips.opt (-mvirt): Update description.
27789 * doc/invoke.texi (-mvirt): Likewise.
27791 2017-04-10 Richard Biener <rguenther@suse.de>
27793 PR middle-end/80362
27794 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27795 looking for NEGATE_EXPR in -A / -B to A / B folding.
27797 2017-04-10 Martin Liska <mliska@suse.cz>
27799 PR gcov-profile/80224
27800 * gcov.c (print_usage): Fix usage string.
27801 (get_gcov_intermediate_filename): Remove.
27802 (output_gcov_file): Use both for normal and intermediate format.
27803 (generate_results): Do not initialize special file for
27804 intermediate format.
27806 2017-04-10 Richard Biener <rguenther@suse.de>
27808 PR tree-optimization/80304
27809 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27812 2017-04-10 Nathan Sidwell <nathan@acm.org>
27815 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27816 (rs6000_init_builtins): Use it.
27818 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27820 * config/arm/arm.md (<mrc>): Add mode to SET source.
27821 (<mrrc>): Likewise.
27823 2017-04-10 Richard Biener <rguenther@suse.de>
27825 PR middle-end/80344
27826 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27828 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27831 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27832 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27833 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27834 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27835 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27836 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27837 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27838 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27839 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27840 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27841 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27842 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27843 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27844 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27845 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27846 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27847 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27848 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27849 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27850 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27851 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27852 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27853 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27855 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27857 PR rtl-optimization/70478
27858 * lra-constraints.c: Reverse the last patch.
27860 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27862 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27863 Add comment for WCHAR_T.
27865 2017-04-08 Martin Liska <mliska@suse.cz>
27868 2017-04-07 Martin Liska <mliska@suse.cz>
27871 * ipa-split.c (split_function): Add function part to a same comdat
27874 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27877 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27879 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27881 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27882 (vec_pairing): ...to this.
27883 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27884 (rs6000_sched_init): Adjust for name change.
27885 (struct rs6000_sched_context): Likewise.
27886 (rs6000_init_sched_context): Likewise.
27887 (rs6000_set_sched_context): Likewise.
27889 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27895 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27897 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27898 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27899 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27901 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27903 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27905 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27907 PR rtl-optimization/70703
27908 * ira-color.c (update_conflict_hard_regno_costs): Use
27909 int64_t instead of HOST_WIDE_INT.
27911 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27913 PR rtl-optimization/70478
27914 * lra-constraints.c (process_alt_operands): Disfavor alternative
27915 insn memory operands.
27917 2017-04-07 Jeff Law <law@redhat.com>
27919 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27920 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27922 2017-04-07 Martin Liska <mliska@suse.cz>
27925 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27926 Show error message instead of an ICE.
27928 2017-04-07 Martin Liska <mliska@suse.cz>
27931 * ipa-split.c (split_function): Add function part to a same comdat
27934 2017-04-07 Richard Biener <rguenther@suse.de>
27936 PR middle-end/80341
27937 * tree.c (get_unwidened): Also handle ! for_type case for
27939 * convert.c (do_narrow): Split out from ...
27940 (convert_to_integer_1): ... here. Do not pass final truncation
27941 type to get_unwidened for TRUNC_DIV_EXPR.
27943 2017-04-07 Richard Biener <rguenther@suse.de>
27945 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27947 (aff_combination_const): Adjust.
27948 (aff_combination_scale): Likewise.
27949 (aff_combination_add_elt): Likewise.
27950 (aff_combination_add_cst): Likewise.
27951 (aff_combination_convert): Likewise.
27952 (add_elt_to_tree): Likewise. Remove unused argument.
27953 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27955 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27957 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27959 * config/arm/arm.c (arm_default_short_enums): Use
27960 ARM_DEFAULT_SHORT_ENUMS.
27961 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27963 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27966 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27967 members with redundant out-of-class redeclaration.
27969 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27972 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27973 * config/i386/i386.md (*zero_extendsidi2):
27974 Add (?*x,*x) and (?*v,*v) alternatives.
27976 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27979 * config/i386/i386.c (ix86_expand_builtin)
27980 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27981 mode from insn data. Convert operands to insn operand mode.
27982 Copy operands that don't satisfy insn predicate to a register.
27984 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27986 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27989 2017-04-06 Richard Biener <rguenther@suse.de>
27991 PR tree-optimization/80334
27992 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27993 preserve alignment of accesses.
27995 2017-04-06 Richard Biener <rguenther@suse.de>
27997 PR tree-optimization/80262
27998 * tree-sra.c (build_ref_for_offset): Preserve address-space
28000 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28001 Drop useless address-space information on MEM_REF offsets.
28003 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28005 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28007 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28009 PR rtl-optimization/70703
28010 * ira-color.c (update_conflict_hard_regno_costs): Use
28011 HOST_WIDE_INT instead of long.
28013 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28016 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28017 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28019 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28020 for x86_64 target. Handle -m3dnowa option.
28022 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28024 PR rtl-optimization/70703
28025 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28026 (update_conflict_hard_regno_costs): Use long instead of unsigned
28027 arithmetic for cost calculation.
28029 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28030 Bernd Edlinger <bernd.edlinger@hotmail.de>
28033 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28036 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28039 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28040 ptr_mode with Pmode throughout.
28041 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28042 into probe_stack_range and use DImode.
28044 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28047 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28048 call_eh_return is true.
28050 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28052 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28053 Initialize last_match_fntype_index.
28055 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28058 * tree-nvr.c: Include internal-fn.h.
28059 (pass_return_slot::execute): Ignore internal calls without
28062 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28063 Richard Biener <rguenther@suse.de>
28066 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28067 captures used multiple times, except for the last use.
28068 * generic-match-head.c: Include gimplify.h.
28070 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28072 PR tree-optimization/79390
28073 * target.h (struct noce_if_info): Declare.
28074 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28075 * target.def (noce_conversion_profitable_p): New target hook.
28076 * ifcvt.h (struct noce_if_info): New type, moved from ...
28077 * ifcvt.c (struct noce_if_info): ... here.
28078 (noce_conversion_profitable_p): Renamed to ...
28079 (default_noce_conversion_profitable_p): ... this. No longer
28081 (noce_try_store_flag_constants, noce_try_addcc,
28082 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28083 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28084 instead of noce_conversion_profitable_p.
28085 * config/i386/i386.c: Include ifcvt.h.
28086 (ix86_option_override_internal): Don't override
28087 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28088 (ix86_noce_conversion_profitable_p): New function.
28089 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28090 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28091 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28092 * doc/tm.texi: Regenerated.
28094 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28096 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28099 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28102 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28103 instructions for small multiply cores.
28105 2017-04-04 Jeff Law <law@redhat.com>
28107 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28109 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28110 that are not set by the loop over the elements.
28112 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28115 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28116 int mode, convert_modes it to mode as unsigned, otherwise use
28117 lowpart_subreg to mode rather than SImode.
28118 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28119 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28120 Use DImode instead of SImode for the shift count operand.
28121 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28124 2017-04-04 Richard Biener <rguenther@suse.de>
28126 PR middle-end/80281
28127 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28128 arithmetic done for the negate or the plus. Simplify.
28129 (A - (-B) -> A + B): Likewise.
28130 * fold-const.c (split_tree): Make sure to not negate pointers.
28132 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28134 PR rtl-optimization/60818
28135 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28136 a compare of comparisons with the thing compared if this results
28137 in a different machine mode.
28139 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28141 * alias.c (base_alias_check): Fix typo in comment.
28142 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28143 * cgraphunit.c (symbol_table::compile): Likewise.
28144 * collect2.c (maybe_run_lto_and_relink): Likewise.
28145 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28146 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28147 * config/avr/avr.c (avr_map_op_t): Likewise.
28148 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28149 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28150 * config/epiphany/epiphany.md (movcc): Likewise.
28151 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28152 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28154 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28155 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28156 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28157 * config/sh/sh.c (sh_rtx_costs): Likewise.
28158 * fold-const.c (fold_truth_andor): Likewise.
28159 * genautomata.c (collapse_flag): Likewise.
28160 * gengtype.h (struct type::u::s): Likewise.
28161 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28162 * input.c (FORMAT_AMOUNT): Likewise.
28163 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28164 (known_aggs_to_agg_replacement_list): Likewise.
28165 * ipa-inline-analysis.c: Likewise.
28166 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28167 * ipa-polymorphic-call.c
28168 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28169 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28170 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28172 * modulo-sched.c (apply_reg_moves): Likewise.
28173 * omp-expand.c (build_omp_regions_1): Likewise.
28174 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28175 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28176 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28177 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28178 * value-prof.c: Likewise.
28179 * var-tracking.c (val_reset): Likewise.
28181 2017-04-03 Richard Biener <rguenther@suse.de>
28183 PR tree-optimization/80275
28184 * fold-const.c (split_address_to_core_and_offset): Handle
28187 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28189 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28190 descriptors is at least equal to that of functions.
28192 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28194 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28196 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28199 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28200 (mov<IMOD4:mode>): New expander.
28201 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28203 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28205 PR rtl-optimization/79405
28206 * fwprop.c (propagations_left): New variable.
28207 (forward_propagate_into): Decrement it.
28208 (fwprop_init): Initialize it.
28209 (fw_prop): If the variable has reached zero, stop propagating.
28210 (fwprop_addr): Ditto.
28212 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28215 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28216 a FUNCTION_DECL, pass it as decl instead of origin to
28219 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28221 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28224 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28227 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28228 TARGET_VSX_SMALL_INTEGER.
28230 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28232 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28233 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28235 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28237 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28238 extraction from odd-numbered MSA register.
28240 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28242 PR middle-end/80173
28243 * expmed.c (store_bit_field_1): Don't attempt to create
28244 a word subreg out of hard registers wider than word if they
28245 have HARD_REGNO_NREGS of 1 for their mode.
28247 PR middle-end/80163
28248 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28249 conversions to integer types wider than word and pointer.
28252 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28253 (rtx_equal_for_cselib_p): Pass 0 to it.
28254 * cselib.c (cselib_hasher::equal): Likewise.
28255 (rtx_equal_for_cselib_1): Add depth argument. If depth
28256 is 128, don't look up VALUE locs and punt. Increment
28257 depth in recursive calls when walking VALUE locs.
28259 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28261 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28262 (make_gcov_file_name): Use the canonical path name for generating
28264 (read_line): Fix handling of files with ascii null bytes.
28266 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28268 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28269 to initialise a vector register instead
28270 of using a const_int.
28272 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28274 PR translation/80189
28275 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28276 diagnostic messages.
28278 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28281 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28282 (dfp_diex_<mode>): Update mode of operand 1.
28283 * doc/extend.texi (dxex, dxexq): Document change to return type.
28284 (diex, diexq): Document change to argument type.
28286 2017-03-30 Martin Jambor <mjambor@suse.cz>
28289 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28290 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28291 it reflects the signature changes performed at the callee side.
28292 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28293 to cgraph_build_function_type_skip_args.
28294 (build_function_decl_skip_args): Adjust call to the above function.
28296 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28299 * config/i386/sse.md
28300 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28301 register as dest whenever it is a MEM not rtx_equal_p to the
28302 corresponding dup operand, and when forcing into reg move the
28303 reg into the memory afterwards.
28304 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28305 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28306 for the force_reg mode.
28307 (avx512vl_vextractf128<mode>): Use register as dest either
28308 always when a MEM, or when it is a MEM not rtx_equal_p to the
28309 corresponding dup operand, or even not when it is a CONST_VECTOR
28310 depending on the mode and lo vs. hi.
28311 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28313 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28314 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28315 Likewise. Require that operands[2] is even.
28316 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28317 Remove extraneous parens. Require that operands[2] is a multiple
28319 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28320 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28321 disallow memory then.
28323 2017-03-30 Richard Biener <rguenther@suse.de>
28325 PR tree-optimization/77498
28326 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28327 to non-constants over backedges.
28329 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28331 PR rtl-optimization/80233
28332 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28333 as last_combined_insn. Do not test for BARRIER_P separately.
28335 2017-03-29 Andreas Schwab <schwab@suse.de>
28338 * calls.c (prepare_call_address): Convert funexp to Pmode before
28339 copying to temp reg.
28341 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28343 PR tree-optimization/80158
28344 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28345 Handle possible future case of more than one alternate
28347 (replace_rhs_if_not_dup): Likewise.
28348 (replace_one_candidate): Likewise.
28350 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28352 PR rtl-optimization/80193
28353 * ira.c (ira): Do not check allocation for LRA.
28355 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28357 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28358 (nvptx_output_simt_exit): Declare.
28359 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28360 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28361 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28362 (nvptx_declare_function_name): ...here. Emit declaration of local
28363 memory space buffer for omp_simt_enter insn.
28364 (nvptx_output_unisimt_switch): New.
28365 (nvptx_output_softstack_switch): New.
28366 (nvptx_output_simt_enter): New.
28367 (nvptx_output_simt_exit): New.
28368 * config/nvptx/nvptx.h (struct machine_function): New fields
28369 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28370 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28371 (UNSPECV_SIMT_EXIT): Ditto.
28372 (omp_simt_enter_insn): New insn.
28373 (omp_simt_enter): New expansion.
28374 (omp_simt_exit): New insn.
28375 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28377 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28378 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28379 (expand_GOMP_SIMT_EXIT): New.
28380 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28381 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28382 (GOMP_SIMT_EXIT): Ditto.
28383 * target-insns.def (omp_simt_enter): New insn.
28384 (omp_simt_exit): Ditto.
28385 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28387 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28388 (lower_rec_input_clauses): Likewise.
28389 (lower_lastprivate_clauses): Handle SIMT privatization.
28391 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28392 (ompdevlow_adjust_simt_enter): New.
28393 (find_simtpriv_var_op): New.
28394 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28395 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28397 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28398 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28399 (copy_decl_for_dup_finish): Ditto.
28401 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28403 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28406 * config/i386/i386.c (ix86_option_override_internal): Always
28407 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28409 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28411 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28413 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28415 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28416 mark new edge's irreducible flag accordign to it.
28417 (vect_do_peeling): Check loop preheader edge's irreducible flag
28418 and pass it to function slpeel_add_loop_guard.
28420 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28422 PR tree-optimization/80218
28423 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28424 Update block frequencies and counts.
28426 2017-03-28 Richard Biener <rguenther@suse.de>
28428 PR tree-optimization/78644
28429 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28430 of a simplification result we may not use it at all.
28432 2017-03-28 Richard Biener <rguenther@suse.de>
28435 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28436 without arguments, generate default definition of a SSA name.
28438 2017-03-28 Richard Biener <rguenther@suse.de>
28440 PR middle-end/80222
28441 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28442 TYPE_REF_CAN_ALIAS_ALL references.
28443 * fold-const.c (fold_indirect_ref_1): Likewise.
28445 2017-03-28 Martin Liska <mliska@suse.cz>
28448 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28449 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28451 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28452 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28454 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28455 (EXTRA_SPECS): Define.
28456 (SUBTARGET_EXTRA_SPECS): Likewise.
28457 (SUBTARGET_CPP_SPEC): Likewise.
28458 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28459 SUBTARGET_EXTRA_SPECS.
28460 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28462 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28464 * config/arc/simdext.md (vst64_insn): Update pattern.
28465 (vld32wh_insn): Likewise.
28466 (vld32wl_insn): Likewise.
28467 (vld64_insn): Likewise.
28468 (vld32_insn): Likewise.
28470 2017-03-28 Marek Polacek <polacek@redhat.com>
28473 * fold-const.c (fold_comparison): Use protected_set_expr_location
28474 instead of SET_EXPR_LOCATION.
28476 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28478 * tree.c (add_expr): Avoid name lookup warning.
28480 2017-03-27 Jeff Law <law@redhat.com>
28482 PR tree-optimization/80216
28483 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28484 function name. Limit recursion depth.
28485 (record_temporary_equivalences): Corresponding changes.
28487 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28489 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28492 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28495 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28497 * cfgcleanup.c (reg_note_cfa_p): New array.
28498 (insns_have_identical_cfa_notes): New function.
28499 (old_insns_match_p): Don't cross-jump in between /f
28500 and non-/f instructions. If both i1 and i2 are frame related,
28501 verify all CFA notes, their order and content.
28503 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28506 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28507 HImode and SImode with zero extend to DImode to one insn.
28508 (bswap<mode>2_extenddi): Likewise.
28509 (bswapsi2_extenddi): Likewise.
28510 (bswaphi2_extendsi): Likewise.
28511 (bswaphi2): Combine bswap HImode and SImode into one insn.
28512 Separate memory insns from swapping register.
28513 (bswapsi2): Likewise.
28514 (bswap<mode>2): Likewise.
28515 (bswaphi2_internal): Delete, no longer used.
28516 (bswapsi2_internal): Likewise.
28517 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28518 store, and gpr<-gpr swap insns.
28519 (bswap<mode>2_store): Likewise.
28520 (bswaphi2_reg): Register only splitter, combine with the splitter.
28521 (bswaphi2 splitter): Likewise.
28522 (bswapsi2_reg): Likewise.
28523 (bswapsi2 splitter): Likewise.
28524 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28525 the insns into load, store, and register/register insns.
28526 (bswapdi2_ldbrx): Likewise.
28527 (bswapdi2_load): Likewise.
28528 (bswapdi2_store): Likewise.
28529 (bswapdi2_reg): Likewise.
28531 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28533 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28534 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28536 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28539 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28541 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28542 special handling for target option conflicts between dform
28543 options (-mpower9-dform, -mpower9-dform-vector,
28544 -mpower9-dform-scalar) and -mno-direct-move.
28546 2017-03-27 Richard Biener <rguenther@suse.de>
28548 PR tree-optimization/80181
28549 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28551 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28553 * config/arc/predicates.md (move_double_src_operand): Replace the
28554 call to move_double_src_operand with a call to address_operand.
28556 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28558 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28559 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28560 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28562 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28564 * config/arc/predicates.md (long_immediate_loadstore_operand):
28565 Consider scaled addresses cases.
28567 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28569 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28570 restored when in interrupt.
28571 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28572 doesn't have delay slot.
28574 2017-03-27 Richard Biener <rguenther@suse.de>
28577 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28578 inlined thunk clones.
28580 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28583 * asan.c (instrument_derefs): Copy over last operand from
28584 original COMPONENT_REF to the new COMPONENT_REF with
28585 DECL_BIT_FIELD_REPRESENTATIVE.
28586 * ubsan.c (instrument_object_size): Likewise.
28588 2017-03-27 Richard Biener <rguenther@suse.de>
28590 PR tree-optimization/80170
28591 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28592 sure DR/SCEV didnt fold in constants we do not see when looking
28593 at the reference base alignment.
28595 2017-03-27 Richard Biener <rguenther@suse.de>
28597 PR middle-end/80171
28598 * gimple-fold.c (fold_ctor_reference): Properly guard against
28599 NULL return value from canonicalize_constructor_val.
28601 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28604 * config/i386/i386.c (ix86_expand_builtin)
28605 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28606 flags reg setting and flags reg using instructions.
28607 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28608 clobbering instructions to zero extend op2.
28610 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28612 * doc/install.texi (Configuration) <--with-aix-soname>:
28613 Update link to AIX ld.
28615 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28617 PR rtl-optimization/80160
28618 PR rtl-optimization/80159
28619 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28620 reg_alternate_class into account.
28622 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28625 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28626 to consider in curr_insn_transform.
28628 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28630 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28631 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28632 and emit_mode_inner.
28634 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28636 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28637 argument to the overloaded builtin variants. Use the new flag to
28638 deprecate certain builtin variants.
28639 * config/s390/s390-builtin-types.def: Add new builtin types.
28640 * config/s390/s390-builtins.h: Support new flags field for
28641 overloaded builtins.
28642 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28643 (s390_macro_to_expand): Enable vector float data type.
28644 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28645 builtins by incrementing the __VEC__ version number.
28646 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28648 (s390_resolve_overloaded_builtin): Emit error messages depending
28649 on the builtin flags.
28650 * config/s390/s390.c (s390_expand_builtin): Support additional
28651 flags argument. Change error message to match the messages
28652 emitted in s390-c.c.
28653 * config/s390/s390.md: New UNSPEC_* constants.
28654 (op_type): Add new instruction types.
28655 * config/s390/vecintrin.h: Add new builtins and test data class
28657 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28658 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28659 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28660 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28661 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28662 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28664 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28665 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28666 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28667 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28669 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28670 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28671 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28672 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28674 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28675 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28676 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28677 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28678 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28679 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28680 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28682 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28683 ("vec_scatter_element<V_HW_4:mode>_DI")
28684 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28685 ("vec_fpint<mode>", "vflls")
28686 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28687 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28688 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28689 ("*vec_cmphe<mode>_cc"): ... these.
28691 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28692 mode constant instead of magic value.
28694 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28696 * config/s390/s390.c (s390_expand_vec_compare): Support other
28697 vector floating point modes than just V2DF.
28698 (s390_expand_vcond): Likewise.
28699 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28700 (s390_cannot_change_mode_class): Prevent mode changes between TF
28701 and V1TF in vector registers.
28702 * config/s390/s390.md (DF, SF): New mode attributes.
28703 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28704 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28705 SFmode support for VRs.
28706 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28708 (VFT, VF_HW): New mode iterators.
28709 (vw, sdx): New mode attributes.
28710 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28711 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28712 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28713 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28714 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28715 also the new vector floating point modes. Renaming to ...
28717 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28718 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28719 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28720 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28721 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28722 ("vec_unordered<mode>"): ... these.
28724 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28725 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28726 ("*vec_extendv2df"): New insn definitions.
28728 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28730 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28731 ("mulditi3_2", "*muldi3_sign"): New patterns.
28732 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28733 rename the pattern definition.
28735 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28737 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28739 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28741 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28743 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28744 instruction if possible.
28745 * config/s390/vector.md (vec_halfnumelts): New mode
28747 ("*vec_vllezlf<mode>"): New pattern.
28749 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28751 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28752 ("popcountv4si2", "popcountv2di2"): Rename to ...
28753 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28754 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28756 ("popcount<mode>2_vxe"): New pattern.
28758 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28760 * common/config/s390/s390-common.c (processor_flags_table): Add
28762 * config.gcc: Add arch12.
28763 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28764 Default to arch12 for unknown CPU model numbers.
28765 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28766 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28767 PROCESSOR_max sanity check.
28768 * config/s390/s390-opts.h (enum processor_type): Add
28770 * config/s390/s390.c (processor_table): Add arch12.
28771 (s390_expand_builtin): Add check for B_VXE flag.
28772 (s390_issue_rate): Add PROCESSOR_ARCH12.
28773 (s390_get_sched_attrmask): Likewise.
28774 (s390_get_unit_mask): Likewise.
28775 (s390_sched_score): Enable z13 scheduling for arch12.
28776 (s390_sched_reorder): Likewise.
28777 (s390_sched_variable_issue): Likewise.
28778 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28780 (s390_tune_attr): Use z13 scheduling also for arch12.
28781 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28782 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28783 (TARGET_VXE_P): New macros.
28784 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28785 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28786 * config/s390/s390.opt: Add arch12 as processor_type.
28788 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28790 * config/s390/s390.md
28791 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28792 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28793 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28795 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28796 Rename expanders to ...
28798 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28799 ("fixuns_truncdddi2_emu"): ... these.
28801 ("fixuns_trunc<mode>si2_emu"): New expander.
28803 ("*fixuns_truncdfdi2_z13"): Rename to ...
28804 ("*fixuns_truncdfdi2_vx"): ... this.
28806 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28808 * config/s390/2964.md: Remove the single element vector compare
28809 instructions which are no longer used.
28810 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28812 (s390_canonicalize_comparison): Remove handling of DFmode
28814 (s390_expand_vec_compare_scalar): Remove function.
28815 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28816 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28818 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28820 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28822 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28824 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28825 will anyway by matched by mov<mode>_64dfp.
28827 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28829 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28830 vlef/vstef. Add missing operand to vleif.
28832 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28834 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28835 pair for all vector types with 64 bit elements.
28836 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28837 * config/s390/vector.md (V_HW_64): ... here.
28838 (V_128_NOSINGLE): New mode iterator.
28839 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28840 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28841 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28842 ("*vec_load_pairv2di"): Change to ...
28843 ("*vec_load_pair<mode>"): ... this one.
28845 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28847 * config/s390/constraints.md: Add comments.
28848 (jKK): Reject element sizes > 8 bytes.
28849 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28851 * config/s390/s390.md: Add the s_operand checks formerly in
28852 s390_split_ok_p to various splitters where they are still
28854 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28855 for 128 bit vectors. Plus two splitters.
28857 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28859 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28862 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28865 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28866 error if the boundary argument is not constant.
28868 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28870 PR rtl-optimization/80112
28871 * loop-doloop.c (doloop_condition_get): Don't check condition
28872 if cmp isn't SET with IF_THEN_ELSE src.
28874 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28876 PR tree-optimization/80158
28877 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28878 replacing a candidate statement, also replace it for the
28879 candidate's alternate interpretation.
28880 (replace_rhs_if_not_dup): Likewise.
28881 (replace_one_candidate): Likewise.
28883 2017-03-24 Richard Biener <rguenther@suse.de>
28885 PR tree-optimization/80167
28886 * graphite-isl-ast-to-gimple.c
28887 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28889 (translate_isl_ast_to_gimple::get_rename): Likewise.
28891 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28893 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28894 handling of certain combinations of target options, including the
28895 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28896 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28898 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28901 * config/arm/arm.md (*load_multiple): Add reload_completed to
28902 matching condition.
28904 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28905 Richard Biener <rguenth@suse.de>
28907 PR tree-optimization/79908
28908 PR tree-optimization/80136
28909 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28910 been cast away, gimplify_and_add suffices.
28912 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28914 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28916 2017-03-23 Richard Biener <rguenther@suse.de>
28918 PR tree-optimization/80032
28919 * gimplify.c (gimple_push_cleanup): Forced unconditional
28920 cleanups still have to go to the conditional_cleanups
28923 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28925 PR tree-optimization/80072
28926 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28928 (next_operand_entry_id): Change type to unsigned int.
28929 (sort_by_operand_rank): Make sure to return the right return value
28930 even if unsigned fields are bigger than INT_MAX.
28931 (struct oecount): Change cnt and id type to unsigned int.
28932 (oecount_hasher::equal): Formatting fix.
28933 (oecount_cmp): Make sure to return the right return value
28934 even if unsigned fields are bigger than INT_MAX.
28935 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28938 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28939 TREE_READONLY on result if writing it more than once.
28942 * doc/invoke.texi (-fsanitize=thread): Document that with
28943 -fnon-call-exceptions atomics are not able to throw
28947 * tsan.c: Include tree-eh.h.
28948 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28949 maybe_clean_or_replace_eh_stmt where needed.
28950 (instrument_memory_accesses): Add cfg_changed argument.
28951 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28952 if it returned true.
28953 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28955 PR rtl-optimization/63191
28956 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28957 wrapper function, moved the whole old content into ...
28958 (ix86_delegitimize_address_1): ... this. New inline function.
28959 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28960 true as last argument instead of ix86_delegitimize_address.
28962 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28964 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28965 cortexa57_branch_cost.
28967 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28969 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28971 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28974 * doc/md.texi (Constraints): Document wA constraint.
28975 * config/rs6000/constraints.md (wA): New.
28976 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28977 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28978 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28979 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28981 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28984 * gimplify.c (is_oacc_declared): New function.
28985 (oacc_default_clause): Use it to set default flags for acc declared
28986 variables inside parallel regions.
28987 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28988 declared variables.
28989 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28990 declare attribute to any decl as necessary.
28992 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28995 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28996 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28997 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28998 (arm_arch_lpae): This.
28999 * config/arm/arm.c (arm_arch7ve): Rename into ...
29000 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29001 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29004 2017-03-22 Martin Liska <mliska@suse.cz>
29007 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29008 error message instead of an ICE.
29010 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29012 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29014 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29016 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29018 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29021 2017-03-21 Martin Sebor <msebor@redhat.com>
29023 * doc/extend.texi: Use "cannot" instead of "can't."
29024 * doc/hostconfig.texi: Same.
29025 * doc/install.texi: Same.
29026 * doc/invoke.texi: Same.
29027 * doc/loop.texi: Same.
29028 * doc/md.texi: Same.
29029 * doc/objc.texi: Same.
29030 * doc/rtl.texi: Same.
29031 * doc/tm.texi: Same.
29032 * doc/tm.texi.in: Same.
29033 * doc/trouble.texi: Same.
29035 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29038 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29039 (collect_checksum_attributes): Set it.
29040 (die_checksum_ordered): Use it.
29042 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29044 PR tree-optimization/79908
29045 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29046 change: For a VA_ARG whose LHS has been cast away, use
29047 force_gimple_operand to construct the side effects.
29049 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29051 PR translation/80001
29052 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29053 more amenable to translation.
29054 (oacc_loop_auto_partitions): Likewise.
29056 2017-03-21 Marek Polacek <polacek@redhat.com>
29057 Martin Sebor <msebor@redhat.com>
29059 PR tree-optimization/80109
29060 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29061 on INTEGRAL_TYPE_P.
29063 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29064 Segher Boessenkool <segher@kernel.crashing.org>
29067 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29068 check reg_used_between_p between insn and one of succ or succ2
29069 depending on if succ is artificial insn not inserted into insn
29072 2017-03-21 Martin Liska <mliska@suse.cz>
29074 PR gcov-profile/80081
29075 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29076 * doc/gcc.texi: Include gcov-dump stuff.
29077 * doc/gcov-dump.texi: New file.
29079 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29081 PR rtl-optimization/79150
29082 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29083 conditional jump, if the jump is the last insn of the loop.
29085 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29086 Richard Biener <rguenth@suse.de>
29088 PR tree-optimization/79908
29089 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29090 been cast away, use force_gimple_operand to construct the side
29093 2017-03-21 Martin Liska <mliska@suse.cz>
29095 PR libfortran/79956
29096 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29099 2017-03-21 Brad Spengler <spender@grsecurity.net>
29102 * plugin.c (htab_hash_plugin): New function.
29103 (add_new_plugin): Use it and adjust.
29104 (parse_plugin_arg_opt): Adjust.
29105 (init_one_plugin): Likewise.
29107 2017-03-21 Richard Biener <rguenther@suse.de>
29109 PR tree-optimization/80032
29110 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29111 if set force the cleanup to happen unconditionally.
29112 (gimplify_target_expr): Push inserted clobbers with force_uncond
29113 to avoid them being removed by control-dependent DCE.
29115 2017-03-21 Richard Biener <rguenther@suse.de>
29117 PR tree-optimization/80122
29118 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29119 va_arg_pack_len when the inlined call stmt requires pack
29121 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29123 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29126 * tsan.c (instrument_builtin_call): If the memory model argument
29127 is not a constant, assume it is valid.
29130 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29133 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29135 PR rtl-optimization/79910
29136 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29137 if its dest is used by an insn before I2 (other than the combined
29138 insns themselves, which are properly handled already).
29140 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29143 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29145 * combine.c (record_used_regs): New static function.
29146 (try_combine): Handle situations where there is an additional
29147 instruction between I2 and I3 which needs to have a LOG_LINK
29151 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29153 * combine.c (try_combine): Delete redundant i1 test. Call
29154 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29156 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29159 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29160 alternatives 13/14.
29162 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29164 PR tree-optimization/80054
29165 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29166 the optimization if a PHI or any of its arguments is not dominated
29167 by the candidate's basis. Use gphi* rather than gimple* as
29169 (replace_profitable_candidates): Clean up a gimple* variable that
29170 should be a gphi* variable.
29172 2017-03-20 Martin Sebor <msebor@redhat.com>
29175 * doc/extend.texi (attribute constructor): Document present limitation.
29177 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29180 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29181 __POWER9_VECTOR__ #ifdef control, change template definition to
29182 use Power9-specific built-in function.
29183 (vec_any_eq): Likewise.
29184 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29185 to control outcomes from this test.
29186 (vector_ae_<mode>p): For VEC_F modes, likewise.
29188 2017-03-20 Ian Lance Taylor <iant@google.com>
29190 * config/i386/i386.c (ix86_function_regparm): Save an extra
29191 register for -fsplit-stack with DECL_STATIC_CHAIN.
29193 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29196 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29197 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29199 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29201 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29203 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29206 2017-03-20 Marek Polacek <polacek@redhat.com>
29209 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29211 2017-03-20 Richard Biener <rguenther@suse.de>
29213 PR tree-optimization/80113
29214 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29215 allocate extra SSA name for PHI def.
29216 (add_close_phis_to_outer_loops): Likewise.
29217 (add_close_phis_to_merge_points): Likewise.
29218 (copy_loop_close_phi_args): Likewise.
29219 (copy_cond_phi_nodes): Likewise.
29221 2017-03-20 Martin Liska <mliska@suse.cz>
29223 PR middle-end/79753
29224 * tree-chkp.c (chkp_build_returned_bound): Do not build
29225 returned bounds for a LHS that's not a BOUNDED_P type.
29227 2017-03-20 Martin Liska <mliska@suse.cz>
29231 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29232 COMPLEX_CST and VECTOR_CST.
29234 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29237 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29238 target operand. A new splitter adds the clobber statement in case
29239 the target operand is dead anyway.
29241 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29243 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29244 to age-old versions of binutils and glibc.
29246 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29248 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29250 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29252 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29254 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29256 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29257 requirement for binutils 2.13.
29259 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29261 * combine.c (try_combine): Delete redundant i1 test. Call
29262 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29264 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29266 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29267 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29269 <riscv64-*-elf>: Re-arrange section
29270 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29271 <riscv32-*-linux>: Likewise.
29272 <riscv64-*-elf>: Likewise
29273 <riscv64-*-linux>: Likewise.
29275 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29278 * aarch64.opt(verbose-cost-dump): Fix typo.
29280 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29283 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29284 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29286 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29288 * reload.c (find_reloads): When reloading a nonoffsettable address,
29289 use RELOAD_OTHER for it and its address reloads.
29291 PR rtl-optimization/79910
29292 * combine.c (record_used_regs): New static function.
29293 (try_combine): Handle situations where there is an additional
29294 instruction between I2 and I3 which needs to have a LOG_LINK
29297 2017-03-17 Jeff Law <law@redhat.com>
29299 PR tree-optimization/71437
29300 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29301 conditional in the hash table first.
29302 (vrp_dom_walker::before_dom_children): Extract condition from
29303 ASSERT_EXPR. Record condition, its inverion and any implied
29304 conditions as well.
29306 2017-03-17 Marek Polacek <polacek@redhat.com>
29307 Markus Trippelsdorf <markus@trippelsdorf.de>
29309 PR tree-optimization/80079
29310 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29313 2017-03-17 Richard Biener <rguenther@suse.de>
29315 PR middle-end/80075
29316 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29317 Properly verify the LHS before the RHS possibly claims to be
29319 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29322 2017-03-17 Martin Jambor <mjambor@suse.cz>
29324 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29325 (List of -O2 options): Likewise.
29326 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29329 2017-03-17 Tom de Vries <tom@codesourcery.com>
29331 * gcov-dump.c (print_usage): Print bug_report_url.
29333 2017-03-17 Richard Biener <rguenther@suse.de>
29335 PR middle-end/80050
29336 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29337 (parser::peek): Likewise.
29339 2017-03-17 Richard Biener <rguenther@suse.de>
29341 PR tree-optimization/80048
29342 * sese.c (free_sese_info): Properly release rename_map and
29343 copied_bb_map elements.
29345 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29347 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29348 Add linked-list forward and backlinks. Insert on
29349 construction, remove on destruction.
29350 (class pass_store_merging): Add m_stores_head field.
29351 (pass_store_merging::terminate_and_process_all_chains):
29352 Iterate over m_stores_head list.
29353 (pass_store_merging::terminate_all_aliasing_chains):
29355 (pass_store_merging::execute): Check for debug stmts first.
29356 Push new chains onto the m_stores_head stack.
29358 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29361 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29362 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29363 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29365 2017-03-16 Jeff Law <law@redhat.com>
29367 PR tree-optimization/71437
29368 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29369 member function. Implementation moved into after_dom_children
29370 member function and into the threader's thread_outgoing_edges
29372 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29373 some code into new thread_outgoing_edges.
29374 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29375 definition. Simplify marker handling (do it here). Assume we always
29376 have the available expression and the const/copies tables.
29377 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29379 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29380 * tree-vrp.c (equiv_stack): No longer file scoped.
29381 (vrp_dom_walker): New class.
29382 (vrp_dom_walker::before_dom_children): New member function.
29383 (vrp_dom_walker::after_dom_children): Likewise.
29384 (identify_jump_threads): Setup domwalker. Use it rather than
29385 walking edges in a random order by hand. Simplify setup/finalization.
29386 (finalize_jump_threads): Remove.
29387 (vrp_finalize): Do not call identify_jump_threads here.
29388 (execute_vrp): Do it here instead and call thread_through_all_blocks
29391 PR tree-optimization/71437
29392 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29394 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29396 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29397 (dom_opt_dom_walker::thread_across_edge): Remove
29398 handle_dominating_asserts argument. All callers changed.
29399 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29400 changes. Remove calls to lhs_of_dominating_assert. Other
29401 uses of handle_dominating_asserts turn into unconditional code
29402 (simplify_control_stmt_condition_1): Likewise.
29403 (simplify_control_stmt_condition): Likewise.
29404 (thread_through_normal_block, thread_across_edge): Likewise.
29405 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29406 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29407 object if it is not an SSA_NAME.
29408 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29409 before calling into the VRP specific simplifiers.
29410 (identify_jump_threads): Remove handle_dominating_asserts
29413 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29416 * tree-diagnostic.c (default_tree_printer): No longer static.
29417 * tree-diagnostic.h (default_tree_printer): New prototype.
29419 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29421 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29422 Change ins into fmov.
29424 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29426 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29427 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29428 Use h_con constraint for operand 1.
29429 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29430 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29432 2017-03-15 Jeff Law <law@redhat.com>
29434 PR tree-optimization/71437
29435 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29436 (record_temporary_equivalences): Use it.
29438 PR tree-optimization/71437
29439 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29440 tree-ssa-scopedtables.
29441 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29442 (record_conditions, record_cond, vuse_eq): Likewise.
29443 (record_edge_info): Adjust to API tweak of record_conditions.
29444 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29445 (record_temporary_equivalences, optimize_stmt): Likewise.
29446 (eliminate_redundant_computations): Likewise.
29447 (record_equivalences_from_stmt): Likewise.
29448 * tree-ssa-scopedtables.c: Include options.h and params.h.
29449 (vuse_eq): New function, moved from tree-ssa-dom.c
29450 (build_and_record_new_cond): Likewise.
29451 (record_conditions): Likewise. Accept vector of conditions rather
29452 than edge_equivalence structure for first argument.
29453 for the first argument.
29454 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29455 from tree-ssa-dom.c.
29456 (avail_exprs_stack::record_cond): Likewise.
29457 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29458 from tree-ssa-dom.c.
29459 (avail_exprs_stack): Add new member functions lookup_avail_expr
29461 (record_conditions): Declare.
29463 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29466 * lra-constraints.c (process_alt_operands): Increase reject for
29467 reloading an input/output operand.
29469 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29472 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29473 insns to convert from signed/unsigned char/short to IEEE 128-bit
29475 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29477 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29480 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29481 subreg of inner mode for values already in registers.
29483 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29485 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29486 iteration reg is used after the loop.
29488 2017-03-14 Martin Sebor <msebor@redhat.com>
29490 PR tree-optimization/79800
29491 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29492 precision in negative-positive range.
29493 (format_floating): Call non-const overload with adjusted precision.
29495 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29498 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29501 2017-03-14 Martin Sebor <msebor@redhat.com>
29503 PR middle-end/80020
29504 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29505 * builtins.def (aligned_alloc): Use it.
29508 * Makefile.in (GTFILES): Add calls.c.
29509 * calls.c: Include "gt-calls.h".
29511 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29513 PR rtl-optimization/79728
29514 * regs.h (struct target_regs): New field
29515 x_contains_allocatable_regs_of_mode.
29516 (contains_allocatable_regs_of_mode): New macro.
29517 * reginfo.c (init_reg_sets_1): Initialize it, and change
29518 contains_reg_of_mode so it includes global regs as well.
29519 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29520 rather than contains_regs_of_mode.
29522 2017-03-14 Martin Liska <mliska@suse.cz>
29524 * doc/invoke.texi: Document options that can't be combined with
29525 -fcheck-pointer-bounds.
29527 2017-03-14 Martin Liska <mliska@suse.cz>
29529 PR middle-end/79831
29530 * doc/invoke.texi (-Wchkp): Document the option.
29532 2017-03-14 Martin Liska <mliska@suse.cz>
29534 * Makefile.in: Install gcov-dump.
29536 2017-03-14 Martin Liska <mliska@suse.cz>
29538 * multiple_target.c (expand_target_clones): Bail out for
29539 an invalid attribute.
29541 2017-03-14 Richard Biener <rguenther@suse.de>
29543 * alias.c (struct alias_set_entry): Pack properly.
29544 * cfgloop.h (struct loop): Likewise.
29545 * cse.c (struct set): Likewise.
29546 * ipa-utils.c (struct searchc_env): Likewise.
29547 * loop-invariant.c (struct invariant): Likewise.
29548 * lra-remat.c (struct cand): Likewise.
29549 * recog.c (struct change_t): Likewise.
29550 * rtl.h (struct address_info): Likewise.
29551 * symbol-summary.h (function_summary): Likewise.
29552 * tree-loop-distribution.c (struct partition): Likewise.
29553 * tree-object-size.c (struct object_size_info): Likewise.
29554 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29555 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29556 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29557 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29558 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29559 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29560 (struct _stmt_vec_info): Likewise.
29562 2017-03-14 Martin Liska <mliska@suse.cz>
29565 * multiple_target.c (create_dispatcher_calls): Check that
29566 a target can create a function dispatcher.
29568 2017-03-14 Martin Liska <mliska@suse.cz>
29571 * multiple_target.c (expand_target_clones): Drop local.local
29572 flag for default implementation.
29574 2017-03-14 Richard Biener <rguenther@suse.de>
29576 PR tree-optimization/80030
29577 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29579 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29581 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29582 gcc_fallthrough() instead of __attribute__((fallthrough));
29584 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29586 * doc/gcc.texi: Remove "up" link to (DIR).
29587 * doc/gccint.texi: Ditto.
29589 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29591 * doc/install.texi (Specific) <avr>: Remove reference to
29594 2017-03-13 Jeff Law <law@redhat.com>
29596 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29597 attribute rather than comments.
29599 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29600 match_scratch operand is highest.
29602 2017-03-13 Martin Liska <mliska@suse.cz>
29604 PR middle-end/78339
29605 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29606 is a CHKP clone, use original declaration.
29608 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29610 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29611 (arc_conditional_register_usage): Use a different allocation order
29612 when optimizing for size.
29613 * common/config/arc/arc-common.c (arc_option_optimization_table):
29614 Section anchors default on when optimizing for size.
29616 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29618 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29620 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29622 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29623 * config/arc/arc.md (cpu_facility): Add cd variant.
29624 (*movqi_insn): Add code density variant.
29625 (*movhi_insn): Likewise.
29626 (*movqi_insn): Likewise.
29627 (*addsi3_mixed): Likewise.
29628 (subsi3_insn): Likewise.
29630 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29632 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29634 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29636 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29637 expressions with MINUS and UNARY ops.
29639 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29642 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29644 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29645 between vec_select and vector argument.
29646 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29647 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29648 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29649 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29650 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29651 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29653 2017-03-13 Richard Biener <rguenther@suse.de>
29656 * params.def (vect-max-peeling-for-alignment): Fix typo.
29658 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29660 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29661 issue that only occurred with binutils below 2.18.
29663 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29665 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29666 refer to binutils 2.11/2.12 minimum.
29668 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29670 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29671 ftp.kernel.org and simplify binutils requirement.
29673 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29675 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29677 (Optimize Options): Ditto. Also remove redundancy.
29679 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29681 PR translation/79848
29682 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29684 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29685 to G_ to avoid double translation.
29687 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29689 PR translation/79923
29690 * auto-profile.c (get_combined_location): Convert leading
29691 character of diagnostics to lower case and remove trailing period.
29692 (read_profile): Likewise for various diagnostics.
29693 * config/arm/arm.c (arm_option_override): Remove trailing period
29694 from various diagnostics.
29695 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29696 (msp430_expand_delay_cycles): Likewise.
29698 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29701 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29702 full command-line argument, rather than just "str".
29703 (aarch64_validate_march): Likewise.
29704 (aarch64_validate_mtune): Likewise.
29706 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29708 PR rtl-optimization/78911
29709 * lra-assigns.c (must_not_spill_p): New function.
29710 (spill_for): Use it.
29712 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29714 PR tree-optimization/79981
29715 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29716 ATOMIC_COMPARE_EXCHANGE ifn result.
29717 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29718 IFN_ATOMIC_COMPARE_EXCHANGE.
29720 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29723 * opts.c (parse_sanitizer_options): Add missing question mark to
29724 "did you mean" message.
29726 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29728 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29730 (VMULEUH_UNS): Likewise.
29731 (VMULOUB_UNS): Likewise.
29732 (VMULOUH_UNS): Likewise.
29733 * config/rs6000/rs6000.c (builtin_function_type): Remove
29734 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29736 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29739 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29740 x with result of extra_parsing_for_operand_code_0.
29741 (function_reader::extra_parsing_for_operand_code_0): Convert
29742 return type from void to rtx, returning x. When reading
29743 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29744 larger size containing struct block_symbol.
29746 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29748 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29749 -mfloat128-hardware without -m64.
29751 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29754 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29755 entries to the case statement that marks unsigned arguments to
29756 overloaded functions.
29758 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29760 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29761 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29763 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29766 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29767 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29769 2017-03-10 Martin Liska <mliska@suse.cz>
29773 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29774 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29777 2017-03-10 Olivier Hainque <hainque@adacore.com>
29779 * tree-switch-conversion (array_value_type): Start by resetting
29780 candidate type to it's main variant.
29782 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29784 PR rtl-optimization/79909
29785 * combine.c (try_combine): Use simplify_replace_rtx on individual
29786 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29787 of the whole CALL_INSN_FUNCTION_USAGE.
29789 PR tree-optimization/79972
29790 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29791 get_range_info on SSA_NAMEs. Formatting fixes.
29793 2017-03-10 Richard Biener <rguenther@suse.de>
29794 Jakub Jelinek <jakub@redhat.com>
29796 PR tree-optimization/77975
29797 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29798 edge to be constant.
29799 (get_val_for): For constant x return it. Formatting fix.
29800 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29801 would use the same bases as the current one.
29803 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29805 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29806 instead of vec_select for V1TImode.
29807 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29809 (VSX_LE_128): Add V1TI to this mode iterator.
29810 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29811 (*vsx_le_perm_store_<mode>): Likewise.
29812 (pre-reload splitter for VSX stores): Likewise.
29813 (post-reload splitter for VSX stores): Likewise.
29814 (*vsx_xxpermdi2_le_<mode>): Likewise.
29815 (*vsx_lxvd2x2_le_<mode>): Likewise.
29816 (*vsx_stxvd2x2_le_<mode>): Likewise.
29818 2017-03-09 Michael Eager <eager@eagercon.com>
29820 Correct failures with --enable-checking=yes,rtl.
29822 * config/microblaze/microblaze.c (microblaze_expand_shift):
29823 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29824 test for const0_rtx.
29825 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29826 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29828 2017-03-09 Richard Biener <rguenther@suse.de>
29830 PR tree-optimization/79977
29831 * graphite-scop-detection.c (scop_detection::merge_sese):
29832 Handle the case of extra exits to blocks dominating the entry.
29834 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29836 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29839 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29841 PR rtl-optimization/79949
29842 * lra-constraints.c (process_alt_operands): Check memory when
29843 trying to predict a cycle. Print about the overall increase.
29845 2017-03-09 Richard Biener <rguenther@suse.de>
29847 PR middle-end/79971
29848 * gimple-expr.c (useless_type_conversion_p): Preserve
29849 TYPE_SATURATING for fixed-point types.
29851 2017-03-09 Richard Biener <rguenther@suse.de>
29854 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29855 alignment of BLKmode params.
29857 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29860 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29861 (VALL_NO_V2Q): Likewise.
29863 * config/aarch64/aarch64-simd.md
29864 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29866 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29867 VALL_NO_V2Q mode iterator.
29868 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29870 2017-03-09 Martin Liska <mliska@suse.cz>
29872 PR tree-optimization/79631
29873 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29874 tree_int_cst_sign_bit just for INTEGER constants.
29876 2017-03-09 Martin Liska <mliska@suse.cz>
29880 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29883 2017-03-09 Marek Polacek <polacek@redhat.com>
29886 * tree.c (inchash::add_expr): Handle TREE_VEC.
29888 2017-03-09 Martin Liska <mliska@suse.cz>
29891 (chkp_narrow_size_and_offset): New function.
29892 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29893 (void chkp_parse_bit_field_ref): New function.
29894 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29895 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29897 2017-03-09 Martin Liska <mliska@suse.cz>
29900 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29901 (chkp_find_bounds_1): Remove gcc_unreachable.
29903 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29906 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29907 BUILT_IN_SYNC*, determine the access type from the size suffix and
29908 always build a MEM_REF with that type. Handle forgotten
29909 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29912 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29913 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29914 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29915 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29916 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29917 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29918 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29919 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29920 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29921 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29922 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29923 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29924 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29925 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29926 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29927 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29928 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29929 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29930 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29931 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29932 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29933 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29934 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29935 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29936 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29937 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29938 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29939 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29940 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29941 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29942 definitions outside of __OPTIMIZE__ guarded section.
29945 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29946 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29947 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29948 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29951 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29953 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29954 ("vfenez<mode>"): Add missing constraints.
29956 2017-03-08 Martin Sebor <msebor@redhat.com>
29959 * config/nds32/nds32.c (nds32_option_override):
29960 Fix misspelled diagnostic.
29962 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29965 * gimplify.c (gimplify_omp_for): Replace index var in outer
29966 taskloop statement with an artificial variable and add
29967 OMP_CLAUSE_PRIVATE clause for it.
29969 2017-03-08 Richard Biener <rguenther@suse.de>
29971 PR tree-optimization/79955
29972 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29973 for accesses that are completely outside of the variable.
29975 2017-03-08 Andrew Haley <aph@redhat.com>
29977 PR tree-optimization/79943
29978 * tree-ssa-loop-split.c (compute_new_first_bound): When
29979 calculating the new upper bound, (END-BEG) should be added, not
29982 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29984 * config/avr/avr.md (setmemhi): Make sure match_dup
29985 operand number comes before match_scratch.
29987 2017-03-08 Richard Biener <rguenther@suse.de>
29989 PR tree-optimization/79920
29990 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29991 with ncopies == 1 to ...
29992 (vect_transform_slp_perm_load): ... here. Properly compute
29993 all element loads by iterating VF times over the group. Do
29994 not handle ncopies (computed in a broken way) in
29995 vect_create_mask_and_perm.
29997 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30000 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30001 is a uniform vector, use uniform_vector_p return value instead of
30002 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30004 2017-03-07 Marek Polacek <polacek@redhat.com>
30006 PR middle-end/79809
30007 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30008 (alloca_call_type): Likewise.
30010 2017-03-07 Martin Liska <mliska@suse.cz>
30012 * gcov.c (process_args): Put comment to correct location.
30014 2017-03-07 Martin Liska <mliska@suse.cz>
30016 PR middle-end/68270
30017 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30018 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30019 (chkp_narrow_bounds_for_field): Likewise.
30020 (chkp_parse_array_and_component_ref): Pass one more argument to
30023 2017-03-07 Richard Biener <rguenther@suse.de>
30025 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30028 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30030 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30031 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30033 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30036 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30037 to end of description.
30038 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30040 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30042 PR rtl-optimization/79901
30043 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30045 (*avx512f_<code><mode>3<mask_name>): ... this.
30046 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30047 iterator instead of VI8_AVX2_AVX512BW.
30049 PR rtl-optimization/79901
30050 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30051 min/max expander, expand it using expand_vec_cond_expr.
30054 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30057 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30060 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30061 to void * for PCH reasons.
30062 * dwarf2out.c (output_loc_operands, output_die): Cast
30063 v.val_vec.array to unsigned char *.
30065 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30068 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30071 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30073 PR rtl-optimization/79571
30074 * lra-constraints.c (process_alt_operands): Calculate static
30075 reject and subtract it from overall when only addresses will be
30078 2017-03-06 Julia Koval <julia.koval@intel.com>
30081 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30082 incoming stack boundary to 128 for 64-bit targets.
30084 2017-03-06 Richard Biener <rguenther@suse.de>
30086 PR tree-optimization/79894
30087 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30088 to NULL after folding it.
30090 2017-03-06 Richard Biener <rguenther@suse.de>
30092 PR tree-optimization/79824
30093 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30094 check disabling peeling for gaps.
30096 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30098 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30099 attributes): Document gettimeofday.
30101 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30103 * config/s390/s390.c (s390_option_override_internal): Set
30104 PARAM_MIN_VECT_LOOP_BOUND
30106 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30108 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30109 * config/s390/s390.md: Likewise.
30111 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30114 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30115 (<avx2_avx512>_perm<mode>): Rename to ...
30116 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30118 (<avx512>_perm<mode>_mask): Rename to ...
30119 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30121 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30122 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30123 instead of VI8F_256_512.
30124 (avx512f_perm<mode>): New define_expand.
30125 (avx512f_perm<mode>_mask): Likewise.
30126 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30127 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30129 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30131 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30132 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30134 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30136 2017-03-06 Martin Liska <mliska@suse.cz>
30139 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30140 when having a SSA NAME w/o VAR_DECL assigned to it.
30142 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30144 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30145 msa_dpsub_<su>_d): Fix MODE for vec_select.
30147 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30149 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30151 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30153 2017-03-06 Richard Biener <rguenther@suse.de>
30155 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30156 * plugin.c (register_plugin_info): Likewise.
30157 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30159 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30161 * config/i386/sse.md (sse_storehps, sse_storelps,
30162 avx_<castmode><avxsizesuffix>_<castmode>,
30163 avx512f_<castmode><avxsizesuffix>_<castmode>,
30164 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30165 in condition that at least one operand is not a MEM.
30167 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30169 PR middle-end/79805
30170 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30171 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30173 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30174 gimple_call_nothrow_p flag based on whether original builtin can throw.
30175 If it can, emit following stmts on the fallthrough edge.
30176 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30177 don't create new bb if inserting just debug stmts on the edge, try to
30178 insert them on the fallthru bb or just reset debug stmts.
30180 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30183 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30184 restore recog_data (including the operand rtxes inside it) around
30185 the call to get_insn_template.
30187 2017-03-03 Martin Sebor <msebor@redhat.com>
30189 PR tree-optimization/79699
30190 * context.c (context::~context): Free MPFR caches to avoid
30191 a memory leak on program exit.
30193 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30195 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30196 Use wide_int::ulow () instead of .elt (0).
30198 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30200 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30201 (*pushxf): Limit oF constraint to 32bit targets and add oC
30202 constraint for 64bit targets.
30203 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30204 (*pushdf): Change rmF constraint to rmC.
30206 2017-03-03 Martin Liska <mliska@suse.cz>
30208 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30209 Remove unused variable.
30211 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30214 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30215 is a memory operand, increase num_memory.
30216 (ix86_expand_args_builtin): Likewise.
30218 2017-03-03 Jan Hubicka <jh@suse.cz>
30221 * ipa-devirt.c (maybe_record_node): Properly handle
30222 __cxa_pure_virtual visibility.
30224 2017-03-03 Martin Liska <mliska@suse.cz>
30226 PR tree-optimization/79803
30227 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30229 (pass_loop_prefetch::execute): Disabled optimization if an
30230 assumption about L1 cache size is not met.
30232 2017-03-03 Martin Liska <mliska@suse.cz>
30234 PR rtl-optimization/79574
30235 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30236 (hash_scan_set): Likewise.
30237 (dump_hash_table): Likewise.
30238 (hoist_code): Likewise.
30240 2017-03-03 Richard Biener <rguenther@suse.de>
30242 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30243 in place of uhigh ().
30244 (fixed_convert_from_real): Likewise.
30246 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30249 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30251 2017-03-03 Richard Biener <rguenther@suse.de>
30253 PR middle-end/79818
30254 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30255 TYPE_OVERFLOW_UNDEFINED check.
30257 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30259 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30261 (vector_ae_<mode>_p): Likewise.
30262 (vector_nez_<mode>_p): Likewise.
30263 (vector_ne_v2di_p): Likewise.
30264 (vector_ae_v2di_p): Likewise.
30265 (vector_ne_<mode>_p): Likewise.
30266 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30268 (vsx_tsqrt<mode>2_fe): Likewise.
30270 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30273 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30275 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30277 PR rtl-optimization/79780
30278 * cprop.c (one_cprop_pass): When second and further conditional trap
30279 in a single basic block is turned into an unconditional trap, turn it
30280 into a deleted note to avoid RTL verification failures.
30282 2017-03-02 Richard Biener <rguenther@suse.de>
30284 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30286 2017-03-02 Richard Biener <rguenther@suse.de>
30288 PR tree-optimization/79345
30290 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30291 param and abort the walk, returning -1 if it is hit.
30292 (walk_aliased_vdefs): Take a limit param and pass it on.
30293 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30294 defaulting to 0 and return a signed int.
30295 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30296 (check_defs): New helper.
30297 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30298 about uninitialized memory.
30299 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30300 bogus uninitialized warning.
30301 (fixed_convert_from_real): Likewise.
30303 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30305 PR tree-optimization/66768
30306 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30307 iv_use if base object can't be determined.
30309 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30311 PR tree-optimization/79345
30312 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30313 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30314 (get_pattern_stats): Initialize it.
30315 * genemit.c (gen_expand): Verify match_scratch numbers come after
30316 match_operand/match_dup numbers.
30317 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30318 match_scratch numbers.
30319 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30321 * config/s390/s390.md (trunctdsd2): Likewise.
30323 2017-03-02 Richard Biener <rguenther@suse.de>
30325 * wide-int.h (wide_int_storage::operator=): Implement in terms
30328 2017-03-02 Richard Biener <rguenther@suse.de>
30330 PR tree-optimization/79777
30331 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30332 the to insert expression to sth existing.
30334 2017-03-01 Martin Sebor <msebor@redhat.com>
30336 PR middle-end/79692
30337 * gimple-ssa-sprintf.c
30338 (directive::known_width_and_precision): New function.
30339 (format_integer): Use it.
30340 (get_mpfr_format_length): Consider the full range of precision
30341 when computing %g output with the # flag. Set the likely byte
30342 count to 3 rather than 1 when precision is indeterminate.
30343 (format_floating): Correct the lower bound of precision.
30345 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30347 * doc/invoke.texi: Document default code model for 64-bit Linux.
30349 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30352 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30353 udiv rather than div since input pattern is unsigned.
30355 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30357 * config/i386/i386.c (print_reg): Warn for values of
30358 unsupported size in integer register.
30360 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30363 * config/rs6000/predicates.md (current_file_function_operand): Do
30364 not allow self calls to be local if the function is replaceable.
30366 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30369 * config/rs6000/altivec.h (vec_ctz and others): Change the
30370 preprocessor macro that controls conditional compilation from
30371 _ARCH_PWR9 to __POWER9_VECTOR__.
30372 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30373 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30374 control (instead of _ARCH_PWR9 control) so that template
30375 definition uses power9-specific function.
30376 (vec_any_eq): Likewise.
30377 (vec_all_ne): Change macro definition to use a power9-specific
30378 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30379 _ARCH_PWR9 control).
30380 (vec_any_eq) Likewise.
30381 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30382 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30383 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30384 support for xvcmpnedp instruction.
30385 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30386 macro expansion so that Power9 implementation of vec_all_ne does
30387 not use the AltiVec predicate framework.
30388 (VCMPNEH_P): Likewise.
30389 (VCMPNEW_P): Likewise.
30390 (VCMPNED_P): Likewise.
30391 (VCMPNEFP_P): Likewise.
30392 (VCMPNEDP_P): Likewise.
30393 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30394 implementation of vec_any_eq to not use AltiVec predicate
30396 (VCMPAEH_P): Likewise.
30397 (VCMPAEW_P): Likewise.
30398 (VCMPAED_P): Likewise.
30399 (VCMPAEFP_P): Likewise.
30400 (VCMPAEDP_P): Likewise.
30401 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30402 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30403 not use the AltiVec predicate framework.
30404 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30405 of vec_any_eq to not use AltiVec predicate framework.
30406 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30407 support for predefined __POWER9_VECTOR__ macro to indicate that
30408 Power9 instruction selection is enabled.
30409 (altivec_overloaded_builtins): Remove extraneous
30410 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30411 function argument types RS6000_BTI_bool_V16QI and
30412 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30413 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30414 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30415 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30416 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30417 Power9 for implementations of vec_cmpne. Change the signature for
30418 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30419 (representing vec_all_ne) to remove the previously described first
30420 argument of type RS6000_BTI_INTSI, as this was an artifact of
30421 reliance on the AltiVec predicate framework, which is no longer
30422 used in the implementation of these functions. Add
30423 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30424 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30425 since, unlike the AltiVec predicate framework implementation, we
30426 do not share function descriptors between vec_alle and vec_anyeq.
30427 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30428 set of modes that receive special treatment even when
30429 TARGET_P9_VECTOR is true. The special treatment emits code that
30430 does not depend on Power9 instructions.
30431 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30432 define_expand to not rely on AltiVec predicate framework.
30433 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30435 (vector_ne_v2di_p): Change this define_expand to not rely on
30436 AltiVec predicate framework.
30437 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30439 (vector_ne_<mode>_p): Change this define_expand to not rely on
30440 AltiVec predicate framework.
30441 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30443 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30444 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30445 define_insn pattern.
30446 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30447 define_insn pattern because the xvcmpne<VSs>. instruction is not
30449 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30450 instruction is not supported.
30452 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30454 * config/nvptx/nvptx.c: Include intl.h.
30456 2017-03-01 Martin Jambor <mjambor@suse.cz>
30459 * ipa-prop.h (ipa_bits): Removed field known.
30460 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30461 to pointers. Adjusted their comments to warn about their sharing.
30462 (ipcp_transformation_summary): Change bits to a vector of pointers.
30463 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30464 (ipa_get_ipa_bits_for_value): Declare.
30465 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30466 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30467 (ipa_bits_hash_table): Likewise.
30468 (ipa_vr_ggc_hash_traits): Likewise.
30469 (ipa_vr_hash_table): Likewise.
30470 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30471 being pointers and vr_known being removed.
30472 (ipa_set_jf_unknown): Likewise.
30473 (ipa_get_ipa_bits_for_value): New function.
30474 (ipa_set_jfunc_bits): Likewise.
30475 (ipa_get_value_range): New overloaded functions.
30476 (ipa_set_jfunc_vr): Likewise.
30477 (ipa_compute_jump_functions_for_edge): Use the above functions to
30478 construct bits and vr parts of jump functions.
30479 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30480 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30482 (ipcp_grow_transformations_if_necessary): Also allocate
30483 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30485 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30486 them. Fix too long lines.
30487 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30488 vr_known being removed.
30489 (ipa_read_jump_function): Use new setter functions to construct bits
30490 and vr parts of jump functions or set them to NULL.
30491 (write_ipcp_transformation_info): Adjust for bits being pointers.
30492 (read_ipcp_transformation_info): Likewise.
30493 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30495 Include gt-ipa-prop.h.
30496 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30498 (ipcp_store_bits_results): Likewise.
30499 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30500 Do not write to existing jump functions but use a temporary instead.
30502 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30505 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30506 attempt to use its first operand as BIT_FIELD_REF base.
30508 2017-03-01 Richard Biener <rguenther@suse.de>
30510 PR middle-end/79721
30511 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30512 interpolating formula in wrapping arithmetic.
30513 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30515 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30517 PR tree-optimization/79734
30518 * tree-vect-generic.c (expand_vector_condition): Optimize
30519 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30520 Handle VEC_COND_EXPR where comparison has different inner width from
30521 type's inner width.
30523 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30525 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30526 markup, and similar issues. Remove @opindex entries for things
30527 that aren't options. Add missing -mmpy-option entries.
30529 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30531 PR tree-optimization/79737
30532 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30533 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30534 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30535 instead of byte_size. Formatting fix.
30536 (shift_bytes_in_array_right): Formatting fix.
30538 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30541 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30542 condition on optimize for the leaf function test.
30544 2017-02-28 Martin Liska <mliska@suse.cz>
30547 * read-rtl-function.c (function_reader::handle_unknown_directive):
30548 Bail out when one uses -flto.
30550 2017-02-28 Martin Liska <mliska@suse.cz>
30552 * common.opt: Replace space with tabular for options of <number>
30554 * config/i386/i386.opt: Show <number> value for
30555 -mlarge-data-threshold.
30556 * opts.c (print_filtered_help): Do not display number in hexadecimal
30559 2017-02-28 Martin Liska <mliska@suse.cz>
30561 * common.opt: Fix --help=option -Q for options which are of
30564 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30566 * config/i386/i386.c (print_reg): Error out for values
30567 of 8-bit size in invalid integer register.
30569 2017-02-28 Martin Sebor <msebor@redhat.com>
30571 PR tree-optimization/79691
30572 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30574 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30577 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30578 gcc_unreachable with output_operand_lossage.
30580 2017-02-28 Richard Biener <rguenther@suse.de>
30582 PR tree-optimization/79740
30583 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30585 (visit_nary_op): Insert the nary into the hashtable if we
30586 pattern-matched sth.
30587 * tree-ssa-pre.c (eliminate_insert): Robustify.
30589 2017-02-28 Richard Biener <rguenther@suse.de>
30591 PR middle-end/79731
30592 * fold-const.c (decode_field_reference): Reject out-of-bound
30595 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30597 * config/i386/i386.c: Include intl.h.
30598 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30599 instead of just cond ? "..." : "...".
30600 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30601 * coverage.c (read_counts_file): Likewise.
30602 * omp-offload.c: Include intl.h.
30603 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30604 of just cond ? "..." : "...".
30605 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30606 of just cond ? "..." : "...".
30608 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30611 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30613 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30614 'tune for' CPU name.
30615 * config/arm/arm-cpu-data.h: Regenerated.
30617 2017-02-28 Richard Biener <rguenther@suse.de>
30619 PR tree-optimization/79732
30620 * tree-inline.c (expand_call_inline): Do not shadow var.
30622 2017-02-28 Richard Biener <rguenther@suse.de>
30624 PR tree-optimization/79723
30625 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30626 address-space properly.
30628 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30630 * doc/optinfo.texi (Optimization groups): Fix option used for
30632 * doc/invoke.texi (-fopt-info): Document "omp".
30633 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30634 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30635 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30636 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30637 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30639 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30641 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30642 OPTGROUP_OMP with "omp".
30644 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30647 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30648 for arithmetic shift of unsigned V2DI.
30650 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30652 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30653 arc/linux.h headers.
30654 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30655 (LINK_SPEC): Likewise.
30656 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30657 (EXTRA_SPECS): Likewise.
30658 (STARTFILE_SPEC): Likewise.
30659 (ENDFILE_SPEC): Likewise.
30660 (LIB_SPEC): Likewise.
30661 (TARGET_SDATA_DEFAULT): Likewise.
30662 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30663 (MULTILIB_DEFAULTS): Likewise.
30664 (DWARF2_UNWIND_INFO): Likewise.
30665 * config/arc/big.h: New file.
30666 * config/arc/elf.h: Likewise.
30667 * config/arc/linux.h: Likewise.
30668 * config/arc/t-uClibc: Remove.
30670 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30672 PR tree-optimization/77536
30673 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30674 (tree_transform_and_unroll_loop): Use above function to compute the
30675 estimated niter of unrolled loop and use it when scaling profile.
30676 Also use count info rather than frequency if it's non-zero.
30677 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30678 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30679 (vect_transform_loop): Call above function.
30681 2017-02-27 Richard Biener <rguenther@suse.de>
30683 PR tree-optimization/45397
30684 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30685 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30686 (visit_nary_op): Add pattern matching for CSEing sign-changed
30687 or truncated operations with wider ones.
30689 2017-02-27 Richard Biener <rguenther@suse.de>
30691 PR tree-optimization/79690
30692 * tree-vect-stmts.c (vectorizable_store): Use vector type
30693 built from the DR with address-space.
30695 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30697 * doc/invoke.texi (Optimize Options): Refine the description
30698 of asan-use-after-return.
30700 2017-02-25 Alan Modra <amodra@gmail.com>
30702 PR rtl-optimization/79584
30703 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30704 base, not ad->base_term, the reg within base. Remove assertion
30705 that ad->base == ad->base_term. Replace gen_int_mode using
30706 bogus mode with const0_rtx.
30708 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30710 PR middle-end/79396
30711 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30712 FMA_EXPR like tcc_binary or tcc_unary.
30714 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30717 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30719 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30720 (output_loc_operands): Handle DW_OP_call_ref and
30721 DW_OP_GNU_variable_value.
30722 (struct variable_value_struct): New type.
30723 (struct variable_value_hasher): Likewise.
30724 (variable_value_hash): New variable.
30725 (string_types): Remove.
30726 (copy_loc_descr): New function.
30727 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30728 (prepend_loc_descr_to_each): New function.
30729 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30730 instead of add_loc_descr_to_each if the first argument is single
30731 location list and the second has multiple.
30732 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30733 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30734 when looking for variable value which doesn't have other location info.
30735 (loc_list_from_tree): Formatting fix.
30736 (gen_array_type_die): Simplify DW_AT_string_length handling.
30737 (adjust_string_types): Remove.
30738 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30739 string_types. Call resolve_variable_values.
30740 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30741 (resolve_addr_in_expr): Likewise. Add A argument.
30742 (copy_deref_exprloc): Remove deref argument. Adjust for the
30743 original expression being DW_OP_GNU_variable_value with optionally
30744 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30745 optionally after it.
30746 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30747 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30748 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30749 (variable_value_hasher::hash, variable_value_hasher::equal): New
30751 (resolve_variable_value_in_expr, resolve_variable_value,
30752 resolve_variable_values, note_variable_value_in_expr,
30753 note_variable_value): New functions.
30754 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30757 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30760 * opts.h (handle_generated_option): Add GENERATED_P argument.
30761 * opts-common.c (handle_option): Adjust function comment.
30762 (handle_generated_option): Add GENERATED_P argument, pass it to
30764 (control_warning_option): Pass false to handle_generated_option
30766 * opts.c (maybe_default_option): Pass true to handle_generated_option
30768 * optc-gen.awk: Likewise.
30770 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30772 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30773 a REG, look at the REG it is a SUBREG of.
30774 (splitter for cmpeqsi_t): Ditto.
30776 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30778 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30779 the special USEs with the pattern of the insn, not the insn itself.
30781 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30784 * doc/invoke.texi: Document -mload-store-pairs.
30786 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30787 Sandra Loosemore <sandra@codesourcery.com>
30789 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30790 argument isn't a CONST_INT.
30791 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30792 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30793 (nios2_validate_compare): Bypass alternate compare logic if *op2
30794 is not a CONST_INT.
30795 (ldstwm_operation_p): Return false if first_base is not a REG or
30796 if first_offset is not a CONST_INT.
30798 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30800 * config/cris/cris.md: Use correct operand in a define_peephole2.
30802 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30804 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30806 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30808 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30809 this_insn if it is an INSN or JUMP_INSN.
30810 (force_offsettable): Look at base, not at addr.
30811 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30812 on things that aren't necessarily CONST_INTs.
30814 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30816 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30817 -mfpmath=sse is the default also for x86-32 targets with SSE2
30818 instruction set when @option{-ffast-math} is enabled
30820 2017-02-24 Jeff Law <law@redhat.com>
30822 PR rtl-optimizatoin/79286
30823 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30826 2017-02-24 Richard Biener <rguenther@suse.de>
30828 PR tree-optimization/79389
30829 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30832 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30834 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30835 function comment to reflect reality.
30836 (loop_exits_before_overflow): Fix typo in function description.
30838 2017-02-24 Richard Biener <rguenther@suse.de>
30840 PR tree-optimization/79389
30841 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30842 properly that a threading opportunity exists. Detect conditional
30843 copy/constant propagation opportunities.
30845 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30847 * config/visium/visium.md (type): Add trap.
30848 (b): New mode attribute.
30849 (*btst): Rename into...
30850 (*btst<mode>): ...this and adjust.
30851 (*cbranchsi4_btst_insn): Rename into...
30852 (*cbranch<mode>4_btst_insn): ...this and adjust.
30853 (trap): New define_insn.
30855 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30857 PR tree-optimization/79389
30858 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30859 (noce_reversed_cond_code): New function.
30860 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30861 reversed_comparison_code. Formatting fix.
30862 (noce_try_store_flag): Test rev_cond != NULL in addition to
30863 reversed_comparison_code.
30864 (noce_try_store_flag_constants): Likewise.
30865 (noce_try_store_flag_mask): Likewise.
30866 (noce_try_addcc): Use rev_cond if non-NULL instead of
30867 reversed_comparison_code.
30868 (noce_try_cmove_arith): Likewise. Formatting fixes.
30869 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30870 (noce_find_if_block): Initialize rev_cond.
30871 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30872 instead of false as last argument never attempt to reverse it
30875 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30877 PR tree-optimization/79663
30878 * tree-predcom.c (combine_chains): Process refs in reverse order
30879 only for ZERO length chains, and add explaining comment.
30881 2017-02-23 Jeff Law <law@redhat.com>
30883 PR tree-optimization/79578
30884 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30885 in call to operand_equal_p.
30887 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30890 * config/i386/cpuid.h: Fix another undefined behavior.
30892 2017-02-23 Richard Biener <rguenther@suse.de>
30894 PR tree-optimization/79683
30895 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30896 vector types for data-refs.
30898 2017-02-23 Martin Liska <mliska@suse.cz>
30900 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30902 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30904 PR middle-end/79665
30905 * internal-fn.c (get_range_pos_neg): Moved to ...
30906 * tree.c (get_range_pos_neg): ... here. No longer static.
30907 * tree.h (get_range_pos_neg): New prototype.
30908 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30909 are known to be in between 0 and signed maximum inclusive, try to
30910 expand both unsigned and signed divmod and use the cheaper one from
30913 2017-02-22 Jeff Law <law@redhat.com>
30915 PR tree-optimization/79578
30916 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30917 to compare base operands.
30919 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30922 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30923 gpc_reg_operand instead of fpr_reg_operand.
30925 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30927 * config/mips/mips.c (mips_return_in_memory): Force FP
30928 vector types to be returned in memory for o32 ABI.
30930 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30932 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30933 instead of DW_TAG_member for static data member declarations and don't
30934 set no_linkage_name for static inline data members.
30935 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30938 2017-02-22 Martin Liska <mliska@suse.cz>
30940 * doc/invoke.texi: Replace inequality signs with square brackets
30943 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30946 * lra-constraints.c (simplify_operand_subreg): Handle
30947 WORD_REGISTER_OPERATIONS targets.
30949 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30952 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30953 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30954 elimination by swapping fld*.
30956 2017-02-22 Richard Biener <rguenther@suse.de>
30958 PR tree-optimization/79673
30959 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30960 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30961 irrelevant address-space qualifiers and avoiding a
30962 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30964 2017-02-22 Richard Biener <rguenther@suse.de>
30966 PR tree-optimization/79666
30967 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30968 to not symbolically negate if that may introduce undefined
30971 2017-02-22 Martin Liska <mliska@suse.cz>
30974 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30975 * data-streamer-out.c (streamer_write_gcov_count_stream):
30977 * value-prof.c (stream_out_histogram_value): Make assert more
30978 precise based on type of counter.
30980 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30983 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30984 Use nonimmediate_operand instead of memory_operand for operand 1.
30985 (float-extend standard_x87sse_constant_load splitter): Ditto.
30987 2017-02-21 Jeff Law <law@redhat.com>
30989 PR tree-optimization/79621
30990 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30991 blocks with edges to themselves.
30993 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30996 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30997 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30998 Use gimple_call_builtin_p.
31001 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31002 on temporarily removed DEBUG_INSNs.
31004 PR tree-optimization/79649
31005 * tree-loop-distribution.c (classify_partition): Give up on
31006 non-generic address space loads/stores.
31008 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31010 * doc/loop.texi (Loop manipulation): Remove nonexistent
31011 tree_ssa_loop_version from the documentation.
31012 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31014 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31017 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31018 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31019 * config/rs6000/rs6000.c: Include except.h.
31020 (rs6000_expand_split_stack_prologue): Call
31021 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31023 2017-02-21 Martin Jambor <mjambor@suse.cz>
31026 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31027 have been analyzed.
31029 2017-02-21 Martin Jambor <mjambor@suse.cz>
31031 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31032 for backward compatibility only.
31033 * doc/invoke.texi (Option Summary): Remove all references to
31034 -fipa-cp-alignment.
31036 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31040 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31042 * lra-constraints.c (curr_insn_transform): Handle
31043 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31045 2017-02-21 Martin Liska <mliska@suse.cz>
31047 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31049 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31051 PR translation/79638
31052 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31054 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31057 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31058 (arm_function_ok_for_sibcall): Return false for an indirect call by
31059 descriptor if all the argument registers are used.
31060 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31061 alignment of the function.
31063 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31065 PR tree-optimization/61441
31066 * simplify-rtx.c (simplify_const_unary_operation): For
31067 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31068 the sNaN unmodified.
31070 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31072 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31073 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31074 instead of SYSTEM_HEADER_DIR.
31076 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31077 Martin Liška <mliska@suse.cz>
31079 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31080 Fix typos and grammar, use active voice, and clarify.
31082 2017-02-20 Marek Polacek <polacek@redhat.com>
31084 PR middle-end/79537
31085 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31088 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31090 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31093 * config/i386/i386.c (ix86_expand_builtin): Handle
31094 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31095 ix86_builtins_isa[fcode].isa as a requirement of those
31096 flags and any other flag in the bitmask.
31097 (ix86_init_mmx_sse_builtins): Use 0 instead of
31098 ~OPTION_MASK_ISA_64BIT as mask.
31099 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31100 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31101 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31102 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31104 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31107 * lra-constraints.c (split_reg): Check requested split mode
31108 is supported by the register.
31110 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31112 * lra-constraints.c (simplify_operand_subreg): Remove early
31115 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31118 * lra-constraints.c (curr_insn_transform): Tighten condition
31119 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31121 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31124 * lra-constraints.c (curr_insn_transform): Handle
31125 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31127 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31130 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31132 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31134 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31137 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31140 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31142 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31145 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31147 * final.c (last_columnnum, override_columnnum): New variables.
31148 (final_start_function): Set last_columnnum, pass it to begin_prologue
31149 hook and pass 0 to dwarf2out_begin_prologue.
31150 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31151 to source_line debug hook.
31152 (notice_source_line): Compute last_columnnum and for debug_column_info
31153 return true on column changes.
31154 * debug.h (struct gcc_debug_hooks): Add column argument to
31155 source_line and begin_prologue hooks.
31156 (debug_nothing_int_charstar_int_bool): Remove prototype.
31157 (debug_nothing_int_int_charstar,
31158 debug_nothing_int_int_charstar_int_bool): New prototypes.
31159 (dwarf2out_begin_prologue): Add column argument.
31160 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31161 begin_prologue hooks.
31162 (debug_nothing_int_charstar_int_bool): Remove.
31163 (debug_nothing_int_int_charstar,
31164 debug_nothing_int_int_charstar_int_bool): New functions.
31165 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31166 through to dwarf2out_source_line.
31167 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31168 (dwarf2out_source_line): Add column argument, emit it if requested.
31169 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31171 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31172 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31173 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31174 through to dwarf2out_begin_prologue.
31175 (vmsdbgout_source_line): Add column argument, pass it through to
31176 dwarf2out_source_line.
31177 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31178 dbxout_source_line caller.
31179 (dbxout_source_line): Add column argument.
31181 * common.opt (gno-column-info, gcolumn-info): New options.
31182 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31183 (check_die): Also test for multiple DW_AT_decl_column attributes.
31184 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31185 DW_AT_decl_column if requested.
31186 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31188 (gen_variable_die): Likewise.
31189 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31190 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31193 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31194 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31195 (ix86_handle_option): Handle OPT_m3dnowa.
31196 * doc/invoke.texi (-m3dnowa): Document.
31197 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31198 -m3dnowa instead of -m3dnow -march=athlon.
31201 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31202 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31204 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31207 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31208 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31209 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31210 generator for vsx_xxpermdi_<mode>_be.
31211 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31212 force big-endian semantics.
31213 (vsx_xxpermdi_<mode>_be): New define_expand with same
31214 implementation as previous version of vsx_xxpermdi_<mode>.
31216 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31218 PR tree-optimization/79327
31219 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31220 variable, its initialization and use.
31222 2017-02-17 Julia Koval <julia.koval@intel.com>
31224 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31225 (OPTION_MASK_ISA_PKU_UNSET): New.
31226 (ix86_handle_option): Handle -mrdpid.
31227 * config/i386/cpuid.h (bit_RDPID): New.
31228 * config/i386/driver-i386.c (host_detect_local_cpu):
31229 Detect RDPID feature.
31230 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31231 * config/i386/i386-c.c (ix86_target_macros_internal):
31233 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31234 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31235 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31236 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31237 * config/i386/i386.md (define_insn "rdpid"): New.
31238 * config/i386/i386.opt Add -mrdpid.
31239 * config/i386/immintrin.h (_rdpid_u32): New.
31241 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31243 PR rtl-optimization/79541
31244 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31245 instead of transforming it into USE.
31247 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31249 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31250 If HONOR_SNANS (SFmode) force the input to a register.
31251 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31252 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31253 an frsp or similar insn.
31255 2017-02-17 Martin Liska <mliska@suse.cz>
31257 PR rtl-optimization/79577
31258 * params.def (selsched-max-sched-times): Increase minimum to 1.
31260 2017-02-17 Martin Liska <mliska@suse.cz>
31262 PR rtl-optimization/79574
31263 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31265 2017-02-17 Martin Liska <mliska@suse.cz>
31267 PR tree-optimization/79529
31268 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31269 ssa_defined_default_def_p to handle cases which are implicitly
31271 * tree-ssa.c (ssa_defined_default_def_p): New function.
31272 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31273 which are implicitly defined.
31274 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31276 2017-02-17 Richard Biener <rguenther@suse.de>
31278 PR middle-end/79576
31279 * params.def (max-ssa-name-query-depth): Limit to 10.
31281 2017-02-17 Richard Biener <rguenther@suse.de>
31283 PR tree-optimization/79552
31284 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31287 2017-02-17 Richard Biener <rguenther@suse.de>
31290 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31292 2017-02-17 Marek Polacek <polacek@redhat.com>
31294 PR middle-end/79536
31295 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31296 (fold_negate_expr): New wrapper.
31298 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31300 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31301 Correct terminology and de-emphasize pre-standard behavior.
31303 2017-02-16 Alan Modra <amodra@gmail.com>
31305 PR rtl-optimization/79286
31306 * ira.c (def_dominates_uses): New function.
31307 (update_equiv_regs): Don't create an equivalence for insns that
31308 may trap where the register def does not dominate the use.
31310 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31312 PR rtl-optimization/78127
31313 * lra.c (lra): Call lra_eliminate before finish the loop after
31316 2017-02-16 Richard Biener <rguenther@suse.de>
31318 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31320 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31321 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31322 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31323 (isl_val_int_from_wi): New function.
31324 (extract_affine_gmp): Rename to ...
31325 (extract_affine_wi): ... this, take a widest_int.
31326 (extract_affine_int): Just wrap extract_affine_wi.
31327 (add_param_constraints): Use isl_val_int_from_wi.
31328 (add_loop_constraints): Likewise, and extract_affine_wi.
31330 2017-02-15 Jeff Law <law@redhat.com>
31332 PR middle-end/79521
31333 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31334 ira_init_register_move_cost_if_necessary.
31336 2017-02-15 Martin Sebor <msebor@redhat.com>
31338 PR middle-end/32003
31339 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31340 removed in a prior commit.
31342 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31344 PR tree-optimization/79347
31345 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31346 counters during peeling.
31348 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31350 * Makefile.in (site.exp): Remove "set ISLVER".
31352 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31355 * real.c (real_from_integer): Call real_convert even for decimal.
31357 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31360 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31362 2017-02-14 Andrew Pinski <apinski@cavium.com>
31364 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31365 cores and change the partno/implementer to be correct.
31366 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31367 the 'B" as the implementer.
31368 * config/aarch64/aarch64-tune.md: Regenerate.
31370 2017-02-14 Carl Love <cel@us.ibm.com>
31372 * config/rs6000/rs6000.c: Add case statement entry to make the
31373 xvcvuxdsp built-in argument unsigned.
31374 * config/rs6000/vsx.md: Fix the source and return operand types so they
31375 match the instruction definitions from the ISA document. Fix typo
31376 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31379 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31382 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31383 member early_clobber_alts.
31384 * lra-lives.c (reg_early_clobber_p): New.
31385 (process_bb_lives): Use it.
31386 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31387 (debug_operand_data): Initialize early_clobber_alts.
31388 (setup_operand_alternative): Set up early_clobber_alts.
31389 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31390 alternatives to new_insn_reg.
31391 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31393 (lra_update_insn_regno_info): Pass the new arg.
31395 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31397 PR middle-end/79505
31398 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31399 (new_oacc_loop_raw): Don't clear already cleared fields.
31402 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31403 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31404 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31406 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31409 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31411 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31414 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31415 the extra instruction to the right place to store 128-bit constant
31418 2017-02-14 Martin Sebor <msebor@redhat.com>
31420 PR middle-end/79448
31421 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31422 warning for strings of unknown length.
31424 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31426 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31428 2017-02-14 Jeff Law <law@redhat.com>
31431 * ira-costs.c (scan_one_insn): Initialize register move costs
31432 for pseudos seen in USE/CLOBBER insns.
31434 PR tree-optimization/79095
31435 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31436 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31437 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31438 if the operands are known to be not equal, then the resulting range
31440 (intersect_ranges): If the new range is ~[0,0] and the old range is
31441 wide, then prefer ~[0,0].
31442 * tree-vrp.c (overflow_comparison_p_1): New function.
31443 (overflow_comparison_p): New function.
31444 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31445 if NAME is used in an overflow test.
31446 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31447 overflow check that can be expressed as an equality test, then adjust
31448 ops to be that equality test.
31450 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31452 * config/s390/s390-builtin-types.def: Remove flags argument.
31453 * config/s390/s390.c (s390_init_builtins): Likewise.
31455 2017-02-14 Martin Liska <mliska@suse.cz>
31457 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31458 vector. Fix trailing white spaces.
31460 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31462 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31465 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31467 PR rtl-optimization/68664
31468 * config/arm/arm.c (arm_sched_can_speculate_insn):
31469 New function. Declare prototype.
31470 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31472 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31474 PR rtl-optimization/68664
31475 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31477 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31479 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31481 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31482 max skip bytes for function, loop and jump.
31484 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31486 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31487 ABS_EXPR for gimple dump.
31489 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31492 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31494 PR tree-optimization/79408
31495 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31496 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31497 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31498 also if rhs1 is INTEGER_CST.
31500 2017-02-14 Richard Biener <rguenther@suse.de>
31502 PR middle-end/79432
31503 * tree-into-ssa.c (insert_phi_nodes): When the function can
31504 have abnormal edges rewrite SSA names with broken use-def
31505 dominance out of SSA and register them for PHI insertion.
31507 2017-02-13 Martin Sebor <msebor@redhat.com>
31509 PR middle-end/79496
31510 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31511 clearing info.nowrite flag when snprintf size argument is a range.
31513 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31515 * cprop.c (cprop_jump): Add missing space in string literal.
31516 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31517 (get_constraint_for_component_ref): Likewise.
31518 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31519 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31520 * lra-constraints.c (process_alt_operands): Likewise.
31521 * ipa-inline.c (inline_small_functions): Likewise.
31522 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31523 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31524 * trans-mem.c (diagnose_tm_1_op): Likewise.
31525 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31526 (grid_parallel_clauses_gridifiable): Likewise.
31528 * config/nvptx/mkoffload.c (process): Add space in between
31531 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31532 "MOD4_SSE_REGS" and "ALL_REGS".
31534 * spellcheck.c (test_data): Add , in between "foo" and "food".
31536 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31539 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31540 boundary crossing check and subsequent code generation agree.
31542 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31544 * config/aarch64/aarch64.c (has_memory_op): Delete.
31545 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31548 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31550 PR rtl-optimization/79388
31551 PR rtl-optimization/79450
31552 * combine.c (distribute_notes): When removing TEM_INSN for which
31553 corresponding dest has last value recorded, invalidate that last
31556 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31558 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31559 of explicit '@'. Add missing assembly comment marker on branch costs
31562 2017-02-13 Nathan Sidwell <nathan@acm.org>
31564 * gengtype-lex.l (<in_struct>): Add '/'.
31566 2017-02-13 Martin Liska <mliska@suse.cz>
31569 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31571 2017-02-13 Richard Biener <rguenther@suse.de>
31573 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31575 * configure: Re-generate.
31576 * config.in: Likewise.
31577 * graphite-dependences.c: Simplify as if
31578 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31579 * graphite-isl-ast-to-gimple.c: Likewise.
31580 * graphite-optimize-isl.c: Likewise.
31581 * graphite-poly.c: Likewise.
31582 * graphite-sese-to-poly.c: Likewise.
31583 * graphite.h: Likewise.
31584 * toplev.c: Include isl/version.h and use isl_version () for
31585 printing the ISL version.
31586 * doc/install.texi: Update ISL requirement.
31588 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31590 * doc/standards.texi (Standards): Update reference to
31593 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31595 * doc/extend.texi (Named Address Spaces): sourceware.org now
31597 * doc/install.texi (Binaries): Ditto.
31600 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31602 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31603 standard terminology "stringize"/"stringizing" throughout.
31604 * doc/cppinternals.texi: Likewise.
31606 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31608 * doc/extend.texi: Fix some spelling mistakes and typos.
31609 * doc/invoke.texi: Likewise.
31611 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31614 * params.def (inline-min-speedup) Change from 10 to 8.
31616 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31618 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31621 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31624 * ipa-inline-analysis.c (get_minimal_bb): New function.
31625 (record_modified): Use it.
31626 (remap_edge_change_prob): Handle also ancestor functions.
31628 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31630 * doc/contrib.texi (Contributors): Remove broken link into
31631 the Mauve CVS repository.
31633 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31635 PR middle-end/79454
31636 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31637 result computation whenever lhs doesn't have vector mode, not
31638 just when it has BLKmode.
31640 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31642 * doc/makefile.texi (profiledbootstrap): Refer to the
31643 installation instructions only in textual form.
31645 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31648 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31650 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31652 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31653 (Specific): Update mingw-w64 reference.
31655 (Specific): Remove broken link to Renesas RX processor.
31657 2017-02-10 Richard Biener <rguenther@suse.de>
31659 * toplev.c (process_options): Do not mention obsolete graphite
31660 options when printing sorry message about missing graphite support.
31661 Mention -floop-nest-optimize.
31663 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31665 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31666 (vtst_p16): Likewise.
31667 (vtstq_p8): Likewise.
31668 (vtstq_p16): Likewise.
31670 (vtstq_p64): Likewise.
31671 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31672 (vset_lane_p64): New.
31673 (vsetq_lane_p64): New.
31675 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31677 PR tree-optimization/79411
31678 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31679 stmt operands are SSA_NAMEs used in abnormal phis.
31680 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31683 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31686 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31689 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31691 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31693 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31695 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31699 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31700 not arbitrary TREE_CONSTANT.
31703 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31704 "omp declare target link" attribute unless is_global_var.
31705 * omp-offload.c (find_link_var_op): Likewise.
31707 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31708 Chung-Lin Tang <cltang@codesourcery.com>
31710 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31712 (gimplify_adjust_omp_clauses): Don't delete TILE.
31713 (gimplify_omp_for): Deal with TILE.
31714 * internal-fn.c (expand_GOACC_TILE): New function.
31715 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31717 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31718 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31720 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31721 avoid DIV for outermost collapse var.
31722 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31723 Remove out of date comments, fix whitespace.
31724 * omp-general.c (omp_extract_for_data): Deal with tiling.
31725 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31726 adjust OLF_DIM_BASE value.
31727 (struct omp_for_data): Add tiling field.
31728 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31729 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31730 for auto loops. Remove default auto determining, moved to
31731 oacc_loop_fixed_partitions.
31732 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31733 stmts, add e_mask field.
31734 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31735 (oacc_thread_numbers): Use oacc_dim_call.
31736 (oacc_xform_tile): New.
31737 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31738 (finish_oacc_loop): Adjust for ifns vector.
31739 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31740 add case for GOACC_TILE fns.
31741 (oacc_loop_xform_loop): Delete.
31742 (oacc_loop_process): Iterate over call list directly, and add
31743 handling for GOACC_TILE fns.
31744 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31746 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31747 vector partitioning to outer loops. Assign 2 partitions to loops
31748 when available. Add TILE handling.
31749 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31750 (execite_oacc_device_lower): Process GOACC_TILE fns,
31751 ignore unknown specs.
31752 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31753 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31754 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31756 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31758 * configure.ac (ACX_BUGURL): Update.
31759 * configure: Regenerate.
31761 2017-02-09 Richard Biener <rguenther@suse.de>
31763 PR tree-optimization/69823
31764 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31765 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31767 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31769 * config/arc/arc-c.def: Add __NPS400__ definition.
31770 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31771 (TARGET_NPS400): Define.
31773 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31775 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31777 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31778 pointer, arch_info.
31779 (arc_cpu_types): Fill the arch_info field with a pointer into the
31780 arc_arch_types table.
31781 (arc_selected_cpu): Declare.
31782 * config/arc/arc.c (arc_selected_cpu): Make global.
31783 (arc_selected_arch): Delete.
31784 (arc_base_cpu): Delete.
31785 (arc_override_options): Remove references to deleted variables,
31786 update access to arch information.
31787 (ARC_OPT): Update access to arch information.
31788 (ARC_OPTX): Likewise.
31789 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31790 (TARGET_ARC600): Update access to arch information.
31791 (TARGET_ARC601): Likewise.
31792 (TARGET_ARC700): Likewise.
31793 (TARGET_EM): Likewise.
31794 (TARGET_HS): Likewise.
31795 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31798 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31801 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31802 condition/operands for integer GE/LE/GEU/LEU operations.
31804 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31806 PR translation/79397
31807 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31810 2017-02-08 Martin Jambor <mjambor@suse.cz>
31813 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31814 whether allocation happened.
31815 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31816 nothing was allocated.
31818 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31820 PR tree-optimization/79408
31821 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31822 constant, but SSA_NAME with a known integer range, use the minimum
31823 of that range instead of op1 to determine if modulo can be replaced
31824 with its first operand.
31826 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31828 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31830 2017-02-08 Richard Biener <rguenther@suse.de>
31832 PR tree-optimization/71824
31833 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31834 Check all loops contained in the merged region.
31836 2017-02-07 Andrew Pinski <apinski@cavium.com>
31838 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31840 2017-02-07 Andrew Pinski <apinski@cavium.com>
31842 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31843 (thunderxt88): Likewise.
31844 (thunderxt81): Disable LSE and change v8.1 to v8.
31845 (thunderxt83): Likewise.
31847 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31848 Richard Biener <rguenther@suse.de>
31850 PR middle-end/79399
31851 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31852 type from int to size_t.
31853 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31855 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31857 PR rtl-optimization/79386
31858 * cprop.c (bypass_conditional_jumps): Initialize
31859 bypass_last_basic_block already before splitting bbs after
31860 unconditional traps...
31861 (bypass_conditional_jumps): ... rather than here.
31864 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31865 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31866 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31867 fix -masm=intel patterns.
31869 2017-02-07 Richard Biener <rguenther@suse.de>
31871 PR tree-optimization/79256
31872 PR middle-end/79278
31873 * builtins.c (get_object_alignment_2): Use min_align_of_type
31874 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31875 and ADJUST_FIELD_ALIGN.
31877 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31879 * doc/tm.texi: Regenerate.
31880 * stor-layout.c (layout_decl): Adjust.
31881 (update_alignment_for_field): Likewise.
31882 (place_field): Likewise.
31883 (min_align_of_type): Likewise.
31884 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31885 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31886 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31887 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31888 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31889 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31890 * config/i386/i386.c (x86_field_alignment): Likewise.
31891 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31892 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31893 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31894 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31895 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31896 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31900 2017-01-30 Richard Biener <rguenther@suse.de>
31902 PR tree-optimization/79256
31903 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31904 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31907 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31909 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31910 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31911 builtins to SImode and emit a zero-extend, if necessary.
31913 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31915 * docs/invoke.texi (RISC-V Options): Alphabetize.
31917 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31919 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31922 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31924 * config/riscv/riscv.c: New file.
31925 * common/config/riscv/riscv-common.c: Likewise.
31926 * config.gcc: Likewise.
31927 * config/riscv/constraints.md: Likewise.
31928 * config/riscv/elf.h: Likewise.
31929 * config/riscv/generic.md: Likewise.
31930 * config/riscv/linux.h: Likewise.
31931 * config/riscv/multilib-generator: Likewise.
31932 * config/riscv/peephole.md: Likewise.
31933 * config/riscv/pic.md: Likewise.
31934 * config/riscv/predicates.md: Likewise.
31935 * config/riscv/riscv-builtins.c: Likewise.
31936 * config/riscv/riscv-c.c: Likewise.
31937 * config/riscv/riscv-ftypes.def: Likewise.
31938 * config/riscv/riscv-modes.def: Likewise.
31939 * config/riscv/riscv-opts.h: Likewise.
31940 * config/riscv/riscv-protos.h: Likewise.
31941 * config/riscv/riscv.h: Likewise.
31942 * config/riscv/riscv.md: Likewise.
31943 * config/riscv/riscv.opt: Likewise.
31944 * config/riscv/sync.md: Likewise.
31945 * config/riscv/t-elf-multilib: Likewise.
31946 * config/riscv/t-linux: Likewise.
31947 * config/riscv/t-linux-multilib: Likewise.
31948 * config/riscv/t-riscv: Likewise.
31949 * configure.ac: Likewise.
31950 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31951 Waterman as RISC-V maintainers.
31952 * doc/install.texi: Add RISC-V entries.
31953 * doc/invoke.texi: Add RISC-V options section.
31954 * doc/md.texi: Add RISC-V constraints section.
31955 * configure: Regenerated.
31957 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31960 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31961 false values to be constant vectors with all 0 or all 1 bits set.
31962 (vcondu<mode><mode>): Likewise.
31963 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31965 (fpmask_comparison_operator): Update comment.
31966 (vecint_comparison_operator): New predicate.
31967 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31968 vector conditionals when the true and false values are constant
31969 vectors with all 0 bits or all 1 bits set.
31971 2017-02-06 Martin Sebor <msebor@redhat.com>
31973 PR tree-optimization/79376
31974 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31976 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31978 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31979 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31980 to simplify split condition.
31982 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31984 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31985 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31988 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31990 PR rtl-optimization/68664
31991 * target.def (can_speculate_insn): New hook.
31992 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31993 * doc/tm.texi: Regenerate.
31994 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31995 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31996 (rs6000_sched_can_speculate_insn): New function.
31998 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32000 PR tree-optimization/79284
32001 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32002 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32003 vectorizable_mask_load_store, vectorizable_operation,
32004 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32005 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32006 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32007 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32008 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32009 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32010 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32011 is_gimple_assign (stmt). Replace another such test with
32012 is_gimple_assign (stmt).
32014 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32017 * config/avr/avr.c (rtl-iter.h): Include it.
32018 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32019 (avr_legitimate_combined_insn): ...and implementation.
32021 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32023 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32024 * config/s390/s390.c (s390_const_operand_ok)
32025 (s390_canonicalize_comparison, s390_extract_part)
32026 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32027 (s390_contiguous_bitmask_p, s390_rtx_costs)
32028 (legitimize_pic_address): Likewise.
32029 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32030 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32031 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32032 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32033 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32035 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32037 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32038 REGNO($0) == REGNO($1).
32040 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32042 * config/s390/linux.h(SIZE_TYPE): Add comment.
32044 2017-02-06 Julian Brown <julian@codesourcery.com>
32045 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32046 Virendra Pathak <virendra.pathak@broadcom.com>
32048 * config/aarch64/aarch64-cores.def: Change the scheduler
32050 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32051 * config/aarch64/thunderx2t99.md: New file.
32053 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32055 * doc/standards.texi (Go Language): Update link to language
32058 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32060 * tree-eh.c (lower_resx): Sanitize profile.
32061 (cleanup_empty_eh_move_lp): Likewise.
32063 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32066 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32068 * cfgloopmanip.h (loop_version): Update prototype.
32069 * modulo-sched.c (sms_schedule): Update call of loop_version.
32070 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32071 * tree-parloops.c (gen_parallel_loop): Likewise.
32072 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32073 * tree-ssa-loop-split.c (split_loop): Likewise.
32074 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32075 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32077 2017-02-05 Martin Liska <mliska@suse.cz>
32080 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32082 (print_operand_address): Initialize a struct to zero.
32084 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32086 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32087 garbage collector only in textual form.
32089 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32091 * doc/extend.texi (x86 specific memory model extensions for
32092 transactional memory): Simplify a phrase.
32094 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32097 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32098 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32099 (atomic_storedi_1): Likewise.
32101 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32103 PR tree-optimization/79338
32104 * tree-parloops.c (gather_scalar_reductions): Don't call
32105 vect_analyze_loop_form for loop->inner before destroying loop's
32108 2017-02-03 Martin Sebor <msebor@redhat.com>
32110 PR tree-optimization/79327
32111 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32112 when precision has resulted in leading zeros.
32113 (format_integer): Adjust the likely counter to assume an unknown
32114 argument that may be zero is non-zero.
32116 2017-02-03 Jason Merrill <jason@redhat.com>
32119 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32120 avoid copying non-taken branch.
32122 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32124 PR tree-optimization/79340
32125 * tree-vect-loop.c (vectorizable_reduction): Release
32126 vec_defs elements after safe_splicing them into other vectors.
32129 PR tree-optimization/79327
32130 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32131 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32133 (format_integer): Use wide_int_to_tree instead of build_int_cst
32134 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32135 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32136 of shortest and longest sequence.
32138 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32140 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32141 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32143 2017-02-03 Walter Lee <walt@tilera.com>
32146 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32147 after initial stackframe link reg save.
32148 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32150 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32153 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32154 wu for stxssp alternative.
32156 2017-02-03 Martin Sebor <msebor@redhat.com>
32158 PR tree-optimization/79352
32159 * gimple-fold.c (get_range_strlen): Add argument.
32160 (get_range_strlen): Change return type to bool.
32161 (get_maxval_strlen): Pass in a dummy argument.
32162 * gimple-fold.h (get_range_strlen): Change return type to bool.
32163 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32164 * tree.h (array_at_struct_end_p): Add argument.
32165 * tree.c (array_at_struct_end_p): Handle it.
32167 2017-02-03 Martin Liska <mliska@suse.cz>
32170 * multiple_target.c (create_dispatcher_calls): Redirect edge
32171 from a caller of a dispatcher.
32172 (expand_target_clones): Make the clones local.
32173 (ipa_target_clone): Do both target clones and resolvers.
32174 (ipa_dispatcher_calls): Remove the pass.
32175 (pass_dispatcher_calls::gate): Likewise.
32176 (make_pass_dispatcher_calls): Likewise.
32177 * passes.def (pass_target_clone): Put as very first IPA early
32180 2017-02-03 Martin Liska <mliska@suse.cz>
32182 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32183 in case of a function with ifunc attribute.
32185 2017-02-03 Martin Liska <mliska@suse.cz>
32187 * cgraph.c (cgraph_node::dump): Dump function version info.
32188 * symtab.c (symtab_node::dump_base): Add missing new line.
32190 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32192 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32193 (ifcombine_ifandif): Use it.
32195 2017-02-03 Martin Liska <mliska@suse.cz>
32197 * doc/invoke.texi: Document default value for
32198 use-after-scope-direct-emission-threshold.
32200 2017-02-03 Martin Liska <mliska@suse.cz>
32202 PR tree-optimization/79339
32203 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32204 (format_floating): Likewise.
32206 2017-02-03 Martin Liska <mliska@suse.cz>
32209 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32211 (ipa_node_params_t::remove): Likewise.
32212 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32213 initialization from removed ipa_node_params_t::insert.
32214 (ipa_node_params::~ipa_node_params): Move from removed
32215 ipa_node_params_t::release.
32216 * symbol-summary.h (symbol_summary::m_released): New member.
32217 Do not release a summary twice. Do not allow to call finalizer
32218 for types of a summary that live in GGC memory.
32220 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32222 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32225 2017-02-02 Martin Sebor <msebor@redhat.com>
32227 PR middle-end/79275
32228 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32229 (format_string): Tighten up the range of output for non-constant
32230 strings and correct the expected range for wide non-constant strings.
32232 2017-02-02 Martin Sebor <msebor@redhat.com>
32234 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32236 PR middle-end/32003
32237 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32239 (-fdump-tree-@var): Add to index and document how to come up
32240 with pass-specific option and dump file names.
32241 (-fdump-passes): Clarify where to look for output.
32243 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32245 PR middle-end/77445
32246 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32247 statistics of the analyzed path; allow threading for speed when
32248 any of BBs along the path are optimized for speed.
32250 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32252 PR middle-end/78468
32253 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32254 settings of the virtual registers.
32257 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32259 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32260 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32263 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32265 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32266 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32268 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32270 * config/s390/s390.md: Add missing comments with the expanded
32272 * config/s390/vector.md: Likewise.
32273 * config/s390/vx-builtins.md: Likewise.
32275 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32278 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32279 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32280 conditions on a single line.
32282 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32284 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32285 __S390_VX__ to __VX__.
32287 2017-02-01 Andrew Pinski <apinski@cavium.com>
32289 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32290 stmt_info to record_stmt_cost.
32291 (vect_get_known_peeling_cost): Pass stmt_info if known to
32293 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32294 cpu_vector_cost field into
32295 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32296 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32297 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32298 splitting of scalar_stmt_cost and vec_stmt_cost.
32299 (thunderx_vector_cost): Likewise.
32300 (cortexa57_vector_cost): LIkewise.
32301 (exynosm1_vector_cost): Likewise.
32302 (xgene1_vector_cost): Likewise.
32303 (thunderx2t99_vector_cost): Improve after the splitting of the two
32305 (aarch64_builtin_vectorization_cost): Update for the splitting of
32306 scalar_stmt_cost and vec_stmt_cost.
32308 2017-02-01 Torvald Riegel <triegel@redhat.com>
32309 Richard Henderson <rth@redhat.com>
32311 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32312 conditional on existance of a fast atomic load.
32313 * optabs-query.c (can_atomic_load_p): New function.
32314 * optabs-query.h (can_atomic_load_p): Declare it.
32315 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32316 no fast atomic load is available for the particular size of access.
32317 (expand_atomic_compare_and_swap): Likewise.
32318 (expand_atomic_load): Likewise.
32319 (expand_atomic_store): Likewise.
32320 (expand_atomic_fetch_op): Likewise.
32321 * testsuite/lib/target-supports.exp
32322 (check_effective_target_sync_int_128): Remove x86 because it provides
32323 no fast atomic load.
32324 (check_effective_target_sync_int_128_runtime): Likewise.
32326 2017-02-01 Richard Biener <rguenther@suse.de>
32328 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32329 (graphite_transform_loops): Provide opt-info for optimized nests.
32330 * tree-parloop.c (parallelize_loops): Provide opt-info for
32331 parallelized loops.
32333 2017-02-01 Richard Biener <rguenther@suse.de>
32335 PR middle-end/79315
32336 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32337 was not set before.
32339 2017-02-01 Richard Biener <rguenther@suse.de>
32341 PR tree-optimization/71824
32342 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32343 Verify the loops are valid in the merged SESE region.
32344 (scop_detection::can_represent_loop_1): Check analyzing the
32345 evolution of the number of iterations in the region succeeds.
32347 2017-01-31 Ian Lance Taylor <iant@golang.org>
32349 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32350 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32352 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32354 PR preprocessor/79210
32355 * input.c (get_substring_ranges_for_loc): Replace line_width
32356 assertion with error-handling.
32358 2017-01-31 Richard Biener <rguenther@suse.de>
32360 PR tree-optimization/77318
32361 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32362 (create_pw_aff_from_tree): Take loop parameter.
32363 (add_condition_to_pbb): Pass loop of the condition to
32364 create_pw_aff_from_tree.
32366 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32368 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32369 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32371 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32375 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32377 (convert_int_to_float128): Likewise.
32378 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32379 (convert_int_to_float128): Likewise.
32380 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32381 (UNSPEC_IEEE128_CONVERT): Likewise.
32382 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32383 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32384 Use local variables for IBM extended format.
32385 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32386 (fix_trunc<mode>si2_fprs): Likewise.
32387 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32388 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32389 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32390 to know that we can now have integers of all sizes in vector
32392 (fix<uns>_<mode>di2_hw): Likewise.
32393 (float<uns>_<mode>si2_hw): Likewise.
32394 (fix_<mode>si2_hw): Likewise.
32395 (fixuns_<mode>si2_hw): Likewise.
32396 (float<uns>_<mode>di2_hw): Likewise.
32397 (float_<mode>di2_hw): Likewise.
32398 (float_<mode>si2_hw): Likewise.
32399 (floatuns_<mode>di2_hw): Likewise.
32400 (floatuns_<mode>si2_hw): Likewise.
32401 (xscvqp<su>wz_<mode>): Delete, no longer used.
32402 (xscvqp<su>dz_<mode>): Likewise.
32403 (xscv<su>dqp_<mode>): Likewise.
32404 (ieee128_mfvsrd_64bit): Likewise.
32405 (ieee128_mfvsrd_32bit): Likewise.
32406 (ieee128_mfvsrwz): Likewise.
32407 (ieee128_mtvsrw): Likewise.
32408 (ieee128_mtvsrd_64bit): Likewise.
32409 (ieee128_mtvsrd_32bit): Likewise.
32411 2017-01-31 Martin Liska <mliska@suse.cz>
32414 * ipa-prop.c (ipa_free_all_node_params): Call release method
32415 instead of ~sumbol_summary to not to trigger double times
32418 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32420 PR tree-optimization/71691
32421 * bitmap.h (class auto_bitmap): New.
32422 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32423 is_maybe_undefined instead of ssa_undefined_value_p.
32425 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32427 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32428 __S390_ARCH_LEVEL__ to __ARCH__.
32430 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32432 PR tree-optimization/79267
32433 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32434 if should_remove_lhs_p is true.
32436 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32439 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32440 (add_alignment_attribute): New.
32441 (base_type_die): Add alignment attribute.
32442 (subrange_type_die): Likewise.
32443 (modified_type_die): Likewise.
32444 (gen_array_type_die): Likewise.
32445 (gen_descr_array_type_die: Likewise.
32446 (gen_enumeration_type_die): Likewise.
32447 (gen_subprogram_die): Likewise.
32448 (gen_variable_die): Likewise.
32449 (gen_field_die): Likewise.
32450 (gen_ptr_to_mbr_type_die): Likewise.
32451 (gen_struct_or_union_type_die): Likewise.
32452 (gen_subroutine_type_die): Likewise.
32453 (gen_typedef_die): Likewise.
32454 (base_type_cmp): Compare alignment attribute.
32456 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32459 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32460 (setb_unsigned) New pattern for setb with CCUNS.
32461 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32462 subfc./subfe sequence to avoid overflow problems. Generate a
32463 shorter sequence with cmpld/setb for power9.
32464 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32465 for generating subfc. instruction.
32466 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32467 now uses this instruction.
32469 2017-01-30 Ian Lance Taylor <iant@google.com>
32472 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32473 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32475 2017-01-30 Martin Sebor <msebor@redhat.com>
32477 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32478 Move constant to the right of a relational operator.
32479 (get_mpfr_format_length, format_character, format_string): Ditto.
32480 (should_warn_p, maybe_warn): Same.
32482 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32484 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32487 * asan.c (get_translation_unit_decl): Remove function.
32488 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32490 2017-01-30 Martin Liska <mliska@suse.cz>
32492 PR gcov-profile/79259
32493 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32494 -fprofile-generate.
32496 2017-01-30 Martin Liska <mliska@suse.cz>
32499 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32500 Initialize variables with NULL value.
32502 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32505 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32507 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32509 2017-01-30 Richard Biener <rguenther@suse.de>
32511 PR tree-optimization/79276
32512 * tree-vrp.c (process_assert_insertions): Properly adjust common
32513 when removing a duplicate.
32515 2017-01-30 Richard Biener <rguenther@suse.de>
32517 PR tree-optimization/79256
32518 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32519 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32521 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32523 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32526 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32527 ("*r<noxa>sbg_<mode>_sll_bitmask")
32528 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32529 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32530 Use contiguous_bitmask_nowrap_operand.
32532 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32535 * config/rs6000/altivec.h (vec_xl): Revise #define.
32536 (vec_xst): Likewise.
32538 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32540 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32542 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32544 PR rtl-optimization/79194
32545 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32546 traps before call to bypass_conditional_jumps.
32548 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32550 PR tree-optimization/71374
32551 * lra-constraints.c (check_conflict_input_operands): New.
32552 (match_reload): Use it.
32554 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32557 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32558 account to calculate conflict_set.
32560 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32562 PR rtl-optimization/78559
32563 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32564 other_insn in combine.
32566 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32568 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32569 uint16_type_node for BT_UINT16.
32571 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32573 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32574 "RTL Tests" to menu.
32575 (GIMPLE Tests): New node.
32576 (RTL Tests): New node.
32578 2017-01-27 Richard Biener <rguenther@suse.de>
32580 PR tree-optimization/79245
32581 * tree-loop-distribution.c (distribute_loop): Apply cost
32582 modeling also to detected patterns.
32584 2017-01-27 Richard Biener <rguenther@suse.de>
32586 PR tree-optimization/71433
32587 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32588 (compare_assert_loc): New function.
32589 (process_assert_insertions): Sort and optimize assert locations
32590 to remove duplicates and push down identical assertions on
32591 edges to their destination block.
32593 2017-01-27 Richard Biener <rguenther@suse.de>
32595 PR tree-optimization/79244
32596 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32597 out SSA names even if abnormal.
32599 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32601 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32602 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32603 instead of MPFR_RNDN.
32605 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32608 * arm.c (arm_option_override): Don't call build_target_option_node
32609 until after doing all option overrides.
32610 (arm_valid_target_attribute_tree): Likewise.
32612 2017-01-27 Martin Liska <mliska@suse.cz>
32614 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32615 for {cd}tors and C++ {cd}tors.
32617 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32619 * config/s390/s390.md ("*setmem_long_and")
32620 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32622 2017-01-26 Martin Sebor <msebor@redhat.com>
32624 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32627 2017-01-26 Martin Sebor <msebor@redhat.com>
32629 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32630 HAVE_DFmode before using XFmode or DFmode.
32631 (parse_directive): Avoid using the z length modifier to avoid
32632 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32634 PR middle-end/78703
32635 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32636 to accept adjustment as an array.
32637 (get_int_range): New function.
32638 (struct directive): Make width and prec arrays.
32639 (directive::set_width, directive::set_precision): Call get_int_range.
32640 (format_integer, format_floating): Handle width and precision ranges.
32641 (format_string, parse_directive): Same.
32643 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32646 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32647 comdat_type_p set, just clone them, but keep the children in the
32651 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32652 which have direct callers with -fvar-tracking-assignments enabled
32654 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32655 inside of type units.
32657 2017-01-26 Martin Sebor <msebor@redhat.com>
32659 PR middle-end/78703
32660 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32662 (struct format_result): Replace number_chars, number_chars_min,
32663 and number_chars_max with a single member of struct result_range.
32665 (format_result::operator+=): Adjust.
32666 (struct fmtresult): Remove bounded. Handle likely and unlikely
32668 (fmtresult::adjust_for_width_or_precision): New function.
32669 (fmtresult:type_max_digits): New function.
32670 (bytes_remaining): Handle likely and unlikely counters.
32671 (min_bytes_remaining): Remove.
32672 (format_percent): Simplify.
32673 (format_integer, format_floating): Set likely and unlikely counters.
32674 (get_string_length, format_character, format_string): Same.
32675 (format_plain, should_warn_p): New function.
32676 (maybe_warn): Call should_warn_p. Update diagnostic messages
32677 and handle those for all directives, including plain strings.
32678 (format_directive): Handle likely and unlikely counters.
32679 Remove unnecessary quoting from diagnostics. Add an informational
32681 (add_bytes): Remove.
32682 (pass_sprintf_length::compute_format_length): Simplify.
32683 (try_substitute_return_value): Handle likely and unlikely counters.
32685 2017-01-26 Carl Love <cel@us.ibm.com>
32687 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32688 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32690 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32693 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32694 endianess for subregs into account.
32695 * lra-constraints.c (lra_constraints): Do risky transformations
32696 always on the first iteration.
32697 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32698 last_call_used_reg_set.
32699 (process_bb_lives): Define and use last_call_used_reg_set.
32700 * lra.c (lra): Always continue after lra_constraints on the first
32703 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32705 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32707 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32709 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32711 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32712 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32713 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32714 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32715 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32716 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32717 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32718 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32719 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32721 2017-01-26 Marek Polacek <polacek@redhat.com>
32724 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32725 for the third operand.
32727 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32729 PR middle-end/79236
32730 * omp-low.c (struct omp_context): Add simt_stmt field.
32731 (scan_omp_for): Return omp_context *.
32732 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32733 context to the _simt_ SIMD stmt.
32734 (lower_omp_for): For combined SIMD with sibling _simt_
32735 SIMD, make sure to use the same decls in _looptemp_
32736 clauses as in the sibling.
32738 2017-01-26 David Sherwood <david.sherwood@arm.com>
32740 PR middle-end/79212
32741 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32744 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32747 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32748 emit fld b; fld a; if possible.
32750 * brig-builtins.def: Update copyright years.
32751 * config/arm/arm_acle_builtins.def: Update copyright years.
32753 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32756 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32757 constraint instead of o for the stxsd instruction.
32759 2017-01-25 Carl Love <cel@us.ibm.com>
32761 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32762 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32764 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32766 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32768 2017-01-25 Richard Biener <rguenther@suse.de>
32770 PR tree-optimization/69264
32771 * target.def (vector_alignment_reachable): Improve documentation.
32772 * doc/tm.texi: Regenerate.
32773 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32775 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32776 earlier changes with respect to TYPE_USER_ALIGN.
32777 (vector_alignment_reachable_p): Likewise. Improve dumping.
32779 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32782 * config/arm/arm.md (xordi3): Force constant operand into a register
32785 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32787 * doc/invoke.texi (-fstore-merging): Correct default optimization
32788 levels at which it is enabled.
32789 (-O): Move -fstore-merging from list to...
32792 2017-01-25 Richard Biener <rguenther@suse.de>
32795 * omp-expand.c: Include debug.h.
32796 (expand_omp_taskreg): Make sure to generate early debug before
32797 outlining anything from a function.
32798 (expand_omp_target): Likewise.
32799 (grid_expand_target_grid_body): Likewise.
32801 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32804 * asan.c (get_translation_unit_decl): New function.
32805 (asan_add_global): Extract modules file name from globals
32806 TRANSLATION_UNIT_DECL name.
32808 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32811 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32812 for long calls with APCS frame and VFP.
32814 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32816 * cfg.c (original_copy_tables_initialized_p): New function.
32817 * cfg.h (original_copy_tables_initialized_p): New decl.
32818 * cfgrtl.c (relink_block_chain): Guard the call to
32819 free_original_copy_tables with a call to
32820 original_copy_tables_initialized_p.
32821 * cgraph.h (symtab_node::native_rtl_p): New decl.
32822 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32823 (symtab_node::needed_p): Don't assert for early assembly output
32824 for __RTL functions.
32825 (cgraph_node::finalize_function): Set "force_output" for __RTL
32827 (cgraph_node::analyze): Bail out early for __RTL functions.
32828 (analyze_functions): Update assertion to support __RTL functions.
32829 (cgraph_node::expand): Bail out early for __RTL functions.
32830 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32832 * function.h (struct function): Update comment for field
32834 * gimple-expr.c: Include "tree-pass.h".
32835 (gimple_has_body_p): Return false for __RTL functions.
32836 * Makefile.in (OBJS): Add run-rtl-passes.o.
32837 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32839 (gcc::pass_manager::get_clean_slate): New accessor.
32840 * passes.c: Include "insn-addr.h".
32841 (should_skip_pass_p): Add logging. Update logic for running
32842 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32843 property-provider override so it is only done for gimple passes.
32845 (skip_pass): New function.
32846 (execute_one_pass): Call skip_pass when skipping passes.
32847 * read-md.c (md_reader::read_char): Support filtering
32848 the input to a subset of line numbers.
32849 (md_reader::md_reader): Initialize fields
32850 m_first_line and m_last_line.
32851 (md_reader::read_file_fragment): New function.
32852 * read-md.h (md_reader::read_file_fragment): New decl.
32853 (md_reader::m_first_line): New field.
32854 (md_reader::m_last_line): New field.
32855 * read-rtl-function.c (function_reader::create_function): Only
32856 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32857 curr_properties. Set DECL_INITIAL to a dummy block.
32858 (read_rtl_function_body_from_file_range): New function.
32859 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32861 * run-rtl-passes.c: New file.
32862 * run-rtl-passes.h: New file.
32864 2017-01-24 Jeff Law <law@redhat.com>
32866 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32869 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32871 PR tree-optimization/79159
32872 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32873 (record_nonwrapping_iv): Improve boundary using above function if no
32874 value range information.
32876 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32877 Martin Jambor <mjambor@suse.cz>
32879 * brig-builtins.def: New file.
32880 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32881 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32882 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32883 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32884 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32885 * builtin-types.def (BT_INT8): New.
32886 (BT_INT16): Likewise.
32887 (BT_UINT8): Likewise.
32888 (BT_UINT16): Likewise.
32889 (BT_FN_ULONG): Likewise.
32890 (BT_FN_UINT_INT): Likewise.
32891 (BT_FN_UINT_ULONG): Likewise.
32892 (BT_FN_UINT_LONG): Likewise.
32893 (BT_FN_UINT_PTR): Likewise.
32894 (BT_FN_ULONG_PTR): Likewise.
32895 (BT_FN_INT8_FLOAT): Likewise.
32896 (BT_FN_INT16_FLOAT): Likewise.
32897 (BT_FN_UINT32_FLOAT): Likewise.
32898 (BT_FN_UINT16_FLOAT): Likewise.
32899 (BT_FN_UINT8_FLOAT): Likewise.
32900 (BT_FN_UINT64_FLOAT): Likewise.
32901 (BT_FN_UINT16_UINT32): Likewise.
32902 (BT_FN_UINT32_UINT16): Likewise.
32903 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32904 (BT_FN_INT_PTR_INT): Likewise.
32905 (BT_FN_UINT_PTR_UINT): Likewise.
32906 (BT_FN_LONG_PTR_LONG): Likewise.
32907 (BT_FN_ULONG_PTR_ULONG): Likewise.
32908 (BT_FN_VOID_UINT64_UINT64): Likewise.
32909 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32910 (BT_FN_INT8_INT8_INT8): Likewise.
32911 (BT_FN_INT16_INT16_INT16): Likewise.
32912 (BT_FN_INT_INT_INT): Likewise.
32913 (BT_FN_UINT_FLOAT_UINT): Likewise.
32914 (BT_FN_FLOAT_UINT_UINT): Likewise.
32915 (BT_FN_ULONG_UINT_UINT): Likewise.
32916 (BT_FN_ULONG_UINT_PTR): Likewise.
32917 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32918 (BT_FN_UINT_UINT_UINT): Likewise.
32919 (BT_FN_VOID_UINT_PTR): Likewise.
32920 (BT_FN_UINT_UINT_PTR: Likewise.
32921 (BT_FN_UINT32_UINT64_PTR): Likewise.
32922 (BT_FN_INT_INT_UINT_UINT): Likewise.
32923 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32924 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32925 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32926 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32927 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32928 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32929 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32930 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32931 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32932 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32933 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32934 * doc/frontends.texi: List BRIG FE.
32935 * doc/install.texi (Testing): Add BRIG tesring requirements.
32936 * doc/invoke.texi (Overall Options): Mention BRIG.
32937 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32939 2017-01-24 Richard Biener <rguenther@suse.de>
32941 PR translation/79208
32942 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32944 2017-01-24 Martin Jambor <mjambor@suse.cz>
32947 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32948 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32949 and known_contexts.
32951 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32953 PR middle-end/79123
32954 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32955 casts from signed to unsigned really don't have a range.
32957 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32959 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32960 GMP_RNDx for compatiblity.
32962 2017-01-24 Martin Liska <mliska@suse.cz>
32965 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32966 that would prevent us to call alloca with -1 as argument.
32968 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32970 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32971 Avoid -Wformat-security warning.
32973 2017-01-23 Andrew Pinski <apinski@cavium.com>
32975 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32978 2017-01-23 Martin Sebor <msebor@redhat.com>
32980 PR middle-end/78703
32981 * gimple-ssa-sprintf.c (warn_level): New global.
32982 (format_integer): Use it here and throughout the rest of the file.
32983 Use the same switch to compute sign as base.
32984 (maybe_warn): New function.
32985 (format_directive): Factor out warnings into maybe_warn.
32986 Add debugging output. Use warn_level.
32987 (add_bytes): Use warn_level.
32988 (pass_sprintf_length::compute_format_length): Add debugging output.
32989 (try_substitute_return_value): Same.
32990 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32992 PR middle-end/78703
32993 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32994 (struct fmtresult, format_integer, format_floating): Adjust.
32995 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32996 (get_string_length, format_string,format_directive): Same.
32997 (pass_sprintf_length::compute_format_length): Same.
32998 (try_substitute_return_value): Simplify slightly.
33000 PR middle-end/78703
33001 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33002 (fmtresult::operator+=): Outlined.
33003 (struct fmtresult): Add ctors.
33004 (struct conversion_spec): Rename...
33005 (struct directive): ...to this. Add and remove data members.
33006 (directive::set_width, directive::set_precision): New functions.
33007 (format_percent): Use fmtresult ctor.
33008 (get_width_and_precision): Remove.
33009 (format_integer): Make naming changes. Avoid computing width and
33011 (format_floating): Same. Adjust indentation.
33012 (format_character, format_none): New functions.
33013 (format_string): Moved character handling to format_character.
33014 (format_directive): Remove arguments, change return type.
33015 (parse_directive): New function.
33016 (pass_sprintf_length::compute_format_length): Move directive
33017 parsing to parse_directive.
33019 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33021 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33022 (assign_assembler_name_if_needed): ... this.
33023 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33024 (assign_assembler_name_if_needed): ... this.
33025 (free_lang_data_in_cgraph): Adjust callers.
33026 * cgraphunit.c (cgraph_node::analyze): Likewise.
33027 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33030 2017-01-23 Richard Biener <rguenther@suse.de>
33032 PR tree-optimization/79088
33033 PR tree-optimization/79188
33034 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33035 resetting loop bounds after last path deletion. Reset loop
33036 bounds of the target loop, make code match the comments.
33037 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33038 Make sure loops need no fixups.
33040 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33042 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33043 exponent support with double type for first argument.
33044 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33045 type returned by __builtin_vec_extract_sig,
33046 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33047 functions from "vector int" to "vector unsigned int" or from
33048 "vector long long int" to "vector unsigned long long int".
33049 Changed type returned by __builtin_vec_extract_exp,
33050 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33051 functions from "vector int" to "vector unsigned int" or from
33052 "vector long long int" to "vector unsigned long long int".
33053 Changed return type of __builtin_vec_test_data_class,
33054 __builtin_vec_test_data_class_sp, and
33055 __builtin_vec_test_data_class_dp from "vector int" to
33056 "vector bool int" or from "vector long long int" to "vector bool
33057 long long int" and changed second argument type from "unsigned
33058 int" to "int". Added new overloaded function forms "vector float
33059 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33060 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33061 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33062 double, vector unsigned long long int)" and "vector double
33063 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33064 long int)". Changed return type of
33065 __builtin_scalar_test_data_class and
33066 __builtin_scalar_test_data_class_sp and
33067 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33068 int" and changed second argument from "unsigned int" to "int".
33069 Changed type returned by __builtin_scalar_test_neg,
33070 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33071 from "int" to "bool int". Added new overloaded function form
33072 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33073 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33074 exponent double-precision with floating point first argument.
33075 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33076 documentation of scalar_test_data_class, scalar_test_neg,
33077 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33078 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33079 vec_test_data_class built-in functions to reflect refinements in
33080 their type signatures.
33082 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33084 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33086 (aarch64_elf_asm_destructor): Likewise.
33088 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33090 PR rtl-optimization/78634
33091 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33092 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33093 * ifcvt.c (noce_try_cmove): Add missing cost check.
33095 PR rtl-optimization/71724
33096 * combine.c (if_then_else_cond): Look for situations where it is
33097 beneficial to undo the work of one of the recursive calls.
33099 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33101 PR tree-optimization/70754
33102 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33103 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33104 combined stmt before it if not NULL.
33105 (combine_chains): Process refs reversely and compute dominance point
33108 2017-01-23 Martin Liska <mliska@suse.cz>
33110 PR tree-optimization/79196
33111 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33112 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33114 (strlen_optimize_stmt): Call the renamed function.
33116 2017-01-23 Michael Matz <matz@suse.de>
33118 PR tree-optimization/78384
33119 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33121 2017-01-23 Richard Biener <rguenther@suse.de>
33123 PR tree-optimization/79186
33124 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33125 both incoming edges before moving an assert.
33127 2017-01-23 Martin Jambor <mjambor@suse.cz>
33129 * ipa-prop.c (load_from_param_1): Removed.
33130 (load_from_unmodified_param): Bits from load_from_param_1 put back
33132 (load_from_param): Removed.
33133 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33134 with stmt. Reverted back to use of load_from_unmodified_param.
33136 2017-01-23 Martin Jambor <mjambor@suse.cz>
33139 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33140 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33141 field a pointer to garbage collected vector, mark lattices and
33142 ipcp_orig_node with GTY((skip)).
33143 (ipa_get_param_count): Adjust to descriptors being a pointer.
33144 (ipa_get_param): Likewise.
33145 (ipa_get_type): Likewise.
33146 (ipa_get_param_move_cost): Likewise.
33147 (ipa_set_param_used): Likewise.
33148 (ipa_get_controlled_uses): Likewise.
33149 (ipa_set_controlled_uses): Likewise.
33150 (ipa_is_param_used): Likewise.
33151 (ipa_node_params_t): Move into garbage collector. New methods insert
33153 (ipa_node_params_sum): Annotate wth GTY(()).
33154 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33156 (ipa_load_from_parm_agg): Adjust declaration.
33157 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33158 * ipa-profile.c (ipa_profile): Likewise.
33159 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33160 (ipa_populate_param_decls): Make descriptors parameter garbage
33162 (ipa_dump_param): Adjust to descriptors being a pointer.
33163 (ipa_alloc_node_params): Likewise.
33164 (ipa_initialize_node_params): Likewise.
33165 (load_from_param_1): Make descriptors parameter garbage collected.
33166 (load_from_unmodified_param): Likewise.
33167 (load_from_param): Likewise.
33168 (ipa_load_from_parm_agg): Likewise.
33169 (ipa_node_params::~ipa_node_params): Removed.
33170 (ipa_free_all_node_params): Remove call to delete operator.
33171 (ipa_node_params_t::insert): New.
33172 (ipa_node_params_t::remove): Likewise.
33173 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33174 copy known_csts and known_contexts vectors.
33175 (ipa_read_node_info): Adjust to descriptors being a pointer.
33176 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33178 (ipcp_transform_function): Make descriptors variable garbage
33181 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33183 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33184 * config/i386/avx512dqintrin.h: Ditto.
33185 * config/i386/avx512fintrin.h: Ditto.
33186 * config/i386/i386.c: Handle new builtins.
33187 * config/i386/i386-builtin.def: Add new builtins.
33188 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33189 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33191 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33192 Martin Liska <mliska@suse.cz>
33194 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33195 * asan.c (asan_expand_poison_ifn): Support stores and use
33196 appropriate ASAN report function.
33197 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33198 * internal-fn.def (ASAN_POISON_USE): Declare.
33199 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33200 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33201 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33202 ASAN_POISON calls w/o LHS.
33203 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33204 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33205 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33206 * gimplify.c (asan_poison_variables): Add attribute
33207 use_after_scope_memory to variables that really needs to live
33209 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33210 having the attribute.
33212 2017-01-23 Martin Liska <mliska@suse.cz>
33214 * asan.c (create_asan_shadow_var): New function.
33215 (asan_expand_poison_ifn): Likewise.
33216 * asan.h (asan_expand_poison_ifn): New declaration.
33217 * internal-fn.c (expand_ASAN_POISON): Likewise.
33218 * internal-fn.def (ASAN_POISON): New builtin.
33219 * sanopt.c (pass_sanopt::execute): Expand
33220 asan_expand_poison_ifn.
33221 * tree-inline.c (copy_decl_for_dup_finish): Make function
33223 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33224 * tree-ssa.c (is_asan_mark_p): New function.
33225 (execute_update_addresses_taken): Rewrite local variables
33226 (identified just by use-after-scope as addressable) into SSA.
33228 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33230 * doc/install.texi (Specific): opensource.apple.com uses https
33231 now. Remove trailing slash.
33233 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33235 * README.Portability: Remove note on an Irix compatibility issue.
33237 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33239 * gcov.c (INCLUDE_ALGORITHM): Define.
33240 (INCLUDE_VECTOR): Define.
33241 No longer include <vector> and <algorithm> directly.
33243 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33245 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33247 * doc/invoke.texi (Code Gen Options): Ditto.
33249 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33252 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33254 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33256 rtl-optimization/79125
33257 * cprop.c (local_cprop_pass): Handle cases where we make an
33258 unconditional trap.
33260 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33264 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33265 read from, for big endian.
33267 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33269 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33270 register pauth builtins for LP64 only.
33272 2017-01-20 Marek Polacek <polacek@redhat.com>
33275 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33278 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33280 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33282 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33283 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33284 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33286 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33289 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33290 in big-endian mode when they are not a single duplicated value.
33292 2017-01-20 Richard Biener <rguenther@suse.de>
33294 * BASE-VER: Bump to 7.0.1.
33296 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33298 * omp-low.c (omplow_simd_context): New struct. Use it...
33299 (lower_rec_simd_input_clauses): ...here and...
33300 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33301 references to idx, lane, max_vf, is_simt.
33303 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33305 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33308 2017-01-20 Martin Jambor <mjambor@suse.cz>
33310 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33311 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33313 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33314 (GTFILES): Rename hsa.c to hsa-common.c.
33315 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33316 * hsa-dump.c: Likewise.
33317 * hsa-gen.c: Likewise.
33318 * hsa-regalloc.c: Likewise.
33319 * ipa-hsa.c: Likewise.
33320 * omp-expand.c: Likewise.
33321 * omp-low.c: Likewise.
33322 * toplev.c: Likewise.
33324 2017-01-20 Marek Polacek <polacek@redhat.com>
33327 * doc/invoke.texi: Document -Wduplicated-branches.
33328 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33329 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33330 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33331 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33332 return 0 only when not OEP_LEXICOGRAPHIC.
33333 (fold_build_cleanup_point_expr): Use the expression
33334 location when building CLEANUP_POINT_EXPR.
33335 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33336 * tree.c (add_expr): Handle error_mark_node.
33338 2017-01-20 Martin Liska <mliska@suse.cz>
33341 * tree-profile.c (init_ic_make_global_vars): Do not call
33343 (gimple_init_gcov_profiler): Likewise.
33345 2017-01-20 Martin Liska <mliska@suse.cz>
33348 * cgraph.h (maybe_create_reference): Remove argument and
33350 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33352 * ipa-cp.c (create_specialized_node): Likewise.
33353 * symtab.c (symtab_node::maybe_create_reference): Handle
33354 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33356 2017-01-20 Martin Liska <mliska@suse.cz>
33358 * read-rtl-function.c (function_reader::create_function): Use
33359 build_decl instread of build_decl_stat.
33361 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33363 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33364 * config/i386/avx512dqintrin.h: Ditto.
33365 * config/i386/avx512fintrin.h: Ditto.
33366 * config/i386/i386-builtin-types.def: Add new types.
33367 * config/i386/i386.c: Handle new types.
33368 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33369 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33370 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33371 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33372 (__builtin_ia32_kshiftridi): New.
33373 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33375 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33379 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33380 define to rs6000_init_stack_protect_guard.
33381 (rs6000_init_stack_protect_guard): New function.
33383 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33384 Yunqiang Su <yunqiang.su@imgtec.com>
33386 * config.gcc (supported_defaults): Add madd4.
33387 (with_madd4): Add validation.
33388 (all_defaults): Add madd4.
33389 * config/mips/mips.opt (mmadd4): New option.
33390 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33392 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33394 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33395 (ISA_HAS_FUSED_MADD4): Likewise.
33396 * doc/invoke.texi (-mmadd4): Document the new option.
33397 * doc/install.texi (--with-madd4): Document the new option.
33399 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33401 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33402 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33403 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33404 (aarch64_init_pauth_hint_builtins): New.
33405 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33406 (aarch64_expand_builtin): Expand new builtins.
33408 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33410 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33411 * combine-stack-adj.c (no_unhandled_cfa): Handle
33412 REG_CFA_TOGGLE_RA_MANGLE.
33413 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33414 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33415 info for return address signing.
33416 (aarch64_expand_epilogue): Likewise.
33418 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33420 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33421 * config/aarch64/aarch64-protos.h
33422 (aarch64_return_address_signing_enabled): New declaration.
33423 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33425 (aarch64_expand_prologue): Sign return address before it's pushed onto
33427 (aarch64_expand_epilogue): Authenticate return address fetched from
33429 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33430 (aarch64_attributes): New function attributes for "sign-return-address".
33431 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33432 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33433 ("*do_return"): Generate combined instructions according to key index.
33434 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33435 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33437 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33438 * config/aarch64/aarch64.opt (msign-return-address=): New.
33439 * doc/extend.texi (AArch64 Function Attributes): Documents
33440 "sign-return-address=".
33441 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33443 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33445 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33446 overall option summary.
33448 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33450 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33451 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33452 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33453 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33455 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33457 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33458 -mpower9-minmax by default for -mcpu=power9.
33459 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33460 128-bit floating point.
33462 2017-01-20 Alan Modra <amodra@gmail.com>
33464 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33465 optimizing for size.
33467 2017-01-20 Alan Modra <amodra@gmail.com>
33470 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33471 for strcmp and strncmp from corresponding builtin decl.
33473 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33475 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33476 instead of i386/rtems-64.h.
33477 * config/i386/rtems-64.h: Remove.
33479 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33483 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33485 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33487 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33489 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33490 Change int to HOST_WIDE_INT.
33491 * config/aarch64/aarch64-protos.h
33492 (aarch64_simd_gen_const_vector_dup): Likewise.
33493 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33495 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33497 * langhooks-def.h (lhd_type_for_size): New decl.
33498 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33499 * langhooks.c (lhd_type_for_size): New function, taken from
33502 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33504 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33505 define_bypass for CR latency.
33506 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33507 (power9-alu2): Add define_bypass for CR latency.
33509 (power9-mul): Update insn latency.
33510 (power9-mul-compare): Update insn latency, bypass latency and remove
33513 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33515 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33517 * config/aarch64/aarch64.md
33518 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33519 aarch64_nopcrelative_literal_loads.
33520 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33522 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33524 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33525 TARGET_LOONGSON_3A.
33526 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33528 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33531 * config.gcc (supported_defaults): Add lxc1-sxc1.
33532 (with_lxc1_sxc1): Add validation.
33533 (all_defaults): Add lxc1-sxc1.
33534 * config/mips/mips.opt (mlxc1-sxc1): New option.
33535 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33537 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33538 __mips_no_lxc1_sxc1.
33539 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33540 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33541 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33543 2017-01-19 Richard Biener <rguenther@suse.de>
33545 PR tree-optimization/72488
33546 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33547 sure to restore SSA info.
33548 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33550 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33552 PR rtl-optimization/79121
33553 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33554 of the inner type when shifting an extended value.
33556 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33559 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33560 interposable aliases.
33562 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33565 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33566 Use the evmergelohi instruction.
33567 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33568 (mov_sitf_e500_subreg8_2_be): Likewise.
33569 (mov_sitf_e500_subreg12_2_le): Likewise.
33570 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33571 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33572 (mov_sitf_e500_subreg8_2_le): Likewise.
33573 (mov_sitf_e500_subreg12_2_be): Likewise.
33575 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33577 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33578 attribute from vecsimple to vecperm.
33579 (altivec_vbpermq2): Likewise.
33581 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33584 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33586 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33587 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33588 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33589 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33590 case where N arg is SIZE_MAX.
33591 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33592 (cmpstrsi): Add pattern.
33594 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33596 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33597 __builtin_vec_revb builtins.
33598 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33599 built-in functions to support generation of the ISA 3.0 XXBR<x>
33600 vector byte reverse instructions.
33601 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33602 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33603 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33604 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33605 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33606 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33607 (P9V_BUILTIN_VEC_REVB): Likewise.
33608 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33609 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33610 (p9_xxbrq_v16qi): Likewise.
33611 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33612 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33613 (p9_xxbrh_v8hi): Likewise.
33614 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33615 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33616 vec_revb built-in functions.
33618 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33620 PR rtl-optimization/78952
33621 * config/i386/i386.md (any_extract): New code iterator.
33622 (*insvqi_2): Use any_extract for source operand.
33623 (*insvqi_3): Use any_shiftrt for source operand.
33625 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33627 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33629 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33631 2017-01-18 Matthias Klose <doko@ubuntu.com>
33633 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33635 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33637 * config/rs6000/altivec.h (vec_bperm): Change #define.
33638 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33639 (altivec_vbpermq2): New define_insn.
33640 (altivec_vbpermd): Likewise.
33641 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33642 function interface.
33643 (VBPERMD): Likewise.
33644 (VBPERM): New polymorphic function interface.
33645 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33646 Add entries for P9V_BUILTIN_VEC_VBPERM.
33647 * doc/extend.texi: Add interfaces for vec_bperm.
33649 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33651 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33652 first letter of error messages.
33653 (s390_resolve_overloaded_builtin): Likewise.
33654 * config/s390/s390.c (s390_expand_builtin): Likewise.
33655 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33656 (s390_valid_target_attribute_inner_p): Likewise.
33657 * config/s390/s390.md ("tabort"): Likewise.
33659 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33661 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33662 (ISA_AVOID_DIV_HILO): New macro.
33663 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33664 (ISA_HAS_DDIV): Likewise.
33666 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33668 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33670 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33672 * doc/invoke.texi (fabi-version): Spelling fix.
33674 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33677 * doc/invoke.texi (fabi-version): Mention mangling fix for
33680 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33683 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33685 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33688 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33689 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33691 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33693 (stack_protect_test): Ditto.
33694 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33695 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33697 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33698 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33699 -mstack-protector-guard-offset=.
33700 (RS/6000 and PowerPC Options): Ditto.
33702 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33704 * config/i386/i386.h (MASK_CLASS_P): New define.
33705 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33706 there are no registers from different register sets also when
33707 mask registers are used. Update function comment.
33708 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33709 to (*k/*r) and (*k/*km) alternatives.
33711 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33713 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33714 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33715 (EH_RETURN_HANDLER_RTX): New define.
33716 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33717 Force frame pointer in EH return functions.
33718 (aarch64_expand_epilogue): Add barrier for eh_return.
33719 (aarch64_final_eh_return_addr): Remove.
33720 (aarch64_eh_return_handler_rtx): New function.
33721 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33723 (aarch64_eh_return_handler_rtx): New prototype.
33725 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33727 * config/rs6000/altivec.h (vec_rlmi): New #define.
33728 (vec_vrlnm): Likewise.
33729 (vec_rlnm): Likewise.
33730 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33731 (UNSPEC_VRLNM): Likewise.
33732 (VIlong): New mode iterator.
33733 (altivec_vrl<VI_char>mi): New define_insn.
33734 (altivec_vrl<VI_char>nm): Likewise.
33735 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33737 (VRLDNM): Likewise.
33738 (RLNM): New polymorphic function entry.
33739 (VRLWMI): New monomorphic function entry.
33740 (VRLDMI): Likewise.
33741 (RLMI): New polymorphic function entry.
33742 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33743 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33744 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33747 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33750 * dwarf2out.c (field_byte_offset): Restore the
33751 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33752 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33753 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33756 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33759 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33761 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33764 * dwarf2out.c (add_data_member_location_attribute): For constant
33765 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33766 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33767 DW_AT_byte_size attributes.
33769 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33771 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33772 after forcing to constant memory when the code model is medium.
33774 2017-01-17 Julia Koval <julia.koval@intel.com>
33777 * config/i386/avx512fintrin.h
33778 (_mm512_i32gather_ps): Change __addr type to void const*.
33779 (_mm512_mask_i32gather_ps): Ditto.
33780 (_mm512_i32gather_pd): Ditto.
33781 (_mm512_mask_i32gather_pd): Ditto.
33782 (_mm512_i64gather_ps): Ditto.
33783 (_mm512_mask_i64gather_ps): Ditto.
33784 (_mm512_i64gather_pd): Ditto.
33785 (_mm512_mask_i64gather_pd): Ditto.
33786 (_mm512_i32gather_epi32): Ditto.
33787 (_mm512_mask_i32gather_epi32): Ditto.
33788 (_mm512_i32gather_epi64): Ditto.
33789 (_mm512_mask_i32gather_epi64): Ditto.
33790 (_mm512_i64gather_epi32): Ditto.
33791 (_mm512_mask_i64gather_epi32): Ditto.
33792 (_mm512_i64gather_epi64): Ditto.
33793 (_mm512_mask_i64gather_epi64): Ditto.
33794 (_mm512_i32scatter_ps): Change __addr type to void*.
33795 (_mm512_mask_i32scatter_ps): Ditto.
33796 (_mm512_i32scatter_pd): Ditto.
33797 (_mm512_mask_i32scatter_pd): Ditto.
33798 (_mm512_i64scatter_ps): Ditto.
33799 (_mm512_mask_i64scatter_ps): Ditto.
33800 (_mm512_i64scatter_pd): Ditto.
33801 (_mm512_mask_i64scatter_pd): Ditto.
33802 (_mm512_i32scatter_epi32): Ditto.
33803 (_mm512_mask_i32scatter_epi32): Ditto.
33804 (_mm512_i32scatter_epi64): Ditto.
33805 (_mm512_mask_i32scatter_epi64): Ditto.
33806 (_mm512_i64scatter_epi32): Ditto.
33807 (_mm512_mask_i64scatter_epi32): Ditto.
33808 (_mm512_i64scatter_epi64): Ditto.
33809 (_mm512_mask_i64scatter_epi64): Ditto.
33810 * config/i386/avx512pfintrin.h
33811 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33812 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33813 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33814 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33815 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33816 (_mm512_prefetch_i32scatter_ps): Ditto.
33817 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33818 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33819 (_mm512_prefetch_i64scatter_pd): Ditto.
33820 (_mm512_prefetch_i64scatter_ps): Ditto.
33821 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33822 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33823 * config/i386/avx512vlintrin.h
33824 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33825 (_mm_mmask_i32gather_ps): Ditto.
33826 (_mm256_mmask_i32gather_pd): Ditto.
33827 (_mm_mmask_i32gather_pd): Ditto.
33828 (_mm256_mmask_i64gather_ps): Ditto.
33829 (_mm_mmask_i64gather_ps): Ditto.
33830 (_mm256_mmask_i64gather_pd): Ditto.
33831 (_mm_mmask_i64gather_pd): Ditto.
33832 (_mm256_mmask_i32gather_epi32): Ditto.
33833 (_mm_mmask_i32gather_epi32): Ditto.
33834 (_mm256_mmask_i32gather_epi64): Ditto.
33835 (_mm_mmask_i32gather_epi64): Ditto.
33836 (_mm256_mmask_i64gather_epi32): Ditto.
33837 (_mm_mmask_i64gather_epi32): Ditto.
33838 (_mm256_mmask_i64gather_epi64): Ditto.
33839 (_mm_mmask_i64gather_epi64): Ditto.
33840 (_mm256_i32scatter_ps): Change __addr type to void*.
33841 (_mm256_mask_i32scatter_ps): Ditto.
33842 (_mm_i32scatter_ps): Ditto.
33843 (_mm_mask_i32scatter_ps): Ditto.
33844 (_mm256_i32scatter_pd): Ditto.
33845 (_mm256_mask_i32scatter_pd): Ditto.
33846 (_mm_i32scatter_pd): Ditto.
33847 (_mm_mask_i32scatter_pd): Ditto.
33848 (_mm256_i64scatter_ps): Ditto.
33849 (_mm256_mask_i64scatter_ps): Ditto.
33850 (_mm_i64scatter_ps): Ditto.
33851 (_mm_mask_i64scatter_ps): Ditto.
33852 (_mm256_i64scatter_pd): Ditto.
33853 (_mm256_mask_i64scatter_pd): Ditto.
33854 (_mm_i64scatter_pd): Ditto.
33855 (_mm_mask_i64scatter_pd): Ditto.
33856 (_mm256_i32scatter_epi32): Ditto.
33857 (_mm256_mask_i32scatter_epi32): Ditto.
33858 (_mm_i32scatter_epi32): Ditto.
33859 (_mm_mask_i32scatter_epi32): Ditto.
33860 (_mm256_i32scatter_epi64): Ditto.
33861 (_mm256_mask_i32scatter_epi64): Ditto.
33862 (_mm_i32scatter_epi64): Ditto.
33863 (_mm_mask_i32scatter_epi64): Ditto.
33864 (_mm256_i64scatter_epi32): Ditto.
33865 (_mm256_mask_i64scatter_epi32): Ditto.
33866 (_mm_i64scatter_epi32): Ditto.
33867 (_mm_mask_i64scatter_epi32): Ditto.
33868 (_mm256_i64scatter_epi64): Ditto.
33869 (_mm256_mask_i64scatter_epi64): Ditto.
33870 (_mm_i64scatter_epi64): Ditto.
33871 (_mm_mask_i64scatter_epi64): Ditto.
33872 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33873 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33874 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33875 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33876 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33877 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33878 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33879 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33880 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33881 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33882 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33883 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33884 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33885 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33886 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33887 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33888 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33889 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33890 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33891 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33892 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33893 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33894 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33895 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33896 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33897 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33898 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33899 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33900 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33901 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33902 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33903 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33904 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33905 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33906 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33907 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33908 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33909 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33910 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33911 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33912 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33913 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33914 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33915 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33916 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33917 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33918 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33919 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33920 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33921 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33922 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33923 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33924 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33925 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33926 definitions accordingly.
33928 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33929 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33932 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33935 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33938 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33939 hard regs for inner regno.
33941 2017-01-17 Martin Liska <mliska@suse.cz>
33944 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33945 assumption and add comment.
33947 2017-01-17 Nathan Sidwell <nathan@acm.org>
33949 * ipa-visibility.c (localize_node): New function, broken out of ...
33950 (function_and_variable_visibility): ... here. Call it.
33952 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33954 PR middle-end/77445
33955 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33956 correctly set frequency of oudgoing edge.
33957 (duplicate_thread_path): Fix profile updating.
33959 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33962 * configure.ac: Add GCC_BASE_VER.
33963 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33964 version from BASE-VER file.
33965 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33966 (gcc.o): Depend on $(BASEVER).
33967 * common.opt (dumpfullversion): New option.
33968 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33969 * doc/invoke.texi: Document -dumpfullversion.
33970 * doc/install.texi: Document --with-gcc-major-version-only.
33971 * configure: Regenerated.
33973 2017-01-17 Richard Biener <rguenther@suse.de>
33975 PR tree-optimization/71433
33976 * tree-vrp.c (register_new_assert_for): Merge same asserts
33977 on all incoming edges.
33978 (process_assert_insertions_for): Handle insertions at the
33981 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33983 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33984 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33986 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33989 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33992 2017-01-17 Alan Modra <amodra@gmail.com>
33995 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33996 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33997 symbolic stack limit when pic.
33999 2017-01-16 Martin Sebor <msebor@redhat.com>
34001 PR tree-optimization/78608
34002 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34004 2017-01-16 Jeff Law <law@redhat.com>
34007 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34008 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34009 for several include directories that may be relative to sysroot.
34010 * config/i386/x-mingw32 (gplus_includedir): Define.
34011 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34012 (native_system_includedir): Likewise.
34013 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34014 override if TARGET_SYSTEM_ROOT is defined.
34015 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34017 PR tree-optimization/79090
34018 PR tree-optimization/33562
34019 PR tree-optimization/61912
34020 PR tree-optimization/77485
34021 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34022 and computed trims into the dump file.
34024 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34026 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34028 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34031 * gimplify.c (gimplify_init_constructor): If want_value and
34032 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34036 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34037 sequence. Formatting fixes.
34038 (doloop_optimize): Formatting fixes.
34041 * gcc.c (debug_level_greater_than_spec_func): New function.
34042 (static_spec_functions): Add debug-level-gt spec function.
34043 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34045 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34046 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34047 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34048 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34049 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34050 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34052 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34054 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34055 QImode fixups to general and mask registers only.
34057 2017-01-16 Carl Love <cel@us.ibm.com>
34059 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34060 for built-in functions
34061 vector signed char vec_nabs (vector signed char)
34062 vector signed short vec_nabs (vector signed short)
34063 vector signed int vec_nabs (vector signed int)
34064 vector signed long long vec_nabs (vector signed long long)
34065 vector float vec_nabs (vector float)
34066 vector double vec_nabs (vector double)
34067 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34069 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34070 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34071 * doc/extend.texi: Update the documentation file for the new built-in
34074 2017-01-16 Martin Sebor <msebor@redhat.com>
34076 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34079 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34081 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34082 UNSPEC_VSX__XXSPLTD to require special splat handling.
34084 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34087 * system.h: Poison strndup.
34089 2017-01-16 Alan Modra <amodra@gmail.com>
34092 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34095 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34097 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34099 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34101 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34102 call recog here. Assert that INSN_CODE (insn) is non-negative.
34104 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34107 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34109 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34110 in the currently scheduled RTL fragment.
34112 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34114 PR rtl-optimization/78751
34115 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34118 2017-01-14 Jeff Law <law@redhat.com>
34120 PR tree-optimization/79090
34121 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34122 variable length stores.
34123 (compute_trims): Delete dead assignment to *trim_tail.
34124 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34127 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34129 PR rtl-optimization/78626
34130 PR rtl-optimization/78727
34131 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34132 of a block, and split such blocks after everything else is finished.
34134 2017-01-14 Alan Modra <amodra@gmail.com>
34137 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34138 target legitimate_combined_insn.
34139 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34140 (rs6000_legitimate_combined_insn): New function.
34141 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34143 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34144 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34145 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34147 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34149 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34151 2017-01-13 Jeff Law <law@redhat.com>
34153 PR tree-optimization/33562
34154 PR tree-optimization/61912
34155 PR tree-optimization/77485
34156 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34158 (delete_dead_assignment): Likewise.
34159 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34160 statement to delete_dead_call and delete_dead_assignment.
34162 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34165 * substring-locations.c (format_warning_va): Strengthen case 1 so
34166 that both endpoints of the substring must be within the format
34167 range for just the substring to be printed.
34169 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34171 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34172 * config/i386/i386.c (ix86_target_string): Add missing options
34173 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34174 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34175 flags_other and ix86_target_other to flags2_other. Display unknown
34177 (ix86_valid_target_attribute_inner_p): Add missing options and
34178 reorder options by implied ISAs, as in ix86_target_string.
34180 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34182 * hash-table.h (hash_table::too_empty_p): New function.
34183 (hash_table::expand): Use it.
34184 (hash_table::traverse): Likewise.
34185 (hash_table::empty_slot): Use sizeof (value_type) instead of
34186 sizeof (PTR) to convert bytes to elements. Shrink the table
34187 if the current size is excessive for the current number of
34190 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34192 * ira-costs.c (record_reg_classes): Break from the inner loop
34193 early once alt_fail is known to be true. Update outer loop
34194 handling accordingly.
34196 2017-01-13 Jeff Law <law@redhat.com>
34198 * tree-ssa-dse.c (decrement_count): New function.
34199 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34200 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34201 when we know the partially dead statement is a mem* function.
34203 PR tree-optimization/61912
34204 PR tree-optimization/77485
34205 * tree-ssa-dse.c: Include expr.h.
34206 (maybe_trim_constructor_store): New function.
34207 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34209 PR tree-optimization/33562
34210 PR tree-optimization/61912
34211 PR tree-optimization/77485
34212 * doc/invoke.texi: Document new dse-max-object-size param.
34213 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34214 * tree-ssa-dse.c: Include params.h.
34215 (dse_store_status): New enum.
34216 (initialize_ao_ref_for_dse): New, partially extracted from
34218 (valid_ao_ref_for_dse, normalize_ref): New.
34219 (setup_live_bytes_from_ref, compute_trims): Likewise.
34220 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34221 (maybe_trim_partially_dead_store): Likewise.
34222 (maybe_trim_complex_store): Likewise.
34223 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34224 Track what bytes live from the original store. Return tri-state
34225 for dead, partially dead or live.
34226 (dse_dom_walker): Add constructor, destructor and new private members.
34227 (delete_dead_call, delete_dead_assignment): New extracted from
34229 (dse_optimize_stmt): Make a member of dse_dom_walker.
34230 Use initialize_ao_ref_for_dse.
34232 PR tree-optimization/33562
34233 PR tree-optimization/61912
34234 PR tree-optimization/77485
34235 * sbitmap.h (bitmap_count_bits): Prototype.
34236 (bitmap_clear_range, bitmap_set_range): Likewise.
34237 * sbitmap.c (bitmap_clear_range): New function.
34238 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34240 2017-01-13 Martin Liska <mliska@suse.cz>
34243 * function.c (set_cfun): Add new argument force.
34244 * function.h (set_cfun): Likewise.
34245 * ipa-inline-transform.c (inline_call): Use the function when
34246 strict alising from is dropped for function we inline to.
34248 2017-01-13 Richard Biener <rguenther@suse.de>
34250 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34251 for dumping GIMPLE INTEGER_CSTs.
34253 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34255 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34256 to 201112L since C++17.
34258 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34261 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34262 if -fsanitize=kernel-address is present.
34264 2017-01-13 Richard Biener <rguenther@suse.de>
34266 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34267 as _Literal ( type ) number in case usual suffixes do not
34268 preserve all information.
34270 2017-01-13 Richard Biener <rguenther@suse.de>
34272 PR tree-optimization/77283
34273 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34274 and ssa-iterators.h.
34275 (is_feasible_trace): Implement a cost model based on joiner
34278 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34281 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34282 char or short to __float128/_Float128 directly.
34284 2017-01-12 Martin Sebor <msebor@redhat.com>
34286 to -Wformat-overflow.
34287 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34288 (min_bytes_remaining): Same.
34289 (get_string_length): Same.
34290 (format_string): Same.
34291 (format_directive): Same.
34293 (pass_sprintf_length::handle_gimple_call): Same.
34295 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34297 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34298 info.nowrite calls with no lhs that can't throw. Return bool
34299 whether gsi_remove has been called or not.
34300 (pass_sprintf_length::handle_gimple_call): Return bool whether
34301 try_substitute_return_value called gsi_remove. Formatting fix.
34302 (pass_sprintf_length::execute): Don't use gsi_remove if
34303 handle_gimple_call returned true.
34306 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34307 be removed due to side-effects, don't remove following barrier nor
34308 turn the successor edge into fallthru edge.
34310 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34313 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34314 element-reversing loads and stores as not swappable.
34316 2017-01-12 Nathan Sidwell <nathan@acm.org>
34317 Nicolai Stange <nicstange@gmail.com>
34319 * combine.c (try_combine): Don't ignore result of overlap checking
34320 loop. Combine overlap & asm check into single loop.
34322 2017-01-12 Richard Biener <rguenther@suse.de>
34324 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34325 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34327 2017-01-12 Richard Biener <rguenther@suse.de>
34329 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34330 and TS_TARGET_OPTION directly derive from TS_BASE.
34331 * tree-core.h (tree_optimization_option): Derive from tree_base.
34332 (tree_target_option): Likewise.
34334 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34336 * config/i386/i386.c (memory_address_length): Increase len
34337 only when rip_relative_addr_p returns false.
34339 2017-01-11 Julia Koval <julia.koval@intel.com>
34341 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34342 (OPTION_MASK_ISA_SGX_SET): New.
34343 (ix86_handle_option): Handle OPT_msgx.
34344 * config.gcc: Added sgxintrin.h.
34345 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34346 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34347 * config/i386/i386.c (ix86_target_string): Add -msgx.
34349 (ix86_option_override_internal): Handle new options.
34350 (ix86_valid_target_attribute_inner_p): Add sgx.
34351 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34352 * config/i386/i386.opt: Add msgx.
34353 * config/i386/sgxintrin.h: New file.
34354 * config/i386/x86intrin.h: Add sgxintrin.h.
34356 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34359 * fold-const.c (maybe_nonzero_address): Return 1 for function
34361 (tree_single_nonzero_warnv_p): Don't handle function local objects
34365 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34368 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34371 * opts.c: Include "spellcheck.h"
34372 (struct string_fragment): New struct.
34373 (struct edit_distance_traits<const string_fragment &>): New
34375 (get_closest_sanitizer_option): New function.
34376 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34378 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34380 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34382 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34383 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34384 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34385 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34386 for initial die_offset if dwarf_split_debug_info.
34387 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34388 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34390 (output_skeleton_debug_sections): Formatting fix. Use
34391 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34392 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34394 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34396 * config/arm/cortex-a53.md: Add bypasses for
34397 cortex_a53_r2f_cvt.
34398 (cortex_a53_r2f): Only use for transfers.
34399 (cortex_a53_f2r): Likewise.
34400 (cortex_a53_r2f_cvt): Add reservation for conversions.
34401 (cortex_a53_f2r_cvt): Likewise.
34403 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34405 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34406 to all inlined functions, change static to extern.
34408 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34411 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34413 (arm_assemble_integer): Likewise.
34415 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34417 * config.gcc: Use new awk script to check CPU, FPU and architecture
34418 parameters for --with-... options.
34419 * config/arm/parsecpu.awk: New file
34420 * config/arm/arm-cpus.in: New file.
34421 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34423 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34425 * config/arm/t-arm: Update dependency rules.
34426 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34427 of processing .def files.
34428 * config/arm/genopt.sh: Deleted.
34429 * config/arm/gentune.sh: Deleted.
34430 * config/arm/arm-cores.def: Deleted.
34431 * config/arm/arm-arches.def: Deleted.
34432 * config/arm/arm-fpus.def: Deleted.
34433 * config/arm/arm-tune.md: Regenerated.
34434 * config/arm/arm-tables.opt: Regenerated.
34435 * config/arm/arm-cpu.h: New generated file.
34436 * config/arm/arm-cpu-data.h: New generated file.
34437 * config/arm/arm-cpu-cdata.h: New generated file.
34439 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34442 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34444 (input_varpool_node): Unpack dynamically_initialized bit.
34446 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34448 PR rtl-optimization/79032
34449 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34450 the alignment of the adjusted memory reference against that of MODE,
34451 instead of the alignment of the original memory reference.
34453 2017-01-11 Martin Jambor <mjambor@suse.cz>
34455 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34457 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34458 decorated functions.
34460 2017-01-11 Richard Biener <rguenther@suse.de>
34462 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34463 set range/nonnull info for PHI results. Do not set it on
34464 stmts marked for removal.
34466 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34468 * expr.c (store_field): In the bitfield case, fetch the return value
34469 from the registers before applying a single big-endian adjustment.
34470 Always do a final load for a BLKmode value not larger than a word.
34472 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34475 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34476 that we correctly handle column numbers greater than
34477 LINE_MAP_MAX_COLUMN_NUMBER.
34479 2017-01-10 Martin Sebor <msebor@redhat.com>
34481 PR middle-end/78245
34482 * gimple-ssa-sprintf.c (get_destination_size): Call
34483 {init,fini}object_sizes.
34484 * tree-object-size.c (addr_object_size): Adjust.
34485 (pass_through_call): Adjust.
34486 (pass_object_sizes::execute): Adjust.
34487 * tree-object-size.h (fini_object_sizes): Declare.
34489 2017-01-10 Martin Sebor <msebor@redhat.com>
34491 PR tree-optimization/78775
34492 * builtins.c (get_size_range): Move...
34493 * calls.c: ...to here.
34494 (alloc_max_size): Accept zero argument.
34495 (operand_signed_p): Remove.
34496 (maybe_warn_alloc_args_overflow): Call get_size_range.
34497 * calls.h (get_size_range): Declare.
34499 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34501 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34502 from TI's devices.csv file as of September 2016.
34503 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34505 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34507 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34508 * doc/invoke.texi: Likewise.
34509 * doc/md.texi: Likewise.
34510 * doc/objc.texi: Likewise.
34512 2017-01-10 Joshua Conner <joshconner@google.com>
34514 * config/arm/fuchsia-elf.h: New file.
34515 * config/fuchsia.h: New file.
34516 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34517 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34519 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34521 2016-01-10 Richard Biener <rguenther@suse.de>
34523 PR tree-optimization/79034
34524 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34525 Propagate out degenerate PHIs in the joiner.
34527 2017-01-10 Martin Liska <mliska@suse.cz>
34529 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34530 (sort_congruence_classes_by_decl_uid): Likewise.
34531 (sort_congruence_class_groups_by_decl_uid): Likewise.
34532 (sem_item_optimizer::merge_classes): Sort class, groups in these
34533 classes and members in the groups by DECL_UID of declarations.
34534 This would make merge operations stable.
34536 2017-01-10 Martin Liska <mliska@suse.cz>
34538 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34539 usage of m_classes_vec.
34540 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34541 (sem_item_optimizer::get_group_by_hash): Likewise.
34542 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34543 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34544 (sem_item_optimizer::verify_classes): Likewise.
34545 (sem_item_optimizer::process_cong_reduction): Likewise.
34546 (sem_item_optimizer::dump_cong_classes): Likewise.
34547 (sem_item_optimizer::merge_classes): Likewise.
34548 * ipa-icf.h (congruence_class_hash): Rename from
34549 congruence_class_group_hash. Remove declaration of m_classes_vec.
34551 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34553 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34554 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34555 * config.gcc: Add avx512vpopcntdqintrin.h.
34556 * config/i386/avx512vpopcntdqintrin.h: New.
34557 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34558 * config/i386/i386-builtin-types.def: Add new types.
34559 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34560 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34561 __builtin_ia32_vpopcountq_v8di_mask): New.
34562 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34563 __AVX512VPOPCNTDQ__.
34564 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34565 (PTA_AVX512VPOPCNTDQ): Define.
34566 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34567 TARGET_AVX512VPOPCNTDQ_P): Define.
34568 * config/i386/i386.opt: Add mavx512vpopcntdq.
34569 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34570 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34572 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34574 PR middle-end/77484
34575 * predict.def (PRED_CALL): Set to 67.
34577 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34579 * expr.c (store_field): In the bitfield case, if the value comes from
34580 a function call and is of an aggregate type returned in registers, do
34581 not modify the field mode; extract the value in all cases if the mode
34582 is BLKmode and the size is not larger than a word.
34584 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34587 * config/i386/cpuid.h: Fix undefined behavior.
34589 2017-01-04 Jeff Law <law@redhat.com>
34591 PR tree-optimization/79007
34592 PR tree-optimization/67955
34593 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34594 conservative for pt.null when flag_non_call_exceptions is on.
34596 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34598 PR translation/79019
34599 PR translation/79020
34600 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34601 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34602 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34604 * config/avr/avr.opt (maccumulate-args): Likewise.
34605 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34606 * common.opt (freport-bug): Likewise.
34607 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34608 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34609 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34610 translatable string.
34611 * config/i386/i386.c (function_value_32): Likewise.
34612 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34613 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34615 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34616 * common/config/msp430/msp430-common.c (msp430_handle_option):
34618 * symtab.c (symtab_node::verify_base): Likewise.
34619 * opts.c (set_debug_level): Likewise.
34620 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34621 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34622 missing whitespace to translatable strings.
34623 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34624 * config/sh/superh.h: Likewise.
34625 * config/i386/xopintrin.h: Likewise.
34626 * config/i386/znver1.md: Likewise.
34627 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34628 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34629 * double-int.h (struct double_int): Likewise.
34630 * double-int.c (div_and_round_double): Likewise.
34631 * wide-int.cc: Likewise.
34632 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34633 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34634 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34635 (crossjumps_occurred): ... this.
34636 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34639 PR tree-optimization/78899
34640 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34641 returning bool return struct loop *, NULL for failure and the new
34643 (versionable_outer_loop_p): Don't version outer loop if it has
34644 dont_vectorized bit set.
34645 (tree_if_conversion): When versioning outer loop, ensure
34646 tree_if_conversion is performed also on the inner loop of the
34647 non-vectorizable outer loop copy.
34648 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34649 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34650 prevent vectorization of it.
34651 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34652 the outer loop vectorization of the non-scalar version is attempted
34653 before vectorization of the inner loop in scalar version. If
34654 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34655 vectorization of its inner loop.
34656 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34657 has 2 inner loops, rename also on edges from bb whose single pred
34658 is outer_loop->header. Fix typo in function comment.
34660 2017-01-09 Martin Sebor <msebor@redhat.com>
34663 * asan.c (asan_emit_stack_protection): Increase local buffer size
34664 to avoid snprintf truncation warning.
34666 2017-01-09 Andrew Pinski <apinski@cavium.com>
34668 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34669 to reference thunderx2t99 for the tuning structure
34670 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34672 (thunderx2t99_extra_costs): This.
34673 * config/aarch64/aarch64-tune.md: Regenerate.
34674 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34675 (vulcan_addrcost_table): This.
34676 (vulcan_regmove_cost): Rename to ...
34677 (thunderx2t99_regmove_cost): This.
34678 (vulcan_vector_cost): Rename to ...
34679 (thunderx2t99_vector_cost): this.
34680 (vulcan_branch_cost): Rename to ...
34681 (thunderx2t99_branch_cost): This.
34682 (vulcan_tunings): Rename to ...
34683 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34684 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34686 2017-01-09 Martin Jambor <mjambor@suse.cz>
34690 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34691 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34692 (propagate_vr_accross_jump_function): Use the above function for all
34693 value range computations for pass-through jump functions and type
34694 converasion from explicit value range values.
34695 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34696 parameters from TYPE_ARG_TYPES.
34697 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34698 (ipa_write_node_info): Stream type of the actual argument.
34699 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34701 2017-01-09 Martin Liska <mliska@suse.cz>
34704 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34705 (lookup_compiler): Do not show error message with have_E.
34707 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34709 PR tree-optimization/78938
34710 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34711 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34712 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34713 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34716 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34718 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34721 2017-01-09 Richard Biener <rguenther@suse.de>
34723 PR tree-optimization/78997
34724 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34725 name condition properly.
34727 2017-01-09 Richard Biener <rguenther@suse.de>
34730 * dwarf2out.c (is_cxx): New overload with context.
34731 (is_naming_typedef_decl): Use it.
34733 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34735 * invoke.texi (Option Summary): Correct spacing in option lists
34736 and add line breaks to fix over-long lines.
34738 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34740 PR middle-end/17660
34742 * extend.texi (Common Variable Attributes): Add xref to GCC
34743 Internals manual to explain mode attribute keywords.
34745 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34748 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34749 and Preprocessor Options.
34750 (Options for Linking): Document -pthread here....
34751 (RS/6000 and PowerPC Options): ...not here.
34752 (Solaris 2 Options): ...or here.
34753 * doc/cppopts.texi: Document -pthread.
34755 2017-01-08 Martin Sebor <msebor@redhat.com>
34757 PR middle-end/77708
34758 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34759 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34760 New member functions.
34761 (format_directive): Used them.
34763 (pass_sprintf_length::handle_gimple_call): Same.
34764 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34765 to avoid truncation for any argument.
34766 (extract_affine_mul): Same.
34767 * tree.c (get_file_function_name): Same.
34769 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34771 PR middle-end/77484
34772 * predict.def (PRED_INDIR_CALL): Set to 86.
34774 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34776 PR preprocessor/54124
34777 * doc/cppopts.texi: Reformat -d subtable to list the full name
34778 of the options. Add cross-reference to the docs for the general
34779 compiler -d options.
34780 * doc/invoke.texi (Developer Options): Add cross-reference to the
34781 preprocessor-specific -d option documentation.
34783 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34785 PR preprocessor/13498
34786 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34787 redudant material, and reflect new command-line options.
34788 (System Headers): Likewise.
34790 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34792 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34793 -isystem, and -idirafter. Copy-edit.
34794 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34795 default for -ftrack-macro-expansion. Delete obsolete and
34796 badly-formatted implementation details about -fdebug-cpp output.
34797 * doc/cppwarnopts.texi: Copy-edit.
34799 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34802 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34803 that the transition from a max line width >= 1<<10 to narrower
34804 lines works correctly.
34806 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34808 * doc/options.texi (PerFunction): New.
34809 * opt-functions.awk (switch_flags): Map both Optimization and
34810 PerFunction to CL_OPTIMIZATION.
34811 * opth-gen.awk: Test for PerFunction flag along with
34813 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34814 it only when the latter is present. Skip those that don't in
34815 the hash function generator.
34816 * common.opt (fvar-tracking): Mark as PerFunction instead of
34818 (fvar-tracking-assignments): Likewise.
34819 (fvar-tracking-assignments-toggle): Likewise.
34820 (fvar-tracking-uninit): Likewise.
34822 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34824 PR translation/79018
34825 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34828 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34831 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34832 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34833 TARGET_LONG_JUMP_TABLE_OFFSETS.
34834 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34835 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34836 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34837 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34838 * config/m68k/m68k.md (tablejump expander): Likewise.
34839 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34840 TARGET_LONG_JUMP_TABLE_OFFSETS.
34841 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34842 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34844 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34845 David Holsgrove <david.holsgrove@xilinx.com>
34847 * common/config/microblaze/microblaze-common.c
34848 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34849 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34851 * config/microblaze/microblaze.c (microblaze_must_save_register)
34852 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34854 (microblaze_eh_return): New function.
34855 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34856 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34857 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34858 * config/microblaze/microblaze.md (eh_return): New pattern.
34860 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34862 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34863 GCC_DIAGNOSTIC_STRINGIFY): Define.
34865 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34867 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34869 * config/arm/arm.md (<mcrr>): New.
34871 * config/arm/arm.c (arm_arch5te): New.
34872 (arm_option_override): Set arm_arch5te.
34873 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34875 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34876 (arm_mcrr_qualifiers): ... this. New.
34877 (MRRC_QUALIFIERS): Define to...
34878 (arm_mrrc_qualifiers): ... this. New.
34879 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34881 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34882 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34883 (MRRCI, mrrc, MRRC): New.
34884 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34885 VUNSPEC_MRRC2): New.
34887 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34889 * config/arm/arm.md (<mcr>): New.
34891 * config/arm/arm.c (arm_coproc_builtin_available): Add
34892 support for mcr, mrc, mcr2 and mrc2.
34893 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34894 (arm_mcr_qualifiers): ... this. New.
34895 (MRC_QUALIFIERS): Define to ...
34896 (arm_mrc_qualifiers): ... this. New.
34897 (MCR_QUALIFIERS): Define to ...
34898 (arm_mcr_qualifiers): ... this. New.
34899 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34901 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34902 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34903 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34904 VUNSPEC_MRC2): New.
34906 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34908 * config/arm/arm.md (*ldc): New.
34912 * config/arm/arm.c (arm_coproc_builtin_available): Add
34913 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34914 (arm_coproc_ldc_stc_legitimate_address): New.
34915 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34916 'qualifier_const_pointer'.
34917 (LDC_QUALIFIERS): Define to...
34918 (arm_ldc_qualifiers): ... this. New.
34919 (STC_QUALIFIERS): Define to...
34920 (arm_stc_qualifiers): ... this. New.
34921 * config/arm/arm-protos.h
34922 (arm_coproc_ldc_stc_legitimate_address): New.
34923 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34924 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34925 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34926 stc2, stcl, stc2l): New.
34927 * config/arm/constraints.md (Uz): New.
34928 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34929 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34930 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34931 VUNSPEC_STC2L): New.
34933 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34935 * config/arm/arm.md (<cdp>): New.
34936 * config/arm/arm.c (neon_const_bounds): Rename this ...
34937 (arm_const_bounds): ... this.
34938 (arm_coproc_builtin_available): New.
34939 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34940 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34941 (CDP_QUALIFIERS): Define to...
34942 (arm_cdp_qualifiers): ... this. New.
34944 (arm_expand_builtin_args): Add case for 6 arguments.
34945 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34946 (arm_const_bounds): ... this.
34947 (arm_coproc_builtin_available): New.
34948 * config/arm/arm_acle.h (__arm_cdp): New.
34950 * config/arm/arm_acle_builtins.def (cdp): New.
34952 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34953 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34954 'arm_const_bounds'.
34955 * config/arm/types.md (coproc): New.
34956 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34957 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34958 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34959 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34961 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34963 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34964 (UBINOP_QUALIFIERS): New.
34966 (acle_builtin_data): New. Change comment.
34967 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34968 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34969 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34970 arm_acle_builtins.def.
34971 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34972 (arm_init_acle_builtins): New.
34973 (CRC32_BUILTIN): Remove.
34974 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34975 crc32cb, crc32ch and crc32cw.
34976 (arm_init_crc32_builtins): Remove.
34977 (arm_init_builtins): Use arm_init_acle_builtins rather
34978 than arm_init_crc32_builtins.
34979 (arm_expand_acle_builtin): New.
34980 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34981 * config/arm/arm_acle_builtins.def: New.
34983 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34985 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34986 (arm_builtin_datum): ... this.
34987 (arm_init_neon_builtin): Rename to ...
34988 (arm_init_builtin): ... this. Add a new parameters PREFIX
34989 and USE_SIG_IN_NAME.
34990 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34991 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34992 'arm_builtin_datum'.
34993 (arm_init_vfp_builtins): Likewise.
34994 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34995 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34996 (arm_expand_neon_args): Rename to ...
34997 (arm_expand_builtin_args): ... this. Rename builtin_arg
34998 enum values and differentiate between ARG_BUILTIN_MEMORY
34999 and ARG_BUILTIN_NEON_MEMORY.
35000 (arm_expand_neon_builtin_1): Rename to ...
35001 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35002 values, arm_expand_builtin_args and add bool parameter NEON.
35003 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35004 (arm_expand_vfp_builtin): Likewise.
35005 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35007 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35009 PR middle-end/77484
35010 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35011 * predict.c (tree_estimate_probability_bb): Reverse direction of
35012 polymorphic call predictor.
35014 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35016 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35017 (determine_pass_name_match): ...this new function and...
35018 (should_skip_pass_p): ...this new function.
35020 2017-01-06 Nathan Sidwell <nathan@acm.org>
35022 * ipa-visibility.c (function_and_variable_visibility): Reformat
35023 comments and long lines. Remove extrneous if.
35024 * symtab.c (symtab_node::make_decl_local): Fix code format.
35025 (symtab_node::set_section_for_node): Fix comment typo.
35027 2017-01-06 Martin Liska <mliska@suse.cz>
35030 * lra-constraints.c: Rename invariant to lra_invariant.
35031 * predict.c (set_even_probabilities): Initialize e to NULL.
35033 2017-01-05 Martin Sebor <msebor@redhat.com>
35035 PR tree-optimization/78910
35036 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35037 (format_integer): Correct off-by-one error in the handling
35038 of precision with negative numbers in signed conversions..
35040 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35042 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35044 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35046 PR tree-optimization/71016
35047 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35048 factor_out_conditional_conversion. Formatting fix.
35049 (factor_out_conditional_conversion): Add cond_stmt argument.
35050 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35051 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35054 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35056 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35057 read-rtl-function.o, and selftest-rtl.o.
35058 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35059 (selftest::aarch64_test_loading_full_dump): New function.
35060 (selftest::aarch64_run_selftests): New function.
35061 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35062 selftest::aarch64_run_selftests.
35063 * config/i386/i386.c
35064 (selftest::ix86_test_loading_dump_fragment_1): New function.
35065 (selftest::ix86_test_loading_call_insn): New function.
35066 (selftest::ix86_test_loading_full_dump): New function.
35067 (selftest::ix86_test_loading_unspec): New function.
35068 (selftest::ix86_run_selftests): Call the new functions.
35069 * emit-rtl.c (maybe_set_max_label_num): New function.
35070 * emit-rtl.h (maybe_set_max_label_num): New decl.
35071 * function.c (instantiate_decls): Guard call to
35072 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35073 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35075 * gensupport.c (gen_reader::gen_reader): Pass "false"
35076 for new "compact" param of rtx_reader.
35077 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35078 rather than an empty string for NULL strings.
35079 * read-md.c: Potentially include config.h rather than bconfig.h.
35080 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35081 (have_error): New global, copied from errors.c.
35082 (md_reader::read_name): Rename to...
35083 (md_reader::read_name_1): ...this, adding "out_loc" param,
35084 and converting "missing name or number" to returning false, rather
35086 (md_reader::read_name): Reimplement in terms of read_name_1.
35087 (md_reader::read_name_or_nil): New function.
35088 (md_reader::read_string): Handle "(nil)" by returning NULL.
35089 (md_reader::md_reader): Add new param "compact".
35090 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35091 (md_reader::read_file): New method.
35092 * read-md.h (md_reader::md_reader): Add new param "compact".
35093 (md_reader::read_file): New method.
35094 (md_reader::is_compact): New accessor.
35095 (md_reader::read_name): Convert return type from void to file_location.
35096 (md_reader::read_name_or_nil): New decl.
35097 (md_reader::read_name_1): New decl.
35098 (md_reader::m_compact): New field.
35099 (noop_reader::noop_reader): Pass "false" for new "compact" param
35101 (rtx_reader::rtx_reader): Add new "compact" param.
35102 (rtx_reader::read_rtx_operand): Make virtual and convert return
35103 type from void to rtx.
35104 (rtx_reader::read_until): New decl.
35105 (rtx_reader::handle_any_trailing_information): New virtual function.
35106 (rtx_reader::postprocess): New virtual function.
35107 (rtx_reader::finalize_string): New virtual function.
35108 (rtx_reader::m_in_call_function_usage): New field.
35109 (rtx_reader::m_reuse_rtx_by_id): New field.
35110 * read-rtl-function.c: New file.
35111 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35112 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35113 (selftest::verify_three_block_rtl_cfg): New decl.
35114 * read-rtl-function.h: New file.
35115 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35116 For host, include function.h, memmodel.h, and emit-rtl.h.
35117 (one_time_initialization): New function.
35118 (struct compact_insn_name): New struct.
35119 (compact_insn_names): New array.
35120 (find_code): Handle insn codes in compact dumps.
35121 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35122 (bind_subst_iter_and_attr): Likewise.
35123 (add_condition_to_string): Likewise.
35124 (add_condition_to_rtx): Likewise.
35125 (apply_attribute_uses): Likewise.
35126 (add_current_iterators): Likewise.
35127 (apply_iterators): Likewise.
35128 (initialize_iterators): Guard usage of apply_subst_iterator with
35129 #ifdef GENERATOR_FILE.
35130 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35131 (md_reader::read_mapping): Likewise.
35132 (add_define_attr_for_define_subst): Likewise.
35133 (add_define_subst_attr): Likewise.
35134 (read_subst_mapping): Likewise.
35135 (check_code_iterator): Likewise.
35136 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35138 (one_time_initialization): New function.
35139 (rtx_reader::read_until): New method.
35140 (read_flags): New function.
35141 (parse_reg_note_name): New function.
35142 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35143 Handle reuse_rtx ids.
35144 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35145 Add parsing support for RTL dumps, mirroring the special-cases in
35146 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35147 values, and calling handle_any_trailing_information.
35148 (rtx_reader::read_rtx_operand): Convert return type from void
35149 to rtx, returning return_rtx. Handle case 'e'. Call
35150 finalize_string on XSTR and XTMPL fields.
35151 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35152 "(nil)" values were omitted. Call the postprocess vfunc on the
35154 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35155 class ctor. Initialize m_in_call_function_usage. Call
35156 one_time_initialization.
35157 * rtl-tests.c (selftest::test_uncond_jump): Call
35158 set_new_first_and_last_insn.
35159 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35160 * selftest-rtl.c: New file.
35161 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35162 (selftest::get_insn_by_uid): New decl.
35163 * selftest-run-tests.c (selftest::run_tests): Call
35164 read_rtl_function_c_tests.
35165 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35166 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35169 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35171 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35172 operands in a special way. Assert that pos+len <= mode precision.
35174 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35176 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35177 3 argument Alias with unlimited for the negative form.
35178 (fno-vect-cost-model): Removed.
35180 2017-01-05 Martin Liska <mliska@suse.cz>
35182 * hsa-gen.c (gen_hsa_divmod): New function.
35183 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35185 2017-01-05 Martin Liska <mliska@suse.cz>
35188 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35191 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35193 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35194 small constant length operands.
35196 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35198 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35199 between loop iterations.
35201 2017-01-05 Martin Liska <mliska@suse.cz>
35204 * gimplify.c (gimplify_decl_expr): Compare to
35205 asan_poisoned_variables instread of checking flags.
35206 (gimplify_target_expr): Likewise.
35207 (gimplify_expr): Likewise.
35208 (gimplify_function_tree): Conditionally initialize
35209 asan_poisoned_variables.
35211 2017-01-04 Jeff Law <law@redhat.com>
35213 PR tree-optimizatin/78812
35214 * rtl.h (contains_mem_rtx_p): Prototype.
35215 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35216 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35217 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35218 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35219 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35221 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35223 * input.c (assert_char_at_range): Default-initialize actual_range.
35225 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35227 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35228 to match the caller.
35230 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35232 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35233 insns after final jump in test to emit dummy move.
35235 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35237 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35238 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35240 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35242 * multiple_target.c (create_dispatcher_calls): Init e_next.
35243 * tree-ssa-loop-split.c (split_loop): Init border.
35244 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35247 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35252 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35253 (altivec_register_operand): Do not return true if the operand
35254 contains a SUBREG mixing SImode and SFmode.
35255 (vsx_register_operand): Likewise.
35256 (vsx_reg_sfsubreg_ok): New predicate.
35257 (vfloat_operand): Do not return true if the operand contains a
35258 SUBREG mixing SImode and SFmode.
35259 (vint_operand): Likewise.
35260 (vlogical_operand): Likewise.
35261 (gpc_reg_operand): Likewise.
35262 (int_reg_operand): Likewise.
35263 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35264 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35265 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35267 (rs6000_emit_move_si_sf_subreg): New helper function.
35268 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35269 fixup SUBREGs involving SImode and SFmode.
35270 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35271 numbers for the new peephole2 optimization.
35272 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35273 the GLIBC math library that do AND/IOR/XOR operations on single
35274 precision floating point.
35275 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35276 target macros to say whether we need to avoid SUBREGs mixing
35278 (TARGET_ALLOW_SF_SUBREG): Likewise.
35279 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35280 (UNSPEC_SI_FROM_SF): Likewise.
35281 (iorxor): Change spacing.
35282 (and_ior_xor): New iterator for AND, IOR, and XOR.
35283 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35284 (movdi_from_sf_zero_ext): Likewise.
35285 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35286 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35287 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35288 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35289 (fms<mode>4): Likewise.
35290 (fnma<mode>4): Likewise.
35291 (fnms<mode>4): Likewise.
35292 (nfma<mode>4): Likewise.
35293 (nfms<mode>4): Likewise.
35295 2017-01-04 Marek Polacek <polacek@redhat.com>
35298 * doc/invoke.texi: Document -Wpointer-compare.
35300 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35302 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35305 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35306 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35309 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35312 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35313 documentation of the powerpc_popcntb_ok attribute.
35314 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35315 code to issue warning messages if a requested CPU configuration is
35316 not supported by the binary (assembler and loader) toolchain.
35317 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35318 made to define a built-in function that has been disabled.
35319 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35320 made to define a built-in function that has been disabled.
35321 (altivec_init_builtins): Add comment explaining why definition
35322 of the DST built-in functions is not preceded by an assertion
35323 check. Add assertions to prevent ICE if attempts are made to
35324 define an altivec predicate or an abs* built-in function that has
35326 (htm_init_builtins): Add comment explaining why definition of the
35327 htm built-in functions is not preceded by an assertion check.
35329 2017-01-04 Jeff Law <law@redhat.com>
35331 PR tree-optimizatin/67955
35332 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35333 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35334 the points-to solution does not include pt_null. Use DECL_PT_UID
35337 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35339 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35340 Use gen_int_mode instead of gen_lopwart for const_int operands.
35342 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35344 PR tree-optimization/71563
35345 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35346 out of range value - has low bits known to be zero.
35348 2017-01-04 Alan Modra <amodra@gmail.com>
35350 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35351 * configure: Regenerate.
35352 * config.in: Regenerate.
35354 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35357 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35358 a substring of the message, but strcmp with the whole message. Ifdef
35359 ENABLE_NLS, translate the message first using dgettext.
35361 2017-01-03 Jeff Law <law@redhat.com>
35363 PR tree-optimizatin/78856
35364 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35365 (mark_threaded_blocks): Remove code to truncate thread paths that
35366 cross multiple loop headers. Instead invalidate the cached loop
35367 iteration information and handle case of a thread path walking
35368 into an irreducible region.
35370 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35373 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35374 assertions. Add support for doing the signbit if the IEEE 128-bit
35375 floating point value is in a GPR.
35376 * config/rs6000/rs6000.md (Fsignbit): Delete.
35377 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35378 Update the length attribute if the value is in a GPR.
35379 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35380 the sign or zero extension instruction, since the value is always 0/1.
35381 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35384 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35385 extracting SImode to a GPR register so that we can generate a
35386 store, limit the vector to be in a traditional Altivec register
35387 for the vextuwrx instruction.
35389 2017-01-03 Ian Lance Taylor <iant@google.com>
35391 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35393 2017-01-03 Martin Sebor <msebor@redhat.com>
35395 PR tree-optimization/78696
35396 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35397 precision. Use MPFR for %f for greater fidelity. Correct handling
35399 (pass_sprintf_length::compute_format_length): Set width and precision
35400 specified by asrerisk to void_node for vararg functions.
35401 (try_substitute_return_value): Adjust dump output.
35403 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35405 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35407 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35409 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35410 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35411 -mlra/-mno-lra was passed to the compiler.
35413 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35415 PR rtl-optimization/65618
35416 * emit-rtl.c (try_split): Move initialization of "before" and
35417 "after" to just before the call to emit_insn_after_setloc.
35419 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35421 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35423 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35425 * dwarf2out.c (gen_enumeration_type_die): When
35426 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35428 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35430 PR tree-optimization/78965
35431 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35432 Change first argument from const call_info & to call_info &. For %n
35433 set info.nowrite to false.
35435 PR middle-end/78901
35436 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35437 possibly throwing calls.
35439 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35440 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35441 and fns handling, rather than in a separate case SSA_NAME.
35443 2017-01-02 Jeff Law <law@redhat.com>
35445 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35446 fixes for first_period and second_period variables.
35448 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35451 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35452 (*insvqi_1): New insn pattern.
35453 (*insvqi_1_mem_rex64): Ditto.
35454 (*insvqi_2): Ditto.
35455 (*insvqi_3): Rename from *insvqi.
35457 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35459 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35461 * doc/cfg.texi (Edges): Remove reference to Java.
35462 (Maintaining the CFG): Ditto.
35464 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35466 PR middle-end/77674
35467 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35468 transparent aliases.
35470 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35472 PR middle-end/77484
35473 * predict.def (PRED_CALL): Update hitrate.
35474 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35475 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35476 into direct/indirect/polymorphic variants.
35478 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35480 Update copyright years.
35482 * gcc.c (process_command): Update copyright notice dates.
35483 * gcov-dump.c (print_version): Ditto.
35484 * gcov.c (print_version): Ditto.
35485 * gcov-tool.c (print_version): Ditto.
35486 * gengtype.c (create_file): Ditto.
35487 * doc/cpp.texi: Bump @copying's copyright year.
35488 * doc/cppinternals.texi: Ditto.
35489 * doc/gcc.texi: Ditto.
35490 * doc/gccint.texi: Ditto.
35491 * doc/gcov.texi: Ditto.
35492 * doc/install.texi: Ditto.
35493 * doc/invoke.texi: Ditto.
35495 Copyright (C) 2017 Free Software Foundation, Inc.
35497 Copying and distribution of this file, with or without modification,
35498 are permitted in any medium without royalty provided the copyright
35499 notice and this notice are preserved.