1 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5 with the required version of automake 1.11.6.
7 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
9 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
11 (DU_C2_3_power9): Correct reservation combinations.
12 (FP_DIV_power9, VEC_DIV_power9): New.
13 (power9-alu): Split out rotate/shift...
14 (power9-rot): ...to here, correct dispatch resource.
15 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
17 (power9-fp): Correct latency.
18 (power9-sdiv): Add div/sqrt resource.
19 (power9-ddiv): Correct latency, add div/sqrt resource.
20 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
21 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
23 (power9-qpdiv, power9-qpmul): Adjust resource usage.
25 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
28 switch statement mapping KF built-ins to TF built-ins if we don't
29 have the proper ISA 3.0 assembler support.
31 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
33 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
34 (lower_emutls_function_body): Do not compute it.
36 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
38 * ipa-split.c (split_bb_info): Turn time to sreal.
39 (split_point): Likewise.
40 (dump_split_point): Likewise.
41 (fine_split_points): Likewise.
42 (execute_split_functions): Only zero split_bbs; turn time to sreals.
44 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
46 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
49 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
51 * predict.c (combine_predictions_for_bb): Preserve zero predicted
53 (expensive_function_p): Remove useless assert.
54 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
56 2017-11-16 Martin Sebor <msebor@redhat.com>
58 PR tree-optimization/82588
59 PR tree-optimization/82583
60 * tree-vrp.c (check_array_ref): Handle flexible array members,
61 string literals, and inner indices.
62 (search_for_addr_array): Add detail to diagnostics.
64 2017-11-16 Nathan Sidwell <nathan@acm.org>
68 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
69 (SET_DECL_ASSEMBLER_NAME): Forward to
70 overwrite_decl_assembler_name.
71 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
72 (overwrite_decl_assembler_name): Declare.
73 * tree.c (overwrite_decl_assembler_name): New.
74 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
75 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
76 (LANG_HOOKS_INITIALIZER): Add it.
77 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
78 * langhooks.c (lhd_set_decl_assembler_name): Use
79 SET_DECL_ASSEMBLER_NAME.
80 (lhd_overwrite_decl_assembler_name): Default implementation.
82 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
83 Jackson Woodruff <jackson.woodruff@arm.com>
85 PR tree-optimization/71026
86 * match.pd: Canonicalize constant multiplies in division.
88 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
90 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
91 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
92 Remove OPT_fomit_frame_pointer entry.
93 * common/config/arc/arc-common.c: Likewise.
94 * common/config/arm/arm-common.c: Likewise.
95 * common/config/avr/avr-common.c: Likewise.
96 * common/config/c6x/c6x-common.c: Likewise.
97 * common/config/cr16/cr16-common.c: Likewise.
98 * common/config/cris/cris-common.c: Likewise.
99 * common/config/epiphany/epiphany-common.c: Likewise.
100 * common/config/fr30/fr30-common.c: Likewise.
101 * common/config/frv/frv-common.c: Likewise.
102 * common/config/ia64/ia64-common.c: Likewise.
103 * common/config/iq2000/iq2000-common.c: Likewise.
104 * common/config/lm32/lm32-common.c: Likewise.
105 * common/config/m32r/m32r-common.c: Likewise.
106 * common/config/mcore/mcore-common.c: Likewise.
107 * common/config/microblaze/microblaze-common.c: Likewise.
108 * common/config/mips/mips-common.c: Likewise.
109 * common/config/mmix/mmix-common.c: Likewise.
110 * common/config/mn10300/mn10300-common.c: Likewise.
111 * common/config/nios2/nios2-common.c: Likewise.
112 * common/config/pa/pa-common.c: Likewise.
113 * common/config/pdp11/pdp11-common.c: Likewise.
114 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
115 * common/config/riscv/riscv-common.c: Likewise.
116 * common/config/rs6000/rs6000-common.c: Likewise.
117 * common/config/rx/rx-common.c: Likewise.
118 * common/config/s390/s390-common.c: Likewise.
119 * common/config/sh/sh-common.c: Likewise.
120 * common/config/sparc/sparc-common.c: Likewise.
121 * common/config/tilegx/tilegx-common.c: Likewise.
122 * common/config/tilepro/tilepro-common.c: Likewise.
123 * common/config/v850/v850-common.c: Likewise.
124 * common/config/visium/visium-common.c: Likewise.
125 * common/config/xstormy16/xstormy16-common.c: Likewise.
126 * common/config/xtensa/xtensa-common.c: Likewise.
127 * invoke.texi (-fomit-frame-pointer): Update documentation.
129 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
131 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
133 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
135 * cfg.c (scale_bbs_frequencies_int,
136 cale_bbs_frequencies_gcov_type): Remove.
137 * cfg.h (scale_bbs_frequencies_int,
138 cale_bbs_frequencies_gcov_type): Remove.
140 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
142 * tree-ssa-loop-manip.c
143 (scale_dominated_blocks_in_loop): Update to profile counts.
144 (tree_transform_and_unroll_loop): Likewise.
146 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
148 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
149 scale_bbs_frequencies_int.
151 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
153 * final.c (compute_alignments): Use counts rather than frequencies.
155 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
157 * cfgloopanal.c: Include sreal.h
158 (average_num_loop_insns): Use counts and sreal for accounting.
160 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
162 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
165 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
166 Kito Cheng <kito.cheng@gmail.com>
168 * config/nds32/constraints.md: Provide more constraints.
169 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
170 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
171 support constraints usage.
173 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
175 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
177 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
178 Kito Cheng <kito.cheng@gmail.com>
180 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
181 * config/nds32/nds32.opt: Refine the layout.
182 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
183 TARGET_EXT_STRING): Support new options.
184 * config/nds32/nds32.h: Likewise.
185 * config/nds32/nds32.md: Likewise.
186 * config/nds32/nds32-predicates.c: Likewise.
187 * config/nds32/constraints.md: Likewise.
188 * common/config/nds32/nds32-common.c: Likewise.
190 2017-11-16 Julia Koval <julia.koval@intel.com>
193 * config/i386/gfniintrin.h: Add sse check.
194 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
196 2017-11-16 Julia Koval <julia.koval@intel.com>
198 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
199 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
200 (ix86_handle_option): Handle -mavx512vbmi2.
201 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
202 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
203 * config/i386/i386-c.c (__AVX512VBMI2__): New.
204 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
205 (ix86_valid_target_attribute_inner_p): Ditto.
206 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
207 * config/i386/i386.opt (mavx512vbmi2): New option.
208 * doc/invoke.texi: Add new option.
210 [2017-11-16 Julia Koval <julia.koval@intel.com>
212 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
213 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
214 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
215 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
216 _mm512_gf2p8mul_epi8): New intrinsics.
217 * config/i386/i386-builtin-types.def
218 (V64QI_FTYPE_V64QI_V64QI): New type.
219 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
220 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
221 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
222 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
223 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
224 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
226 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
228 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
229 explicitly as a stream of bytes.
231 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
233 * config/rs6000/altivec.h (vec_xst_be): New #define.
234 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
235 and externalize from *altivec_vperm_<mode>_internal.
236 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
238 (XL_BE_V8HI): Likewise.
239 (XL_BE_V4SI): Likewise.
240 (XL_BE_V4SI): Likewise.
241 (XL_BE_V2DI): Likewise.
242 (XL_BE_V4SF): Likewise.
243 (XL_BE_V2DF): Likewise.
244 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
245 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
246 all array entries with these keys: VSX_BUILTIN_VEC_XL,
247 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
248 VSX_BUILTIN_VEC_XST_BE.
249 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
250 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
252 (altivec_init_builtins): Replace conditional calls to def_builtin
253 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
254 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
255 with unconditional calls. Remove calls to def_builtin for
256 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
257 __builtin_vec_xst_be.
258 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
259 to define_expand, and add alternate RTL generation for P8.
260 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
262 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
263 add alternate RTL generation for P8.
264 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
265 vsx_ld_elemrev_v16qi.
266 (vsx_st_elemrev_v8hi): Convert define_insn
267 to define_expand, and add alternate RTL generation for P8.
268 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
270 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
271 add alternate RTL generation for P8.
272 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
273 vsx_st_elemrev_v16qi.
275 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
278 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
279 TARGET_AVX512ER check.
280 (ix86_option_override_internal): Set MASK_VZEROUPPER if
281 neither -mzeroupper nor -mno-zeroupper is used and
282 TARGET_EMIT_VZEROUPPER is set.
283 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
284 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
286 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
288 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
289 folding of vector compares.
290 (fold_build_vec_cmp): New helper function.
291 (fold_compare_helper): New helper function.
292 (builtin_function_type): Add compare builtins to the list of functions
293 having unsigned arguments. Cosmetic updates to comment indentation.
294 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
295 the not+eq combination.
297 2017-11-15 Bin Cheng <bin.cheng@arm.com>
299 PR tree-optimization/82726
300 PR tree-optimization/70754
301 * tree-predcom.c (order_drefs_by_pos): New function.
302 (combine_chains): Move code setting has_max_use_after to...
303 (try_combine_chains): ...here. New parameter. Sort combined chains
304 according to position information.
305 (tree_predictive_commoning_loop): Update call to above function.
306 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
308 2017-11-15 Bin Cheng <bin.cheng@arm.com>
310 PR tree-optimization/82726
312 2017-01-23 Bin Cheng <bin.cheng@arm.com>
314 PR tree-optimization/70754
315 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
316 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
317 combined stmt before it if not NULL.
318 (combine_chains): Process refs reversely and compute dominance point
322 2017-02-23 Bin Cheng <bin.cheng@arm.com>
324 PR tree-optimization/79663
325 * tree-predcom.c (combine_chains): Process refs in reverse order
326 only for ZERO length chains, and add explaining comment.
328 2017-11-15 Tamar Christina <tamar.christina@arm.com>
330 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
331 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
332 * doc/invoke.texi (ARM Options): Add armv8.3-a.
334 2017-11-15 Tamar Christina <tamar.christina@arm.com>
336 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
338 2017-11-15 Martin Liska <mliska@suse.cz>
340 * tree-cfg.c (pass_warn_function_return::execute):
341 Compare warn_return_type for greater than zero.
343 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
347 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
348 to return true on Xeon and not on Xeon Phi.
349 (ix86_check_avx256_register): Changed to ...
350 (ix86_check_avx_upper_register): ... this. Add extra check for
351 VALID_AVX512F_REG_OR_XI_MODE.
352 (ix86_avx_u128_mode_needed): Changed
353 ix86_check_avx256_register to ix86_check_avx_upper_register.
354 (ix86_check_avx256_stores): Changed to ...
355 (ix86_check_avx_upper_stores): ... this. Changed
356 ix86_check_avx256_register to ix86_check_avx_upper_register.
357 (ix86_avx_u128_mode_after): Changed
358 avx_reg256_found to avx_upper_reg_found. Changed
359 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
360 (ix86_avx_u128_mode_entry): Changed
361 ix86_check_avx256_register to ix86_check_avx_upper_register.
362 (ix86_avx_u128_mode_exit): Ditto.
363 * config/i386/i386.h: (host_detect_local_cpu): New define.
365 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
367 * config/arm/xgene1.md (xgene1): Split into automatons
368 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
369 (xgene1_f_load): Adjust reservations and/or types.
370 (xgene1_f_store): Likewise.
371 (xgene1_load_pair): Likewise.
372 (xgene1_store_pair): Likewise.
373 (xgene1_fp_load1): Likewise.
374 (xgene1_load1): Likewise.
375 (xgene1_store1): Likewise.
376 (xgene1_move): Likewise.
377 (xgene1_alu): Likewise.
378 (xgene1_simd): Likewise.
379 (xgene1_bfm): Likewise.
380 (xgene1_neon_load1): Likewise.
381 (xgene1_neon_store1): Likewise.
382 (xgene1_neon_logic): Likewise.
383 (xgene1_neon_st1): Likewise.
384 (xgene1_neon_ld1r): Likewise.
385 (xgene1_alu_cond): Added.
386 (xgene1_shift_reg): Likwise.
387 (xgene1_bfx): Likewise.
388 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
390 2017-11-15 Jakub Jelinek <jakub@redhat.com>
393 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
395 (can_widen_mult_without_libcall): New function.
396 (expand_mul_overflow): If only checking unsigned mul overflow,
397 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
398 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
399 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
400 (expand_DIVMOD): Formatting fix.
401 * expmed.h (expand_mult): Add NO_LIBCALL argument.
402 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
403 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
405 PR tree-optimization/82977
406 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
407 constructed temporary to strlen_to_stridx.put.
409 2017-11-15 Martin Liska <mliska@suse.cz>
411 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
412 * configure: Regenerate.
414 2017-11-15 Martin Liska <mliska@suse.cz>
417 * config/sh/sh-mem.cc: Use proper probability for
420 2017-11-14 Jeff Law <law@redhat.com>
422 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
423 the red zone for stack_clash_protection_final_dynamic_probe targets
424 when the total dynamic stack size is zero bytes.
426 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
427 blocks is post order.
429 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
431 * dumpfile.h (TDF_COMPARE_DEBUG): New.
432 * final.c (rest_of_clean_state): Set it for the
433 -fcompare-debug dump.
434 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
435 class when TDF_COMPARE_DEBUG is set.
437 * dwarf2out.c (gen_producer_string): Discard
440 2017-11-15 Joseph Myers <joseph@codesourcery.com>
443 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
444 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
445 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
446 (__TGMATH_REAL_2_3): Remove macros.
447 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
448 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
450 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
452 (remquo): Define using __TGMATH_REAL_3.
454 2017-11-14 Jeff Law <law@redhat.com>
456 * vr-values.c: New file with contents extracted from tree-vrp.c.
457 * Makefile.in (OBJS): Add vr-values.o
458 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
459 (set_value_range, set_and_canonicalize_value_range): Likewise.
460 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
461 (value_range_constant_singleton, symbolic_range_p): Likewise.
462 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
463 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
464 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
465 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
466 (find_case_label_range, find_case_label_index): Likewise.
467 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
468 (range_int_cst_singleton_p, value_inside_range): Likewise.
469 (get_single_symbol): Likewise.
470 (switch_update): Move structure definition here.
471 (to_remove_edges, to_update_switch_stmts): Provide externs.
472 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
473 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
474 (vrp_val_is_min, set_value_range): Likewise.
475 (set_and_canonicalize_value_range, copy_value_range): Likewise.
476 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
477 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
478 (range_is_nonnull, range_int_cst_p): Likewwise.
479 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
480 (get_single_symbol, operand_less_p): Likewise
481 (compare_values_warnv, compare_values): Likewise.
482 (value_inside_range, value_range_constant_singleton): Likewise.
483 (zero_nonzero_bitgs_from_vr): Likewise.
484 (extract_range_from_binary_expr_1): Likewise.
485 (overflow_comparison_p): Likewise.
486 (to_remove_edges, to_update_switch_stmts): Likewise.
487 (find_case_label-index, find_case_label_range): Likewise.
488 (switch_update, set_value_range_to_nonnegative): Remove.
489 (set_value_range_to_truthvalue): Likewise.
490 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
491 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
492 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
493 (find_case_label_ranges, test_for_singularity): Likewise.
494 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
495 (x_vr_values): Move to its remaining use site.
497 2017-11-10 Jeff Law <law@redhat.com>
499 * vr-values.h (VR_INITIALIZER): Move #define here.
500 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
501 * Makefile.in (OBJS): Add tree-evrp.o
502 * tree-vrp.h (assert_info): Move structure definition here.
503 (set_value_range_to_varying): Prototype.
504 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
505 (infer_value_range, register_edge_assert_for): Likewise.
506 (stmt_interesting_for_vrp): Likewise.
507 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
508 (set_value_range_to_varying): No longer static.
509 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
510 (infer_value_range, register_edge_assert_for): Likewise.
512 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
514 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
515 generate the XXBRD instruction.
517 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
518 (rs6000_builtin_type_compatible): Treat _Float128 and long double
519 as being compatible if -mabi=ieeelongdouble.
520 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
521 to setup float128 built-ins with hardware support.
522 (BU_FLOAT128_HW_2): Likewise.
523 (BU_FLOAT128_HW_3): Likewise.
524 (BU_FLOAT128_HW_VSX_1): Likewise.
525 (BU_FLOAT128_HW_VSX_2): Likewise.
526 (scalar_extract_expq): Change float128 built-in functions to
527 accommodate having both KFmode and TFmode functions. Use the
528 KFmode variant as the default.
529 (scalar_extract_sigq): Likewise.
530 (scalar_test_neg_qp): Likewise.
531 (scalar_insert_exp_q): Likewise.
532 (scalar_insert_exp_qp): Likewise.
533 (scalar_test_data_class_qp): Likewise.
534 (sqrtf128_round_to_odd): Delete processing the round to odd
535 built-in functions as special built-in functions, and define them
536 as float128 built-ins. Use the KFmode variant as the default.
537 (truncf128_round_to_odd): Likewise.
538 (addf128_round_to_odd): Likewise.
539 (subf128_round_to_odd): Likewise.
540 (mulf128_round_to_odd): Likewise.
541 (divf128_round_to_odd): Likewise.
542 (fmaf128_round_to_odd): Likewise.
543 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
544 support for KFmode and TFmode xststdcqp calls.
545 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
546 point, switch the built-in handlers for the get/set float128
547 exponent, get float128 mantissa, float128 test built-ins, and the
548 float128 round to odd built-in functions. Eliminate creating the
549 float128 round to odd built-in functions as special built-ins.
550 (rs6000_init_builtins): Eliminate special creation of the float128
551 round to odd built-in functions.
552 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
553 function insns to support both TFmode and KFmode variants.
554 (xsxsigqp_<mode>): Likewise.
555 (xsiexpqpf_<mode>): Likewise.
556 (xsiexpqp_<mode>): Likewise.
557 (xststdcqp_<mode>): Likewise.
558 (xststdcnegqp_<mode>): Likewise.
559 (xststdcqp_<mode>): Likewise.
561 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
563 * tree-ssa-threadupdate.c (compute_path_counts): Remove
564 unused path_in_freq_ptr parameter.
565 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
567 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
569 * ipa-inline.c (edge_badness): Dump sreal frequency.
570 (compute_inlined_call_time): Match natural implementaiton ...
571 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
572 forgotten division by CGRAPH_FREQ_BASE.
574 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
576 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
577 Solaris 11. Update comment.
578 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
580 * config/sol2.h (STARTFILE_SPEC): Likewise.
581 * configure: Regenerate.
583 2017-11-14 Carl Love <cel@us.ibm.com>
585 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
586 le_ and be_ prefixes to swap* variables. Remove
587 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
590 2017-11-14 Jason Merrill <jason@redhat.com>
592 Support GTY((cache)) on hash_map.
593 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
594 (ggc_cache_remove): Override it instead of ggc_mx.
595 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
596 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
597 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
598 (simple_cache_map_traits): Override maybe_mx.
599 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
600 (hash_map): Friend gt_cleare_cache.
601 (gt_cleare_cache): New.
602 * tree.h (tree_cache_traits): New hash_map traits class.
603 (tree_cache_map): New typedef.
605 2017-11-14 Richard Biener <rguenther@suse.de>
607 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
608 paramter and handling.
609 (cleanup_control_flow_bb): Likewise.
610 (cleanup_control_flow_pre): New helper performing a DFS walk
611 to call cleanup_control_flow_bb in PRE order.
612 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
613 via cleanup_control_flow_pre.
615 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
617 * config/aarch64/aarch64-simd.md
618 (aarch64_simd_bsl<mode>_internal): Remove DImode.
619 (*aarch64_simd_bsl<mode>_alt): Likewise.
620 (aarch64_simd_bsldi_internal): New.
621 (aarch64_simd_bsldi_alt): Likewise.
623 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
625 * tracer.c (better_p): Do not compare frequencies.
626 * reg-stack.c (better_edge): Likewise.
627 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
630 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
632 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
633 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
634 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
635 * lto-streamer-in.c (input_function): Use update_max_bb_count.
636 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
637 * predict.c (maybe_hot_frequency_p): Inline to ...
638 (maybe_hot_count_p): ... here; rewrite to counts.
639 (counts_to_freqs): Rename to ...
640 (update_max_bb_count): ... this one.
641 (expensive_function_p): Use counts.
642 (estimate_bb_frequencies): Update.
643 (rebuild_frequencies): Update.
644 * predict.h (counts_to_freqs): Rename to ...
645 (update_max_bb_count): ... this one.
646 * profile.c (compute_branch_probabilities): Add debug info
647 * tree-inline.c (expand_call_inline): Update debug info.
648 (optimize_inline_calls): Use update_max_bb_count..
649 (tree_function_versioning): Use update_max_bb_count..
650 * value-prof.c (gimple_value_profile_transformations):
651 Do not use update_max_bb_count.
653 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
655 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
656 always use frequencies.
658 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
660 * bb-reorder.c: Remove frequencies from comments.
661 (better_edge_p): Use profile counts.
662 (find_traces): Dump profile counts.
663 (rotate_loop): Use profile counts.
664 (find_traces_1_round): Likewise.
665 (connect_better_edge_p): Use counts instead of probabilities for
667 (copy_bb_p): Drop early check for non-0 frequency.
668 (sanitize_hot_paths): Update comments.
670 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
672 * ipa-split.c (struct split_point): Add count.
673 (consider_split): Do not compute incoming frequency; compute incoming
674 count and store it to split_point.
675 (split_function): Set count of the call to split part correctly.
677 2017-11-13 Carl Love <cel@us.ibm.com>
679 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
681 2017-11-13 Tom Tromey <tom@tromey.com>
683 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
685 2017-11-13 Carl Love <cel@us.ibm.com>
687 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
688 Add support for builtins:
689 unsigned int vec_first_{,miss}_match_{,or_eos}index,
690 vector {un,}signed {char,int,short},
691 vector {un,}signed {char,int,short}) arguments.
692 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
693 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
694 Add BU_P9V_AV_2 expansions for the builtins.
695 * config/rs6000/altivec.h (vec_first_match_index,
696 vec_first_mismatch_index, vec_first_match_or_eos_index,
697 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
698 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
699 new extern declaration.
700 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
701 * config/rs6000/vsx.md (first_match_index_<mode>,
702 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
703 first_mismatch_or_eos_index_<mode>): Add define expand.
704 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
705 * doc/extend.texi: Update the built-in documenation file for the new
708 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
710 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
711 into the min/max operations for _Float<N> and _Float<N>X types.
713 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
716 * dwarf2out.c (do_eh_frame): New static variable.
717 (dwarf2out_begin_prologue): Set it.
718 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
720 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
722 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
724 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
726 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
729 * debug.h (dwarf2out_do_eh_frame): Declare.
730 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
731 (dwarf2out_do_frame): Use it.
732 (dwarf2out_do_cfi_asm): Likewise.
733 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
734 (dwarf2out_assembly_start): Likewise.
735 (dwarf2out_begin_prologue): Fix comment.
736 * toplev.c (compile_file): Always call dwarf2out_frame_finish
737 if the target needs either debug or unwind DWARF2 info.
738 * lto-opts.c (lto_write_options): Do not save -fexceptions,
739 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
740 -ftrapping-math, -ftrapv and -fwrapv.
742 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
744 * cgraph.c (cgraph_edge::sreal_frequency): New function.
745 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
746 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
747 (estimate_edge_size_and_time): Likewise.
748 (ipa_merge_fn_summary_after_inlining): Likewise.
749 * ipa-inline.c (cgraph_freq_base_rec): Remove.
750 (compute_uninlined_call_time): Use sreal_frequency.
751 (compute_inlined_call_time): Likewise.
752 (ipa_inline): Do not initialize cgraph_freq_base_rec.
753 * profile-count.c: Include sreal.h.
754 (profile_count::to_sreal_scale): New.
755 * profile-count.h: Forward declare sreal.
756 (profile_count::to_sreal_scale): Declare.
758 2017-11-13 Nathan Sidwell <nathan@acm.org>
760 * diagnostic.c (maybe_line_and_column): New.
761 (diagnostic_get_location_text): Use it.
762 (diagnostic_report_current_module): Likewise.
763 (test_diagnostic_get_location_text): Add tests.
765 2017-11-13 Luis Machado <luis.machado@linaro.org>
767 * doc/md.texi (Specifying processor pipeline description): Fix
768 incorrect latency for the div instruction example.
770 2017-11-13 Jakub Jelinek <jakub@redhat.com>
772 PR tree-optimization/78821
773 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
774 that bit_not_p is the same.
775 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
776 (split_group): Count precisely bit_not_p bits in each statement.
777 (invert_op): New function.
778 (imm_store_chain_info::output_merged_store): Use invert_op to
779 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
780 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
782 2017-11-13 Martin Liska <mliska@suse.cz>
784 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
785 (struct source_info): Likewise.
786 (add_branch_counts): Likewise.
787 (add_line_counts): Likewise.
788 (function_summary): Likewise.
789 (output_intermediate_line): Likewise.
790 (generate_results): Likewise.
792 2017-11-13 Martin Liska <mliska@suse.cz>
794 * gcov.c (struct block_info): Remove typedef for block_t.
795 (struct line_info): Likewise.
796 (line_info::has_block): Likewise.
797 (EXIT_BLOCK): Likewise.
800 (get_cycles_count): Likewise.
801 (process_file): Likewise.
802 (read_graph_file): Likewise.
803 (solve_flow_graph): Likewise.
804 (find_exception_blocks): Likewise.
805 (add_line_counts): Likewise.
806 (accumulate_line_info): Likewise.
807 (output_line_details): Likewise.
809 2017-11-13 Martin Liska <mliska@suse.cz>
811 * gcov.c (struct arc_info): Remove typedef for arc_t.
812 (struct line_info): Likewise.
813 (add_branch_counts): Likewise.
814 (output_branch_count): Likewise.
815 (function_info::~function_info): Likewise.
817 (output_intermediate_line): Likewise.
818 (read_graph_file): Likewise.
819 (solve_flow_graph): Likewise.
820 (find_exception_blocks): Likewise.
821 (add_line_counts): Likewise.
822 (accumulate_line_info): Likewise.
823 (output_line_details): Likewise.
824 (output_function_details): Likewise.
826 2017-11-13 Martin Liska <mliska@suse.cz>
828 * gcov.c (struct function_info): Remove typedef for function_t.
829 (struct source_info): Likewise.
830 (source_info::get_functions_at_location): Likewise.
831 (solve_flow_graph): Likewise.
832 (find_exception_blocks): Likewise.
833 (add_line_counts): Likewise.
834 (output_intermediate_file): Likewise.
835 (process_file): Likewise.
836 (generate_results): Likewise.
837 (release_structures): Likewise.
838 (read_graph_file): Likewise.
839 (read_count_file): Likewise.
840 (accumulate_line_counts): Likewise.
841 (output_lines): Likewise.
843 2017-11-13 Martin Liska <mliska@suse.cz>
845 * gcov.c (function_info::function_info): Remove num_counts
846 and add vector<gcov_type>.
847 (function_info::~function_info): Use the vector.
848 (process_file): Likewise.
849 (read_graph_file): Likewise.
850 (read_count_file): Likewise.
851 (solve_flow_graph): Likewise.
853 2017-11-13 Martin Liska <mliska@suse.cz>
855 * gcov.c (function_info::is_artificial): New function.
856 (process_file): Erase all artificial early.
857 (generate_results): Skip as all artificial are already
860 2017-11-13 Martin Liska <mliska@suse.cz>
862 * gcov.c (read_graph_file): Store to global vector of functions.
863 (read_count_file): Iterate the vector.
864 (process_file): Likewise.
865 (generate_results): Likewise.
866 (release_structures): Likewise.
868 2017-11-13 Jakub Jelinek <jakub@redhat.com>
870 PR tree-optimization/82954
871 * gimple-ssa-store-merging.c
872 (imm_store_chain_info::coalesce_immediate_stores): If
873 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
875 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
877 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
878 Upddate call to ENDIAN_LANE_N.
879 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
880 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
881 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
882 and use aarch64_endian_lane_rtx.
883 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
885 2017-11-12 Tom de Vries <tom@codesourcery.com>
887 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
889 2017-11-12 Tom de Vries <tom@codesourcery.com>
891 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
893 2017-11-12 Tom de Vries <tom@codesourcery.com>
895 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
897 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
898 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
899 * defaults.h (ASM_OUTPUT_LABELREF): Same.
901 2017-11-11 Martin Sebor <msebor@redhat.com>
904 * doc/extend.texi (attribute nonstring): Remove spurious argument.
907 * prefic.c (translate_name): Replace strncpy with memcpy to
908 avoid -Wstringop-truncation.
910 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
912 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
914 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
916 * predict.c (maybe_hot_frequency_p): Do not use cfun.
918 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
920 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
923 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
925 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
926 enable generating XXBRH if the value is in a vector register.
927 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
928 value is in a vector register.
929 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
930 register to register bswap64's instead of doing the GPR sequence
931 used on previous machines.
932 (bswapdi2_xxbrd): New insn.
933 (bswapdi2_reg): Disallow on ISA 3.0.
934 (register to register bswap64 splitter): Do not split the insn on
935 ISA 3.0 systems that use XXBRD.
937 2017-11-10 Martin Sebor <msebor@redhat.com>
940 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
942 (find_subframework_file): Same.
944 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
946 * auto-profile.c (afdo_indirect_call): Drop frequency.
947 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
948 (cgraph_node::create_edge): Drop frequency argument.
949 (cgraph_node::create_indirect_edge): Drop frequency argument.
950 (cgraph_edge::make_speculative): Drop frequency arguments.
951 (cgraph_edge::resolve_speculation): Do not update frequencies
952 (cgraph_edge::dump_edge_flags): Do not dump frequency.
953 (cgraph_node::dump): Check consistency in IPA mode.
954 (cgraph_edge::maybe_hot_p): Use IPA counter.
955 (cgraph_edge::verify_count_and_frequency): Rename to ...
956 (cgraph_edge::verify_count): ... this one; drop frequency checking.
957 (cgraph_node::verify_node): Update.
958 * cgraph.h (struct cgraph_edge): Drop frequency.
959 (cgraph_edge::frequency): New function.
960 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
962 (cgraph_edge::rebuild_edges): Likewise.
963 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
964 (duplicate_thunk_for_node): Do not pass frequency.
965 (cgraph_node::create_clone): Scale only counts.
966 (cgraph_node::create_virtual_clone): Do not pass frequency.
967 (cgraph_node::create_edge_including_clones): Do not pass frequency.
968 (cgraph_node::create_version_clone): Do not pass frequency.
969 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
970 (cgraph_node::expand_thunk): Do not pass frequency.
971 (cgraph_node::create_wrapper): Do not pass frequency.
972 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
974 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
975 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
976 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
977 (gather_caller_stats): Use frequency function.
978 (ipcp_cloning_candidate_p): Use frequency function.
979 (ipcp_propagate_stage): Use frequency function.
980 (get_info_about_necessary_edges): Use frequency function.
981 (update_profiling_info): Update only IPA profile.
982 (update_specialized_profile): Use frequency functoin.
983 (perhaps_add_new_callers): Update only IPA profile.
984 * ipa-devirt.c (ipa_devirt): Use IPA profile.
985 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
986 (dump_ipa_call_summary): Use frequency function.
987 (estimate_edge_size_and_time): Use frequency function.
988 (ipa_merge_fn_summary_after_inlining): Use frequency function.
989 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
990 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
991 (update_noncloned_counts): ... ths one; scale counts only.
992 (clone_inlined_nodes): Do not scale frequency.
993 (inline_call): Do not pass frequency.
994 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
995 (compute_inlined_call_time): Use IPA profile.
996 (want_inline_small_function_p): Use IPA profile.
997 (want_inline_self_recursive_call_p): Use IPA profile.
998 (edge_badness): Use IPA profile.
999 (lookup_recursive_calls): Use IPA profile.
1000 (recursive_inlining): Do not pass frequency.
1001 (resolve_noninline_speculation): Do not update frequency.
1002 (inline_small_functions): Collect max of IPA profile.
1003 (dump_overall_stats): Dump IPA porfile.
1004 (dump_inline_stats): Dump IPA porfile.
1005 (ipa_inline): Collect IPA stats.
1006 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1007 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1008 (ipa_propagate_frequency): Use frequency function.
1009 (ipa_profile): Cleanup.
1010 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1011 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1012 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1013 (input_node): Do not stream frequency.
1014 (input_edge): Do not stream frequency.
1015 (merge_profile_summaries): Scale only IPA profiles.
1016 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1017 * predict.c (drop_profile): Do not recompute frequency.
1018 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1019 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1020 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1021 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1022 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1023 * tree-inline.c (copy_bb): Do not scale frequency.
1024 (expand_call_inline): Do not scale frequency.
1025 (tree_function_versioning): Do not scale frequency.
1026 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1028 2017-11-10 Julia Koval <julia.koval@intel.com>
1030 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1031 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1032 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1033 (_mm256_maskz_gf2p8affine_epi64_epi8)
1034 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1035 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1036 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1037 (__builtin_ia32_vgf2p8affineqb_v32qi)
1038 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1039 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1041 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1045 (arm_option_override): Refactor.
1046 (arm_option_reconfigure_globals): New.
1047 (arm_options_perform_arch_sanity_checks): New.
1048 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1050 (arm_options_perform_arch_sanity_checks): Likewise
1052 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1054 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1055 (power9-qpmul): New.
1056 * rs6000/rs6000.md ("type" attr): Add qmul.
1057 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1058 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1059 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1061 2017-11-10 Martin Sebor <msebor@redhat.com>
1064 * builtins.c (compute_objsize): Handle arrays that
1065 compute_builtin_object_size likes to fail for. Make extern.
1066 * builtins.h (compute_objsize): Declare.
1067 (check_strncpy_sizes): New function.
1068 (expand_builtin_strncpy): Call check_strncpy_sizes.
1069 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1070 -Wstringop-truncation.
1071 (gimple_fold_builtin_strncat): Same.
1072 * gimple.c (gimple_build_call_from_tree): Set call location.
1073 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1074 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1075 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1076 (handle_builtin_strlen): Use strlen_to_stridx.
1077 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1079 Use strlen_to_stridx.
1080 (pass_strlen::execute): Release strlen_to_stridx.
1081 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1082 (-Wstringop-truncation): Document new option.
1084 2017-11-10 Martin Liska <mliska@suse.cz>
1086 PR gcov-profile/82702
1087 * gcov.c (main): Handle intermediate files in a different
1089 (get_gcov_intermediate_filename): New function.
1090 (output_gcov_file): Remove support of intermediate files.
1091 (generate_results): Allocate intermediate file.
1092 (release_structures): Clean-up properly fn_end.
1093 (output_intermediate_file): Start iterating with line 1.
1095 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1097 PR tree-optimization/82929
1098 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1099 ops_swapped_p non-static data member.
1100 (store_immediate_info::store_immediate_info): Clear it.
1101 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1102 ops set ops_swapped_p.
1103 (count_multiple_uses): Handle ops_swapped_p.
1105 2017-11-10 Martin Liska <mliska@suse.cz>
1107 * coverage.c (coverage_init): Stream information about
1108 support of has_unexecuted_blocks.
1109 * doc/gcov.texi: Document that.
1110 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1111 * gcov.c (read_graph_file): Likewise.
1112 (output_line_beginning): Fix a small issue with
1115 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1117 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1118 reference of trivial component.
1120 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1123 * gimple-ssa-store-merging.c
1124 (pass_store_merging::terminate_all_aliasing_chains): For
1125 gimple_store_p stmts also call refs_output_dependent_p.
1127 PR rtl-optimization/82913
1128 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1131 2017-11-09 Jeff Law <law@redhat.com>
1133 * vr-values.h: New file with vr_values class.
1134 * tree-vrp.c: Include vr-values.h
1135 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1136 data objects into the vr_values class.
1137 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1138 (get_value_range): Make it a member function within vr_values class.
1139 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1140 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1141 (op_with_constant_singleton_value_range): Likewise.
1142 (extract_range_for_var_from_comparison_expr): Likewise.
1143 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1144 (extract_range_from_binary_expr): Likewise.
1145 (extract_range_from_unary_expr): Likewise.
1146 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1147 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1148 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1149 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1150 (compare_name_with_value, compare_names): Likewise.
1151 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1152 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1153 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1154 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1155 (extract_range_from_phi_node): Likewise.
1156 (simplify_truth_ops_using_ranges): Likewise.
1157 (simplify_div_or_mod_using_ranges): Likewise.
1158 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1159 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1160 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1161 (simplify_float_conversion_using_ranges): Likewise.
1162 (simplify_internal_call_using_ranges): Likewise.
1163 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1164 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1165 poitner via x_vr_values for calls into gimple folder.
1166 (vrp_initialize_lattice): Make this the vr_values ctor.
1167 (vrp_free_lattice): Make this the vr_values dtor.
1168 (set_vr_value): New function.
1169 (class vrp_prop): Add vr_values data member. Add various member
1170 functions as well as member functions that delegate to vr_values.
1171 (check_array_ref): Make a member function within vrp_prop class.
1172 (search_for_addr_array, vrp_initialize): Likewise.
1173 (vrp_finalize): Likewise. Revamp to avoid direct access to
1174 vr_value, values_propagated, etc.
1175 (check_array_bounds): Extract vrp_prop class instance pointer from
1176 walk info structure. Use it to call member functions.
1177 (check_all_array_refs): Make a member function within vrp_prop class.
1178 Smuggle class instance pointer via walk info structure.
1179 (x_vr_values): New local static.
1180 (vrp_valueize): Use x_vr_values to get class instance.
1181 (vr_valueize_1): Likewise.
1182 (class vrp_folder): Add vr_values data member. Add various member
1183 functions as well as member functions that delegate to vr_values.
1184 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1185 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1186 class instance from vr_values. Use it to call member functions.
1187 (vrp_dom_walker): Add vr_values data member.
1188 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1189 instance via x_vr_values.
1190 (identify_jump_threads): Accept vr_values as argument. Store
1191 it into the walker structure.
1192 (evrp_dom_walker): Add vr_values class data member. Add various
1194 (evrp_dom_walker::try_find_new_range): Use vr_values data
1195 member to access the memory allocator.
1196 (evrp_dom_walker::before_dom_children): Store vr_values class
1197 instance into the vrp_folder class.
1198 (evrp_dom_walker::push_value_range): Rework to avoid direct
1199 access to num_vr_values and vr_value.
1200 (evrp_dom_walker::pop_value_range): Likewise.
1201 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1202 Use vr_values to get to dump_all_value_ranges member function.
1203 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1204 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1205 Pass vr_values class instance down to identify_jump_threads.
1206 Remove call to vrp_free_lattice.
1207 (debug_all_value_ranges): Remove.
1209 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1210 (vrp_folder): Likewise.
1212 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1213 Get it from the existing bitmap instead.
1214 (vrp_intersect_ranges_1): Likewise.
1216 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1218 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1220 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1221 set bit_not_p to it.
1222 (imm_store_chain_info::coalesce_immediate_stores): Break group
1223 if bit_not_p is different.
1224 (count_multiple_uses, split_group,
1225 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1226 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1227 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1228 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1229 to store_immediate_info ctor.
1231 2017-11-09 Jim Wilson <jimw@sifive.com>
1233 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1234 (scan_prog_file): Likewise.
1236 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1238 * bb-reorder.c (max_entry_frequency): Remove.
1239 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1240 connect_traces, push_to_next_round_p): Remove prototypes.
1241 (find_traces_1_round): Use counts only.
1242 (push_to_next_round_p): Likewise.
1243 (find_traces): Likewise.
1244 (rotate_loop): Likewise.
1245 (find_traces_1_round): Likewise.
1246 (connect_traces): Likewise.
1247 (edge_order): Likewise.
1249 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1251 * config/arm/arm.c (output_return_instruction): Add comments to
1252 indicate requirement for cmse_nonsecure_entry return to account
1253 for the size of clearing instruction output here.
1254 (thumb_exit): Likewise.
1255 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1256 return in hardfloat mode.
1258 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1260 * config/rs6000/rs6000.c (machine_function): Add a bool,
1261 "toc_is_wrapped_separately".
1262 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1263 if it wasn't explicitly set or unset, we are optimizing for speed, and
1264 doing separate shrink-wrapping.
1265 (rs6000_get_separate_components): Enable the TOC component if
1266 saving the TOC register in the prologue.
1267 (rs6000_components_for_bb): Handle the TOC component.
1268 (rs6000_emit_prologue_components): Store the TOC register where needed.
1269 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1270 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1272 2017-11-09 Martin Jambor <mjambor@suse.cz>
1274 * ipa-param-manipulation.c: New file.
1275 * ipa-param-manipulation.h: Likewise.
1276 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1277 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1278 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1279 (ipa_parm_adjustment): Likewise.
1280 (ipa_parm_adjustment_vec): Likewise.
1281 (ipa_get_vector_of_formal_parms): Moved declaration to
1282 ipa-param-manipulation.h.
1283 (ipa_get_vector_of_formal_parm_types): Likewise.
1284 (ipa_modify_formal_parameters): Likewise.
1285 (ipa_modify_call_arguments): Likewise.
1286 (ipa_combine_adjustments): Likewise.
1287 (ipa_dump_param_adjustments): Likewise.
1288 (ipa_modify_expr): Likewise.
1289 (ipa_get_adjustment_candidate): Likewise.
1290 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1291 ipa-param-manipulation.c.
1292 (ipa_get_vector_of_formal_parm_types): Likewise.
1293 (ipa_modify_formal_parameters): Likewise.
1294 (ipa_modify_call_arguments): Likewise.
1295 (ipa_modify_expr): Likewise.
1296 (get_ssa_base_param): Likewise.
1297 (ipa_get_adjustment_candidate): Likewise.
1298 (index_in_adjustments_multiple_times_p): Likewise.
1299 (ipa_combine_adjustments): Likewise.
1300 (ipa_dump_param_adjustments): Likewise.
1301 * tree-sra.c: Also include ipa-param-manipulation.h
1302 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1305 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1306 Alan Hayward <alan.hayward@arm.com>
1307 David Sherwood <david.sherwood@arm.com>
1309 * doc/sourcebuild.texi (vect_masked_store): Document.
1311 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1312 Alan Hayward <alan.hayward@arm.com>
1313 David Sherwood <david.sherwood@arm.com>
1315 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1317 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1318 Alan Hayward <alan.hayward@arm.com>
1319 David Sherwood <david.sherwood@arm.com>
1321 * doc/sourcebuild.texi (vect_variable_length): Document.
1323 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1324 Alan Hayward <alan.hayward@arm.com>
1325 David Sherwood <david.sherwood@arm.com>
1327 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1329 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1330 Alan Hayward <alan.hayward@arm.com>
1331 David Sherwood <david.sherwood@arm.com>
1333 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1335 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1336 Alan Hayward <alan.hayward@arm.com>
1337 David Sherwood <david.sherwood@arm.com>
1339 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1340 previously undocumented selectors.
1341 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1343 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1345 * doc/rtl.texi (const_vector): Say that elements can be
1346 const_wide_ints too.
1347 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1348 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1349 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1350 * optabs.c (expand_vector_broadcast): Likewise.
1352 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1353 Alan Hayward <alan.hayward@arm.com>
1354 David Sherwood <david.sherwood@arm.com>
1356 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1357 scaled index even if the unscaled address was invalid.
1358 Don't increase the complexity of using a scale in that case.
1360 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1361 Alan Hayward <alan.hayward@arm.com>
1362 David Sherwood <david.sherwood@arm.com>
1364 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1365 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1366 UNITS_PER_WORD bytes.
1367 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1368 based on REGMODE_NATURAL_SIZE of the inner mode.
1369 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1370 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1371 * expmed.c (lowpart_bit_field_p): Divide the value up into
1372 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1373 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1374 whether something is likely to occupy more than one register.
1376 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1379 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1380 profile_count::adjust_for_ipa_scaling.
1381 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1382 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1384 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1386 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1388 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1389 (split_group): Add total_orig and total_new arguments, estimate the
1390 number of statements related to the store group without store merging
1391 and with store merging.
1392 (imm_store_chain_info::output_merged_store): Adjust split_group
1393 callers, punt if estimated number of statements with store merging
1394 is not smaller than estimated number of statements without it.
1396 (handled_load): Remove has_single_use checks.
1397 (pass_store_merging::process_store): Likewise.
1399 2017-11-09 Richard Biener <rguenther@suse.de>
1401 PR tree-optimization/82902
1402 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1404 2017-11-09 Martin Liska <mliska@suse.cz>
1407 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1410 2017-11-09 Martin Liska <mliska@suse.cz>
1412 PR tree-optimization/82669
1413 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1415 2017-11-09 Martin Liska <mliska@suse.cz>
1417 PR gcov-profile/48463
1418 * coverage.c (coverage_begin_function): Output also end locus
1419 of a function and information whether the function is
1421 * gcov-dump.c (tag_function): Parse and print the information.
1422 * gcov.c (INCLUDE_MAP): Add include.
1423 (INCLUDE_SET): Likewise.
1424 (struct line_info): Move earlier in the source file because
1425 of vector<line_info> in function_info structure.
1426 (line_info::line_info): Likewise.
1427 (line_info::has_block): Likewise.
1428 (struct source_info): Add new member index.
1429 (source_info::get_functions_at_location): New function.
1430 (function_info::group_line_p): New function.
1431 (output_intermediate_line): New function.
1432 (output_intermediate_file): Use the mentioned function.
1433 (struct function_start): New.
1434 (struct function_start_pair_hash): Likewise.
1435 (process_file): Add code that identifies group functions.
1436 Assign lines either to global or function scope.
1437 (generate_results): Skip artificial functions.
1438 (find_source): Assign index for each source file.
1439 (read_graph_file): Read new flag artificial and end_line.
1440 (add_line_counts): Assign it either to global of function scope.
1441 (accumulate_line_counts): Isolate core of the function to
1442 accumulate_line_info and call it for both function and global
1444 (accumulate_line_info): New function.
1445 (output_line_beginning): Fix GNU coding style.
1446 (print_source_line): New function.
1447 (output_line_details): Likewise.
1448 (output_function_details): Likewise.
1449 (output_lines): Iterate both source (global) scope and function
1451 (struct function_line_start_cmp): New class.
1452 * doc/gcov.texi: Reflect changes in documentation.
1454 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1457 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1458 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1459 and similarly for not instead of neg.
1461 2017-11-08 Andi Kleen <ak@linux.intel.com>
1463 * config/i386/i386.opt: Add -mforce-indirect-call.
1464 * config/i386/predicates.md: Check for flag_force_indirect_call.
1465 * doc/invoke.texi: Document -mforce-indirect-call
1467 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1469 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1471 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1472 config/riscv/riscv.c (predict.h): New include.
1473 (riscv_slow_unaligned_access_p): No longer static.
1474 (riscv_block_move_straight): Add require.
1475 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1477 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1480 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1481 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1482 nonimmediate_operand predicate for operand 1 instead of
1485 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1487 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1489 * config/aarch64/constraints.md (Uml): New constraint.
1490 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1493 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1495 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1496 of two vec_duplicates into a vec_concat.
1498 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1500 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1501 Simplify vec_merge of vec_duplicate and vec_concat.
1502 * config/aarch64/constraints.md (Utq): New constraint.
1503 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1506 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1508 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1509 Simplify vec_merge of vec_duplicate and const_vector.
1510 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1512 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1513 mode iterator. Update predicate on operand 1 to
1514 handle non-const_vec constants. Delete constraints.
1515 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1517 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1519 PR tree-optimization/78821
1520 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1522 (store_operand_info::store_operand_info): Initialize it to false.
1523 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1524 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1525 store in the group, and if chain_info is non-NULL, to ignore altogether
1527 (compatible_load_p): Fail if bit_not_p does not match.
1528 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1529 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1530 (pass_store_merging::process_store): Adjust
1531 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1532 call terminate_all_aliasing_chains newly when adding a store into
1533 a chain with non-NULL chain_info.
1535 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1537 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1539 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1541 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1543 (aarch64_layout_frame): Initialise emit_frame_chain.
1544 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1545 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1547 2017-11-08 Martin Liska <mliska@suse.cz>
1549 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1550 of gimple_call_internal_p.
1552 2017-11-07 Tom de Vries <tom@codesourcery.com>
1554 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1556 2017-11-07 Tom de Vries <tom@codesourcery.com>
1558 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1561 2017-11-08 Martin Liska <mliska@suse.cz>
1564 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1566 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1568 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1570 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1573 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1574 (*cmp<mode>_ccz_1): New insn with $k alternative.
1577 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1578 RTX_COMM_COMPARE as commutative as well.
1579 (ix86_binary_operator_ok): Formatting fix.
1580 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1581 *<code><mode>3<mask_name><round_saeonly_name>,
1582 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1583 *<s>mul<mode>3_highpart<mask_name>,
1584 *vec_widen_umult_even_v16si<mask_name>,
1585 *vec_widen_umult_even_v8si<mask_name>,
1586 *vec_widen_umult_even_v4si<mask_name>,
1587 *vec_widen_smult_even_v16si<mask_name>,
1588 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1589 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1590 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1591 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1592 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1593 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1594 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1595 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1596 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1597 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1598 ix86_binary_operator_ok. Formatting fixes.
1599 (*<plusminus_insn><mode>3<mask_name><round_name>,
1600 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1603 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1605 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1606 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1607 eq and ne if TARGET_ISEL.
1608 (cmp): New code_iterator.
1609 (UNS, UNSU_, UNSIK): New code_attrs.
1610 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1611 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1613 ("eq<mode>3"): ... this.
1614 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1616 ("ne<mode>3"): ... this.
1618 2017-11-07 Julia Koval <julia.koval@intel.com>
1621 * common/config/i386/i386-common.c
1622 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1623 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1624 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1625 * config/i386/i386.opt: Ditto.
1626 * config/i386/i386.c (ix86_target_string): Ditto.
1627 (ix86_option_override_internal): Ditto.
1628 (ix86_init_mpx_builtins): Move MPX to args2.
1629 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1630 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1631 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1632 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1633 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1634 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1635 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1637 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1640 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1641 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1642 (zero-extendsidi peephole2): Remove peephole.
1644 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1647 * stor-layout.c: Include attribs.h.
1648 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1649 explicit lookup of "aligned" attribute.
1651 2017-11-07 Andrew Waterman <andrew@sifive.com>
1653 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1654 (riscv_expand_block_move): Likewise.
1655 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1657 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1658 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1659 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1660 (riscv_adjust_block_mem): Likewise.
1661 (riscv_block_move_loop): Likewise.
1662 (riscv_expand_block_move): Likewise.
1663 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1665 2017-11-07 Michael Clark <michaeljclark@mac.com>
1667 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1668 (MUSL_DYNAMIC_LINKER): Likewise.
1670 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1672 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1673 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1676 2017-11-07 Richard Biener <rguenther@suse.de>
1678 * match.pd: Fix build.
1680 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1681 Jackson Woodruff <jackson.woodruff@arm.com>
1683 PR tree-optimization/71026
1684 * match.pd: Canonicalize negate in division.
1686 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1689 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1691 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1693 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1694 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1696 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1698 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1699 non-scalar integral types.
1700 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1701 (-(A-B), -(~A)): New transformations.
1703 2017-11-07 Tom de Vries <tom@codesourcery.com>
1705 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1706 semicolon after "do {} while (0)".
1707 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1708 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1709 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1710 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1711 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1713 2017-11-07 Tom de Vries <tom@codesourcery.com>
1715 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1716 after "do {} while (0)".
1717 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1718 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1719 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1720 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1721 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1723 2017-11-07 Tom de Vries <tom@codesourcery.com>
1726 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1728 (arm_rtx_costs_internal): Add missing semicolon after
1729 HANDLE_NARROW_SHIFT_ARITH call.
1731 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1733 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1734 disable isel if it was not set explicitly.
1736 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1738 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1740 (add_type_attribute) likewise.
1742 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1744 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1745 of struct ix86_frame.
1746 (ix86_initial_elimination_offset): Likewise.
1747 (ix86_expand_split_stack_prologue): Likewise.
1749 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1751 * tree-vrp.h (enum value_range_type): Update stale comment.
1753 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1754 Alan Hayward <alan.hayward@arm.com>
1755 David Sherwood <david.sherwood@arm.com>
1757 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1758 (aarch64_expand_vec_perm_const): Take the number of units too.
1759 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1760 (aarch64_expand_vec_perm_const): Likewise.
1761 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1762 (vec_perm<mode>): Update accordingly.
1764 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1765 Alan Hayward <alan.hayward@arm.com>
1766 David Sherwood <david.sherwood@arm.com>
1768 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1769 Take the number of units too.
1770 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1771 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1772 but check for a vector mode before rather than after the call.
1773 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1774 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1775 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1776 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1777 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1778 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1779 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1780 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1781 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1782 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1783 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1784 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1785 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1786 (aarch64_sqdmull2_n<mode>): Update accordingly.
1788 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1789 Alan Hayward <alan.hayward@arm.com>
1790 David Sherwood <david.sherwood@arm.com>
1792 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1793 the number of units too.
1794 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1795 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1796 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1797 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1798 (vec_store_lanesxi<mode>): Update accordingly.
1800 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1801 Alan Hayward <alan.hayward@arm.com>
1802 David Sherwood <david.sherwood@arm.com>
1804 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1805 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1806 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1807 of units rather than the mode.
1808 * config/aarch64/iterators.md (nunits): New mode attribute.
1809 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1810 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1811 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1812 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1813 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1814 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1815 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1816 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1817 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1818 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1819 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1820 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1821 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1822 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1823 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1824 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1825 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1826 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1827 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1828 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1829 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1830 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1831 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1832 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1833 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1834 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1835 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1836 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1837 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1838 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1839 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1840 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1841 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1842 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1843 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1844 (aarch64_simd_vec_setv2di): Likewise.
1846 2017-11-06 Carl Love <cel@us.ibm.com>
1848 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1850 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1851 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1853 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1855 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1856 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1857 Add power 8 macro expansions.
1858 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1859 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1860 power 8 instructions. (VSX_XXBR): Add iterator.
1862 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1864 * config/arm/arm.md (predicable_short_it): Change default to "no",
1865 improve documentation, remove uses that are identical to the default.
1866 (enabled_for_depr_it): Rename to enabled_for_short_it.
1867 * gcc/config/arm/arm-fixed.md (predicable_short_it):
1868 Remove default uses.
1869 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1870 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1871 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1872 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1874 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1877 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1878 float128 helper macros, which are no longer used after deleting
1879 the old 'q' built-in functions, and moving the round to odd
1880 built-in functions to being special built-in functions.
1881 (BU_FLOAT128_2): Likewise.
1882 (BU_FLOAT128_1_HW): Likewise.
1883 (BU_FLOAT128_2_HW): Likewise.
1884 (BU_FLOAT128_3_HW): Likewise.
1885 (FABSQ): Delete old 'q' built-in functions.
1886 (COPYSIGNQ): Likewise.
1887 (SQRTF128_ODD): Move round to odd built-in functions to be
1888 special built-in functions, so that we can handle
1889 -mabi=ieeelongdouble.
1890 (TRUNCF128_ODD): Likewise.
1891 (ADDF128_ODD): Likewise.
1892 (SUBF128_ODD): Likewise.
1893 (MULF128_ODD): Likewise.
1894 (DIVF128_ODD): Likewise.
1895 (FMAF128_ODD): Likewise.
1896 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1897 built-in names to 'f128'.
1898 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1899 old 'q' built-in functions, as the machine independent code for
1900 'f128' built-in functions handles this.
1901 (rs6000_expand_builtin): Add expansion for float128 round to odd
1902 functions, keying off on -mabi=ieeelongdouble of whether to use
1903 the KFmode or TFmode variant.
1904 (rs6000_init_builtins): Initialize the _Float128 round to odd
1906 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1907 _Float128 'q' built-in functions are now mapped into the new
1908 'f128' built-in functions.
1910 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1912 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1913 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1915 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1918 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1919 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1920 * toplev.c: Include "ipa-fnsummary.h".
1921 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1923 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1925 PR tree-optimization/82838
1926 * gimple-ssa-store-merging.c
1927 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1928 on a separate gimple_seq which is then appended to seq.
1930 2017-11-06 Jeff Law <law@redhat.com>
1933 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1934 (get_probe_interval): New functions.
1935 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1936 (ix86_adjust_stack_and_probe): Likewise.
1937 (output_adjust_stack_and_probe): Likewise.
1938 (ix86_emit_probe_stack_range): Likewise.
1939 (ix86_expand_prologue): Likewise.
1941 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1943 PR tree-optimization/82816
1944 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1945 if the modes of the two types are the same.
1946 (convert_plusminus_to_widen): Likewise.
1948 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1950 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1952 (usadv16qi): New define_expand.
1953 (usadv8hi): New define_expand.
1955 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1958 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1960 (inline_transform): Likewise.
1961 * predict.c (counts_to_freqs): Remove useless conditional.
1962 * profile-count.h (profile_count::apply_scale): Move sanity check.
1963 * tree-inline.c (copy_bb): Always scale.
1964 (copy_cfg_body): Likewise.
1966 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1969 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1971 (*cmp_ite0): Add enabled_for_depr_it attribute.
1972 (*cmp_ite1): Likewise.
1974 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1976 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1979 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1981 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1982 return the value by pointer.
1983 (extract_range_from_multiplicative_op_1): Update accordingly.
1984 Return as soon as an operation fails.
1986 2017-11-05 Tom de Vries <tom@codesourcery.com>
1989 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1990 (DEF_SANITIZER_BUILTIN): ... here.
1991 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1992 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1994 2017-11-05 Tom de Vries <tom@codesourcery.com>
1997 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1999 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2000 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2001 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2003 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2004 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2005 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2007 2017-11-05 Tom de Vries <tom@codesourcery.com>
2010 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2013 2017-11-04 Michael Clark <michaeljclark@mac.com>
2015 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2016 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2018 (*addsi3_extended): Likewise.
2019 (*addsi3_extended2): Likewise.
2020 (<optab>si3): Likewise.
2021 (<optab>di3): Likewise.
2022 (<optab><mode>3): Likewise.
2023 (<*optabe>si3_internal): Likewise.
2024 (zero_extendqi<SUPERQI:mode>2): Likewise.
2025 (*add<mode>hi3): Likewise.
2026 (*xor<mode>hi3): Likewise.
2027 (<optab>di3): Likewise.
2028 (*<optab>si3_extend): Likewise.
2029 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2030 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2031 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2033 2017-11-04 Andrew Waterman <andrew@sifive.com>
2035 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2036 TARGET_STRICT_ALIGN based upon -mtune argument.
2038 2017-11-04 Andrew Waterman <andrew@sifive.com>
2040 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2042 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2044 config/i386/i386.c (choose_basereg): Use optional scratch
2045 register and add assertion.
2046 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2047 needed, and don't allocate stack.
2048 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2049 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2050 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2052 2017-11-03 Jeff Law <law@redhat.com>
2054 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2055 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2056 to probe at the start of a noreturn function.
2058 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2060 PR tree-optimization/78821
2061 * gimple-ssa-store-merging.c: Update the file comment.
2062 (MAX_STORE_ALIAS_CHECKS): Define.
2063 (struct store_operand_info): New type.
2064 (store_operand_info::store_operand_info): New constructor.
2065 (struct store_immediate_info): Add rhs_code and ops data members.
2066 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2067 and op1r arguments to the ctor, initialize corresponding data members.
2068 (struct merged_store_group): Add load_align_base and load_align
2070 (merged_store_group::merged_store_group): Initialize them.
2071 (merged_store_group::do_merge): Update them.
2072 (merged_store_group::apply_stores): Pick the constant for
2073 encode_tree_to_bitpos from one of the two operands, or skip
2074 encode_tree_to_bitpos if neither operand is a constant.
2075 (class pass_store_merging): Add process_store method decl. Remove
2076 bool argument from terminate_all_aliasing_chains method decl.
2077 (pass_store_merging::terminate_all_aliasing_chains): Remove
2078 var_offset_p argument and corresponding handling.
2079 (stmts_may_clobber_ref_p): New function.
2080 (compatible_load_p): New function.
2081 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2082 if there is overlap and rhs_code is not INTEGER_CST. For
2083 non-overlapping stores terminate group if rhs is not mergeable.
2084 (get_alias_type_for_stmts): Change first argument from
2085 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2086 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2087 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2089 (get_location_for_stmts): Change first argument from
2090 auto_vec<gimple *> & to vec<gimple *> &.
2091 (struct split_store): Remove orig_stmts data member, add orig_stores.
2092 (split_store::split_store): Create orig_stores rather than orig_stmts.
2093 (find_constituent_stmts): Renamed to ...
2094 (find_constituent_stores): ... this. Change second argument from
2095 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2096 to info structures rather than the statements.
2097 (split_group): Rename ALLOW_UNALIGNED argument to
2098 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2099 it. Adjust find_constituent_stores caller.
2100 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2101 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2102 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2103 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2104 (mem_valid_for_store_merging): New function.
2105 (handled_load): New function.
2106 (pass_store_merging::process_store): New method.
2107 (pass_store_merging::execute): Use process_store method. Adjust
2108 terminate_all_aliasing_chains caller.
2110 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2112 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2113 Return true for more constants, symbols and label references.
2114 (aarch64_valid_floating_const): Remove unused function.
2116 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2119 * config/aarch64/aarch64.c (aarch64_layout_frame):
2120 Undo forcing of LR at bottom of frame.
2122 2017-11-03 Jeff Law <law@redhat.com>
2125 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2126 for int_registers_saved.
2128 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2129 extracted from tree-ssa-dom.c.
2130 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2131 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2132 (record_equivalences_from_incoming_edge): Add additional argument
2133 to single_pred_edge_ignoring_loop_edges call.
2134 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2135 (uncprop_dom_walker::before_dom_children): Add additional argument
2136 to single_pred_edge_ignoring_loop_edges call.
2137 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2138 single_pred_edge_ignoring_loop_edges rather than open coding.
2139 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2141 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2143 * match.pd (-(-A)): Rewrite.
2145 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2147 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2148 (rs6000_emit_int_cmove): New declaration.
2149 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2150 (rs6000_emit_sISEL): Delete.
2151 (rs6000_emit_int_cmove): Make non-static.
2152 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2153 instead of rs6000_emit_sISEL.
2155 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2157 * asan.c (create_cond_insert_point): Maintain profile.
2158 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2160 * basic-block.h (struct basic_block_def): Remove frequency.
2161 (EDGE_FREQUENCY): Use to_frequency
2162 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2164 (find_traces): Update to use to_frequency.
2165 (find_traces_1_round): Likewise; use only IPA counts.
2166 (bb_to_key): Likewise.
2167 (connect_traces): Use IPA counts only.
2168 (copy_bb_p): Update to use to_frequency.
2169 (fix_up_crossing_landing_pad): Likewise.
2170 (sanitize_hot_paths): Likewise.
2171 * bt-load.c (basic_block_freq): Likewise.
2172 * cfg.c (init_flow): Set count_max to uninitialized.
2173 (check_bb_profile): Remove frequencies; check counts.
2174 (dump_bb_info): Do not dump frequencies.
2175 (update_bb_profile_for_threading): Update counts only.
2176 (scale_bbs_frequencies_int): Likewise.
2177 (MAX_SAFE_MULTIPLIER): Remove.
2178 (scale_bbs_frequencies_gcov_type): Update counts only.
2179 (scale_bbs_frequencies_profile_count): Update counts only.
2180 (scale_bbs_frequencies): Update counts only.
2181 * cfg.h (struct control_flow_graph): Add count-max.
2182 (update_bb_profile_for_threading): Update prototype.
2183 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2184 (find_many_sub_basic_blocks): Likewise.
2185 * cfgcleanup.c (try_forward_edges): Likewise.
2186 (try_crossjump_to_edge): Likewise.
2187 * cfgexpand.c (expand_gimple_cond): Likewise.
2188 (expand_gimple_tailcall): Likewise.
2189 (construct_init_block): Likewise.
2190 (construct_exit_block): Likewise.
2191 * cfghooks.c (verify_flow_info): Check consistency of counts.
2192 (dump_bb_for_graph): Do not dump frequencies.
2193 (split_block_1): Do not update frequencies.
2194 (split_edge): Do not update frequencies.
2195 (make_forwarder_block): Do not update frequencies.
2196 (duplicate_block): Do not update frequencies.
2197 (account_profile_record): Do not update frequencies.
2198 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2199 for global heuristics.
2200 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2201 (expected_loop_iterations_unbounded): Use counts only.
2202 * cfgloopmanip.c (scale_loop_profile): Simplify.
2203 (create_empty_loop_on_edge): Simplify
2205 (duplicate_loop_to_header_edge): Simplify
2206 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2207 (update_br_prob_note): Take care of removing note when profile
2209 (relink_block_chain): Do not dump frequency.
2210 (rtl_account_profile_record): Use to_frequency.
2211 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2212 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2213 (cgraph_update_edges_for_call_stmt_node): Likewise.
2214 (cgraph_edge::verify_count_and_frequency): Update.
2215 (cgraph_node::verify_node): Temporarily disable frequency verification.
2216 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2217 to_cgraph_frequency.
2218 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2219 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2221 (cgraph_node::expand_thunk): Update profile.
2222 * except.c (dw2_build_landing_pads): Do not update frequency.
2223 * final.c (compute_alignments): Use to_frequency.
2224 (dump_basic_block_info): Do not dump frequency.
2225 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2226 (dump_gimple_bb_header): Do not dump frequency.
2227 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2229 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2230 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2231 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2232 (init_before_recovery): Do not update frequency.
2233 (sched_create_recovery_edges): Do not update frequency.
2234 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2235 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2236 (ipa_cp_c_finalize): Set max_count to uninitialized.
2237 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2238 (param_change_prob): Use counts.
2239 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2241 * ipa-split.c (consider_split): Use to_frequency.
2242 (split_function): Use to_frequency.
2243 * ira-build.c (loop_compare_func): Likewise.
2244 (mark_loops_for_removal): Likewise.
2245 (mark_all_loops_for_removal): Likewise.
2246 * loop-doloop.c (doloop_modify): Do not update frequency.
2247 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2249 * lto-streamer-in.c (input_function): Update count_max.
2250 * omp-expand.c (expand_omp_taskreg): Update count_max.
2251 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2252 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2253 (maybe_hot_count_p): Use ipa counts only.
2254 (maybe_hot_bb_p): Simplify.
2255 (maybe_hot_edge_p): Simplify.
2256 (probably_never_executed): Do not take frequency argument.
2257 (probably_never_executed_bb_p): Do not pass frequency.
2258 (probably_never_executed_edge_p): Likewise.
2259 (combine_predictions_for_bb): Check that profile is nonzero.
2260 (propagate_freq): Do not set frequency.
2261 (drop_profile): Simplify.
2262 (counts_to_freqs): Simplify.
2263 (expensive_function_p): Use to_frequency.
2264 (propagate_unlikely_bbs_forward): Simplify.
2265 (determine_unlikely_bbs): Simplify.
2266 (estimate_bb_frequencies): Add hack to silence graphite issues.
2267 (compute_function_frequency): Use ipa counts.
2268 (pass_profile::execute): Update.
2269 (rebuild_frequencies): Use counts only.
2270 (force_edge_cold): Use counts only.
2271 * profile-count.c (profile_count::dump): Dump new count types.
2272 (profile_count::differs_from_p): Check compatiblity.
2273 (profile_count::to_frequency): New function.
2274 (profile_count::to_cgraph_frequency): New function.
2275 * profile-count.h (struct function): Declare.
2276 (enum profile_quality): Add profile_guessed_local and
2277 profile_guessed_global0.
2278 (class profile_proability): Decrease number of bits to 29;
2279 update from_reg_br_prob_note and to_reg_br_prob_note.
2280 (class profile_count: Update comment; decrease number of bits
2281 to 61. Check compatibility.
2282 (profile_count::compatible_p): New private member function.
2283 (profile_count::ipa_p): New member function.
2284 (profile_count::operator<): Handle global zero correctly.
2285 (profile_count::operator>): Handle global zero correctly.
2286 (profile_count::operator<=): Handle global zero correctly.
2287 (profile_count::operator>=): Handle global zero correctly.
2288 (profile_count::nonzero_p): New member function.
2289 (profile_count::force_nonzero): New member function.
2290 (profile_count::max): New member function.
2291 (profile_count::apply_scale): Handle IPA scalling.
2292 (profile_count::guessed_local): New member function.
2293 (profile_count::global0): New member function.
2294 (profile_count::ipa): New member function.
2295 (profile_count::to_frequency): Declare.
2296 (profile_count::to_cgraph_frequency): Declare.
2297 * profile.c (OVERLAP_BASE): Delete.
2298 (compute_frequency_overlap): Delete.
2299 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2300 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2301 * sched-ebb.c (rank): Use counts only.
2302 * shrink-wrap.c (handle_simple_exit): Use counts only.
2303 (try_shrink_wrapping): Use counts only.
2304 (place_prologue_for_one_component): Use counts only.
2305 * tracer.c (find_best_predecessor): Use to_frequency.
2306 (find_trace): Use to_frequency.
2307 (tail_duplicate): Use to_frequency.
2308 * trans-mem.c (expand_transaction): Do not update frequency.
2309 * tree-call-cdce.c: Do not update frequency.
2310 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2311 (gimple_merge_blocks): Likewise.
2312 (gimple_split_edge): Likewise.
2313 (gimple_duplicate_sese_region): Likewise.
2314 (gimple_duplicate_sese_tail): Likewise.
2315 (move_sese_region_to_fn): Likewise.
2316 (gimple_account_profile_record): Likewise.
2317 (insert_cond_bb): Likewise.
2318 * tree-complex.c (expand_complex_div_wide): Likewise.
2319 * tree-eh.c (lower_resx): Update profile.
2320 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2322 (initialize_cfun): Do not initialize frequencies
2323 (freqs_to_counts): Delete.
2324 (copy_cfg_body): Ignore count parameter.
2325 (copy_body): Update.
2326 (expand_call_inline): Update count_max.
2327 (optimize_inline_calls): Update count_max.
2328 (tree_function_versioning): Update count_max.
2329 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2330 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2332 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2333 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2334 (try_peel_loop): Likewise.
2335 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2337 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2338 (tree_transform_and_unroll_loop): Do not use frequencies
2339 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2340 Use reliable prediction only.
2341 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2342 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2343 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2344 probability scaling.
2345 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2346 not update frequency
2347 (any_remaining_duplicated_blocks): Likewise.
2348 (update_profile): Likewise.
2349 (estimated_freqs_path): Delete.
2350 (freqs_to_counts_path): Delete.
2351 (clear_counts_path): Delete.
2352 (ssa_fix_duplicate_block_edges): Likewise.
2353 (duplicate_thread_path): Likewise.
2354 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2355 * tree-tailcall.c (decrease_profile): Do not update frequency.
2356 (eliminate_tail_call): Likewise.
2357 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2358 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2359 (optimize_mask_stores): Likewise.
2360 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2361 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2362 (ubsan_expand_ptr_ifn): Update profile.
2363 * value-prof.c (gimple_ic): Simplify.
2364 * value-prof.h (gimple_ic): Update prototype.
2365 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2366 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2368 (want_inline_self_recursive_call_p): Likewise.
2369 (resolve_noninline_speculation): Only cummulate defined counts.
2370 (inline_small_functions): Use nonzero_p.
2371 (ipa_inline): Do not access freed node.
2373 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2375 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2376 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2378 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2380 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2381 non-legitimate address.
2383 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2385 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2386 (*lt0_<mode>di, *lt0_<mode>si): New.
2388 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2390 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2391 TARGET_PAIRED_FLOAT.
2393 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2394 Jim Wilson <jim.wilson@linaro.org>
2396 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2397 * config/aarch64/aarch64-tune.md: Regenerated.
2398 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2399 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2401 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2403 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2404 (arc_expand_prologue): Restore blink for millicode.
2405 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2407 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2410 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2411 gen_vec_duplicate after forcing the scalar into a register.
2413 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2415 * combine (try_combine): Print the insns input to try_combine to the
2418 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2421 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2422 Remove second argument from aarch64_process_target_attr call.
2423 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2425 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2427 (aarch64_handle_attr_arch): Remove second argument.
2428 (aarch64_handle_attr_cpu): Ditto.
2429 (aarch64_handle_attr_tune): Ditto.
2430 (aarch64_handle_attr_isa_flags): Ditto.
2431 (aarch64_process_one_target_attr): Ditto.
2432 (aarch64_process_target_attr): Ditto.
2433 (aarch64_option_valid_attribute_p): Remove second argument.
2434 on aarch64_process_target_attr call.
2436 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2438 * diagnostic.c: Include "selftest-diagnostic.h".
2439 (selftest::assert_location_text): New function.
2440 (selftest::test_diagnostic_get_location_text): New function.
2441 (selftest::diagnostic_c_tests): Call it.
2443 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2445 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2446 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2447 (class selftest::test_diagnostic_context): Move to...
2448 * selftest-diagnostic.c: New file.
2449 * selftest-diagnostic.h: New file.
2451 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2453 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2454 offset range for FT32B.
2455 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2456 * config/ft32/ft32.md: Add TARGET_NOPM.
2457 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2458 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2460 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2462 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2464 2017-11-02 Jeff Law <law@redhat.com>
2466 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2467 virtual keyword on FINAL OVERRIDE members.
2469 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2470 virtuals together. Add virtual destructor.
2471 (substitute_and_fold_engine): Similarly.
2473 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2475 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2477 2017-11-02 Richard Biener <rguenther@suse.de>
2479 PR tree-optimization/82795
2480 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2482 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2484 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2486 (gcc_GAS_CHECK_FEATURE): Remove.
2487 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2488 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2489 * configure: Regenerate.
2491 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2493 * config/arc/arc.c (hwloop_optimize): Account for empty
2496 2017-11-02 Richard Biener <rguenther@suse.de>
2499 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2500 Truncate ARRAY_REF index and element size.
2502 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2504 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2506 2017-11-01 Jeff Law <law@redhat.com>
2508 * tree-ssa-ccp.c (ccp_folder): New class derived from
2509 substitute_and_fold_engine.
2510 (ccp_folder::get_value): New member function.
2511 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2512 (ccp_fold_stmt): Remove prototype.
2513 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2514 * tree-ssa-copy.c (copy_folder): New class derived from
2515 substitute_and_fold_engine.
2516 (copy_folder::get_value): Renamed from get_value.
2517 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2518 * tree-vrp.c (vrp_folder): New class derived from
2519 substitute_and_fold_engine.
2520 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2521 (vrp_folder::get_value): New member function.
2522 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2523 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2524 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2525 provide a class interface to folder/substitute routines.
2526 (ssa_prop_fold_stmt_fn): Remove typedef.
2527 (ssa_prop_get_value_fn): Likewise.
2528 (subsitute_and_fold): Remove prototype.
2529 (replace_uses_in): Likewise.
2530 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2531 Renamed from replace_uses_in. Call the virtual member function
2532 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2533 (substitute_and_fold_dom_walker): Remove initialization of
2534 data member entries for calbacks. Add substitute_and_fold_engine
2535 member and initialize it.
2536 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2537 member functions for get_value, replace_phi_args_in c
2538 replace_uses_in, and fold_stmt calls.
2539 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2540 substitute_and_fold. Remove assert. Update ctor call.
2542 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2543 (ssa_prop_visit_phi_fn): Likewise.
2544 (class ssa_propagation_engine): New class to provide an interface
2546 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2548 (ssa_prop_visit_phi): Likewise.
2549 (ssa_propagation_engine::simulate_stmt): Moved into class.
2550 Call visit_phi/visit_stmt from the class rather than via
2551 file scoped static variables.
2552 (ssa_propagation_engine::simulate_block): Moved into class.
2553 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2554 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2555 set file scoped statics for the visit_stmt/visit_phi callbacks.
2556 * tree-complex.c (complex_propagate): New class derived from
2557 ssa_propagation_engine.
2558 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2559 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2560 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2562 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2563 ssa_propagation_engine.
2564 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2565 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2566 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2567 * tree-ssa-copy.c (copy_prop): New class derived from
2568 ssa_propagation_engine.
2569 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2570 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2571 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2572 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2573 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2574 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2575 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2577 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2579 PR rtl-optimization/82778
2580 PR rtl-optimization/82597
2581 * compare-elim.c (struct comparison): Add in_a_setter field.
2582 (find_comparison_dom_walker::before_dom_children): Remove killed
2583 bitmap and df_simulate_find_defs call, instead walk the defs.
2584 Compute last_setter and initialize in_a_setter. Merge definitions
2585 with first initialization for a few variables.
2586 (try_validate_parallel): Use insn_invalid_p instead of
2587 recog_memoized. Return insn rather than just the pattern.
2588 (try_merge_compare): Fix up comment. Don't uselessly test if
2589 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2591 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2594 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2595 Alan Hayward <alan.hayward@arm.com>
2596 David Sherwood <david.sherwood@arm.com>
2598 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2599 aarch64_hard_regno_nregs to get the number of registers
2602 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2603 Alan Hayward <alan.hayward@arm.com>
2604 David Sherwood <david.sherwood@arm.com>
2606 * config/aarch64/constraints.md (Upl): Rename to...
2608 * config/aarch64/aarch64.md
2609 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2612 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2613 Alan Hayward <alan.hayward@arm.com>
2614 David Sherwood <david.sherwood@arm.com>
2616 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2617 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2620 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2621 Alan Hayward <alan.hayward@arm.com>
2622 David Sherwood <david.sherwood@arm.com>
2624 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2625 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2626 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2628 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2629 of a single element.
2630 * config/aarch64/iterators.md: Add a comment above the permute
2631 unspecs to say that they are generated directly by
2632 aarch64_expand_vec_perm_const.
2633 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2635 2017-11-01 Nathan Sidwell <nathan@acm.org>
2637 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2639 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2641 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2642 and medany code models, and describe what they do.
2644 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2646 Revert accidental duplicate:
2648 * combine.c (can_change_dest_mode): Reject changes in
2649 REGMODE_NATURAL_SIZE.
2651 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2653 PR rtl-optimization/64682
2654 PR rtl-optimization/69567
2655 PR rtl-optimization/69737
2656 PR rtl-optimization/82683
2657 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2658 register mentioned in the note, drop the note, unless it came from I3,
2659 in which case it should go to I3 again.
2661 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2663 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2664 and return false if not.
2665 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2667 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2669 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2670 range is known to be empty.
2672 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2673 Alan Hayward <alan.hayward@arm.com>
2674 David Sherwood <david.sherwood@arm.com>
2676 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2677 and CONST_VECTOR_NUNITS instead of computing the number of units from
2678 the byte sizes of the vector and element.
2679 (simplify_binary_operation_1): Likewise.
2680 (simplify_const_binary_operation): Likewise.
2681 (simplify_ternary_operation): Likewise.
2683 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2684 Alan Hayward <alan.hayward@arm.com>
2685 David Sherwood <david.sherwood@arm.com>
2687 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2688 (int_mem_offset): ...this new function.
2689 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2690 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2691 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2694 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2695 Alan Hayward <alan.hayward@arm.com>
2696 David Sherwood <david.sherwood@arm.com>
2698 * lower-subreg.c (interesting_mode_p): New function.
2699 (compute_costs, find_decomposable_subregs, decompose_register)
2700 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2701 (resolve_clobber, dump_choices): Use it.
2703 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2704 Alan Hayward <alan.hayward@arm.com>
2705 David Sherwood <david.sherwood@arm.com>
2707 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2709 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2710 Alan Hayward <alan.hayward@arm.com>
2711 David Sherwood <david.sherwood@arm.com>
2713 * alias.c (find_base_value, find_base_term): Only process integer
2714 truncations. Check the precision rather than the size.
2716 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2717 Alan Hayward <alan.hayward@arm.com>
2718 David Sherwood <david.sherwood@arm.com>
2720 * machmode.h (is_narrower_int_mode): New function
2721 * optabs.c (expand_float, expand_fix): Use it.
2722 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2724 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2725 Alan Hayward <alan.hayward@arm.com>
2726 David Sherwood <david.sherwood@arm.com>
2728 * rtl.h (narrower_subreg_mode): New function.
2729 * ira-color.c (update_costs_from_allocno): Use it.
2731 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2732 Alan Hayward <alan.hayward@arm.com>
2733 David Sherwood <david.sherwood@arm.com>
2735 * optabs-query.h (convert_optab_p): New function, split out from...
2736 (convert_optab_handler): ...here.
2737 (widening_optab_handler): Delete.
2738 (find_widening_optab_handler): Remove permit_non_widening parameter.
2739 (find_widening_optab_handler_and_mode): Likewise. Provide an
2740 override that operates on mode class wrappers.
2741 * optabs-query.c (widening_optab_handler): Delete.
2742 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2743 parameter. Assert that the two modes are the same class and that
2744 the "from" mode is narrower than the "to" mode. Use
2745 convert_optab_handler instead of widening_optab_handler.
2746 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2747 instead of widening_optab_handler.
2748 * expr.c (expand_expr_real_2): Update calls to
2749 find_widening_optab_handler.
2750 * optabs.c (expand_widen_pattern_expr): Likewise.
2751 (expand_binop_directly): Take the insn_code as a parameter.
2752 (expand_binop): Only call find_widening_optab_handler for
2753 conversion optabs; use optab_handler otherwise. Update calls
2754 to find_widening_optab_handler and expand_binop_directly.
2755 Use convert_optab_handler instead of widening_optab_handler.
2756 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2757 find_widening_optab_handler and use scalar_mode rather than
2759 (convert_plusminus_to_widen): Likewise.
2761 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2762 Alan Hayward <alan.hayward@arm.com>
2763 David Sherwood <david.sherwood@arm.com>
2765 * machmode.h (fixed_size_mode): New class.
2766 * rtl.h (get_pool_mode): Return fixed_size_mode.
2767 * gengtype.c (main): Add fixed_size_mode.
2768 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2769 (get_raw_arg_mode): Likewise.
2770 * doc/tm.texi: Regenerate.
2771 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2772 * targhooks.c (default_get_reg_raw_mode): Likewise.
2773 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2774 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2775 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2776 (msp430_get_raw_result_mode): Likewise.
2777 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2778 * dbxout.c (dbxout_parms): Require fixed-size modes.
2779 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2780 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2781 * omp-low.c (lower_oacc_reductions): Likewise.
2782 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2783 (simplify_subreg): Update accordingly.
2784 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2785 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2786 that aren't fixed-size.
2787 (get_pool_mode): Return a fixed_size_mode.
2788 (output_constant_pool_2): Take a fixed_size_mode.
2790 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2791 Alan Hayward <alan.hayward@arm.com>
2792 David Sherwood <david.sherwood@arm.com>
2794 * doc/rtl.texi (vec_series): Document.
2795 (const): Say that the operand can be a vec_series.
2796 * rtl.def (VEC_SERIES): New rtx code.
2797 * rtl.h (const_vec_series_p_1): Declare.
2798 (const_vec_series_p): New function.
2799 * emit-rtl.h (gen_const_vec_series): Declare.
2800 (gen_vec_series): Likewise.
2801 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2802 (gen_vec_series): Likewise.
2803 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2804 * simplify-rtx.c (simplify_unary_operation): Handle negations
2806 (simplify_binary_operation_series): New function.
2807 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2808 (test_vector_ops_series): New function.
2809 (test_vector_ops): Call it.
2810 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2811 gen_const_vec_series.
2812 (altivec_lvsr): Likewise.
2813 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2815 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2816 Alan Hayward <alan.hayward@arm.com>
2817 David Sherwood <david.sherwood@arm.com>
2819 * doc/rtl.texi (const): Update description of address constants.
2820 Say that vector constants are allowed too.
2821 * common.md (E, F): Use CONSTANT_P instead of checking for
2823 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2824 checking for CONST_VECTOR.
2825 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2827 * expr.c (expand_expr_real_2): Check for vector modes instead
2828 of checking for CONST_VECTOR.
2829 * rtl.h (const_vec_p): New function.
2830 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2831 (unwrap_const_vec_duplicate): Handle them here too.
2833 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2834 David Malcolm <dmalcolm@redhat.com>
2835 Alan Hayward <alan.hayward@arm.com>
2836 David Sherwood <david.sherwood@arm.com>
2838 * rtl.h (vec_duplicate_p): New function.
2839 * selftest-rtl.c (assert_rtx_eq_at): New function.
2840 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2841 (assert_rtx_eq_at): Declare.
2842 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2843 * selftest-run-tests.c (selftest::run_tests): Call it.
2844 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2845 (simplify_unary_operation_1): Recursively handle vector duplicates.
2846 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2848 (simplify_subreg): Handle subregs of vector duplicates.
2849 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2850 (selftest::simplify_rtx_c_tests): New functions.
2852 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2853 Alan Hayward <alan.hayward@arm.com>
2854 David Sherwood <david.sherwood@arm.com>
2856 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2857 (gen_vec_duplicate): Likewise.
2858 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2860 (gen_const_vector): ...here.
2861 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2862 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2863 whose elements are all equal.
2864 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2865 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2866 (simplify_relational_operation): Likewise.
2867 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2869 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2870 (aarch64_expand_vector_init): Likewise.
2871 * config/arm/arm.c (neon_vdup_constant): Likewise.
2872 (neon_expand_vector_init): Likewise.
2873 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2874 (arm_block_set_unaligned_vect): Likewise.
2875 (arm_block_set_aligned_vect): Likewise.
2876 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2877 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2878 (expand_vec_perm_even_odd_pack): Likewise.
2879 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2880 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2881 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2882 gen_const_vec_duplicate.
2883 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2884 * config/mips/mips.c (mips_gen_const_int_vector): Use
2885 gen_const_vec_duplicate.
2886 (mips_expand_vector_init): Use CONST0_RTX.
2887 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2888 (define_split): Use gen_const_vec_duplicate.
2889 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2890 (define_split): Use gen_const_vec_duplicate.
2891 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2892 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2893 * config/spu/spu.c (spu_const): Likewise.
2895 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2896 Alan Hayward <alan.hayward@arm.com>
2897 David Sherwood <david.sherwood@arm.com>
2899 * combine.c (can_change_dest_mode): Reject changes in
2900 REGMODE_NATURAL_SIZE.
2902 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2904 * configure.ac (--enable-libssp): New.
2905 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2906 trying to determine target-specific default. Adjust indentation.
2907 * configure: Regenerated.
2908 * doc/install.texi (Configuration): Expand --disable-libssp
2911 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2913 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2916 2017-10-31 Martin Jambor <mjambor@suse.cz>
2919 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2921 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2923 * auto-profile.c (autofdo_source_profile::read): Use
2924 UNKNOWN_LOCATION rather than 0.
2925 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2926 (warning_at): ...this overload.
2927 (warning_at_rich_loc_n): Rename to...
2928 (warning_n): ...this overload.
2929 (error_at_rich_loc): Rename to...
2930 (error_at): ...this overload.
2931 (pedwarn_at_rich_loc): Rename to...
2932 (pedwarn): ...this overload.
2933 (permerror_at_rich_loc): Rename to...
2934 (permerror): ...this overload.
2935 (inform_at_rich_loc): Rename to...
2936 (inform): ...this overload.
2937 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2938 (diagnostic_n_impl_richloc): Rename to...
2939 (diagnostic_n_impl): ...this rich_location *-based decl.
2940 (inform_at_rich_loc): Rename to...
2941 (inform): ...this, and add an assertion.
2942 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2943 (warning_at_rich_loc): Rename to...
2944 (warning_at): ...this, and add an assertion.
2945 (warning_at_rich_loc_n): Rename to...
2946 (warning_n): ...this, and add an assertion.
2947 (warning_n): Update location_t-based implementation for removal of
2948 location_t-based diagnostic_n_impl.
2949 (pedwarn_at_rich_loc): Rename to...
2950 (pedwarn): ...this, and add an assertion.
2951 (permerror_at_rich_loc): Rename to...
2952 (permerror): ...this, and add an assertion.
2953 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2954 (error_at_rich_loc): Rename to...
2955 (error_at): ...this, and add an assertion.
2956 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2957 (driver::do_spec_on_infiles): Likewise.
2958 * substring-locations.c (format_warning_va): Update for renaming
2959 of inform_at_rich_loc.
2961 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2963 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2964 _Float<N>X built-in functions so that the variant without the
2965 "__builtin_" prefix is only enabled for the GNU C and Objective C
2966 languages when they are in non-strict ANSI/ISO mode.
2967 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2968 * target.def (floatn_builtin_p): Add a target hook to control
2969 whether _Float<N> and _Float<N>X built-in functions without the
2970 "__builtin_" prefix are enabled, and return true for C and
2971 Objective C in the default hook. Include langhooks.h in
2973 * targhooks.h (default_floatn_builtin_p): Likewise.
2974 * targhooks.c (default_floatn_builtin_p): Likewise.
2975 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2976 floatn_builtin_p target hook.
2977 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2979 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2980 Eric Botcazou <ebotcazou@adacore.com>
2982 PR rtl-optimization/81803
2983 * lra-constraints.c (curr_insn_transform): Also reload the whole
2984 register for a strict subreg no wider than a word if this is for
2985 a WORD_REGISTER_OPERATIONS target.
2987 2017-10-31 Jason Merrill <jason@redhat.com>
2989 * gdbinit.in: Skip over inlines from timevar.h.
2991 2017-10-31 Martin Liska <mliska@suse.cz>
2993 * doc/gcov.texi: Document new option.
2994 * gcov.c (print_usage): Likewise print it.
2995 (process_args): Support the argument.
2996 (format_count): New function.
2997 (format_gcov): Use the function.
2999 2017-10-31 Martin Liska <mliska@suse.cz>
3001 * gcov.c (struct name_map): do not use typedef.
3002 Define operator== and operator<.
3003 (name_search): Remove.
3004 (name_sort): Remove.
3005 (main): Do not allocate names.
3006 (process_file): Add vertical space.
3007 (generate_results): Use std::find.
3008 (release_structures): Do not release memory.
3009 (find_source): Use std::find.
3011 2017-10-31 Martin Liska <mliska@suse.cz>
3013 * gcov.c (struct line_info): Remove it's typedef.
3014 (line_info::line_info): Add proper ctor.
3015 (line_info::has_block): Do not use a typedef.
3016 (struct source_info): Do not use typedef.
3017 (circuit): Likewise.
3018 (get_cycles_count): Likewise.
3019 (output_intermediate_file): Iterate via vector iterator.
3020 (add_line_counts): Use std::vector methods.
3021 (accumulate_line_counts): Likewise.
3022 (output_lines): Likewise.
3024 2017-10-31 Martin Liska <mliska@suse.cz>
3026 * gcov.c (struct source_info): Remove typedef.
3027 (source_info::source_info): Add proper ctor.
3028 (accumulate_line_counts): Use struct, not it's typedef.
3029 (output_gcov_file): Likewise.
3030 (output_lines): Likewise.
3031 (main): Do not allocate an array.
3032 (output_intermediate_file): Use size of vector container.
3033 (process_file): Resize the vector.
3034 (generate_results): Do not preallocate, use newly added vector
3036 (release_structures): Do not release sources.
3037 (find_source): Use vector methods.
3038 (add_line_counts): Do not use typedef.
3040 2017-10-31 Martin Liska <mliska@suse.cz>
3042 * doc/gcov.texi: Document that.
3043 * gcov.c (add_line_counts): Mark lines with a non-executed
3045 (output_line_beginning): Handle such lines.
3046 (output_lines): Pass new argument.
3047 (output_intermediate_file): Print it in intermediate format.
3049 2017-10-31 Martin Liska <mliska@suse.cz>
3051 * color-macros.h: New file.
3052 * diagnostic-color.c: Factor out color related to macros to
3054 * doc/gcov.texi: Document -k option.
3055 * gcov.c (INCLUDE_STRING): Include string.h.
3056 (print_usage): Add -k option.
3057 (process_args): Parse it.
3058 (pad_count_string): New function.
3059 (output_line_beginning): Likewise.
3060 (DEFAULT_LINE_START): New macro.
3061 (output_lines): Support color output.
3063 2017-10-31 Martin Liska <mliska@suse.cz>
3065 PR gcov-profile/82633
3066 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3067 their interaction with GCOV infrastructure.
3068 * configure.ac: Add -fkeep-{inline,static}-functions to
3070 * configure: Regenerate.
3072 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3075 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3077 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3080 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3081 into a register if it does not fit into an immediate offset field.
3083 2017-10-31 Olivier Hainque <hainque@adacore.com>
3085 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3087 2017-10-31 Julia Koval <julia.koval@intel.com>
3089 * config.gcc: Add gfniintrin.h.
3090 * config/i386/gfniintrin.h: New.
3091 * config/i386/i386-builtin-types.def
3092 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3093 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3094 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3095 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3096 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3097 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3098 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3099 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3100 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3101 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3102 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3103 * config/i386/immintrin.h: Include gfniintrin.h.
3104 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3106 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3108 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3110 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3112 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3113 (arm_ashldi3_1bit): Remove pattern.
3114 (ashrdi3): Remove shift by 1 expansion.
3115 (arm_ashrdi3_1bit): Remove pattern.
3116 (lshrdi3): Remove shift by 1 expansion.
3117 (arm_lshrdi3_1bit): Remove pattern.
3118 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3119 cost of ashldi3 by 1.
3120 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3121 (<shift>di3_neon): Likewise.
3123 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3125 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3126 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3127 and (*aarch64_simd_mov<VQ:mode>).
3128 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3129 pattern alternative.
3130 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3131 attributes to match pattern alternative.
3133 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3135 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3136 * config/rs6000/emmintrin.h: New file.
3137 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3139 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3141 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3142 * (movdi_vfp_cortexa8): Remove pattern.
3144 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3146 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3148 (Specific, alpha*-dec-osf5.1): Remove.
3149 (Specific, mips-sgi-irix5): Remove.
3150 (Specific, mips-sgi-irix6): Remove.
3152 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3155 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3156 arguments to clear_bit_region_be.
3158 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3160 * gimplify.c: Include memmodel.h.
3162 2017-10-30 Martin Jambor <mjambor@suse.cz>
3164 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3165 condition whether loop should be executed at all.
3167 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3169 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3170 gimple folding of vec_madd() intrinsics.
3171 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3172 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3173 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3175 2017-10-30 Richard Biener <rguenther@suse.de>
3177 PR tree-optimization/82762
3179 2017-10-23 Richard Biener <rguenther@suse.de>
3181 PR tree-optimization/82129
3183 2017-08-01 Richard Biener <rguenther@suse.de>
3185 PR tree-optimization/81181
3186 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3187 (compute_antic): ... end of iteration here.
3189 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3191 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3192 -std=iso9899:2017 and -std=gnu17.
3193 * doc/standards.texi (C Language): Document C17 support.
3194 * doc/cpp.texi (Overview): Mention -std=c17.
3195 (Standard Predefined Macros): Document C11 and C17 values of
3196 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3197 * doc/extend.texi (Inline): Do not list individual options for
3198 standards newer than C99.
3199 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3201 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3204 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3206 * asan.c (asan_finish_file): Align asan globals array by shadow
3209 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3212 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3213 (struct store_immediate_info): Add bitregion_start and bitregion_end
3215 (store_immediate_info::store_immediate_info): Add brs and bre
3216 arguments and initialize bitregion_{start,end} from those.
3217 (struct merged_store_group): Add bitregion_start, bitregion_end,
3218 align_base and mask fields. Drop unnecessary struct keyword from
3219 struct store_immediate_info. Add do_merge method.
3220 (clear_bit_region_be): Use memset instead of loop storing zeros.
3221 (merged_store_group::do_merge): New method.
3222 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3223 stores as long as the surrounding bitregions have no gaps.
3224 (merged_store_group::merge_overlapping): Use do_merge.
3225 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3226 is byte aligned, rather than requiring that start and width are
3227 byte aligned. Drop unnecessary struct keyword from
3228 struct store_immediate_info. Allocate and populate also mask array.
3229 Make start of the arrays relative to bitregion_start rather than
3230 start and size them according to bitregion_{end,start} difference.
3231 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3232 struct store_immediate_info.
3233 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3234 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3235 struct keyword from struct store_immediate_info.
3236 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3237 between stores as long as the surrounding bitregions have no gaps.
3239 (struct split_store): Add orig non-static data member.
3240 (split_store::split_store): Initialize orig to false.
3241 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3242 if there is exactly a single original stmt. Change stmts argument
3243 to pointer from reference, if NULL, don't push anything to it. Add
3244 first argument, use it to optimize skipping over orig stmts that
3245 are known to be before bitpos already. Simplify.
3246 (split_group): Return unsigned int count how many stores are or
3247 would be needed rather than a bool. Add allow_unaligned argument.
3248 Change split_stores argument from reference to pointer, if NULL,
3249 only do a dry run computing how many stores would be produced.
3250 Rewritten algorithm to use both alignment and misalign if
3251 !allow_unaligned and handle bitfield stores with gaps.
3252 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3253 from bitregion_start instead of start. Compute allow_unaligned
3254 here, if true, do 2 split_group dry runs to compute which one
3255 produces fewer stores and prefer aligned if equal. Punt if
3256 new count is bigger or equal than original before emitting any
3257 statements, rather than during that. Remove no longer needed
3258 new_ssa_names tracking. Replace num_stmts with
3259 split_stores.length (). Use 32-bit stack allocated entries
3260 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3261 if possible. Handle bitfields with gaps.
3262 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3263 Compute bitregion_{start,end} for the stores and construct
3264 store_immediate_info with that. Formatting fixes.
3266 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3269 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3270 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3272 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3274 * gimplify.c: Include tm_p.h.
3276 * common.opt (gcoff): Re-add as ignored option.
3277 (gcoff1, gcoff2, gcoff3): Likewise.
3279 * Makefile.in (OBJS): Delete sdbout.o.
3280 (GTFILES): Delete $(srcdir)/sdbout.c.
3281 * debug.h: Delete sdb_debug_hooks.
3282 * final.c: Delete sdbout.h include.
3283 (final_scan_insn): Delete SDB_DEBUG check.
3284 (rest_of_clean_state): Likewise.
3285 * output.h: Delete sdb_begin_function_line.
3288 * toplev.c: Delete sdbout.h include.
3289 (process_options): Delete SDB_DEBUG check.
3290 * tree-core.h (tree_type_common): Delete pointer field of
3292 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3293 TYPE_SYMTAB_POINTER.
3294 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3295 (TYPE_SYMTAB_IS_POINTER): Delete.
3296 (TYPE_SYMTAB_IS_DIE): Renumber.
3297 * xcoffout.c: Refer to former sdbout.c file.
3298 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3300 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3301 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3302 (Debugging Options): Delete -gcoff.
3303 (-gstabs): Delete SDB reference.
3305 (-gcoff@var{level}): Delete.
3306 * doc/passes.texi (Debugging information output): Delete SDB and
3307 sdbout.c references.
3308 * doc/tm.texi: Regenerate.
3309 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3310 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3312 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3313 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3314 (SDB_DEBUGGING_INFO): Delete.
3315 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3316 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3317 * target.def (output_source_filename): Delete COFF reference.
3319 * common.opt (gcoff): Delete.
3320 (gxcoff+): Update Negative chain.
3321 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3323 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3324 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3325 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3327 (expand_function_start): Change sdb reference to past tense.
3328 (expand_function_end): Change sdb reference to past tense.
3329 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3330 * opts.c (debug_type_names): Delete coff entry.
3331 (common_handle_option): Delete OPT_gcoff case.
3332 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3334 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3335 * config/cris/cris.h: Delete SDB reference in comment.
3336 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3337 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3338 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3339 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3341 (ix86_expand_prologue): Likewise.
3342 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3343 * config/ia64/ia64.h: Likewise.
3344 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3345 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3347 * config/mmix/mmix.h: Likewise.
3348 * config/nds32/nds32.c: Likewise.
3349 * config/stormy/storym16.h: Likewise.
3350 * config/visium/visium.h: Likewise.
3351 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3353 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3355 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3356 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3357 FRAME_POINTER_REGNUM point at high end of local var area.
3359 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3361 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3362 Move comment around. Do not reset best_edge for a copiable
3363 destination if the copy would cause a partition change.
3364 (better_edge_p): Remove redundant check.
3366 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3368 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3370 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3372 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3373 for math functions that have _Float<N> and _Float<N>X variants.
3374 (mathfn_built_in_2): Add support for math functions that have
3375 _Float<N> and _Float<N>X variants.
3376 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3377 (expand_builtin_mathfn_ternary): Add support for fma with
3378 _Float<N> and _Float<N>X variants.
3379 (expand_builtin): Likewise.
3380 (fold_builtin_3): Likewise.
3381 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3382 create math function _Float<N> and _Float<N>X variants as external
3384 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3385 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3386 the __builtin_ prefix and if not strict ansi, without the prefix.
3387 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3388 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3389 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3390 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3391 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3392 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3393 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3394 function signatures for fma _Float<N> and _Float<N>X variants.
3395 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3396 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3397 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3398 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3399 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3400 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3401 * gencfn-macros.c (print_case_cfn): Add support for math functions
3402 that have _Float<N> and _Float<N>X variants.
3403 (print_define_operator_list): Likewise.
3404 (fltfn_suffixes): Likewise.
3406 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3407 for math functions that have _Float<N> and _Float<N>X variants.
3408 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3409 and _Float<N>X variants.
3410 (COPYSIGN): Likewise.
3413 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3414 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3416 (integer_valued_read_call_p): Likewise.
3417 * fold-const-call.c (fold_const_call_ss): Likewise.
3418 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3419 _Float<N> and _Float<N>X variants.
3420 (fold_const_call_ssss): Add support for fma _Float<N> and
3421 _Float<N>X variants.
3422 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3423 support for copysign and fma _Float<N> and _Float<N>X variants.
3424 (backprop::process_builtin_call_use): Likewise.
3425 * tree-call-cdce.c (can_test_argument_range); Add support for
3426 sqrt _Float<N> and _Float<N>X variants.
3427 (edom_only_function): Likewise.
3428 (get_no_error_domain): Likewise.
3429 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3430 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3431 copysign _Float<N> and _Float<N>X variants.
3432 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3433 handled by machine independent code.
3434 (FMAF128): Likewise.
3435 * doc/cpp.texi (Common Predefined Macros): Document defining
3436 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3437 fma _Float<N> and _Float<N>X variants.
3439 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3442 * config/i386/i386-modes.def (CCFPU): Remove definition.
3443 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3444 (ix86_cc_modes_compatible): Ditto.
3445 (ix86_expand_carry_flag_compare): Ditto.
3446 (ix86_expand_int_movcc): Ditto.
3447 (ix86_expand_int_addcc): Ditto.
3448 (ix86_reverse_condition): Ditto.
3449 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3450 Return true/false for unordered/ordered fp comparisons.
3451 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3452 (ix86_prepare_fp_compare_args): Update for rename.
3453 (ix86_expand_fp_compare): Update for rename. Generate unordered
3454 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3455 (ix86_expand_sse_compare_and_jump): Ditto.
3456 * config/i386/predicates.md (fcmov_comparison_operator):
3457 Remove CCFPU mode handling.
3458 (ix86_comparison_operator): Ditto.
3459 (ix86_carry_flag_operator): Ditto.
3460 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3461 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3462 (*cmpu<mode>_cc_i387): Ditto.
3463 (FPCMP): Remove mode iterator.
3464 (unord): Remove mode attribute.
3465 (unord_subst): New define_subst transformation
3466 (unord): New define_subst attribute.
3468 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3469 (*cmpi<unord>xf_i387): Ditto.
3470 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3471 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3472 using unord_subst transformation.
3473 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3474 (round_saeonly): Also handle CCFP mode.
3475 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3476 Remove UNSPEC_SAHF unspec handling.
3478 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3480 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3482 2017-10-27 Jeff Law <law@redhat.com>
3484 * gimple-ssa-sprintf.c: Include domwalk.h.
3485 (class sprintf_dom_walker): New class, derived from dom_walker.
3486 (sprintf_dom_walker::before_dom_children): New function.
3487 (struct call_info): Moved into sprintf_dom_walker class
3488 (compute_formath_length, handle_gimple_call): Likewise.
3489 (sprintf_length::execute): Call the dominator walker rather
3490 than walking the statements.
3492 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3493 gimple statement locations.
3494 (check_array_bounds): Corresponding changes. Get the statement's
3495 location directly from wi->stmt.
3497 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3500 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3502 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3504 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3505 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3507 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3510 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3511 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3512 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3513 maybe_get_pool_constant.
3514 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3517 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3519 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3520 2.26 caveat. Update gas and gld versions.
3521 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3524 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3526 * cgraph.h (set_malloc_flag): Declare.
3527 * cgraph.c (set_malloc_flag_1): New function.
3528 (set_malloc_flag): Likewise.
3529 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3530 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3532 (read_ipa_call_summary): Add support for reading is_return_callee.
3533 (write_ipa_call_summary): Stream is_return_callee.
3534 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3535 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3536 ipa-prop.h, ipa-fnsummary.h.
3537 (pure_const_names): Change to static.
3538 (malloc_state_e): Define.
3539 (malloc_state_names): Define.
3540 (funct_state_d): Add field malloc_state.
3541 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3542 (check_retval_uses): New function.
3543 (malloc_candidate_p): Likewise.
3544 (analyze_function): Add support for malloc attribute.
3545 (pure_const_write_summary): Stream malloc_state.
3546 (pure_const_read_summary): Add support for reading malloc_state.
3547 (dump_malloc_lattice): New function.
3548 (propagate_malloc): New function.
3549 (warn_function_malloc): New function.
3550 (ipa_pure_const::execute): Call propagate_malloc and
3551 ipa_free_fn_summary.
3552 (pass_local_pure_const::execute): Add support for malloc attribute.
3553 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3554 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3556 2017-10-27 Martin Liska <mliska@suse.cz>
3558 PR gcov-profile/82457
3559 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3560 for fork-like functions to be properly instrumented.
3562 2017-10-27 Richard Biener <rguenther@suse.de>
3565 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3566 info when we redirected EH.
3568 2017-10-26 Michael Collison <michael.collison@arm.com>
3570 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3572 (<optab>_trunchf<GPI:mode>2: New pattern.
3573 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3574 * config/aarch64/iterators.md (wv): New mode attribute.
3575 (vf, VF): New mode attributes.
3576 (vgp, VGP): New mode attributes.
3577 (s): Update attribute with SImode and DImode prefixes.
3579 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3581 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3582 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3583 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3584 (nios2_option_overide): Initialize it. Don't allow R0-relative
3585 addressing with PIC.
3586 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3587 (nios2_symbolic_constant_p): Likewise.
3588 (nios2_legitimate_address_p): Likewise.
3589 (nios2_r0rel_section_name_p): New.
3590 (nios2_symbol_ref_in_r0rel_data_p): New.
3591 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3592 (r0rel_constant_p): New.
3593 (nios2_print_operand_address): Handle r0rel_constant_p.
3594 (nios2_cdx_narrow_form_p): Likewise.
3595 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3596 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3597 (Nios II Options): Document -mr0rel-sec.
3599 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3601 * config/nios2/nios2.c: Include xregex.h.
3602 (nios2_gprel_sec_regex): New.
3603 (nios2_option_overide): Initialize it. Don't allow GP-relative
3604 addressing with PIC.
3605 (nios2_small_section_name_p): Check for regex match.
3606 * config/nios2/nios2.opt (mgprel-sec=): New option.
3607 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3608 (Nios II Options): Document -mgprel-sec.
3610 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3612 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3614 2017-10-26 Tom de Vries <tom@codesourcery.com>
3616 PR tree-optimization/82707
3617 * gimple.c (gimple_copy): Fix unsharing of
3618 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3620 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3622 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3623 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3624 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3625 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3626 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3627 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3628 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3629 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3630 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3631 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3632 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3633 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3634 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3635 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3636 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3637 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3638 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3640 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3642 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3644 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3645 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3646 warning to rs6000.c. Remove the Undocumented flag, since it has
3648 (-mabi=ibmlongdouble): Likewise.
3649 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3650 already set, set the default format for long double.
3651 (rs6000_debug_reg_global): Print whether long double is IBM or
3653 (rs6000_option_override_internal): Rework setting long double
3654 format. Only warn if the user is changing the long double default
3655 and they did not use -Wno-psabi.
3656 * doc/invoke.texi (PowerPC options): Update the documentation for
3657 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3659 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3660 Alan Hayward <alan.hayward@arm.com>
3661 David Sherwood <david.sherwood@arm.com>
3663 * rtl.h (wider_subreg_mode): New function.
3664 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3665 rather than an unsigned int *.
3666 * ira-color.c (regno_max_ref_width): Replace with...
3667 (regno_max_ref_mode): ...this new variable.
3668 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3669 Use wider_subreg_mode.
3670 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3671 rather than an unsigned int *.
3672 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3673 (process_alt_operands): Likewise.
3674 (invariant_p): Likewise.
3675 * lra-spills.c (assign_mem_slot): Likewise.
3676 (add_pseudo_to_slot): Likewise.
3677 * lra.c (collect_non_operand_hard_regs): Likewise.
3678 (add_regs_to_insn_regno_info): Likewise.
3679 * reload1.c (regno_max_ref_width): Replace with...
3680 (regno_max_ref_mode): ...this new variable.
3681 (reload): Update accordingly. Update call to
3682 ira_sort_regnos_for_alter_reg.
3683 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3684 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3685 (scan_paradoxical_subregs): Likewise.
3687 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3689 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3690 (aarch64_frame): Add emit_frame_chain boolean.
3691 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3692 Move eh_return case to aarch64_layout_frame.
3693 (aarch64_layout_frame): Initialize emit_frame_chain.
3694 (aarch64_expand_prologue): Use emit_frame_chain.
3696 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3698 * config/aarch64/aarch64.c (aarch64_layout_frame):
3699 Ensure LR is always stored at the bottom of the callee-saves.
3700 Remove rarely used frame layout which saves callee-saves at top of
3701 frame, so the store of LR can be used as a valid probe in all cases.
3703 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3705 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3706 Improve unaligned TImode/TFmode base/offset split.
3708 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3709 Alan Hayward <alan.hayward@arm.com>
3710 David Sherwood <david.sherwood@arm.com>
3712 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3713 * combine.c (find_single_use_1): Likewise.
3714 (expand_field_assignment): Likewise.
3715 (move_deaths): Likewise.
3716 * lra-constraints.c (simplify_operand_subreg): Likewise.
3717 (curr_insn_transform): Likewise.
3718 * lra.c (collect_non_operand_hard_regs): Likewise.
3719 (add_regs_to_insn_regno_info): Likewise.
3720 * rtlanal.c (reg_referenced_p): Likewise.
3721 (covers_regno_no_parallel_p): Likewise.
3723 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3725 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3726 Don't print any bits outside the precision of the value.
3727 * wide-int.cc (test_printing): Add some new tests.
3729 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3731 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3732 supports -xbrace_comment option.
3733 * configure: Regenerate.
3734 * config.in: Regenerate.
3735 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3736 (ASM_CPU_SPEC): Use it.
3738 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3740 * target.def (static_rtx_alignment): New hook.
3741 * targhooks.h (default_static_rtx_alignment): Declare.
3742 * targhooks.c (default_static_rtx_alignment): New function.
3743 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3744 * doc/tm.texi: Regenerate.
3745 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3746 instead of targetm.constant_alignment. Remove call to
3748 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3749 (cris_preferred_mininum_alignment): New function, split out from...
3750 (cris_constant_alignment): ...here.
3751 (cris_static_rtx_alignment): New function.
3752 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3754 (ix86_constant_alignment): ...here.
3755 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3756 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3757 (mmix_static_rtx_alignment): New function.
3758 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3759 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3761 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3764 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3765 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3767 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3769 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3770 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3771 mark as a sign-extending load.
3772 (local_pic_load_u): Define.
3774 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3777 * fold-const.c (operand_equal_for_comparison_p): Also return true
3778 if ARG0 is a simple variant of ARG1 with narrower precision.
3779 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3781 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3783 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3785 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3786 scatter_static, scatter_per_elt.
3787 * x86-tune-costs.h: Add new cost entries.
3789 2017-10-25 Richard Biener <rguenther@suse.de>
3791 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3792 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3793 class pass_fre): Move to ...
3794 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3795 class pass_fre): ... here and adjust for statistics.
3797 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3800 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3801 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3802 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3803 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3806 2017-10-25 Richard Biener <rguenther@suse.de>
3808 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3809 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3810 eliminate_push_avail, eliminate_insert): Move inside...
3811 (class eliminate_dom_walker): ... this class in preparation
3813 (fini_eliminate): Remove by merging with ...
3814 (eliminate): ... this function. Adjust for class changes.
3815 (pass_pre::execute): Remove fini_eliminate call.
3816 (pass_fre::execute): Likewise.
3818 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3821 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3822 (VPERMI2, VPERMI2I): New mode iterators.
3823 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3824 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3826 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3827 mode iterator. Remove 3 old define_insn patterns.
3828 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3829 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3830 VPERMI2 mode iterator, remove the other two expanders.
3831 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3832 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3833 instructions, remove the other two patterns.
3834 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3835 mode iterator, remove the other two patterns.
3836 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3837 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3838 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3839 and adjust argument order accordingly.
3840 (ix86_expand_vec_perm): Adjust caller.
3841 (expand_vec_perm_1): Likewise.
3842 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3843 (expand_vec_perm_vpermt2_vpshub2): ... this.
3844 (ix86_expand_vec_perm_const_1): Adjust caller.
3845 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3848 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3849 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3850 (vec_shl_<mode>): Remove unused expander.
3851 (avx512bw_<shift_insn><mode>3): New define_insn.
3852 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3853 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3855 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3858 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3861 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3863 PR rtl-optimization/82396
3864 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3865 (autopref_multipass_init): Simplify initialization.
3866 (autopref_rank_data): Simplify sort order.
3867 * gcc/sched-int.h (autopref_multipass_data_): Remove
3868 multi_mem_insn_p, min_offset and max_offset.
3870 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3873 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3874 Check special value of flag_omit_frame_pointer.
3875 (aarch64_can_eliminate): Likewise.
3876 (aarch64_override_options_after_change_1): Simplify handling of
3877 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3879 2017-10-24 Richard Biener <rguenther@suse.de>
3881 PR tree-optimization/82697
3882 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3883 zero for conditional load and unconditional store.
3885 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3887 * doc/install.texi: Document bootstrap-cet.
3889 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3892 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3893 ENDBR instruction at function entrance if function is only
3896 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3899 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3900 patterns to better describe from which operation the CF is computed.
3901 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3902 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3903 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3904 is 0, use _0 suffixed expanders instead of emitting a comparison
3907 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3909 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3910 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3912 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3915 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3916 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3917 * loop-iv.c (iv_get_reaching_def): Likewise.
3918 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3919 variable is promoted and the partition contains undefined values.
3921 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3923 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3925 (nios2_address_cost): Define.
3926 (nios2_legitimize_address): Recognize (exp + constant) directly.
3927 (TARGET_ADDRESS_COST): Define.
3929 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3931 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3932 (nios2_symbolic_memory_operand_p): Declare.
3933 (nios2_split_large_constant): Declare.
3934 (nios2_split_symbolic_memory_operand): Declare.
3935 * config/nios2/nios2.c: Adjust includes.
3936 (nios2_symbolic_constant_allowed): New.
3937 (nios2_symbolic_constant_p): New.
3938 (nios2_plus_symbolic_constant_p): New.
3939 (nios2_valid_addr_expr_p): Recognize addresses involving
3941 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3942 (nios2_symbolic_memory_operand_p): New.
3943 (nios2_large_constant_p): New.
3944 (nios2_split_large_constant): New.
3945 (nios2_split_plus_large_constant): New.
3946 (nios2_split_symbolic_memory_operand): New.
3947 (nios2_legitimize_address): Code refactoring. Handle addresses
3948 involving symbolic constants.
3949 (nios2_emit_move_sequence): Likewise.
3950 (nios2_print_operand): Improve error output.
3951 (nios2_print_operand_address): Handle LO_SUM.
3952 (nios2_cdx_narrow_form_p): Likewise.
3953 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3954 operands involving symbolic constants.
3955 (movhi_internal, movsi_internal): Likewise.
3956 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3957 (extendhisi2, extendqi<mode>2): Likewise.
3959 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3961 * tree-pass.h (PROP_rtl_split_insns): Define.
3962 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3964 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3966 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3968 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3971 * target.def (const_not_ok_for_debug_p): Default to
3972 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3973 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3974 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3975 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3976 which targetm.const_not_ok_for_debug_p returned true.
3977 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3979 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3981 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3982 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3983 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3985 (ix86_const_not_ok_for_debug_p): New function.
3986 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3987 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3989 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3992 * system.h: Conditionally include "unique-ptr.h" if
3993 INCLUDE_UNIQUE_PTR is defined.
3994 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3995 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3997 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3999 * config/rl78/rl78.md: New define_expand "subdi3".
4001 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4004 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4005 DF_REF_INSN if DF_REF_INSN_INFO is false.
4007 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4009 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4010 xmm_move instead of sse_move.
4011 (sse_store_index): New function.
4012 (ix86_register_move_cost): Be more sensible about mismatch stall;
4013 model AVX moves correctly; make difference between sse->integer and
4015 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4016 moves; make difference between SSE and AVX.
4017 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4018 and zmm_move. Increase size of sse load and store tables;
4019 add unaligned load and store tables; add ssemmx_to_integer.
4020 * x86-tune-costs.h: Update all entries according to real
4021 move latencies from Agner Fog's manual and chip documentation.
4023 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4026 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4027 * config/i386/constraints.md (Wf): New constraint.
4028 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4029 (cmp<dwi>_doubleword): Removed.
4030 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4031 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4032 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4033 expand with cmp<dwi>_doubleword. For LTU and GEU use
4034 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4036 * common.opt (gcolumn-info): Enable by default.
4037 * doc/invoke.texi (gcolumn-info): Document new default.
4039 2017-10-23 Richard Biener <rguenther@suse.de>
4041 PR tree-optimization/82672
4042 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4043 Fold the stmt if we propagated into it.
4045 2017-10-23 Richard Biener <rguenther@suse.de>
4047 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4048 (bitmap_remove_expr_from_set): ... this. All callers call this
4049 for non-constant values.
4050 (bitmap_set_subtract): Rename to...
4051 (bitmap_set_subtract_expressions): ... this. Adjust and
4053 (bitmap_set_contains_value): Remove superfluous check.
4054 (bitmap_set_replace_value): Inline into single caller ...
4055 (bitmap_value_replace_in_set): ... here and simplify.
4056 (dependent_clean): Merge into ...
4057 (clean): ... this using an overload. Adjust.
4058 (prune_clobbered_mems): Adjust.
4059 (compute_antic_aux): Likewise.
4060 (compute_partial_antic_aux): Likewise.
4062 2017-10-23 Richard Biener <rguenther@suse.de>
4064 PR tree-optimization/82129
4066 2017-08-01 Richard Biener <rguenther@suse.de>
4068 PR tree-optimization/81181
4069 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4070 (compute_antic): ... end of iteration here.
4072 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4074 * target.def (starting_frame_offset): New hook.
4075 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4076 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4077 * doc/tm.texi.in: Regenerate.
4078 * hooks.h (hook_hwi_void_0): Declare.
4079 * hooks.c (hook_hwi_void_0): New function.
4080 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4081 STARTING_FRAME_OFFSET.
4082 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4083 * reload1.c (reload): Likewise.
4084 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4085 instead of STARTING_FRAME_OFFSET.
4086 * function.c (try_fit_stack_local): Likewise.
4087 (assign_stack_local_1): Likewise
4088 (instantiate_virtual_regs): Likewise.
4089 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4090 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4091 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4092 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4093 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4094 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4095 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4096 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4097 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4098 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4099 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4100 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4101 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4102 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4103 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4104 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4105 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4106 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4107 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4108 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4109 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4110 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4111 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4112 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4113 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4114 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4115 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4116 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4117 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4118 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4119 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4120 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4121 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4122 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4123 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4124 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4125 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4126 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4127 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4128 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4129 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4130 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4131 return a HOST_WIDE_INT.
4132 (avr_builtin_setjmp_frame_value): Use it instead of
4133 STARTING_FRAME_OFFSET.
4134 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4135 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4136 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4138 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4139 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4140 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4141 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4142 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4143 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4144 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4145 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4146 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4147 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4148 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4149 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4151 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4152 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4153 * config/mips/mips.c (mips_compute_frame_info): Refer to
4154 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4155 (mips_starting_frame_offset): New function.
4156 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4157 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4158 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4159 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4160 and return a HOST_WIDE_INT.
4161 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4162 (mmix_initial_elimination_offset): Refer to
4163 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4164 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4165 * config/pa/pa.c (pa_starting_frame_offset): New function.
4166 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4167 (pa_expand_prologue): Likewise.
4168 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4169 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4170 !FRAME_GROWS_DOWNWARD handling to...
4171 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4172 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4173 !FRAME_GROWS_DOWNWARD handling to...
4174 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4175 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4176 !FRAME_GROWS_DOWNWARD handling to...
4177 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4178 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4180 (rs6000_starting_frame_offset): New function.
4181 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4182 !FRAME_GROWS_DOWNWARD handling to...
4183 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4184 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4185 !FRAME_GROWS_DOWNWARD handling to...
4186 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4187 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4188 !FRAME_GROWS_DOWNWARD handling to...
4189 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4190 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4191 (rs6000_starting_frame_offset): New function.
4192 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4193 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4194 * config/vax/vax.c (vax_starting_frame_offset): New function.
4195 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4196 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4197 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4198 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4199 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4200 * system.h (STARTING_FRAME_OFFSET): Poison.
4202 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4204 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4205 SCALAR_TYPE_MODE instead of TYPE_MODE.
4207 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4208 Alan Hayward <alan.hayward@arm.com>
4209 David Sherwood <david.sherwood@arm.com>
4211 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4213 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4214 Alan Hayward <alan.hayward@arm.com>
4215 David Sherwood <david.sherwood@arm.com>
4217 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4219 2017-10-23 Richard Biener <rguenther@suse.de>
4221 PR tree-optimization/82129
4222 * tree-ssa-pre.c (bitmap_set_and): Remove.
4223 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4224 canonicalizing expressions in the set to those with lowest
4225 ID rather than taking that from the first edge.
4227 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4229 * combine.c (rtx_equal_for_field_assignment_p): Use
4230 byte_lowpart_offset.
4232 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4233 Alan Hayward <alan.hayward@arm.com>
4234 David Sherwood <david.sherwood@arm.com>
4236 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4237 to a SUBREG_PROMOTED_VAR.
4239 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4240 Alan Hayward <alan.hayward@arm.com>
4241 David Sherwood <david.sherwood@arm.com>
4243 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4244 (expand_debug_source_expr): Likewise.
4245 * combine.c (combine_simplify_rtx): Likewise.
4246 * cse.c (fold_rtx): Likewise.
4247 * optabs.c (expand_float): Likewise.
4248 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4249 (simplify_binary_operation_1): Likewise.
4251 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4252 Alan Hayward <alan.hayward@arm.com>
4253 David Sherwood <david.sherwood@arm.com>
4255 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4256 (record_promoted_value): Likewise.
4257 * expr.c (expand_expr_real_2): Likewise.
4258 * ree.c (update_reg_equal_equiv_notes): Likewise.
4259 (combine_set_extension): Likewise.
4260 * rtlanal.c (low_bitmask_len): Likewise.
4261 * simplify-rtx.c (neg_const_int): Likewise.
4262 (simplify_binary_operation_1): Likewise.
4264 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4265 Alan Hayward <alan.hayward@arm.com>
4266 David Sherwood <david.sherwood@arm.com>
4268 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4269 * regcprop.c (maybe_mode_change): Likewise.
4270 * reload1.c (alter_reg): Likewise.
4272 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4274 * inchash.h (inchash::hash::add_wide_int): New function.
4275 * lto-streamer-out.c (hash_tree): Use it.
4277 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4279 * inchash.h (inchash::hash::add_wide_int): Rename to...
4280 (inchash::hash::add_hwi): ...this.
4281 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4282 (polymorphic_call_target_hasher::hash): Likewise.
4283 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4284 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4285 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4286 * lto-streamer-out.c (hash_tree): Likewise.
4287 * optc-save-gen.awk: Likewise.
4288 * tree.c (add_expr): Likewise.
4290 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4293 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4294 for ordered inequality comparisons even with TARGET_IEEE_FP.
4296 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4299 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4300 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4301 Expand with cmp<dwi>_doubleword.
4303 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4305 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4306 List CET intrinsics.
4307 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4308 specific to -fcf-protection option.
4310 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4312 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4313 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4314 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4315 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4316 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4317 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4318 (extra_objs): Add cet.o for Linux/x86 targets.
4319 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4320 * config/i386/cet.c: New file.
4321 * config/i386/cetintrin.h: Likewise.
4322 * config/i386/t-cet: Likewise.
4323 * config/i386/cpuid.h (bit_SHSTK): New.
4324 (bit_IBT): Likewise.
4325 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4326 pass IBT and SHSTK bits.
4327 * config/i386/i386-builtin-types.def
4328 (VOID_FTYPE_UNSIGNED_PVOID): New.
4329 (VOID_FTYPE_UINT64_PVOID): Likewise.
4330 * config/i386/i386-builtin.def: Add CET intrinsics.
4331 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4332 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4333 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4334 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4336 * config/i386/i386.c (rest_of_insert_endbranch): New.
4337 (pass_data_insert_endbranch): Likewise.
4338 (pass_insert_endbranch): Likewise.
4339 (make_pass_insert_endbranch): Likewise.
4340 (ix86_notrack_prefixed_insn_p): Likewise.
4341 (ix86_target_string): Add -mibt, -mshstk flags.
4342 (ix86_option_override_internal): Add flag_cf_protection
4344 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4345 (ix86_print_operand): Add 'notrack' prefix output.
4346 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4347 (ix86_expand_builtin): Expand CET intrinsics.
4348 (x86_output_mi_thunk): Add 'endbranch' instruction.
4349 * config/i386/i386.h (TARGET_IBT): New.
4350 (TARGET_IBT_P): Likewise.
4351 (TARGET_SHSTK): Likewise.
4352 (TARGET_SHSTK_P): Likewise.
4353 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4354 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4355 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4356 (builtin_setjmp_setup): New pattern.
4357 (builtin_longjmp): Likewise.
4358 (rdssp<mode>): Likewise.
4359 (incssp<mode>): Likewise.
4360 (saveprevssp): Likewise.
4361 (rstorssp): Likewise.
4362 (wrss<mode>): Likewise.
4363 (wruss<mode>): Likewise.
4364 (setssbsy): Likewise.
4365 (clrssbsy): Likewise.
4366 (nop_endbr): Likewise.
4367 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4369 * config/i386/immintrin.h: Include <cetintrin.h>.
4370 * config/i386/linux-common.h
4371 (file_end_indicate_exec_stack_and_cet): New prototype.
4372 (TARGET_ASM_FILE_END): New.
4374 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4376 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4377 latencies instead of having separate table; make difference between
4378 integer and float costs.
4379 * i386.h (processor_costs): Remove scalar_stmt_cost,
4380 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4381 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4383 * x86-tune-costs.h: Remove entries which has been removed in
4384 procesor_costs from all tables; make cond_taken_branch_cost
4385 and cond_not_taken_branch_cost COST_N_INSNS based.
4387 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4389 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4391 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4393 * config/i386/i386.md (isa): Remove fma_avx512f.
4394 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4395 <avx512>_fmadd_<mode>_mask3<round_name>,
4396 <avx512>_fmsub_<mode>_mask<round_name>,
4397 <avx512>_fmsub_<mode>_mask3<round_name>,
4398 <avx512>_fnmadd_<mode>_mask<round_name>,
4399 <avx512>_fnmadd_<mode>_mask3<round_name>,
4400 <avx512>_fnmsub_<mode>_mask<round_name>,
4401 <avx512>_fnmsub_<mode>_mask3<round_name>,
4402 <avx512>_fmaddsub_<mode>_mask<round_name>,
4403 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4404 <avx512>_fmsubadd_<mode>_mask<round_name>,
4405 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4406 (*vec_widen_umult_even_v16si<mask_name>,
4407 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4408 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4410 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4412 * extend.texi: Add 'nocf_check' documentation.
4413 * gimple.texi: Add second parameter to
4414 gimple_build_call_from_tree.
4415 * invoke.texi: Add -fcf-protection documentation.
4416 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4418 2017-10-20 Richard Biener <rguenther@suse.de>
4420 PR tree-optimization/82473
4421 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4422 the largest input type.
4424 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4426 * c-attribs.c (handle_nocf_check_attribute): New function.
4427 (c_common_attribute_table): Add 'nocf_check' handling.
4428 * gimple-parser.c: Add second argument NULL to
4429 gimple_build_call_from_tree.
4430 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4431 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4433 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4434 * common.opt: Add fcf-protection flag.
4435 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4436 * flag-types.h: Add enum cf_protection_level.
4437 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4438 Add 'nocf_check' attribute propagation to gimple call.
4439 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4440 (gimple_build_call_from_tree): Update prototype.
4441 (gimple_call_nocf_check_p): New function.
4442 (gimple_call_set_nocf_check): Likewise.
4443 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4444 * ipa-icf.c: Add nocf_check attribute in statement hash.
4445 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4446 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4447 * toplev.c (process_options): Add flag_cf_protection handling.
4449 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4451 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4453 2017-10-20 Richard Biener <rguenther@suse.de>
4455 PR tree-optimization/82603
4456 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4457 remove false predicated stores.
4459 2017-10-20 Richard Biener <rguenther@suse.de>
4461 * graphite-isl-ast-to-gimple.c
4462 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4463 Remove return value and simplify, dump copied stmt after lhs
4465 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4466 Reduce dump verbosity.
4467 (gsi_insert_earliest): Likewise.
4468 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4469 * graphite.c (print_global_statistics): Adjust dumping.
4470 (print_graphite_scop_statistics): Likewise.
4471 (print_graphite_statistics): Do not dump loops here.
4472 (graphite_transform_loops): But here.
4474 2017-10-20 Nicolas Roche <roche@adacore.com>
4476 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4477 * configure: Regenerate.
4479 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4482 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4483 functions when optimizing replace GIMPLE_RETURN stmts with
4484 calls to __builtin_unreachable ().
4487 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4488 for -fsanitize=thread link of executables.
4489 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4490 link of executables.
4493 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4495 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4496 define_insns for logical vector shifts to use VI248_AVX512BW
4497 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4498 condition, useless isa and prefix attributes. Change the first
4499 2 of these define_insns to ...
4500 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4501 define_insn for avx512vl.
4502 (<shift_insn><mode>3): ... and this, new define_insn without
4503 masking for non-avx512vl.
4506 * config/i386/sse.md (*andnot<mode>3,
4507 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4508 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4509 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4510 not applied use empty suffix even for TARGET_AVX512VL.
4511 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4512 is applied, supply evex,evex or evex,evex,evex instead of just
4515 2017-10-20 Julia Koval <julia.koval@intel.com>
4517 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4518 (OPTION_MASK_ISA_GFNI_UNSET): New.
4519 (ix86_handle_option): Handle OPT_mgfni.
4520 * config/i386/cpuid.h (bit_GFNI): New.
4521 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4522 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4523 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4524 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4525 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4526 * config/i386/i386.opt: Add mgfni.
4528 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4530 * config/msp430/msp430.c (msp430_option_override): Disable
4531 -fdelete-null-pointer-checks.
4532 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4534 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4536 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4537 of x87 and SSE instructions.
4539 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4541 * asan.c (create_cond_insert_point): Do not update edge count.
4542 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4543 (afdo_propagate_circuit): Likewise.
4544 (afdo_calculate_branch_prob): Likewise.
4545 (afdo_annotate_cfg): Likewise.
4546 * basic-block.h (struct edge_def): Remove count.
4547 (edge_def::count): New accessor.
4548 * bb-reorder.c (rotate_loop): Update.
4549 (find_traces_1_round): Update.
4550 (connect_traces): Update.
4551 (sanitize_hot_paths): Update.
4552 * cfg.c (unchecked_make_edge): Update.
4553 (make_single_succ_edge): Update.
4554 (check_bb_profile): Update.
4555 (dump_edge_info): Update.
4556 (update_bb_profile_for_threading): Update.
4557 (scale_bbs_frequencies_int): Update.
4558 (scale_bbs_frequencies_gcov_type): Update.
4559 (scale_bbs_frequencies_profile_count): Update.
4560 (scale_bbs_frequencies): Update.
4561 * cfganal.c (connect_infinite_loops_to_exit): Update.
4562 * cfgbuild.c (compute_outgoing_frequencies): Update.
4563 (find_many_sub_basic_blocks): Update.
4564 * cfgcleanup.c (try_forward_edges): Update.
4565 (try_crossjump_to_edge): Update
4566 * cfgexpand.c (expand_gimple_cond): Update
4567 (expand_gimple_tailcall): Update
4568 (construct_exit_block): Update
4569 * cfghooks.c (verify_flow_info): Update
4570 (redirect_edge_succ_nodup): Update
4571 (split_edge): Update
4572 (make_forwarder_block): Update
4573 (duplicate_block): Update
4574 (account_profile_record): Update
4575 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4576 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4577 * cfgloopmanip.c (scale_loop_profile): Update.
4579 (lv_adjust_loop_entry_edge): Update.
4580 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4581 (force_nonfallthru_and_redirect): Update.
4582 (purge_dead_edges): Update.
4583 (rtl_flow_call_edges_add): Update.
4584 * cgraphunit.c (init_lowered_empty_function): Update.
4585 (cgraph_node::expand_thunk): Update.
4586 * gimple-pretty-print.c (dump_probability): Update.
4587 (dump_edge_probability): Update.
4588 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4589 * haifa-sched.c (sched_create_recovery_edges): Update.
4590 * hsa-gen.c (convert_switch_statements): Update.
4591 * ifcvt.c (dead_or_predicable): Update.
4592 * ipa-inline-transform.c (inline_transform): Update.
4593 * ipa-split.c (split_function): Update.
4594 * ipa-utils.c (ipa_merge_profiles): Update.
4595 * loop-doloop.c (add_test): Update.
4596 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4597 * lto-streamer-in.c (input_cfg): Update.
4598 (input_function): Update.
4599 * lto-streamer-out.c (output_cfg): Update.
4600 * modulo-sched.c (sms_schedule): Update.
4601 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4602 * predict.c (maybe_hot_edge_p): Update.
4603 (unlikely_executed_edge_p): Update.
4604 (probably_never_executed_edge_p): Update.
4605 (dump_prediction): Update.
4606 (drop_profile): Update.
4607 (propagate_unlikely_bbs_forward): Update.
4608 (determine_unlikely_bbs): Update.
4609 (force_edge_cold): Update.
4610 * profile.c (compute_branch_probabilities): Update.
4611 * reg-stack.c (better_edge): Update.
4612 * shrink-wrap.c (handle_simple_exit): Update.
4613 * tracer.c (better_p): Update.
4614 * trans-mem.c (expand_transaction): Update.
4615 (split_bb_make_tm_edge): Update.
4616 * tree-call-cdce.c: Update.
4617 * tree-cfg.c (gimple_find_sub_bbs): Update.
4618 (gimple_split_edge): Update.
4619 (gimple_duplicate_sese_region): Update.
4620 (gimple_duplicate_sese_tail): Update.
4621 (gimple_flow_call_edges_add): Update.
4622 (insert_cond_bb): Update.
4623 (execute_fixup_cfg): Update.
4624 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4625 * tree-complex.c (expand_complex_div_wide): Update.
4626 * tree-eh.c (lower_resx): Update.
4627 (unsplit_eh): Update.
4628 (cleanup_empty_eh_move_lp): Update.
4629 * tree-inline.c (copy_edges_for_bb): Update.
4630 (freqs_to_counts): Update.
4631 (copy_cfg_body): Update.
4632 * tree-ssa-dce.c (remove_dead_stmt): Update.
4633 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4634 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4635 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4636 (unloop_loops): Update.
4637 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4638 * tree-ssa-loop-split.c (connect_loops): Update.
4639 (split_loop): Update.
4640 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4641 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4642 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4643 * tree-ssa-reassoc.c (branch_fixup): Update.
4644 * tree-ssa-tail-merge.c (replace_block_by): Update.
4645 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4646 (compute_path_counts): Update.
4647 (update_profile): Update.
4648 (recompute_probabilities): Update.
4649 (update_joiner_offpath_counts): Update.
4650 (estimated_freqs_path): Update.
4651 (freqs_to_counts_path): Update.
4652 (clear_counts_path): Update.
4653 (ssa_fix_duplicate_block_edges): Update.
4654 (duplicate_thread_path): Update.
4655 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4656 (case_bit_test_cmp): Update.
4657 (collect_switch_conv_info): Update.
4658 (gen_inbound_check): Update.
4659 (do_jump_if_equal): Update.
4660 (emit_cmp_and_jump_insns): Update.
4661 * tree-tailcall.c (decrease_profile): Update.
4662 (eliminate_tail_call): Update.
4663 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4664 (vect_do_peeling): Update.
4665 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4666 * ubsan.c (ubsan_expand_null_ifn): Update.
4667 (ubsan_expand_ptr_ifn): Update.
4668 * value-prof.c (gimple_divmod_fixed_value): Update.
4669 (gimple_mod_pow2): Update.
4670 (gimple_mod_subtract): Update.
4671 (gimple_ic): Update.
4672 (gimple_stringop_fixed_value): Update.
4674 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4677 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4679 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4681 PR rtl-optimization/82395
4682 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4683 based on non_spilled_static_chain_regno_p.
4685 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4687 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4688 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4689 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4691 2017-10-19 Martin Sebor <msebor@redhat.com>
4693 PR tree-optimization/82596
4694 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4696 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4698 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4699 (get_mem_refs_of_builtin_call): Likewise.
4700 * builtins.c (expand_builtin_apply): Adjust call to
4701 allocate_dynamic_stack_space.
4702 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4703 the third argument to allocate_dynamic_stack_space, otherwise -1.
4704 (expand_builtin): Deal with all alloca variants.
4705 (is_inexpensive_builtin): Likewise.
4706 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4707 * calls.c (special_function_p): Deal with all alloca variants.
4708 (initialize_argument_information): Adjust call to
4709 allocate_dynamic_stack_space.
4710 (expand_call): Likewise.
4711 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4712 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4713 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4714 use it for the stack usage computation.
4715 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4716 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4717 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4718 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4719 (in_loop_p): Remove first argument and useless check.
4720 (pass_walloca::execute): Remove useless test and adjust call to above.
4721 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4722 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4723 (gimplify_call_expr): Deal with all alloca variants.
4724 * hsa-gen.c (gen_hsa_alloca): Likewise.
4725 (gen_hsa_insns_for_call): Likewise.
4726 * ipa-pure-const.c (special_builtin_state): Likewise.
4727 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4728 * tree-object-size.c (alloc_object_size): Likewise.
4729 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4730 (call_may_clobber_ref_p_1): Likewise.
4731 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4732 (ccp_fold_stmt): Likewise.
4733 (optimize_stack_restore): Likewise.
4734 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4735 (mark_all_reaching_defs_necessary_1): Likewise.
4736 (propagate_necessity): Likewise.
4737 (eliminate_unnecessary_stmts): Likewise.
4738 * tree.c (build_common_builtin_nodes): Build
4739 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4740 (build_alloca_call_expr): New function.
4741 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4742 (CASE_BUILT_IN_ALLOCA): Likewise.
4743 (build_alloca_call_expr): Declare.
4744 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4746 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4749 * dwarf2out.c (new_die_raw): New static inline function.
4750 (new_die): Use it to create the DIE.
4751 (add_AT_external_die_ref): Likewise.
4752 (clone_die): Likewise.
4753 (clone_as_declaration): Likewise.
4754 (dwarf2out_vms_debug_main_pointer): Likewise.
4755 (base_type_die): Likewise. Remove early return for corner cases.
4756 Do not call add_pubtype on the DIE here.
4757 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4758 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4759 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4760 native order exists for base types, attach the DIE manually and call
4761 add_pubtype on it. Do not equate a reverse order DIE to the type.
4763 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4765 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4766 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4768 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4770 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4772 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4773 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4774 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4775 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4776 * opts.c (sanitizer_opts): Add builtin.
4777 * ubsan.c (instrument_builtin): New function.
4778 (pass_ubsan::execute): Call it.
4779 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4780 * doc/invoke.texi: Document -fsanitize=builtin.
4782 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4783 builtins, store max (log2 (align), 0) into uchar field instead of
4784 align into uptr field.
4785 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4786 store uchar 0 field instead of uptr 0 field.
4787 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4788 instead of passing one address of struct with 2 locations pass
4789 two addresses of structs with 1 location each.
4790 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4791 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4792 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4793 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4794 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4795 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4796 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4797 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4799 2017-10-19 Martin Liska <mliska@suse.cz>
4802 * file-find.c (remove_prefix): Remove.
4803 * file-find.h (remove_prefix): Likewise.
4804 * gcc-ar.c: Remove smartness of lookup.
4806 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4808 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4809 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4810 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4812 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4815 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4816 (setcc + and to xor + setcc): New peephole2.
4818 2017-10-19 Tom de Vries <tom@codesourcery.com>
4820 * doc/sourcebuild.texi (Test Directives, Variants of
4821 dg-require-support): Add dg-require-stack-size.
4823 2017-10-19 Martin Liska <mliska@suse.cz>
4826 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4827 that have a large alignment.
4828 (gimplify_target_expr): Likewise.
4830 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4832 PR rtl-optimization/82602
4833 * ira.c (rtx_moveable_p): Return false for volatile asm.
4835 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4838 * config/i386/i386-modes.def (CCGZ): New CC mode.
4839 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4840 * config/i386/predicates.md (ix86_comparison_operator):
4842 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4843 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4844 with double-word subtraction.
4845 (put_condition_code): Handle CCGZmode.
4847 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4849 * wide-int.cc (debug (const wide_int &)): New.
4850 (debug (const wide_int *)): New.
4851 (debug (const widest_int &)): New.
4852 (debug (const widest_int *)): New.
4854 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4857 * lra-constraints.c (curr_insn_transform): Use non-input operand
4858 instead of output one for matched reload.
4860 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4862 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4863 (tree-ssa-loop-ivopts.h): New header file.
4864 (struct builtin_info): New fields.
4865 (classify_builtin_1): Compute and record base and offset parts for
4866 memset builtin partition by calling strip_offset.
4867 (offset_cmp, fuse_memset_builtins): New functions.
4868 (finalize_partitions): Fuse adjacent memset partitions by calling
4870 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4871 Expose the interface.
4872 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4874 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4876 PR tree-optimization/82574
4877 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4878 that data reference must be executed exactly once per iteration
4879 against the outermost loop in nest.
4880 (classify_partition): Update call to above function.
4882 2017-10-18 Richard Biener <rguenther@suse.de>
4884 PR tree-optimization/82591
4885 * graphite.c (graphite_transform_loops): Move code gen message
4887 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4888 Here. Handle scop_to_isl_ast failing.
4889 (scop_to_isl_ast): Limit the number of ISL operations.
4891 2017-10-18 Richard Biener <rguenther@suse.de>
4893 * graphite-isl-ast-to-gimple.c
4894 (translate_isl_ast_to_gimple::set_rename): Simplify.
4895 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4896 (graphite_copy_stmts_from_block): ... here.
4897 (copy_bb_and_scalar_dependences): Simplify.
4898 (add_parameters_to_ivs_params): Canonicalize.
4899 (generate_entry_out_of_ssa_copies): Simplify.
4900 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4901 by passing in ISL dimension.
4902 (parameter_index_in_region_1): Rename to ...
4903 (parameter_index_in_region): ... this.
4904 (extract_affine): Adjust assert, pass down parameter index.
4905 (add_param_constraints): Use range-info when available.
4906 (build_scop_context): Adjust.
4907 * sese.c (new_sese_info): Adjust.
4908 (free_sese_info): Likewise.
4909 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4910 Remove unused typedefs.
4911 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4913 2017-10-18 Martin Liska <mliska@suse.cz>
4915 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4917 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4919 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4920 (s390_sched_init): Do not reset s390_sched_state if we entered the
4921 current basic block via a fallthru edge and all others are unlikely.
4923 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4925 * config/s390/s390.c (NUM_SIDES): New variable.
4926 (LONGRUNNING_THRESHOLD): New variable.
4927 (LATENCY_FACTOR): New variable.
4928 (s390_sched_score): Decrease score for long-running instructions on
4930 (s390_sched_variable_issue): Perform bookkeeping for long-running
4933 2017-10-18 Richard Biener <rguenther@suse.de>
4935 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4936 Simplify with removal of the parameter rename map.
4937 (set_rename): Likewise.
4938 (should_copy_to_new_region): Likewise.
4939 (graphite_copy_stmts_from_block): Likewise.
4940 (copy_bb_and_scalar_dependences): Remove initialization of
4941 unused copied_bb_map.
4943 (copy_internal_parameters): Likewise.
4944 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4945 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4946 Use INTEGRAL_TYPE_P.
4947 (parameter_index_in_region_1): Rename to ...
4948 (assign_parameter_index_in_region): ... this. Assert we have
4949 a parameter we handle.
4950 (scan_tree_for_params): Adjust.
4951 * sese.h (parameter_rename_map_t): Remove.
4952 (struct sese_info_t): Remove unused parameter_rename_map and
4953 copied_bb_map members.
4954 * sese.c (new_sese_info): Adjust.
4955 (free_sese_info): Likewise.
4957 2017-10-18 Martin Liska <mliska@suse.cz>
4960 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4961 on an abnormal edge.
4963 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4965 * doc/invoke.texi (ffunction-sections and fdata-sections):
4968 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4970 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4971 the use statement can throw internally.
4973 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4975 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4976 any RTX present on the RHS of a SET.
4977 * compare-elim.c (try_eliminate_compare): Restore comment.
4979 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4981 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4982 may be also called on tcc_type nodes.
4983 * langhooks.c (lhd_tree_size): Likewise.
4985 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4987 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4988 format_warning_at_substring.
4989 (maybe_warn): Convert source_range * param to a location_t. Pass
4990 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4991 (format_directive): Remove code to extract source_ranges and
4992 source_range * in favor of just a location_t.
4993 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4995 * substring-locations.c (format_warning_va): Convert
4996 source_range * param to a location_t.
4997 (format_warning_at_substring): Likewise.
4998 * substring-locations.h (format_warning_va): Likewise.
4999 (format_warning_at_substring): Likewise.
5001 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5003 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5005 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5006 and scatter/gather ops.
5008 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5009 vec_gather_load and vec_scatter_store.
5010 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5011 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5012 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5013 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5014 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5015 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5017 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5019 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5020 Detect FTST instruction and handle its register pops. Only pop
5021 second operand if can_pop_second_op is true.
5022 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5023 set can_pop_second_op to false in the compare_for_stack_reg call.
5025 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5026 output_fp_compare for stack register operands.
5027 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5028 instructions here. Do not emit stack register pops here. Assert
5029 that FCOMPP pops next to top stack register. Rewrite function.
5031 2017-10-17 Nathan Sidwell <nathan@acm.org>
5034 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5035 use DECL_ASSEMBLER_NAME_RAW.
5038 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5041 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5042 Wilco Dijkstra <wilco.dijkstra@arm.com>
5044 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5045 converstion to Pmode from the buf_addr.
5047 2017-10-17 Richard Biener <rguenther@suse.de>
5049 * graphite-dependences.c (scop_get_reads_and_writes): Change
5050 output parameters to references.
5052 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5054 PR 71026/tree-optimization
5055 * fold-const.c (distribute_real_division): Removed.
5056 (fold_binary_loc): Remove calls to distribute_real_divison.
5058 2017-10-17 Richard Biener <rguenther@suse.de>
5060 * graphite-scop-detection.c
5061 (scop_detection::stmt_has_simple_data_refs_p): Always use
5062 the full nest as region.
5063 (try_generate_gimple_bb): Likewise.
5064 * sese.c (scalar_evolution_in_region): Simplify now that
5065 SCEV can handle instantiation in regions.
5066 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5067 in the non-loop part of a function if requested.
5069 2017-10-17 Richard Biener <rguenther@suse.de>
5071 PR tree-optimization/82563
5072 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5074 (graphite_regenerate_ast_isl): Call it.
5075 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5077 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5079 PR tree-optimization/82549
5080 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5081 Formatting fixes. Instead of calling make_bit_field_ref with negative
5084 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5086 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5087 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5088 _mm_maskz_reduce_ss): New.
5089 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5090 __builtin_ia32_reducess_mask): Ditto..
5091 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5092 * config/i386/sse.md (reduces<mode>): Renamed to ...
5093 (reduces<mode><mask_scalar_name>): ... this.
5094 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5096 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5097 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5099 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5101 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5102 * selftest-run-tests.c (selftest::run_tests): Call
5103 selftest::unique_ptr_tests_cc_tests.
5104 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5105 * unique-ptr-tests.cc: New file.
5107 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5110 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5112 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5113 (make_hard_regno_born, make_hard_regno_dead): Update
5114 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5116 2017-10-16 Jeff Law <law@redhat.com>
5118 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5120 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5122 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5124 2017-10-16 Olivier Hainque <hainque@adacore.com>
5126 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5127 with_cpu if we were configured for an e500v2 target cpu name.
5129 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5131 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5132 * doc/invoke.texi: Document +nodsp as a valid extension for
5135 2017-10-16 Martin Liska <mliska@suse.cz>
5137 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5138 (test_set_range): Likewise.
5139 (test_range_functions): Rename to ...
5140 (test_bit_in_range): ... this.
5141 (sbitmap_c_tests): Add new test.
5143 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5145 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5147 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5148 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5150 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5152 * config/aarch64/aarch64-builtins.c
5153 (aarch64_types_quadopu_lane_qualifiers): New.
5154 (TYPES_QUADOPU_LANE): New.
5155 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5156 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5157 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5158 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5159 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5160 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5161 (Vdottype, DOTPROD): New.
5162 (sur): Add SDOT and UDOT.
5164 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5166 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5167 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5168 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5170 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5171 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5172 Enable TARGET_DOTPROD.
5173 (cortex-a75.cortex-a55): Likewise.
5174 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5176 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5178 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5179 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5181 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5183 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5184 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5185 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5186 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5187 * config/arm/types.md (neon_dot, neon_dot_q): New.
5188 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5190 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5192 * config/arm/arm.h (TARGET_DOTPROD): New.
5193 * config/arm/arm.c (arm_arch_dotprod): New.
5194 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5195 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5196 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5197 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5198 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5199 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5200 * doc/invoke.texi (armv8.2-a): Document dotprod
5202 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5204 * i386.c (ix86_vec_cost): New function.
5205 (ix86_rtx_costs): Handle vector operations better.
5206 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5207 * x86-tune-costs.h: Add new costs to all tables.
5209 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5211 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5213 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5214 divsd, sqrtss and sqrtsd
5215 * x86-tune-costs.h: Add new entries to all costs.
5216 (znver1_cost): Fix to match real instruction latencies.
5218 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5219 Michael Collison <michael.collison@arm.com>
5221 * compare-elim.c: Include emit-rtl.h.
5222 (can_merge_compare_into_arith): New function.
5223 (try_validate_parallel): Likewise.
5224 (try_merge_compare): Likewise.
5225 (try_eliminate_compare): Call the above when no previous clobber
5227 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5230 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5234 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5235 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5239 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5240 to be any operand_equal_p operands. For & (B - 1) require
5241 B to be power of 2. Recognize
5242 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5244 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5247 * optabs.c (expand_memory_blockage): Fix call of
5248 targetm.have_memory_blockage.
5250 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5253 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5254 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5255 objects to extra_objs instead of overwriting it.
5257 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5259 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5260 Use any_fp_register_operand as operand[3] predicate. Simplify
5261 equality test for operands[2] and operands[4] memory location.
5262 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5263 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5264 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5265 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5266 any_fp_register_operand as operand[1] predicate. Simplify
5267 equality test for operands[0] and operands[3] memory location.
5268 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5269 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5270 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5272 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5274 * target-insns.def: Add memory_blockage.
5275 * optabs.c (expand_memory_blockage): New function.
5276 (expand_asm_memory_barrier): Rename ...
5277 (expand_asm_memory_blockage): ... to this.
5278 (expand_mem_thread_fence): Call expand_memory_blockage
5279 instead of expand_asm_memory_barrier.
5280 (expand_mem_singnal_fence): Ditto.
5281 (expand_atomic_load): Ditto.
5282 (expand_atomic_store): Ditto.
5283 * doc/md.texi (Standard Pattern Names For Generation):
5284 Document memory_blockage instruction pattern.
5286 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5288 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5289 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5290 * config/rl78/rl78.md: New define_expand "adddi3".
5292 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5294 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5297 2017-10-13 Jeff Law <law@redhat.com>
5299 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5301 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5304 * internal-fn.c (expand_mul_overflow): If both operands have
5305 the same highpart of -1 or 0 and the topmost bit of lowpart
5306 is different, overflow is if res <= 0 rather than res < 0.
5308 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5310 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5311 TARGET_P9_VECTOR code for unaligned_load case.
5313 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5315 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5317 2017-10-13 Nathan Sidwell <nathan@acm.org>
5319 * tree-core.h (tree_contains_struct): Make bool.
5320 * tree.c (tree_contains_struct): Likewise.
5321 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5322 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5323 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5324 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5325 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5327 2017-10-13 Richard Biener <rguenther@suse.de>
5329 * graphite-isl-ast-to-gimple.c
5330 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5331 parameters and dominance check.
5332 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5333 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5334 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5335 Do not update SSA form here or do intermediate IL verification.
5336 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5337 (graphite_initialize): Remove check on the number of loops in
5338 the function and inline into graphite_transform_loops.
5339 (graphite_finalize): Inline into graphite_transform_loops.
5340 (graphite_transform_loops): Perform SSA update and IL verification
5342 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5344 2017-10-13 Richard Biener <rguenther@suse.de>
5346 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5347 graphite_expression_type_precision): Avoid global constructor
5349 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5350 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5351 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5352 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5353 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5354 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5356 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5359 * config/i386/i386.h (ix86_red_zone_size): New.
5360 * config/i386/i386.md (push peephole2s): Replace
5361 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5363 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5364 Alan Hayward <alan.hayward@arm.com>
5365 David Sherwood <david.sherwood@arm.com>
5367 * combine.c (can_change_dest_mode): Reject changes in
5368 REGMODE_NATURAL_SIZE.
5370 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5371 Alan Hayward <alan.hayward@arm.com>
5372 David Sherwood <david.sherwood@arm.com>
5374 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5375 (expand_debug_source_expr): Likewise.
5376 * combine.c (combine_simplify_rtx): Likewise.
5377 * cse.c (fold_rtx): Likewise.
5378 * fwprop.c (canonicalize_address): Likewise.
5379 * targhooks.c (default_shift_truncation_mask): Likewise.
5381 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5382 Alan Hayward <alan.hayward@arm.com>
5383 David Sherwood <david.sherwood@arm.com>
5385 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5386 (widened_mode): Likewise.
5387 (expand_unop): Likewise.
5388 * ree.c (transform_ifelse): Likewise.
5389 (merge_def_and_ext): Likewise.
5390 (combine_reaching_defs): Likewise.
5391 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5393 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5394 Alan Hayward <alan.hayward@arm.com>
5395 David Sherwood <david.sherwood@arm.com>
5397 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5398 * combine.c (gen_lowpart_for_combine): Likewise.
5399 * dwarf2out.c (rtl_for_decl_location): Likewise.
5400 * final.c (alter_subreg): Likewise.
5401 * rtlhooks.c (gen_lowpart_general): Likewise.
5402 (gen_lowpart_if_possible): Likewise.
5404 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5405 Alan Hayward <alan.hayward@arm.com>
5406 David Sherwood <david.sherwood@arm.com>
5408 * calls.c (expand_call): Use subreg_lowpart_offset.
5409 * cse.c (cse_insn): Likewise.
5410 * regcprop.c (copy_value): Likewise.
5411 (copyprop_hardreg_forward_1): Likewise.
5413 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5416 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5417 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5418 =Q constraints to +Q and into insn condition add check
5419 that operands[0] and operands[1] are equal.
5420 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5421 =Q constraints to +Q and into insn condition add check
5422 that operands[0] is equal to either operands[1] or operands[2].
5425 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5426 instead of handling MINUS_EXPR twice (once for each argument),
5427 canonicalize operand order and handle just once, use rtype where
5428 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5431 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5432 any values of __C while still being pattern recognizable as a simple
5435 2017-10-13 Richard Biener <rguenther@suse.de>
5437 PR tree-optimization/82451
5439 2017-10-02 Richard Biener <rguenther@suse.de>
5441 PR tree-optimization/82355
5442 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5443 a mapping for the enclosing loop but avoid generating one for
5445 (copy_bb_and_scalar_dependences): Remove premature codegen
5446 error on PHIs in blocks duplicated into multiple places.
5447 * graphite-scop-detection.c
5448 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5449 in the region use it as loop and nest to analyze the DR in.
5450 (try_generate_gimple_bb): Likewise.
5451 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5452 (add_loop_constraints): For blocks in a loop not in the region
5453 create a dimension with a single iteration.
5454 * sese.h (gbb_loop_at_index): Remove assert.
5456 * cfgloop.c (loop_preheader_edge): For the loop tree root
5457 return the single successor of the entry block.
5458 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5459 Reset the SCEV hashtable and niters.
5460 * graphite-scop-detection.c
5461 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5462 assert that we only have POLYNOMIAL_CHREC that vary in loops
5463 contained in the region.
5464 (scop_detection::graphite_can_represent_expr): Adjust.
5465 (scop_detection::stmt_has_simple_data_refs_p): For loops
5466 not in the region set loop to NULL. The nest is now the
5467 entry edge to the region.
5468 (try_generate_gimple_bb): Likewise.
5469 * sese.c (scalar_evolution_in_region): Adjust for
5470 instantiate_scev change.
5471 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5472 Make nest parameter the edge into the region.
5473 (create_data_ref): Likewise.
5474 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5475 entry edge into a region and adjust instantiate_scev calls.
5476 (create_data_ref): Likewise.
5477 (graphite_find_data_references_in_stmt): Likewise.
5478 (find_data_references_in_stmt): Pass the loop preheader edge
5479 from the nest argument.
5480 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5481 parameter the edge into the region.
5482 (instantiate_parameters): Use the loop preheader edge as entry.
5483 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5485 (get_instantiated_value_entry): Make instantiate_below parameter
5486 the edge into the region.
5487 (instantiate_scev_name): Likewise. Adjust dominance checks,
5488 when we cannot use loop-based instantiation instantiate by
5489 walking use-def chains.
5490 (instantiate_scev_poly): Adjust.
5491 (instantiate_scev_binary): Likewise.
5492 (instantiate_scev_convert): Likewise.
5493 (instantiate_scev_not): Likewise.
5494 (instantiate_array_ref): Remove.
5495 (instantiate_scev_3): Likewise.
5496 (instantiate_scev_2): Likewise.
5497 (instantiate_scev_1): Likewise.
5498 (instantiate_scev_r): Do not blindly handle N-operand trees.
5499 Do not instantiate array-refs. Handle all constants and invariants.
5500 (instantiate_scev): Make instantiate_below parameter
5501 the edge into the region.
5502 (resolve_mixers): Use the loop preheader edge for the region
5503 parameter to instantiate_scev_r.
5504 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5506 2017-10-13 Richard Biener <rguenther@suse.de>
5508 PR tree-optimization/82525
5509 * graphite-isl-ast-to-gimple.c
5510 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5512 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5513 Fail code generation when we cannot represent the isl integer.
5514 (binary_op_to_tree): Elide modulo operations that are no-ops
5515 in the type we code generate. Remove now superfluous code
5518 2017-10-13 Richard Biener <rguenther@suse.de>
5520 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5521 (scop_detection::harmful_loop_in_region): Remove premature
5522 IV type restriction.
5523 (scop_detection::graphite_can_represent_scev): We can handle
5524 pointer IVs just fine.
5526 2017-10-13 Alan Modra <amodra@gmail.com>
5528 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5529 "Clobbers and Scratch Registers". Add paragraph on
5530 alternative to clobbers for scratch registers and OpenBLAS
5533 2017-10-13 Alan Modra <amodra@gmail.com>
5535 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5536 example of a memory input for a string of known length. Move
5537 commentary out of table. Add a number of new examples
5538 covering array memory inputs.
5540 2017-10-12 Martin Liska <mliska@suse.cz>
5542 PR tree-optimization/82493
5543 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5544 (test_range_functions): New function.
5545 (sbitmap_c_tests): Likewise.
5546 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5547 * selftest.h (sbitmap_c_tests): New function.
5549 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5551 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5553 * config/rs6000/amo.h: Fix spacing issue.
5555 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5558 * config/i386/i386.md (*ashl<mode>3_mask_1,
5559 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5560 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5563 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5565 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5566 (profile_probability): Set max_probability
5567 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5569 * profile-count.c (profile_probability::differs_from_p): Do not
5570 rely on max_probaiblity == 10000
5572 2017-10-12 Jeff Law <law@redhat.com>
5574 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5577 2017-10-12 Martin Sebor <msebor@redhat.com>
5581 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5582 (handle_alias_pairs): Call it.
5583 * common.opt (-Wattribute-alias): New option.
5584 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5585 * doc/invoke.texi (-Wattribute-alias): Document.
5587 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5590 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5592 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5594 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5595 (make_hard_regno_born, make_hard_regno_dead): Update
5596 bb_killed_pseudos and bb_gen_pseudos.
5598 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5600 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5602 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5604 * config/alpha/alpha.c (alpha_split_conditional_move):
5605 Use std::swap instead of manually swapping.
5606 (alpha_stdarg_optimize_hook): Ditto.
5607 (alpha_canonicalize_comparison): Ditto.
5609 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5611 * tree-loop-distribution.c (struct builtin_info): New struct.
5612 (struct partition): Refactor fields into struct builtin_info.
5613 (partition_free): Free struct builtin_info.
5614 (build_size_arg_loc, build_addr_arg_loc): Delete.
5615 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5616 information from struct builtin_info.
5617 (find_single_drs): New function refactored from classify_partition.
5618 Also moved builtin validity checks to this function.
5619 (compute_access_range, alloc_builtin): New functions.
5620 (classify_builtin_st, classify_builtin_ldst): New functions.
5621 (classify_partition): Refactor code into functions find_single_drs,
5622 classify_builtin_st and classify_builtin_ldst.
5623 (distribute_loop): Don't do runtime alias check when distributing
5625 (find_seed_stmts_for_distribution): New function.
5626 (pass_loop_distribution::execute): Refactor code finding seed
5627 stmts into above function. Support distribution for the innermost
5628 two-level loop nest. Adjust dump information.
5630 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5632 * tree-loop-distribution.c: Adjust the general comment.
5633 (NUM_PARTITION_THRESHOLD): New macro.
5634 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5635 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5636 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5637 in call to build_partition_graph.
5638 (finalize_partitions): New parameter. Make loop distribution more
5639 conservative by fusing more partitions.
5640 (distribute_loop): Don't do runtime alias check in case of loop nest
5642 (find_seed_stmts_for_distribution): New function.
5643 (prepare_perfect_loop_nest): New function.
5644 (pass_loop_distribution::execute): Refactor code finding seed stmts
5645 and loop nest into above functions. Support loop nest distribution.
5646 Adjust dump information accordingly.
5648 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5650 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5651 and set PTYPE_SEQUENTIAL for merged partition.
5653 2017-10-12 Richard Biener <rguenther@suse.de>
5655 PR tree-optimization/69728
5657 2017-09-19 Richard Biener <rguenther@suse.de>
5659 PR tree-optimization/69728
5660 * graphite-sese-to-poly.c (schedule_error): New global.
5661 (add_loop_schedule): Handle empty domain by failing the
5663 (build_original_schedule): Handle schedule_error.
5665 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5666 domain by returning an unchanged schedule.
5668 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5670 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5671 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5673 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5675 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5678 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5681 * expr.c (store_field): Don't optimize away bitsize == 0 store
5682 from CALL_EXPR with addressable return type.
5684 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5686 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5687 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5688 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5689 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5690 TARGET_ISEL instead of TARGET_ISEL<sel>.
5692 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5694 * config/rs6000/rs6000.c
5695 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5697 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5699 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5701 (reg_or_cint_operand): Fix comment.
5702 (reg_or_zero_operand): New predicate.
5703 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5704 * config/rs6000/rs6000.c (output_isel): Delete.
5705 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5706 instead of reg_or_cint_operand. Output instruction directly (not via
5708 (isel_unsigned_<mode>): Ditto.
5709 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5710 gpc_reg_operand. Add an instruction alternative for this. Output
5711 instruction directly.
5712 (*isel_reversed_unsigned_<mode>): Ditto.
5714 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5716 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5717 (TARGET_CANONICALIZE_COMPARISON): Define.
5719 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5722 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5723 Check whether the dest is REG before adding REG_EQUIV note.
5725 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5728 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5730 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5731 (make_hard_regno_born, make_hard_regno_dead): Update
5732 bb_killed_pseudos and bb_gen_pseudos.
5734 2017-10-11 Nathan Sidwell <nathan@acm.org>
5736 * incpath.h (enum incpath_kind): Name enum, prefix values.
5737 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5738 * incpath.c (heads, tails): Use INC_MAX.
5739 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5740 (merge_include_chains, split_quote_chain,
5741 register_include_chains): Update incpath_kind names.
5742 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5743 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5745 (add_framework_path, darwin_register_objc_includes): Likewise.
5746 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5748 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5750 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5751 Do not use float_operator operator predicate.
5752 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5753 * config/i386/predicates.md (float_operator): Remove predicate.
5755 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5757 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5758 (*jccxf_i387): Ditto.
5759 (*jcc<mode>_i387): Ditto.
5760 (*jccu<mode>_i387): Ditto.
5761 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5762 (*jcc_*_i387 splitters): Remove.
5763 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5764 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5765 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5768 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5770 * profile-count.h (slow_safe_scale_64bit): New function.
5771 (safe_scale_64bit): New inline.
5772 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5773 * profile-count.c: Include wide-int.h
5774 (slow_safe_scale_64bit): New.
5776 2017-10-11 Nathan Sidwell <nathan@acm.org>
5778 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5779 HAS_DECL_ASSEMBLER_NAME_P.
5780 * gimple-expr.c (gimple_decl_printable_name: Check
5781 HAS_DECL_ASSEMBLER_NAME_P too.
5782 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5783 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5784 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5785 * passes.c (rest_of_decl_compilation): Check
5786 HAS_DECL_ASSEMBLER_NAME_P too.
5787 * recog.c (verify_changes): Likewise.
5788 * tree-pretty-print.c (dump_decl_name): Likewise.
5789 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5791 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5792 (SET_DECL_ASSEMBLER_NAME): Use it.
5793 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5794 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5795 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5797 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5799 * config.gcc (i386, x86_64): Add extra objects.
5800 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5801 (ix86_min_insn_size): Declare.
5802 (ix86_issue_rate): Declare.
5803 (ix86_adjust_cost): Declare.
5804 (ia32_multipass_dfa_lookahead): Declare.
5805 (ix86_macro_fusion_p): Declare.
5806 (ix86_macro_fusion_pair_p): Declare.
5807 (ix86_bd_has_dispatch): Declare.
5808 (ix86_bd_do_dispatch): Declare.
5809 (ix86_core2i7_init_hooks): Declare.
5810 (ix86_atom_sched_reorder): Declare.
5811 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5812 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5813 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5814 (rip_relative_addr_p): Rename to ...
5815 (ix86_rip_relative_addr_p): ... this one; export.
5816 (memory_address_length): Update.
5817 (ix86_issue_rate): Move to x86-tune-sched.c.
5818 (ix86_flags_dependent): Move to x86-tune-sched.c.
5819 (ix86_agi_dependent): Move to x86-tune-sched.c.
5820 (exact_dependency_1): Move to x86-tune-sched.c.
5821 (exact_store_load_dependency): Move to x86-tune-sched.c.
5822 (ix86_adjust_cost): Move to x86-tune-sched.c.
5823 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5824 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5825 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5826 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5827 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5828 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5829 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5830 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5831 (min_insn_size): Rename to ...
5832 (ix86_min_insn_size): ... this one; export.
5833 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5834 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5835 (core2i7_first_cycle_multipass_backtrack): Move to
5836 x86-tune-sched-core.c.
5837 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5838 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5839 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5840 (ix86_avoid_jump_mispredicts): Update.
5841 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5842 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5843 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5844 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5845 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5846 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5847 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5848 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5849 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5850 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5851 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5852 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5853 (BIG): Move to ix86-tune-sched-bd.c.
5854 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5855 (enum insn_path): Move to ix86-tune-sched-bd.c.
5856 (get_mem_group): Move to ix86-tune-sched-bd.c.
5857 (is_cmp): Move to ix86-tune-sched-bd.c.
5858 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5859 (is_branch): Move to ix86-tune-sched-bd.c.
5860 (is_prefetch): Move to ix86-tune-sched-bd.c.
5861 (init_window): Move to ix86-tune-sched-bd.c.
5862 (allocate_window): Move to ix86-tune-sched-bd.c.
5863 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5864 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5865 (process_end_window): Move to ix86-tune-sched-bd.c.
5866 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5867 (find_constant): Move to ix86-tune-sched-bd.c.
5868 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5869 (has_immediate): Move to ix86-tune-sched-bd.c.
5870 (get_insn_path): Move to ix86-tune-sched-bd.c.
5871 (get_insn_group): Move to ix86-tune-sched-bd.c.
5872 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5873 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5874 (add_insn_window): Move to ix86-tune-sched-bd.c.
5875 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5876 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5877 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5878 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5879 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5880 (do_dispatch): Move to ix86-tune-sched-bd.c.
5881 (has_dispatch): Move to ix86-tune-sched-bd.c.
5882 * config/i386/t-i386: Add new object files.
5883 * config/i386/x86-tune-costs.h: New file.
5884 * config/i386/x86-tune-sched-atom.c: New file.
5885 * config/i386/x86-tune-sched-bd.c: New file.
5886 * config/i386/x86-tune-sched-core.c: New file.
5887 * config/i386/x86-tune-sched.c: New file.
5889 2017-10-11 Liu Hao <lh_mouse@126.com>
5891 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5892 the generic version below instead.
5893 (should_colorize): Recognize Windows consoles as terminals
5895 * pretty-print.c [__MINGW32__] (write_all): New function.
5896 [__MINGW32__] (find_esc_head): Likewise.
5897 [__MINGW32__] (find_esc_terminator): Likewise.
5898 [__MINGW32__] (eat_esc_sequence): Likewise.
5899 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5901 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5904 2017-10-11 Richard Biener <rguenther@suse.de>
5906 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5907 Properly call analyze_scalar_evolution with the loop of the stmt.
5909 2017-10-11 Richard Biener <rguenther@suse.de>
5911 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5912 * tree-core.h (tree_base): Add chrec_var union member.
5913 * tree.h (CHREC_VAR): Remove.
5914 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5915 * tree-chrec.h (build_polynomial_chrec): Adjust.
5916 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5917 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5919 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5921 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5922 * match.pd: ... here.
5923 ((T) X == (T) Y): Relax condition.
5925 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5927 PR tree-optimization/82472
5928 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5930 (break_alias_scc_partitions): Update postorder number.
5932 2017-10-11 Martin Liska <mliska@suse.cz>
5935 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5937 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5939 2017-10-11 Martin Liska <mliska@suse.cz>
5941 * print-rtl.c (print_insn): Move declaration of idbuf
5942 to same scope as name.
5944 2017-10-11 Martin Liska <mliska@suse.cz>
5949 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5952 2017-10-11 Martin Liska <mliska@suse.cz>
5955 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5958 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5960 * config.gcc (default_gnu_indirect_function): Default to yes for
5961 arm*-*-linux* with glibc.
5963 2017-10-11 Richard Biener <rguenther@suse.de>
5965 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5966 default-defs and types we do not want to analyze.
5967 (interpret_loop_phi): Replace unreachable code with an assert.
5968 (compute_scalar_evolution_in_loop): Remove and inline ...
5969 (analyze_scalar_evolution_1): ... here, replacing condition with
5970 what makes the intent clearer. Remove handling of cases
5971 get_scalar_evolution now handles.
5973 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5975 PR rtl-optimization/81434
5976 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5977 comment for main loop. In sched_group_found if, also add checks for
5978 pass and min_cost_group.
5980 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5982 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5983 (rs6000_insn_cost): New function.
5984 * config/rs6000/rs6000.md (cost): New attribute.
5986 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5987 H.J. Lu <hongjiu.lu@intel.com>
5991 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5992 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5993 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5994 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5995 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5996 OPTION_MASK_ISA_MMX and some other bit set require both
5997 mmx and the ISAs without the mmx bit.
5998 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5999 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6000 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6001 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6002 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6003 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6004 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6005 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6006 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6007 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6008 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6009 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6010 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6011 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6012 Add OPTION_MASK_ISA_MMX.
6014 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6016 * config.gcc (armv7*-*-freebsd*): New target.
6017 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6019 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6021 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6022 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6025 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6027 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6028 a template parameter.
6029 (WIDE_INT_REF_FOR): Update accordingly.
6030 * tree.h (wi::int_traits <const_tree>): Delete.
6031 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6032 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6033 (wi::tree_to_wide_ref): New typedef.
6034 (wi::to_wide): New function.
6035 * calls.c (get_size_range): Use wi::to_wide when operating on
6037 * cgraph.c (cgraph_node::create_thunk): Likewise.
6038 * config/i386/i386.c (ix86_data_alignment): Likewise.
6039 (ix86_local_alignment): Likewise.
6040 * dbxout.c (stabstr_O): Likewise.
6041 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6042 * expr.c (const_vector_from_tree): Likewise.
6043 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6044 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6045 (fold_negate_expr_1, int_const_binop_1, const_binop)
6046 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6047 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6048 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6049 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6050 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6051 (fold_not_const, round_up_loc): Likewise.
6052 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6053 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6054 (alloca_call_type): Likewise.
6055 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6056 * godump.c (go_output_typedef): Likewise.
6057 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6058 * internal-fn.c (get_min_precision): Likewise.
6059 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6060 * ipa-polymorphic-call.c
6061 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6062 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6063 (ipa_modify_call_arguments): Likewise.
6064 * match.pd: Likewise.
6065 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6066 * print-tree.c (print_node_brief, print_node): Likewise.
6067 * stmt.c (expand_case): Likewise.
6068 * stor-layout.c (layout_type): Likewise.
6069 * tree-affine.c (tree_to_aff_combination): Likewise.
6070 * tree-cfg.c (group_case_labels_stmt): Likewise.
6071 * tree-data-ref.c (dr_analyze_indices): Likewise.
6072 (prune_runtime_alias_test_list): Likewise.
6073 * tree-dump.c (dequeue_and_dump): Likewise.
6074 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6075 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6076 * tree-pretty-print.c (dump_generic_node): Likewise.
6077 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6078 (simple_iv_with_niters): Likewise.
6079 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6080 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6081 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6082 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6083 (refine_value_range_using_guard, number_of_iterations_ne_max)
6084 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6085 (get_cst_init_from_scev, record_nonwrapping_iv)
6086 (scev_var_range_cant_overflow): Likewise.
6087 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6088 * tree-ssa-pre.c (compute_avail): Likewise.
6089 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6090 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6091 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6092 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6093 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6094 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6095 (dump_case_nodes, try_switch_expansion): Likewise.
6096 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6097 (vect_do_peeling): Likewise.
6098 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6099 * tree-vect-stmts.c (vectorizable_load): Likewise.
6100 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6101 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6102 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6103 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6104 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6105 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6106 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6107 (evrp_dom_walker::before_dom_children): Likewise.
6108 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6109 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6110 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6111 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6112 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6113 (get_range_pos_neg): Likewise.
6114 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6115 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6116 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6117 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6118 * config/avr/avr.c (avr_fold_builtin): Likewise.
6119 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6120 * config/msp430/msp430.c (msp430_attr): Likewise.
6121 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6122 * config/powerpcspe/powerpcspe-c.c
6123 (altivec_resolve_overloaded_builtin): Likewise.
6124 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6125 (rs6000_expand_ternop_builtin): Likewise.
6126 * config/rs6000/rs6000-c.c
6127 (altivec_resolve_overloaded_builtin): Likewise.
6128 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6129 (rs6000_expand_ternop_builtin): Likewise.
6130 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6132 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6134 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6135 when copying loop nest with only one inner loop.
6137 2017-10-10 Richard Biener <rguenther@suse.de>
6139 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6140 blocks if SCEV is active.
6141 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6143 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6144 (scev_initialize): Assert we are not yet initialized.
6146 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6148 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6149 inner loop's exit stmt by making it always exit the loop, otherwise
6150 we would generate an infinite empty loop.
6152 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6154 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6155 renaming variables in new preheader if it's deleted.
6157 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6159 * tree-loop-distribution.c (struct partition): Remove unused field
6160 loops of the structure.
6161 (partition_alloc, partition_free): Ditto.
6162 (build_rdg_partition_for_vertex): Ditto.
6164 2017-10-09 Jeff Law <law@redhat.com>
6166 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6167 return type to match prototype and documentation.
6169 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6171 * config/rs6000/rs6000.c (processor_costs): Move to ...
6172 * config/rs6000/rs6000.h: ... here.
6173 (rs6000_cost): Declare.
6175 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6177 * except.c (setjmp_fn): New global variable.
6178 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6179 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6180 if DONT_USE_BUILTIN_SETJMP is defined.
6182 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6184 * target.def (insn_cost): New hook.
6185 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6186 * doc/tm.texi: Regenerate.
6187 * rtlanal.c (insn_cost): Use the new hook.
6189 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6191 * combine.c (combine_validate_cost): Compute the new insn_cost,
6192 not just pattern_cost.
6193 (try_combine): Adjust comment.
6195 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6197 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6199 * combine.c (uid_insn_cost): Adjust comment.
6200 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6202 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6203 * dse.c (find_shift_sequence): Ditto.
6204 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6205 (bb_valid_for_noce_process_p): Use pattern_cost.
6206 * rtl.h (insn_rtx_cost): Delete.
6207 (pattern_cost): New prototype.
6208 (insn_cost): New prototype.
6209 * rtlanal.c (insn_rtx_cost): Rename to...
6210 (pattern_cost): ... this.
6213 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6215 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6216 (*jcc<mode>_0_r_i387): Ditto.
6217 (*jccxf_r_i387): Ditto.
6218 (*jcc<mode>_r_i387): Ditto.
6219 (*jccu<mode>_r_i387): Ditto.
6220 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6221 (*jcc): Rename from *jcc_1.
6223 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6225 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6226 deferred rescans after the lvx/stvx recombination pre-pass.
6228 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6230 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6231 memory operation instruction support.
6232 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6233 (rs6000-ibm-aix[789]*): Likewise.
6234 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6235 Document new functions.
6237 2017-10-09 Richard Biener <rguenther@suse.de>
6239 PR tree-optimization/82397
6240 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6241 equality only for semantically equal trees.
6243 2017-10-09 Richard Biener <rguenther@suse.de>
6245 PR tree-optimization/82449
6246 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6247 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6248 allow constant addresses.
6249 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6252 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6254 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6257 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6260 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6263 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6266 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6268 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6271 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6272 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6274 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6276 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6277 (WI_BINARY_PREDICATE_RESULT): Likewise.
6278 (wi::binary_traits::operator_result): New type.
6279 (wi::binary_traits::predicate_result): Likewise.
6280 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6281 (generic_wide_int::operator==, generic_wide_int::operator!=)
6282 (generic_wide_int::operator&, generic_wide_int::and_not)
6283 (generic_wide_int::operator|, generic_wide_int::or_not)
6284 (generic_wide_int::operator^, generic_wide_int::operator+
6285 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6287 (operator~, unary operator-, operator==, operator!=, operator&)
6288 (operator|, operator^, operator+, binary operator-, operator*): New
6290 * expr.c (get_inner_reference): Use wi::bit_and_not.
6291 * fold-const.c (fold_binary_loc): Likewise.
6292 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6293 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6294 (bit_value_binop): Likewise.
6295 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6296 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6297 (extract_range_from_binary_expr_1): Likewise.
6298 (masked_increment): Likewise.
6299 (simplify_bit_ops_using_ranges): Likewise.
6301 2017-10-09 Martin Jambor <mjambor@suse.cz>
6304 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6305 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6306 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6308 (hsa_fixup_mov_insn_type): New function.
6309 (hsa_op_with_type::get_in_type): Use it.
6310 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6311 immediates in an assert.
6312 (hsa_op_with_type::extend_int_to_32bit): New method.
6313 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6314 types. Convert to dest type if necessary.
6315 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6316 (reg_for_gimple_ssa): Pass false as min32int to
6317 hsa_type_for_scalar_tree_type.
6318 (gen_hsa_addr): Fixup type when creating addresable temporary.
6319 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6320 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6321 necessary. Call hsa_fixup_mov_insn_type.
6322 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6323 extend operands and convert to dest type if necessary.
6324 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6325 to dest type if necessary.
6326 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6327 if conversion nt necessary and size matches.
6328 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6329 to dest type if necessary.
6330 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6331 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6333 (gen_hsa_clrsb): Likewise.
6334 (gen_hsa_ffs): Likewise.
6335 (gen_hsa_divmod): Extend operands and convert to dest type if
6337 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6339 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6341 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6342 Remove empty default arguments. Use a brace block as output
6344 (conditional return): Ditto.
6346 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6347 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6349 (group_ending_nop): Ditto.
6350 (doloop_end): Ditto.
6351 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6352 (splitters for those): Ditto.
6354 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6356 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6357 a conditional jump (and the compare for it) so that pc_rtx is the
6359 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6360 for the deleted and renamed ctr<mode>_internal[234] patterns.
6361 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6362 Delete second conditional return pattern.
6363 (ctr<mode>_internal2): Delete this second bdnz pattern.
6364 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6365 (ctr<mode>_internal4): Delete this second bdz pattern.
6367 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6369 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6370 (always_initialized_rtx_for_ssa_name_p): New predicate.
6371 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6372 (finish_out_of_ssa): Free new field of SA.
6373 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6374 * tree-ssa-coalesce.c: Include tree-ssa.h.
6375 (get_parm_default_def_partitions): Remove extern keyword.
6376 (get_undefined_value_partitions): New function.
6377 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6378 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6380 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6382 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6384 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6386 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6388 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6389 for targets that preffer 128bit.
6391 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6393 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6395 2017-10-08 Olivier Hainque <hainque@adacore.com>
6397 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6398 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6400 (thumb_set_return_address): Likewise.
6402 2017-10-08 Olivier Hainque <hainque@adacore.com>
6404 * common/config/arm/arm-common.c (arm_except_unwind_info):
6405 Handle DWARF2_UNWIND_INFO.
6407 2017-10-07 Michael Collison <michael.collison@arm.com>
6409 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6412 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6414 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6415 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6416 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6417 defined, force the creation of a new block for a dispatch label.
6419 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6421 * invoke.texi (Wsuggest-attribute=cold): Document.
6422 * common.opt (Wsuggest-attribute=cold): New
6423 * ipa-pure-const.c (warn_function_cold): New function.
6424 * predict.c (compute_function_frequency): Use it.
6425 * predict.h (warn_function_cold): Declare.
6427 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6429 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6432 2017-10-06 Martin Liska <mliska@suse.cz>
6434 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6435 keyword for member functions.
6436 (struct sanopt_tree_couple): New struct.
6437 (struct sanopt_tree_couple_hash): New function.
6438 (struct sanopt_ctx): Add new hash_map.
6439 (has_dominating_ubsan_ptr_check): New function.
6440 (record_ubsan_ptr_check_stmt): Likewise.
6441 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6442 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6443 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6445 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6448 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6449 aarch64_simd_valid_immediate on CONST_VECTORs.
6450 (aarch64_reg_or_bic_imm): Likewise.
6452 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6454 PR rtl-optimization/82396
6455 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6457 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6459 * graphite-dependences.c (scop_get_reads): Move code to...
6460 (scop_get_must_writes): Move code to...
6461 (scop_get_may_writes): Move code to...
6462 (scop_get_reads_and_writes): ... here.
6463 (scop_get_dependences): Call scop_get_reads_and_writes.
6465 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6467 PR tree-optimization/82434
6468 * fold-const.h (can_native_encode_type_p,
6469 can_native_encode_string_p): Remove.
6470 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6471 don't encode anything, just return what would be otherwise returned.
6472 (native_encode_fixed, native_encode_complex, native_encode_vector):
6474 (native_encode_string): Likewise. Inline by hand
6475 can_native_encode_string_p.
6476 (can_native_encode_type_p): Remove.
6477 (can_native_encode_string_p): Remove.
6478 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6479 STRING_CSTs using can_native_encode_string_p, test all
6480 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6481 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6482 argument from native_encode_expr.
6483 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6484 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6487 2017-10-06 Richard Biener <rguenther@suse.de>
6489 PR tree-optimization/82397
6490 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6491 operand_equal_p but rely on data_ref_compare_tree for detecting
6493 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6494 to match up with dr_group_sort_cmp.
6496 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6499 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6501 * config/s390/s390-builtin-types.def: Regenerate.
6503 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6506 * config/s390/s390-builtin-types.def: Regenerate.
6507 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6508 Change flag from B_VXE to B_VX.
6509 (s390_vec_min_dbl): Remove B_VXE flag.
6511 2017-10-06 Richard Biener <rguenther@suse.de>
6513 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6514 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6515 translate_isl_ast_to_gimple::is_valid_rename,
6516 translate_isl_ast_to_gimple::get_rename,
6517 translate_isl_ast_to_gimple::get_def_bb_for_const,
6518 translate_isl_ast_to_gimple::get_new_name,
6519 translate_isl_ast_to_gimple::collect_all_ssa_names,
6520 translate_isl_ast_to_gimple::copy_loop_phi_args,
6521 translate_isl_ast_to_gimple::collect_all_ssa_names,
6522 translate_isl_ast_to_gimple::copy_loop_phi_args,
6523 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6524 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6525 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6526 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6527 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6528 translate_isl_ast_to_gimple::copy_cond_phi_args,
6529 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6530 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6531 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6532 translate_isl_ast_to_gimple::rename_uses,
6533 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6534 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6535 (set_rename_for_each_def): Likewise.
6536 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6537 here. Handle rewriting SCEV analyzable uses here.
6538 (copy_bb_and_scalar_dependences): Generate code for PHI
6540 (graphite_regenerate_ast_isl): Adjust.
6541 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6542 (add_write, add_read): New functions.
6543 (build_cross_bb_scalars_def): Use it and simplify.
6544 (build_cross_bb_scalars_use): Likewise.
6545 (graphite_find_cross_bb_scalar_vars): Inline into...
6546 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6547 simulating out-of-SSA. Compute liveout and add dependencies.
6548 (build_scops): Force an empty entry block.
6549 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6551 (sese_build_liveouts): Declare.
6552 (sese_trivially_empty_bb_p): Likewise.
6553 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6554 compute liveout and debug_liveout.
6555 (sese_bad_liveouts_use): Remove.
6556 (sese_reset_debug_liveouts_bb): Likewise.
6557 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6558 (sese_build_liveouts): Build liveout and debug_liveout and store
6560 (new_sese_info): Adjust.
6561 (free_sese_info): Likewise.
6562 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6563 do not build liveout here.
6564 (move_sese_in_condition): Adjust region entry.
6565 (scev_analyzable_p): Match up with chrec_apply requirements.
6566 (sese_trivially_empty_bb_p): New.
6567 * tree-into-ssa.c (get_reaching_def): Properly support generating
6568 default-defs for incremental rewrite of anonymous names.
6570 2017-10-06 Richard Biener <rguenther@suse.de>
6572 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6573 precision do not perform modulo reduction.
6575 2017-10-06 Richard Biener <rguenther@suse.de>
6577 PR tree-optimization/82436
6578 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6579 conservatively choose the vectorization factor when checking
6580 whether we can perform the required load permutation.
6581 (vect_transform_slp_perm_load): Assert when we may not fail.
6583 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6585 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6586 message for incompatible -msdata=* and -mcall-* options.
6588 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6590 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6591 rate for post-reload scheduling.
6593 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6595 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6597 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6599 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6600 to improve monte carlo in scimark.
6602 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6604 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6605 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6606 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6607 pentium4_cost, nocona_cost): Set reassociation width to 1.
6608 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6609 width to 2 for fp operations and 1 otherwise.
6610 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6612 (atom_cost): Set reassociation width to 2.
6613 (slm_cost, generic_cost): Set fp reassociation width
6614 to 2 and 1 otherwise.
6615 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6616 (core_cost): Set fp reassociation width to 4 and vector to 2.
6617 (ix86_reassociation_width): Rewrite using cost table; special case
6618 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6619 and TARGET_AVX128_OPTIMAL.
6620 * config/i386/i386.h (processor_costs): Add
6621 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6622 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6623 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6624 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6625 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6626 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6628 2017-10-05 Nathan Sidwell <nathan@acm.org>
6630 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6632 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6634 * config/arm/arm.c (arm_test_fpu_data): New.
6635 (arm_run_selftests): Call arm_test_fpu_data.
6637 2017-10-04 Nathan Sidwell <nathan@acm.org>
6639 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6641 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6643 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6645 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6646 check type for aarch64_simd_valid_immediate.
6647 (aarch64_output_simd_mov_immediate): Update prototype.
6648 (aarch64_simd_valid_immediate): Update prototype.
6649 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6650 support for ORR-immediate.
6651 (and<mode>3): modified pattern to add support for BIC-immediate.
6652 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6653 now checks for valid immediate for BIC and ORR based on new enum
6655 (aarch64_output_simd_mov_immediate): Function now used to output
6656 BIC/ORR imm as well based on new enum argument.
6657 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6659 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6660 (aarch64_reg_or_bic_imm): Likewise.
6662 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6664 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6665 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6667 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6671 PR rtl-optimization/82396
6672 * haifa-sched.c (autopref_multipass_init): Simplify
6674 (autopref_rank_data): Simplify sort order.
6675 * sched-int.h (autopref_multipass_data_): Remove
6676 multi_mem_insn_p, min_offset and max_offset.
6678 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6680 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6682 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6684 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6687 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6689 * doc/sourcebuild.texi: Document vect_long_mult.
6691 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6693 PR tree-optimization/82413
6694 * fold-const.c (build_range_check): Use widest_int when comparing
6695 the maximum ETYPE value with HIGH.
6697 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6699 PR rtl-optimization/82396
6700 * haifa-sched.c (autopref_multipass_init): Simplify
6702 (autopref_rank_data): Simplify sort order.
6703 * sched-int.h (autopref_multipass_data_): Remove
6704 multi_mem_insn_p, min_offset and max_offset.
6706 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6708 PR tree-optimization/82381
6709 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6710 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6713 PR tree-optimization/82374
6714 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6715 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6716 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6717 current_function_decl to the new decl.
6719 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6721 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6722 helper macro for IEEE float128 hardware built-in functions.
6723 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6725 (TRUNCF128_ODD): Likewise.
6726 (ADDF128_ODD): Likewise.
6727 (SUBF128_ODD): Likewise.
6728 (MULF128_ODD): Likewise.
6729 (DIVF128_ODD): Likewise.
6730 (FMAF128_ODD): Likewise.
6731 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6732 UNSPEC_TRUNC_ROUND_TO_ODD.
6733 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6734 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6735 floating point round to odd instructions.
6736 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6737 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6738 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6739 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6740 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6741 (trunc<mode>sf2_hw): Change the truncate with round to odd
6742 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6743 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6744 to odd hardware instructions.
6745 (sub<mode>3_odd): Likewise.
6746 (mul<mode>3_odd): Likewise.
6747 (div<mode>3_odd): Likewise.
6748 (sqrt<mode>2_odd): Likewise.
6749 (fma<mode>4_odd): Likewise.
6750 (fms<mode>4_odd): Likewise.
6751 (nfma<mode>4_odd): Likewise.
6752 (nfms<mode>4_odd): Likewise.
6753 (trunc<mode>df2_odd): Change the truncate with round to odd
6754 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6756 * doc/extend.texi (PowerPC built-in functions): Update documentation
6757 for existing IEEE float128-bit built-in functions. Add built-in
6758 functions that generate the IEEE 128-bit floating point round to
6761 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6763 PR rtl-optimization/77729
6764 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6765 to (X&(C1&~C2))|C2 transformations.
6767 2017-10-03 Martin Jambor <mjambor@suse.cz>
6769 PR tree-optimization/82363
6770 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6771 mismatch, mark lacc written regardless of racc.
6773 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6775 PR tree-optimization/82381
6776 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6777 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6778 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6779 fallthrough into reassoc_stmt_dominates_stmt_p.
6782 * combine.c (combine_instructions): Don't combine in unreachable
6785 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6788 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6789 function to not use the have_cpu variable. Do not set cpu_index,
6790 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6792 (rs6000_valid_attribute_p): Remove duplicate initializations of
6793 old_optimize and func_optimize.
6794 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6795 (rs6000_activate_target_options): Make global.
6796 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6799 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6801 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6802 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6803 Set *poffset to 0 and *psize and *pmax_size to -1 if
6804 *poffset + *psize overflows in HOST_WIDE_INT.
6806 PR tree-optimization/82387
6807 PR tree-optimization/82388
6808 PR tree-optimization/82389
6809 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6810 instead of live_bytes non-NULL.
6812 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6815 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6818 2017-10-02 Richard Biener <rguenther@suse.de>
6820 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6821 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6822 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6824 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6826 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6827 requested precision matches the type's.
6828 * calls.c (alloc_max_size): Calculate the new candidate size as
6829 a widest_int and use wi::to_widest when comparing it with the
6830 current candidate size.
6831 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6832 zero rather than integer_zero_node.
6833 * match.pd: Check for a no-op conversion before using wi::add
6834 rather than after. Use tree_to_uhwi when summing small shift
6835 counts into an unsigned int.
6837 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6838 Alan Hayward <alan.hayward@arm.com>
6839 David Sherwood <david.sherwood@arm.com>
6842 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6843 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6844 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6845 POINTER_AND_FP_REGS.
6847 2017-10-02 Richard Biener <rguenther@suse.de>
6849 PR tree-optimization/82355
6850 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6851 a mapping for the enclosing loop but avoid generating one for
6853 (copy_bb_and_scalar_dependences): Remove premature codegen
6854 error on PHIs in blocks duplicated into multiple places.
6855 * graphite-scop-detection.c
6856 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6857 in the region use it as loop and nest to analyze the DR in.
6858 (try_generate_gimple_bb): Likewise.
6859 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6860 (add_loop_constraints): For blocks in a loop not in the region
6861 create a dimension with a single iteration.
6862 * sese.h (gbb_loop_at_index): Remove assert.
6864 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6866 * omp-expand.c (adjust_context_scope): New function.
6867 (expand_parallel_call): Call adjust_context_scope.
6869 2017-10-01 Jeff Law <law@redhat.com>
6871 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6872 dom_opt_dom_walker class with direct access to private members.
6873 Add comments. Call test_for_singularity.
6874 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6875 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6876 m_dummy_cond anymore.
6877 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6879 (pass_dominator:execute): Build the dummy_cond here and pass it
6880 to the dom_opt_dom_walker ctor.
6881 (test_for_singularity): New function.
6883 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6884 Maya Rashish <coypu@sdf.org>
6886 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6887 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6888 (alpha*-*-netbsd*) Use nbsd_tm_file.
6889 (arm*-*-netbsdelf*) Likewise.
6890 (i[34567]86-*-netbsdelf*) Likewise.
6891 (x86_64-*-netbsd*) Likewise.
6892 (mips*-*-netbsd*) Likewise.
6893 (powerpc-*-netbsd*) Likewise.
6894 (sh*-*-netbsd*) Likewise.
6895 (sparc-*-netbsdelf*) Likewise.
6896 (sparc64-*-netbsd*) Likewise.
6897 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6899 (vax-*-netbsdelf*) Likewise.
6900 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6901 (UINT_FAST8_TYPE) Likewise.
6902 (INT_FAST16_TYPE) Check CHAR_FAST16.
6903 (UINT_FAST16_TYPE) Likewise.
6905 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6908 * config/i386/i386.md
6909 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6910 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6911 *divmodsi4_zext_2): New define_insn_and_split.
6912 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6913 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6914 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6915 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6916 New define_insn_and_split.
6917 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6918 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6919 operands[1] having DImode when mode is SImode.
6921 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6922 always SImode for DIV and MOD in REG_EQUAL notes.
6924 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6927 * match.pd: Fix handling of NaNs in pattern.
6929 2017-09-29 Jeff Law <law@redhat.com>
6931 * sbitmap.c (bitmap_bit_in_range_p): New function.
6932 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6933 * tree-ssa-dse.c (live_bytes_read): New function.
6934 (dse_classify_store): Ignore reads of dead bytes.
6936 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6937 typos and whitespace errors.
6938 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6939 * config/s390/s390.c (s390_emit_prologue): Likewise.
6941 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6944 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6945 with a symbol for LRA.
6947 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6949 PR rtl-optimization/82338
6950 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6952 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6954 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6955 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6956 (qsort_chk): Declare.
6957 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6958 (qsort_chk): New function.
6960 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6962 PR tree-optimization/82337
6963 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6964 phi definition if the PHI result appears in an abnormal PHI.
6965 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6966 basis appears in an abnormal PHI.
6968 2017-09-29 Richard Biener <rguenther@suse.de>
6970 * graphite-isl-ast-to-gimple.c
6971 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6972 (binary_op_to_tree): Use it.
6973 (get_rename_from_scev): Likewise.
6974 (copy_loop_phi_nodes): Likewise.
6975 (copy_bb_and_scalar_dependences): Likewise.
6976 (translate_pending_phi_nodes): Likewise.
6978 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6981 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6982 for movabsq $(i32 << shift), r64.
6984 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6986 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6987 index when encoding %esp as %rsp to avoid 0x67 prefix.
6989 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6991 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6992 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6994 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6996 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6997 Extensions with more than 16 double VFP registers.
6998 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6999 to_clear_mask and all code related to it. Replace the remaining
7000 entry by a sbitmap and adapt code accordingly.
7002 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7004 * brig-builtins.def: Change pure attributes to const.
7006 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7008 * config.gcc (default_gnu_indirect_function): Default to yes for
7009 sparc*-*-linux* with glibc.
7011 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7013 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7014 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7015 when creating .init_array and .fini_array sections with priority
7018 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7021 * config/aarch64/aarch64.c
7022 (aarch64_builtin_support_vector_misalignment): Always return false
7023 when misalignment is unknown.
7025 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7027 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7028 this function to return false if the definition used by the swap
7029 instruction is artificial, or if the memory address from which the
7030 constant value is loaded is not represented by a base address held
7031 in a register or if the base address register is a frame or stack
7032 pointer. Additionally, return false if the base address of the
7033 loaded constant is a SYMBOL_REF but is not considered to be a
7035 (replace_swapped_load_constant): New function.
7036 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7037 loaded constant vector with a load of a swapped constant vector.
7039 2017-09-27 Carl Love <cel@us.ibm.com>
7041 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7042 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7043 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7046 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7048 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7049 first, always call autopref_rank_data otherwise.
7051 2017-09-27 Richard Biener <rguenther@suse.de>
7053 * graphite-scop-detection.c (find_scop_parameters): Move
7054 loop bound handling ...
7055 (gather_bbs::before_dom_children): ... here, avoiding the need
7056 to build scop_info->loop_nest.
7057 (record_loop_in_sese): Remove.
7058 * sese.h (sese_info_t::loop_nest): Remove.
7059 * sese.c (new_sese_info): Do not allocate loop_nest.
7060 (free_sese_info): Do not free loop_nest.
7062 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7065 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7066 lhs from calls if the lhs has addressable type.
7068 2017-09-27 Richard Biener <rguenther@suse.de>
7070 * graphite.h (scop::max_alias_set): New member.
7071 * graphite-scop-detection.c: Remove references to non-existing
7072 --param in comments.
7073 (build_alias_sets): Record the maximum alias set used for drs.
7074 (build_scops): Support zero as unlimited for
7075 --param graphite-max-arrays-per-scop.
7076 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7078 (build_poly_sr_1): ... here. Compute alias set based on the
7079 maximum alias set used for drs rather than
7080 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7082 2017-09-27 Richard Biener <rguenther@suse.de>
7084 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7085 --param loop-block-tile-size=0 to disable tiling.
7087 2017-09-27 Richard Biener <rguenther@suse.de>
7089 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7090 (graphite-max-nb-scop-params): Document special value zero.
7091 * domwalk.h (dom_walker::STOP): New symbolical constant.
7092 (dom_walker::dom_walker): Add optional parameter for bb to
7094 (dom_walker::~dom_walker): Declare.
7095 (dom_walker::before_dom_children): Document STOP return value.
7096 (dom_walker::m_user_bb_to_rpo): New member.
7097 (dom_walker::m_bb_to_rpo): Likewise.
7098 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7099 mapping here if not provided by the user.
7100 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7101 provided by the user.
7102 (dom_walker::STOP): Define.
7103 (dom_walker::walk): Do not compute bb to RPO mapping here.
7104 Support STOP return value from before_dom_children to stop
7106 * graphite-optimize-isl.c (optimize_isl): If the schedule
7107 is the same still generate code if -fgraphite-identity
7108 or -floop-parallelize-all are given.
7109 * graphite-scop-detection.c: Include cfganal.h.
7110 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7112 (gather_bbs::before_dom_children): Return STOP for BBs
7114 (build_scops): Compute bb to RPO mapping and pass it to
7115 the domwalk. Treat --param graphite-max-nb-scop-params=0
7116 as not limiting the number of params.
7117 * graphite.c (graphite_initialize): Remove limit on the
7118 number of basic-blocks in a function.
7119 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7120 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7121 default value of 10.
7123 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7125 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7126 Adjust code to eliminate needing to do the shift right 32-bits
7127 operation after XSCVDPSPN.
7129 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7131 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7132 ((X / Y) != 0 -> X >= Y): Likewise.
7134 2017-09-26 Carl Love <cel@us.ibm.com>
7136 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7137 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7138 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7139 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7140 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7141 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7142 definitions and overloading.
7143 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7144 statement for P9V_BUILTIN_XST_LEN_R.
7145 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7146 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7147 define_expand and define_insn for the instructions and builtins.
7148 * doc/extend.texi: Update the built-in documentation file for the new
7150 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7151 define_insn for the instructions
7153 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7156 * gcc/config/netbsd-protos.h: New file.
7157 * gcc/config/netbsd.c: New file.
7158 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7159 * gcc/config/t-netbsd: New file.
7160 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7161 (tmake_file) Add t-netbsd.
7162 (extra_objs) Add netbsd.o.
7164 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7168 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7170 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7172 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7173 sign extension from a vector register to a GPR by doing a 32-bit
7174 direct move and then an EXTSW.
7175 (extendsi<mode>2 splitter): Likewise.
7176 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7177 right or vector extract after doing XSCVDPSPN. Use
7178 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7180 (movdi_from_sf_zero_ext): Likewise.
7181 (reload_gpr_from_vsxsf): Likewise.
7182 (p8_mfvsrd_4_disf): Delete, no longer used.
7183 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7184 then needing to move the SFmode to a GPR to use the XSCVDPSP
7185 instruction instead of FRSP and XSCVDPSPN.
7186 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7187 it is adjacent to the other XSCVSPDP insns.
7188 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7189 SFmode to be in traditional Altivec registers.
7190 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7191 (vsx_xscvspdpn): Likewise.
7192 (vsx_xscvspdpn_scalar): Likewise.
7194 2017-09-26 Martin Jambor <mjambor@suse.cz>
7196 * tree-sra.c (compare_access_positions): Put integral types first,
7197 stabilize sorting of integral types, remove conditions putting
7198 non-full-precision integers last.
7199 (sort_and_splice_var_accesses): Disable scalarization if a
7200 non-integert would be represented by a non-full-precision integer.
7202 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7204 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7205 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7206 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7207 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7208 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7209 conditionals inside the function instead of around it. Call
7210 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7211 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7213 2017-09-26 Richard Biener <rguenther@suse.de>
7215 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7217 (scop_detection::build_scop_breadth): ... this. Removed.
7218 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7219 (scop_detection::harmful_stmt_in_bb): Likewise.
7220 (scop_detection::graphite_can_represent_stmt): Likewise.
7221 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7222 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7223 (scop_detection::can_represent_loop_1): ... this. Removed.
7224 (scop_detection::harmful_loop_in_region): Simplify after inlining
7225 the above and remove more quadraticness.
7226 (build_scops): Adjust.
7227 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7230 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7233 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7234 REGNO (base) == SP_REG if base is a REG.
7237 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7238 if it is different SSA_NAME.
7239 (optimize_range_tests_cmp_bitwise): New function.
7240 (optimize_range_tests): Call it.
7242 2017-09-26 Richard Biener <rguenther@suse.de>
7244 PR tree-optimization/82321
7245 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7246 for the def being inside the loop.
7248 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7250 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7252 * config/s390/s390-builtins.def: Fix constraint on op4.
7254 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7256 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7257 independent expanders.
7258 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7259 ("vec_ordered", "vec_unordered"): New expanders.
7261 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7263 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7266 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7268 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7269 vec_unpacks_lo_v16qi.
7270 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7272 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7274 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7275 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7276 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7278 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7280 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7282 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7284 ("*vec_slb<mode>"): New insn pattern.
7285 ("vec_shr_<mode>"): New expander.
7286 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7287 and force the shift count operand to V16QImode.
7288 ("vec_srb<mode>"): Set shift count mode to V16QI.
7290 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7292 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7293 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7294 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7296 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7299 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7301 2017-09-26 Richard Biener <rguenther@suse.de>
7303 PR tree-optimization/82320
7304 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7307 2017-09-25 Jeff Law <law@redhat.com>
7309 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7310 prototype for new argument.
7311 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7312 mostly extracted from rs6000_emit_allocate_stack.
7313 (rs6000_emit_probe_stack_range_stack_clash): New function.
7314 (rs6000_emit_allocate_stack): Call
7315 rs6000_emit_probe_stack_range_stack_clash as needed.
7316 (rs6000_emit_probe_stack_range): Add additional argument
7317 to call to gen_probe_stack_range{si,di}.
7318 (output_probe_stack_range): New.
7319 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7320 (output_probe_stack_range_stack_clash): New.
7321 (rs6000_emit_prologue): Emit notes into dump file as requested.
7322 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7323 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7324 Add additional operand and pass it to output_probe_stack_range.
7326 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7328 PR tree-optimization/82163
7329 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7330 (checking_verify_loop_closed_ssa): New parameter.
7331 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7332 (check_loop_closed_ssa_stmt): Delete.
7333 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7334 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7335 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7338 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7340 * brig-builtins.def: Treat HSAIL barrier builtins as
7341 setjmp/longjump style functions.
7343 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7345 * target.def (constant_alignment): New hook.
7346 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7347 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7348 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7349 * doc/tm.texi: Regenerate.
7350 * targhooks.h (default_constant_alignment): Declare.
7351 (constant_alignment_word_strings): Likewise.
7352 * targhooks.c (default_constant_alignment): New function.
7353 (constant_alignment_word_strings): Likewise.
7354 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7355 instead of CONSTANT_ALIGNMENT.
7356 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7357 (force_const_mem): Likewise.
7358 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7359 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7360 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7361 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7362 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7364 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7365 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7366 constant_alignment_word_strings.
7367 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7368 (CONSTANT_ALIGNMENT): Likewise.
7369 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7370 (arm_constant_alignment): New function.
7371 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7372 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7373 constant_alignment_word_strings.
7374 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7375 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7376 constant_alignment_word_strings.
7377 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7378 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7379 (cris_constant_alignment): New function.
7380 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7381 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7382 (epiphany_constant_alignment): New function.
7383 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7384 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7385 constant_alignment_word_strings.
7386 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7387 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7388 constant_alignment_word_strings.
7389 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7390 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7391 constant_alignment_word_strings.
7392 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7393 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7394 * config/i386/i386.c (ix86_constant_alignment): Make static.
7395 Use the same interface as the target hook.
7396 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7397 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7398 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7399 constant_alignment_word_strings.
7400 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7401 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7402 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7403 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7404 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7405 constant_alignment_word_strings.
7406 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7407 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7408 constant_alignment_word_strings.
7409 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7410 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7411 constant_alignment_word_strings.
7412 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7413 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7415 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7416 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7417 * config/mips/mips.c (mips_constant_alignment): New function.
7418 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7419 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7420 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7421 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7422 (mmix_constant_alignment): Make static. Use the same interface
7424 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7425 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7426 constant_alignment_word_strings.
7427 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7428 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7429 constant_alignment_word_strings.
7430 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7431 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7432 constant_alignment_word_strings.
7433 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7434 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7435 (rs6000_constant_alignment): New function.
7436 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7437 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7438 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7439 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7440 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7441 (rs6000_constant_alignment): New function.
7442 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7443 * config/s390/s390.c (s390_constant_alignment): New function.
7444 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7445 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7446 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7447 constant_alignment_word_strings.
7448 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7449 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7450 (sparc_constant_alignment): New function.
7451 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7452 * config/spu/spu.c (spu_constant_alignment): New function.
7453 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7454 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7455 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7456 constant_alignment_word_strings.
7457 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7458 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7459 constant_alignment_word_strings.
7460 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7461 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7462 constant_alignment_word_strings.
7463 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7464 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7465 (visium_constant_alignment): New function.
7466 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7467 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7468 (xtensa_constant_alignment): New function.
7469 * system.h (CONSTANT_ALIGNMENT): Poison.
7471 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7473 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7474 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7475 (rs6000_builtin_valid_without_lhs): New helper function.
7476 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7477 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7479 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7481 * target.h (vec_perm_indices): Use unsigned short rather than
7483 (auto_vec_perm_indices): Likewise.
7484 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7485 Use unsigned int rather than unsigned char.
7486 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7488 2017-09-25 Richard Biener <rguenther@suse.de>
7490 * cfgloop.h (sort_sibling_loops): Declare.
7491 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7492 (sort_sibling_loops): New function sorting the sibling loop list
7494 * graphite.c (graphite_transform_loops): Sort sibling loops.
7496 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7498 * target.def (vec_perm_const_ok): Change sel parameter to
7500 * optabs-query.c (can_vec_perm_p): Update accordingly.
7501 * doc/tm.texi: Regenerate.
7502 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7503 auto_vec_perm_indices and remove separate nelt field.
7504 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7505 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7506 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7507 (aarch64_expand_vec_perm_const): Update accordingly.
7508 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7509 to vec_perm_indices.
7510 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7511 auto_vec_perm_indices and remove separate nelt field.
7512 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7513 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7514 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7516 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7517 to vec_perm_indices.
7518 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7519 sel to vec_perm_indices.
7520 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7521 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7522 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7524 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7527 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7530 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7531 on the FUNCTION_DECL function context if it has a DIE that is a
7534 2017-09-25 Richard Biener <rguenther@suse.de>
7536 PR tree-optimization/82285
7537 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7540 2017-09-25 Tom de Vries <tom@codesourcery.com>
7544 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7547 2017-09-25 Richard Biener <rguenther@suse.de>
7549 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7550 ISL errors other than isl_error_quota happen. Dump if the
7551 schedule is the same.
7552 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7553 errors instead of aborting inside ISL.
7555 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7558 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7559 of libgcc_eh for m64.
7560 * config/i386/darwin64.h: Likewise.
7562 2017-09-25 Richard Biener <rguenther@suse.de>
7565 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7566 attribute for incomplete types nor twice for complete ones.
7568 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7571 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7572 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7575 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7578 * config/i386/i386.opt (mprefer-avx256): Use
7579 ix86_target_flags variable.
7580 * config/i386/i386.c (ix86_target_string): Move
7581 -mprefer-avx256 to flag2_opts.
7583 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7586 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7587 and x != -1 | y != -1 into (x & y) != -1.
7589 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7591 * config.gcc: Add new case statement to set
7592 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7593 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7594 s390x-*-linux* case statements. Added aarch64 to the list of
7595 supported architectures.
7597 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7599 PR tree-optimization/82289
7600 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7601 STMT_VINFO_RELEVANT_P.
7603 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7604 Alan Hayward <alan.hayward@arm.com>
7605 David Sherwood <david.sherwood@arm.com>
7607 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7608 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7609 (extract_range_from_binary_expr_1): Don't call
7610 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7612 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7613 Alan Hayward <alan.hayward@arm.com>
7614 David Sherwood <david.sherwood@arm.com>
7616 * target.def (preferred_vector_alignment): New hook.
7617 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7619 * doc/tm.texi: Regenerate.
7620 * targhooks.h (default_preferred_vector_alignment): Declare.
7621 * targhooks.c (default_preferred_vector_alignment): New function.
7622 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7624 (DR_TARGET_ALIGNMENT): New macro.
7625 (aligned_access_p): Update commentary.
7626 (vect_known_alignment_in_bytes): New function.
7627 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7629 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7630 Calculate the misalignment based on the target alignment rather than
7632 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7633 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7634 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7635 the target alignment, rather than masking the element misalignment
7636 with the number of elements in a vector. Also use the target
7637 alignment when calculating the maximum number of peels.
7638 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7639 instead of TYPE_ALIGN_UNIT.
7640 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7641 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7642 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7643 (vect_create_data_ref_ptr): Likewise.
7644 (vect_setup_realignment): Realign by ANDing with
7645 -DR_TARGET_MISALIGNMENT.
7646 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7647 the number of peels based on DR_TARGET_ALIGNMENT.
7648 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7649 with the guaranteed alignment boundary when deciding whether
7651 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7652 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7653 (ensure_base_align): Remove stmt_info parameter. Get the
7654 target base alignment from DR_TARGET_ALIGNMENT.
7655 (vectorizable_store): Update call accordingly. Interpret
7656 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7658 (vectorizable_load): Likewise.
7660 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7661 Alan Hayward <alan.hayward@arm.com>
7662 David Sherwood <david.sherwood@arm.com>
7664 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7665 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7666 (vect_enhance_data_refs_alignment): Likewise.
7668 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7670 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7671 error. Only quit immediately if parsing is complete.
7672 (BEGIN): Initialize fatal_err and parse_done.
7673 (begin fpu, end fpu): Check number of arguments.
7674 (begin arch, end arch): Likewise.
7675 (begin cpu, end cpu): Likewise.
7676 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7677 (optalias): Likewise.
7679 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7681 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7682 * config/arm/arm-isa.h: Delete. Move definitions to ...
7683 * arm-cpus.in: ... here. Use new feature and fgroup values.
7684 * config/arm/arm.c (arm_option_override): Use lower case for feature
7686 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7687 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7688 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7690 (print_isa_bits_for): New function.
7691 (gen_isa): New function.
7692 (gen_comm_data): Use print_isa_bits_for.
7693 (define feature): New keyword.
7694 (define fgroup): New keyword.
7695 * config/arm/t-arm (TM_H): Remove.
7696 (GTM_H): Add arm-isa.h.
7697 (arm-isa.h): Add rule to generate file.
7698 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7699 case for feature bit names.
7701 2017-09-22 Richard Biener <rguenther@suse.de>
7703 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7705 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7706 print of no dependency loops ...
7707 * graphite.c (graphite_transform_loops): ... here.
7708 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7710 (same_close_phi_node, remove_duplicate_close_phi,
7711 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7712 (canonicalize_loop_closed_ssa): ... here and simplify.
7713 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7714 (optimize_isl): Use dump_printf_loc to tell when we stopped
7715 optimizing because of an ISL timeout.
7717 2017-09-22 Richard Biener <rguenther@suse.de>
7719 PR tree-optimization/82291
7720 * tree-if-conv.c (predicate_mem_writes): Make sure to
7721 remove writes in blocks predicated with false.
7723 2017-09-22 Richard Biener <rguenther@suse.de>
7725 * sese.c: Include cfganal.h.
7726 (if_region_set_false_region): Remove.
7727 (create_if_region_on_edge): Likewise.
7728 (move_sese_in_condition): Re-implement without destroying
7731 2017-09-22 Richard Biener <rguenther@suse.de>
7733 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7734 Verify both BBs contain loop PHI nodes before dispatching to
7736 (graphite_regenerate_ast_isl): Do not recompute dominators,
7737 do not verify three times. Restructure for clarity.
7738 * graphite-scop-detection.c (same_close_phi_node,
7739 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7740 defined_in_loop_p, canonicalize_loop_closed_ssa,
7741 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7742 checking and SSA rewrite, move to ...
7743 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7744 (graphite_initialize): Do not pass in ctx, do not reset the
7745 SCEV cache, compute only dominators.
7746 (graphite_transform_loops): Allocate ISL ctx after
7747 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7748 Maintain post-dominators only around build_scops.
7749 * sese.c (if_region_set_false_region): Make static. Free
7750 and recompute dominators.
7751 (move_sese_in_condition): Assert we don't get called with
7752 post-dominators computed.
7753 * sese.h (if_region_set_false_region): Remove.
7755 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7757 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7758 mode attribute for TARGET_AVX512VL.
7760 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7762 * config/i386/i386.opt (mprefer-avx256): New option.
7763 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7765 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7766 for TARGET_PREFER_AVX256.
7767 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7769 2017-09-21 Jeff Law <law@redhat.com>
7771 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7772 Fix dump output if the only stack space is for pushed registers.
7774 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7776 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7779 2017-09-21 Martin Sebor <msebor@redhat.com>
7782 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7783 code (in C++) or code that triggers warnings.
7785 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7787 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7789 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7791 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7792 * sched-rgn.c: Ditto.
7793 * sel-sched-ir.c: Ditto.
7795 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7797 * toplev.h (set_random_seed): Adjust return type.
7798 * toplev.c (init_local_tick): Move eager initialization of random_seed
7799 to get_random_seed. Adjust comment.
7800 (init_random_seed): Inline to get_random_seed, delete.
7801 (get_random_seed): Initialize random_seed lazily.
7802 (set_random_seed): Do not return previous value.
7803 (print_switch_value): Do not call get_random_seed.
7805 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7807 * cgraph.c (delete_function_version): New, broken out from...
7808 (cgraph_node::delete_function_version): ...here. Rename to
7809 cgraph_node::delete_function_version_by_decl. Update all uses.
7810 (cgraph_node::remove): Call delete_function_version.
7812 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7815 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7816 VAR_DECLs to which addressable non-volatile parameters are mapped
7817 and for id->retvar after the return value assignment. Clear
7818 id->retval and id->retbnd after inlining.
7820 2017-09-21 Richard Biener <rguenther@suse.de>
7822 PR tree-optimization/82276
7823 PR tree-optimization/82244
7824 * tree-vrp.c (build_assert_expr_for): Set
7825 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7827 (remove_range_assertions): Revert earlier change.
7829 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7832 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7834 2017-09-21 Richard Biener <rguenther@suse.de>
7836 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7837 Restore valid IL after code generation errors.
7838 * graphite.c (graphite_transform_loops): Diagnose code
7839 generation issues as MSG_MISSED_OPTIMIZATION and continue
7840 with processing SCOPs.
7842 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7843 Alan Hayward <alan.hayward@arm.com>
7844 David Sherwood <david.sherwood@arm.com>
7846 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7847 rather than choosing between plus_constant and gen_rtx_<CODE>.
7848 * expr.c (emit_push_insn): Likewise.
7849 (expand_expr_real_2): Likewise.
7851 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7852 Alan Hayward <alan.hayward@arm.com>
7853 David Sherwood <david.sherwood@arm.com>
7855 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7857 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7858 Alan Hayward <alan.hayward@arm.com>
7859 David Sherwood <david.sherwood@arm.com>
7861 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7862 calling tree_to_uhwi.
7864 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7865 Alan Hayward <alan.hayward@arm.com>
7866 David Sherwood <david.sherwood@arm.com>
7868 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7869 INTEGER_CST rather than a negative test for ADDR_EXPR.
7871 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7872 Alan Hayward <alan.hayward@arm.com>
7873 David Sherwood <david.sherwood@arm.com>
7875 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7876 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7878 2017-09-21 Richard Biener <rguenther@suse.de>
7880 PR tree-optimization/71351
7881 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7882 graphite_create_new_loop_guard): Remove, fold remaining parts
7884 (translate_isl_ast_node_for): ... here and simplify.
7886 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7889 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7890 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7891 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7892 alternative always use QI mode, for -Os imov (=R,R) alternative
7893 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7896 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7897 Jeff Law <law@redhat.com>
7899 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7900 (allocate_stack_space): New function, partially extracted from
7902 (s390_emit_prologue): Track offset to most recent stack probe.
7903 Code to allocate space moved into allocate_stack_space.
7904 Dump actions when no stack is allocated.
7905 (s390_prologue_plus_offset): New function.
7906 (s390_emit_stack_probe): Likewise.
7908 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7910 * common.opt (Wa, Wl, Wp, g, gz=): Add
7912 (gno-column-info): Remove.
7913 (gcolumn-info): Drop RejectNegative.
7915 (gno-record-gcc-switches): Remove.
7916 (grecord-gcc-switches): Drop RejectNegative.
7917 (gno-split-dwarf): Remove.
7918 (gsplit-dwarf): Drop RejectNegative.
7919 (gno-strict-dwarf): Remove.
7920 (gstrict-dwarf): Drop RejectNegative.
7921 * config/darwin.opt (gfull, gused): Add RejectNegative.
7922 * dwarf2out.c (gen_producer_string): Drop
7923 gno-record-gcc-switches handler.
7924 * optc-gen.awk: Add g to prefixes with negative forms.
7925 * opts-common.c (remapping_prefix_p): New.
7926 (find_opt): Check it.
7927 (generate_canonical_option): Test g prefix.
7928 (option_map): Add -gno- mapping.
7929 (add_misspelling_candidates): Check remapping_prefix_p.
7931 2017-09-20 Jeff Law <law@redhat.com>
7933 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7934 thinko in stack clash protection support.
7936 * explow.c (compute_stack_clash_protection_loop_data): Use
7937 CONST_INT_P instead of explicit test. Verify object is a
7938 CONST_INT_P before looking at INTVAL.
7939 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7940 instead of explicit test.
7942 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7945 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7946 address instead of to r1 and r11.
7948 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7950 * config.gcc: Support "knm".
7951 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7952 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7954 * config/i386/i386.c (m_KNM): Define.
7955 (processor_target_table): Add "knm".
7957 (ix86_option_override_internal): Add "knm".
7958 (ix86_issue_rate): Add PROCESSOR_KNM.
7959 (ix86_adjust_cost): Ditto.
7960 (ia32_multipass_dfa_lookahead): Ditto.
7961 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7962 (fold_builtin_cpu): Add M_INTEL_KNM.
7963 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7964 (processor_type): Add PROCESSOR_KNM.
7965 * config/i386/x86-tune.def: Add m_KNM.
7966 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7968 2017-09-20 Richard Biener <rguenther@suse.de>
7970 PR tree-optimization/80213
7971 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7972 are allowed in empty BBs as well.
7973 (canonicalize_loop_closed_ssa): Also look for other complex
7975 (scop_detection::get_sese): Include the loop-closed PHI block
7977 (scop_detection::merge_sese): Remove code adding extra blocks.
7978 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7979 (build_scops): Assert the final returned scop is invalid.
7981 2017-09-20 Richard Biener <rguenther@suse.de>
7983 PR tree-optimization/82264
7984 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7986 (vn_phi_lookup): Likewise.
7987 (vn_phi_insert): Likewise.
7989 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7991 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7992 that fits into uhwi or shwi, add DW_AT_const_value regardless
7993 of early_dwarf without going through RTL, using add_AT_unsigned
7996 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7997 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7998 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7999 (DEBUG_MACRO_SECTION): Likewise.
8000 (DEBUG_LTO_MACRO_SECTION): Likewise.
8001 (DEBUG_STR_DWO_SECTION): Likewise.
8002 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8003 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8004 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8005 (DEBUG_LTO_LINE_STR_SECTION): Define.
8006 (init_sections_and_labels): Initialize debug_line_str_section
8007 variable. Initialize debug_loc_section for -gdwarf-5 to
8008 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8010 2017-09-20 Richard Biener <rguenther@suse.de>
8012 * graphite-sese-to-poly.c (extract_affine): Properly handle
8013 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8015 2017-09-20 Richard Biener <rguenther@suse.de>
8017 PR tree-optimization/81373
8018 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8019 Force SESE live-out defs to be handled even if they are
8022 2017-09-19 Jeff Law <law@redhat.com>
8024 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8025 nothing for stack adjustments with REG_STACK_CHECK.
8026 * sched-deps.c (parse_add_or_inc): Reject insns with
8027 REG_STACK_CHECK from dependency breaking.
8028 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8029 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8030 * reg-notes.def (STACK_CHECK): New note.
8032 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8033 (ix86_expand_prologue): Dump stack clash info as needed.
8034 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8036 * function.c (dump_stack_clash_frame_info): New function.
8037 * function.h (dump_stack_clash_frame_info): Prototype.
8038 (enum stack_clash_probes): New enum.
8040 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8041 flag_stack_clash_protection.
8042 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8043 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8044 (arm_frame_pointer_required): Likewise.
8045 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8046 (ia64_expand_prologue): Likewise.
8047 * config/mips/mips.c (mips_expand_prologue): Likewise.
8048 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8049 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8050 (sparc_flat_expand_prologue): Likewise.
8051 * config/spu/spu.c (spu_expand_prologue): Likewise.
8053 * explow.c: Include "params.h".
8054 (anti_adjust_stack_and_probe_stack_clash): New function.
8055 (get_stack_check_protect): Likewise.
8056 (compute_stack_clash_protection_loop_data): Likewise.
8057 (emit_stack_clash_protection_loop_start): Likewise.
8058 (emit_stack_clash_protection_loop_end): Likewise.
8059 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8060 Use anti_adjust_stack_and_probe_stack_clash.
8061 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8062 (emit_stack_clash_protection_loop_start): Likewise.
8063 (emit_stack_clash_protection_loop_end): Likewise.
8064 * rtl.h (get_stack_check_protect): Prototype.
8065 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8066 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8067 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8069 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8071 * doc/tm.texi: Rebuilt.
8072 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8073 get_stack_check_protect.
8074 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8075 * config/arm/arm.c (arm_expand_prologue): Likewise.
8076 (arm_frame_pointer_required): Likewise.
8077 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8078 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8079 * config/mips/mips.c (mips_expand_prologue): Likewise.
8080 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8081 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8082 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8083 (sparc_flat_expand_prologue): Likewise.
8085 * common.opt (-fstack-clash-protection): New option.
8086 * flag-types.h (enum stack_check_type): Note difference between
8087 -fstack-check= and -fstack-clash-protection.
8088 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8089 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8090 * toplev.c (process_options): Issue warnings/errors for cases
8091 not handled with -fstack-clash-protection.
8092 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8093 (-fstack-check): Note additional problem with -fstack-check=generic.
8094 Note that -fstack-check is primarily for Ada and refer users
8095 to -fstack-clash-protection for stack-clash-protection.
8096 Document new params for stack clash protection.
8098 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8100 * config/i386/i386.c (ix86_split_long_move): Do not handle
8101 address used for LEA in a special way.
8103 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8105 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8107 2017-09-19 Martin Sebor <msebor@redhat.com>
8110 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8111 of incompatible types.
8113 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8115 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8116 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8117 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8118 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8120 2017-09-19 Richard Biener <rguenther@suse.de>
8122 PR tree-optimization/82244
8123 * tree-vrp.c (remove_range_assertions): Do not propagate
8124 a constant to abnormals but replace the assert with a copy.
8126 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8128 PR rtl-optimization/57878
8129 PR rtl-optimization/68988
8130 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8131 avoidance test involving non_reload_pseudos. Move frequency test
8132 below the general fragmentation avoidance test.
8134 2017-09-19 Richard Biener <rguenther@suse.de>
8136 PR tree-optimization/69728
8137 * graphite-sese-to-poly.c (schedule_error): New global.
8138 (add_loop_schedule): Handle empty domain by failing the
8140 (build_original_schedule): Handle schedule_error.
8142 2017-09-19 Richard Biener <rguenther@suse.de>
8144 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8145 Do not iterate to sibling loops but only to siblings of inner
8148 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8151 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8152 that the registers are different.
8154 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8156 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8157 to processor_model and "amdfam17h" to arch_names_table.
8158 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8160 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8163 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8165 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8166 Alan Hayward <alan.hayward@arm.com>
8167 David Sherwood <david.sherwood@arm.com>
8169 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8171 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8173 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8174 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8175 vect_schedule_slp_instance.
8176 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8177 Update call to vect_slp_analyze_node_operations. Simplify return
8179 (vect_slp_analyze_bb_1): Update call accordingly.
8180 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8181 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8182 (vect_schedule_slp): Update call accordingly.
8184 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8185 Alan Hayward <alan.hayward@arm.com>
8186 David Sherwood <david.sherwood@arm.com>
8188 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8189 with types that aren't in fact scalar.
8191 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8193 * tree-vect-slp.c (vect_record_max_nunits): New function,
8195 (vect_build_slp_tree_1): ...here.
8196 (vect_build_slp_tree_2): Call it for phis too.
8198 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8200 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8201 to vect_get_vec_def_for_operand when getting the mask operand.
8203 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8204 Alan Hayward <alan.hayward@arm.com>
8205 David Sherwood <david.sherwood@arm.com>
8207 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8210 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8211 Alan Hayward <alan.hayward@arm.com>
8212 David Sherwood <david.sherwood@arm.com>
8214 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8215 calculation for vector booleans.
8217 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8218 Alan Hayward <alan.hayward@arm.com>
8219 David Sherwood <david.sherwood@arm.com>
8221 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8223 (vect_transform_stmt): ...here.
8224 (vect_analyze_stmt): Use it instead of calling
8225 vectorizable_live_operation directly.
8227 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8229 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8230 non-SIMT targets in acc vector loops.
8232 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8234 * configure.ac: Add arc and check if assembler supports gdwarf2.
8235 * configure: Regenerate.
8237 2017-09-18 Richard Biener <rguenther@suse.de>
8239 PR tree-optimization/82220
8240 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8241 epilogue niters from the min_profitable_iters compute.
8243 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8246 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8247 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8248 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8250 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8253 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8254 switching to a new text section.
8256 2017-09-18 Richard Biener <rguenther@suse.de>
8258 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8260 (build_alias_set): Reject aliases with no access function.
8262 2017-09-18 Richard Biener <rguenther@suse.de>
8264 PR tree-optimization/79622
8265 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8267 (build_cross_bb_scalars_use): Likewise.
8269 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8271 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8273 2017-09-18 Alan Modra <amodra@gmail.com>
8276 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8277 stack_pointer_rtx for count 0. Update comments. Break up
8278 large rtl expression.
8280 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8282 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8283 Increase to 20 bytes.
8284 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8285 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8286 or avx version of the stub.
8288 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8291 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8292 compute the minimum stack alignment. Also update preferred stack
8293 boundary for leaf functions.
8295 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8297 PR tree-optimization/82228
8298 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8301 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8303 * common/config/nds32/nds32-common.c
8304 (nds32_option_optimization_table): Refine formatting.
8305 (nds32_option_optimization_table): Use -fsched-pressure and
8306 -fomit-frame-pointer for specific optimization level.
8308 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8310 * config/nds32/nds32.c: Refine formatting and comments.
8311 * config/nds32/nds32.h: Likewise.
8312 * config/nds32/nds32.md: Likewise.
8313 * config/nds32/nds32-cost.c: Likewise.
8314 * config/nds32/nds32-isr.c: Likewise.
8315 * config/nds32/nds32-md-auxiliary.c: Likewise.
8316 * config/nds32/nds32-multiple.md: Likewise.
8317 * config/nds32/nds32-predicates.c: Likewise.
8319 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8320 Jakub Jelinek <jakub@redhat.com>
8322 Add support for -std=c++2a.
8323 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8325 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8327 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8330 * doc/extend.texi (Common Function Attributes): Add
8331 references to ARM, AArch64, and S/390 specific attributes.
8332 (Function Specific Option Pragmas): Add AArch64 and S/390
8333 to list of back ends that support the target pragma.
8335 2017-09-15 Nathan Sidwell <nathan@acm.org>
8337 * doc/standards.texi: Fix C++17 description. Update URLs for
8340 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8342 * common.opt (Wcast-align=strict): New warning option.
8343 * doc/invoke.texi: Document -Wcast-align=strict.
8345 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8347 * cgraph.h (cgraph_thunk_info): Add comments.
8348 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8349 assert for VIRTUAL_* arguments stricter.
8351 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8353 PR tree-optimization/71026
8354 * match.pd: Move RDIV patterns from fold-const.c
8355 * fold-const.c (distribute_real_division): Removed.
8356 (fold_binary_loc): Remove calls to distribute_real_divison.
8358 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8360 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8361 c++1z and gnu++1z as deprecated. Change other references to
8362 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8363 Change -Wc++1z-compat to -Wc++17-compat.
8364 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8365 * dwarf2out.c (highest_c_language): Handle C++17.
8366 (gen_compile_unit_die): Likewise.
8368 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8370 PR rtl-optimization/82192
8371 * combine.c (make_extraction): Don't look through non-paradoxical
8372 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8375 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8376 Alan Hayward <alan.hayward@arm.com>
8377 David Sherwood <david.sherwood@arm.com>
8379 * target.def (function_arg_offset): New hook.
8380 * targhooks.h (default_function_arg_offset): Declare.
8381 * targhooks.c (default_function_arg_offset): New function.
8382 * function.c (locate_and_pad_parm): Use
8383 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8384 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8385 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8386 * doc/tm.texi: Regenerate.
8387 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8388 * config/spu/spu.c (spu_function_arg_offset): New function.
8389 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8390 * system.h (FUNCTION_ARG_OFFSET): Poison.
8392 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8393 Alan Hayard <alan.hayward@arm.com>
8394 David Sherwood <david.sherwood@arm.com>
8396 * target.def (truly_noop_truncation): New hook.
8397 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8398 than TRULY_NOOP_TRUNCATION.
8399 * hooks.h (hook_bool_uint_uint_true): Declare.
8400 * hooks.c (hook_bool_uint_uint_true): New function.
8401 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8402 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8403 * doc/tm.texi: Regenerate.
8404 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8405 rather than TRULY_NOOP_TRUNCATION in comments.
8406 (simplify_comparison): Likewise.
8407 (record_truncated_value): Likewise.
8408 * expmed.c (extract_bit_field_1): Likewise.
8409 (extract_split_bit_field): Likewise.
8410 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8411 instead of TRULY_NOOP_TRUNCATION.
8412 * function.c (assign_parm_setup_block): Likewise.
8413 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8414 * rtlhooks.c: Include target.h.
8415 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8416 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8417 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8418 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8419 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8420 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8421 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8422 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8423 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8424 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8425 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8426 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8427 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8428 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8429 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8430 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8431 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8432 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8433 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8434 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8435 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8436 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8437 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8438 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8439 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8440 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8441 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8442 rather than TRULY_NOOP_TRUNCATION in comments.
8443 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8444 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8445 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8446 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8447 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8448 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8449 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8450 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8451 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8452 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8453 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8454 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8455 rather than TRULY_NOOP_TRUNCATION in comments.
8456 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8457 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8458 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8459 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8460 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8461 TRULY_NOOP_TRUNCATION condition.
8462 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8463 (TRULY_NOOP_TRUNCATION): Delete.
8464 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8465 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8466 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8467 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8468 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8469 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8470 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8471 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8472 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8473 rather than TRULY_NOOP_TRUNCATION in comments.
8474 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8475 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8476 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8477 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8478 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8479 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8481 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8484 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8485 (*cmp_ior): Likewise.
8486 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8487 (*ior_scc_scc_cmp): Likewise.
8488 (*and_scc_scc): Likewise.
8489 (*and_scc_scc_cmp): Likewise.
8491 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8492 Alan Hayard <alan.hayward@arm.com>
8493 David Sherwood <david.sherwood@arm.com>
8495 * target.def (can_change_mode_class): New hook.
8496 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8497 (hard_regno_nregs): Likewise.
8498 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8499 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8500 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8501 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8502 (LOAD_EXTEND_OP): Update accordingly.
8503 * doc/tm.texi: Regenerate.
8504 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8505 CANNOT_CHANGE_MODE_CLASS.
8506 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8507 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8508 * combine.c (simplify_set): Update accordingly.
8509 * emit-rtl.c (validate_subreg): Likewise.
8510 * recog.c (general_operand): Likewise.
8511 * regcprop.c (mode_change_ok): Likewise.
8512 * reload1.c (choose_reload_regs): Likewise.
8513 (inherit_piecemeal_p): Likewise.
8514 * rtlanal.c (simplify_subreg_regno): Likewise.
8515 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8516 instead of CANNOT_CHANGE_MODE_CLASS.
8517 (reload_cse_simplify_operands): Likewise.
8518 * reload.c (push_reload): Use targetm.can_change_mode_class
8519 instead of CANNOT_CHANGE_MODE_CLASS.
8520 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8521 REG_CANNOT_CHANGE_MODE_P.
8522 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8523 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8524 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8525 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8526 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8527 (arm_can_change_mode_class): New function.
8528 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8529 than CANNOT_CHANGE_MODE_CLASS in comments.
8530 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8531 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8532 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8533 (ix86_can_change_mode_class): ...this new function, inverting the
8534 sense of the return value.
8535 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8536 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8537 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8538 (ia64_can_change_mode_class): New function.
8539 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8540 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8541 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8542 (m32c_can_change_mode_class): ...this new function, inverting the
8543 sense of the return value.
8544 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8545 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8546 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8547 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8548 (mips_can_change_mode_class): ...this new function, inverting the
8549 sense of the return value.
8550 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8551 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8552 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8553 (msp430_can_change_mode_class): New function.
8554 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8555 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8556 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8557 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8558 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8559 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8560 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8561 (pa_cannot_change_mode_class): Replace with...
8562 (pa_can_change_mode_class): ...this new function, inverting the
8563 sense of the return value.
8564 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8565 than CANNOT_CHANGE_MODE_CLASS in comments.
8566 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8567 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8568 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8569 (pdp11_cannot_change_mode_class): Replace with...
8570 (pdp11_can_change_mode_class): ...this new function, inverting the
8571 sense of the return value.
8572 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8573 * config/powerpcspe/powerpcspe-protos.h
8574 (rs6000_cannot_change_mode_class_ptr): Delete.
8575 * config/powerpcspe/powerpcspe.c
8576 (rs6000_cannot_change_mode_class_ptr): Delete.
8577 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8578 (rs6000_option_override_internal): Assign to
8579 targetm.can_change_mode_class instead of
8580 rs6000_cannot_change_mode_class_ptr.
8581 (rs6000_cannot_change_mode_class): Replace with...
8582 (rs6000_can_change_mode_class): ...this new function, inverting the
8583 sense of the return value.
8584 (rs6000_debug_cannot_change_mode_class): Replace with...
8585 (rs6000_debug_can_change_mode_class): ...this new function.
8586 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8587 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8588 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8589 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8590 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8592 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8593 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8594 (rs6000_option_override_internal): Assign to
8595 targetm.can_change_mode_class instead of
8596 rs6000_cannot_change_mode_class_ptr.
8597 (rs6000_cannot_change_mode_class): Replace with...
8598 (rs6000_can_change_mode_class): ...this new function, inverting the
8599 sense of the return value.
8600 (rs6000_debug_cannot_change_mode_class): Replace with...
8601 (rs6000_debug_can_change_mode_class): ...this new function.
8602 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8603 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8604 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8605 (s390_can_change_mode_class): ...this new function, inverting the
8606 sense of the return value.
8607 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8608 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8609 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8610 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8611 (sh_cannot_change_mode_class): Replace with...
8612 (sh_can_change_mode_class): ...this new function, inverting the
8613 sense of the return value.
8614 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8615 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8616 (sparc_can_change_mode_class): New function.
8617 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8618 * config/spu/spu.c (spu_can_change_mode_class): New function.
8619 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8620 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8621 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8622 (visium_can_change_mode_class): New function.
8623 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8625 2017-09-15 Richard Biener <rguenther@suse.de>
8627 PR tree-optimization/82217
8628 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8629 but not undefined case.
8631 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8634 * postreload.c (reload_cse_simplify_operands): Skip
8635 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8637 2017-09-15 Richard Biener <rguenther@suse.de>
8639 PR tree-optimization/68823
8640 * graphite-scop-detection.c (build_alias_set): If we have a
8641 possible dependence check whether we can handle them by just
8642 looking at the DRs DR_ACCESS_FNs.
8643 (build_scops): If build_alias_set fails, fail the SCOP.
8645 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8647 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8648 to support float128 built-in functions that require the ISA 3.0
8650 (BU_FLOAT128_3_HW): Likewise.
8651 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8653 (FMAF128): Likewise.
8655 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8656 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8657 floating point instructions.
8658 (rs6000_invalid_builtin): Likewise.
8659 (rs6000_builtin_mask_names): Likewise.
8660 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8661 (RS6000_BTM_FLOAT128_HW): Likewise.
8662 (RS6000_BTM_COMMON): Likewise.
8663 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8665 * doc/extend.texi (RS/6000 built-in functions): Document the
8666 IEEE 128-bit floating point square root and fused multiply-add
8669 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8671 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8672 reg (r2) isn't in the set of registers defined in the prologue.
8674 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8675 Alan Hayward <alan.hayward@arm.com>
8676 David Sherwood <david.sherwood@arm.com>
8678 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8679 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8680 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8681 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8682 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8684 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8685 max_vectorization_factor.
8686 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8688 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8689 Alan Hayward <alan.hayward@arm.com>
8690 David Sherwood <david.sherwood@arm.com>
8692 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8693 (vect_worthwhile_without_simd_p): Declare.
8694 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8695 (vectorizable_reduction): Use it.
8696 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8697 (vectorizable_operation): Likewise.
8699 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8700 Alan Hayward <alan.hayward@arm.com>
8701 David Sherwood <david.sherwood@arm.com>
8703 * tree-vectorizer.h (vect_get_num_copies): New function.
8704 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8705 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8706 (vectorizable_induction): Likewise.
8707 (vectorizable_live_operation): Likewise.
8708 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8709 (vectorizable_bswap): Likewise.
8710 (vectorizable_call): Likewise.
8711 (vectorizable_conversion): Likewise.
8712 (vectorizable_assignment): Likewise.
8713 (vectorizable_shift): Likewise.
8714 (vectorizable_operation): Likewise.
8715 (vectorizable_store): Likewise.
8716 (vectorizable_load): Likewise.
8717 (vectorizable_condition): Likewise.
8718 (vectorizable_comparison): Likewise.
8719 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8721 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8722 Alan Hayward <alan.hayward@arm.com>
8723 David Sherwood <david.sherwood@arm.com>
8725 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8726 of vect_init_vector.
8728 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8729 Alan Hayward <alan.hayward@arm.com>
8730 David Sherwood <david.sherwood@arm.com>
8732 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8733 an inline wrapper that provides a location.
8734 (gimple_build_vector): Likewise.
8735 * gimple-fold.c (gimple_build_vector_from_val): New function.
8736 (gimple_build_vector): Likewise.
8737 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8738 functions to build the initial value. Always return a gimple value.
8739 (get_initial_defs_for_reduction): Likewise. Only compute
8741 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8742 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8743 (vectorizable_induction): Use gimple_build_vector rather than
8746 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8747 Alan Hayward <alan.hayward@arm.com>
8748 David Sherwood <david.sherwood@arm.com>
8750 * target.h (vec_perm_indices): New typedef.
8751 (auto_vec_perm_indices): Likewise.
8752 * optabs-query.h: Include target.h
8753 (can_vec_perm_p): Take a vec_perm_indices *.
8754 * optabs-query.c (can_vec_perm_p): Likewise.
8755 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8756 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8757 * tree-vect-generic.c (lower_vec_perm): Likewise.
8758 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8759 (vect_grouped_load_supported): Likewise.
8760 (vect_shift_permute_load_chain): Likewise.
8761 (vect_permute_store_chain): Use auto_vec_perm_indices.
8762 (vect_permute_load_chain): Likewise.
8763 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8764 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8765 Update uses of can_vec_perm_p.
8766 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8767 mode with a number of elements. Take a vec_perm_indices *.
8768 (vect_create_epilog_for_reduction): Update accordingly.
8769 Use auto_vec_perm_indices.
8770 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8771 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8772 (vect_transform_slp_perm_load): Likewise.
8773 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8774 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8775 (vect_gen_perm_mask_checked): Likewise.
8776 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8777 (vect_gen_perm_mask_checked): Likewise.
8778 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8779 (vectorizable_store): Likewise.
8780 (vectorizable_load): Likewise.
8781 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8782 (vectorizable_bswap): Likewise.
8784 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8785 Alan Hayward <alan.hayward@arm.com>
8786 David Sherwood <david.sherwood@arm.com>
8788 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8789 * tree.c (build_vector): Likewise.
8790 (build_vector_from_ctor): Update accordingly.
8791 (build_vector_from_val): Likewise.
8792 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8793 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8794 * tree-vect-generic.c (add_rshift): Likewise.
8795 (expand_vector_divmod): Likewise.
8796 (optimize_vector_constructor): Likewise.
8797 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8798 (vect_transform_slp_perm_load): Likewise.
8799 (vect_schedule_slp_instance): Likewise.
8800 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8801 (vectorizable_call): Likewise.
8802 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8803 * expmed.c (make_tree): Likewise.
8804 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8805 a vector passed to build_vector.
8806 (fold_convert_const): Likewise.
8807 (exact_inverse): Likewise.
8808 (fold_ternary_loc): Likewise.
8809 (fold_relational_const): Likewise.
8810 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8811 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8812 (const_unop): Likewise. Store the reduction accumulator in a
8813 variable rather than an array.
8814 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8815 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8816 the new vector, rather than constructing it after the input arrays.
8817 (native_interpret_vector): Use auto_vec<tree> when building
8818 a vector passed to build_vector. Add elements in order.
8819 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8820 auto_vec<tree> when building a vector passed to build_vector.
8821 (vect_create_epilog_for_reduction): Likewise.
8822 (vectorizable_induction): Likewise.
8823 (get_initial_def_for_reduction): Likewise. Fix indentation of
8825 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8827 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8828 passed to build_vector.
8830 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8831 Alan Hayward <alan.hayward@arm.com>
8832 David Sherwood <david.sherwood@arm.com>
8834 * tree-core.h (tree_base::u): Add an "nelts" field.
8835 (tree_vector): Use VECTOR_CST_NELTS as the length.
8836 * tree.c (tree_size): Likewise.
8837 (make_vector): Initialize VECTOR_CST_NELTS.
8838 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8839 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8840 TYPE_VECTOR_SUBPARTS.
8841 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8842 as the number of units, setting it from VECTOR_CST_NELTS.
8843 (const_vector_from_tree): Likewise.
8844 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8845 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8846 (fold_negate_expr_1): Likewise.
8847 (fold_convert_const): Likewise.
8848 (const_binop): Likewise. Differentiate the number of output and
8850 (const_unop): Likewise.
8851 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8852 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8853 in cases that did the opposite.
8855 2017-09-14 Richard Biener <rguenther@suse.de>
8857 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8860 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8862 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8864 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8867 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8868 if and where to split a bb, except for splitting before debug insn
8869 sequences followed by non-label real insn. Delete debug insns
8870 in between basic blocks.
8872 * combine.c (make_compound_operation_int): Formatting fixes.
8874 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8875 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8876 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8877 * config/sol2.h (LINK_EH_SPEC): Likewise.
8878 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8879 * config/s390/linux.h (LINK_SPEC): Likewise.
8880 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8881 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8882 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8883 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8884 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8885 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8886 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8887 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8889 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8891 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8893 (ENDFILE_LINUX_SPEC): Likewise.
8894 (LINK_EH_SPEC): Likewise.
8895 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8896 (LINK_OS_LINUX_SPEC32): Likewise.
8897 (LINK_OS_LINUX_SPEC64): Likewise.
8898 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8899 (LINK_OS_LINUX_SPEC): Likewise.
8901 2017-09-13 Martin Liska <mliska@suse.cz>
8904 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8905 CASE_HIGH is NULL_TREE.
8907 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8908 Alan Hayward <alan.hayward@arm.com>
8909 David Sherwood <david.sherwood@arm.com>
8911 * target.def (secondary_memory_needed): New hook.
8912 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8913 instead of SECONDARY_MEMORY_NEEDED.
8914 (secondary_memory_needed_mode): Likewise.
8915 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8916 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8917 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8918 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8919 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8920 * doc/tm.texi: Regenerate.
8921 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8922 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8923 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8924 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8925 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8926 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8927 mode argument first and change the reg_class arguments to reg_class_t.
8928 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8929 Make static. Update the call to inline_secondary_memory_needed.
8930 (ix86_register_move_cost): Update the call to
8931 inline_secondary_memory_needed.
8932 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8933 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8935 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8936 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8938 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8939 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8940 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8941 and match hook interface. Add comment from mips.h.
8942 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8943 * config/mmix/mmix.md (truncdfsf2): Refer to
8944 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8946 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8947 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8948 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8949 (pa_secondary_memory_needed): New function.
8950 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8951 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8952 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8953 (pdp11_secondary_memory_needed): Make static and match hook interface.
8954 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8955 * config/powerpcspe/powerpcspe-protos.h
8956 (rs6000_secondary_memory_needed_ptr): Delete.
8957 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8959 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8960 (rs6000_option_override_internal): Assign to
8961 targetm.secondary_memory_needed rather than
8962 rs6000_secondary_memory_needed_ptr.
8963 (rs6000_secondary_memory_needed): Match hook interface.
8964 (rs6000_debug_secondary_memory_needed): Likewise.
8965 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8966 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8967 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8968 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8969 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8970 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8972 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8973 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8974 (rs6000_option_override_internal): Assign to
8975 targetm.secondary_memory_needed rather than
8976 rs6000_secondary_memory_needed_ptr.
8977 (rs6000_secondary_memory_needed): Match hook interface.
8978 (rs6000_debug_secondary_memory_needed): Likewise.
8979 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8980 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8981 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8982 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8983 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8984 (sparc_secondary_memory_needed): New function.
8985 * lra-constraints.c (check_and_process_move): Refer to
8986 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8988 (curr_insn_transform): Likewise.
8989 (process_alt_operands): Use targetm.secondary_memory_needed
8990 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8991 (check_secondary_memory_needed_p): Likewise.
8992 (choose_split_class): Likewise.
8993 * reload.c: Unconditionally include code that was previously
8994 conditional on SECONDARY_MEMORY_NEEDED.
8995 (push_secondary_reload): Use targetm.secondary_memory_needed
8996 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8997 (push_reload): Likewise.
8998 * reload1.c: Unconditionally include code that was previously
8999 conditional on SECONDARY_MEMORY_NEEDED.
9000 (choose_reload_regs): Use targetm.secondary_memory_needed
9001 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9002 (gen_reload): Likewise.
9003 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9005 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9006 Alan Hayward <alan.hayward@arm.com>
9007 David Sherwood <david.sherwood@arm.com>
9009 * target.def (secondary_memory_needed_mode): New hook:
9010 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9011 * targhooks.h (default_secondary_memory_needed_mode): New function.
9012 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9013 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9014 * doc/tm.texi: Regenerate.
9015 * lra-constraints.c (check_and_process_move): Use
9016 targetm.secondary_memory_needed_mode instead of
9017 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9018 (curr_insn_transform): Likewise.
9019 * reload.c (get_secondary_mem): Likewise.
9020 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9021 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9023 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9024 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9025 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9026 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9027 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9029 * config/powerpcspe/powerpcspe-protos.h
9030 (rs6000_secondary_memory_needed_mode): Delete.
9031 * config/powerpcspe/powerpcspe.c
9032 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9033 (rs6000_secondary_memory_needed_mode): Make static.
9034 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9035 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9037 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9039 (rs6000_secondary_memory_needed_mode): Make static.
9040 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9041 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9042 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9043 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9044 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9046 (sparc_secondary_memory_needed_mode): New function.
9047 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9049 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9051 * config/aarch64/constraints.md (Umq): New constraint.
9052 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9054 (mov<mode>): Update condition.
9056 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9058 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9059 when bitposition is the same.
9061 2017-09-13 Richard Biener <rguenther@suse.de>
9063 * dwarf2out.c (output_die_symbol): Remove.
9064 (output_die): Do not output a DIEs symbol.
9066 2017-09-13 Richard Biener <rguenther@suse.de>
9069 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9070 default-def to avoid breaking iterator update with the weird
9071 interaction with cgraph_update_edges_for_call_stmt_node.
9073 2017-09-13 Richard Biener <rguenther@suse.de>
9075 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9076 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9077 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9078 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9080 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9082 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9083 Disable pc relative literal load irrespective of
9084 TARGET_FIX_ERR_A53_84341 for default.
9086 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9088 * config/sparc/sparc.c (output_return): Output the source location of
9089 the insn in the delay slot, if any.
9090 (output_sibcall): Likewise.
9092 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9095 * common.opt (-static-pie): New alias.
9096 (shared): Negate static-pie.
9097 (-no-pie): Update help text.
9099 (static-pie): New option.
9100 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9101 -static-pie support.
9102 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9103 (LINK_EH_SPEC): Likewise.
9104 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9105 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9106 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9107 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9108 (init_gcc_specs): Likewise.
9109 (init_spec): Likewise.
9110 (display_help): Update help message for -pie.
9111 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9114 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9116 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9117 (movdi_aarch64): Likewise.
9118 (movti_aarch64): Likewise.
9120 2017-09-12 Simon Wright <simon@pushface.org>
9123 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9124 calculation of the minor version, always output as 0.
9126 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9129 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9130 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9131 on it early, rather than manual conversion late. For
9132 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9133 instead of performing manual conversion.
9135 2017-09-12 Carl Love <cel@us.ibm.com>
9137 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9138 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9140 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9141 VMULOSW): Add definitions.
9142 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9143 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9144 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9145 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9146 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9148 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9150 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9152 (movti_aarch64): Likewise.
9153 (movdf_aarch64): Likewise.
9154 (movtf_aarch64): Likewise.
9155 (load_pairdi): Likewise.
9156 (store_pairdi): Likewise.
9157 (load_pairdf): Likewise.
9158 (store_pairdf): Likewise.
9159 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9160 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9161 (ldr_got_small_<mode>): Likewise.
9162 (ldr_got_small_28k_<mode>): Likewise.
9163 (ldr_got_tiny): Likewise.
9164 * config/aarch64/iterators.md (ldst_sz): New.
9165 (ldpstp_sz): Likewise.
9166 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9168 (thunderx_load): Split load_8 to load_16.
9169 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9171 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9172 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9173 (xgene1_store_pair): Split store_8 to store_16.
9174 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9175 (falkor_st_0_st_sd): Split store_8 to store_16.
9177 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9179 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9180 and store1/2/3/4 to store_4/8/12/16.
9181 * config/aarch64/aarch64.md: Update for rename.
9182 * config/arm/arm.md: Likewise.: Likewise.
9183 * config/arm/arm.c: Likewise.
9184 * config/arm/thumb1.md: Likewise.
9185 * config/arm/thumb2.md: Likewise.
9186 * config/arm/vfp.md: Likewise.
9187 * config/arm/arm-generic.md: Likewise.
9188 * config/arm/arm1020e.md: Likewise.
9189 * config/arm/arm1026ejs.md: Likewise.
9190 * config/arm/arm1136jfs.md: Likewise.
9191 * config/arm/arm926ejs.md: Likewise.
9192 * config/arm/cortex-a15.md: Likewise.
9193 * config/arm/cortex-a17.md: Likewise.
9194 * config/arm/cortex-a5.md: Likewise.
9195 * config/arm/cortex-a53.md: Likewise.
9196 * config/arm/cortex-a57.md: Likewise.
9197 * config/arm/cortex-a7.md: Likewise.
9198 * config/arm/cortex-a8.md: Likewise.
9199 * config/arm/cortex-a9.md: Likewise.
9200 * config/arm/cortex-m4.md: Likewise.
9201 * config/arm/cortex-m7.md: Likewise.
9202 * config/arm/cortex-r4.md: Likewise.
9203 * config/arm/exynos-m1.md: Likewise.
9204 * config/arm/fa526.md: Likewise.
9205 * config/arm/fa606te.md: Likewise.
9206 * config/arm/fa626te.md: Likewise.
9207 * config/arm/fa726te.md: Likewise.
9208 * config/arm/fmp626.md: Likewise.
9209 * config/arm/iwmmxt.md: Likewise.
9210 * config/arm/ldmstm.md: Likewise.
9211 * config/arm/marvell-pj4.md: Likewise.
9212 * config/arm/xgene1.md: Likewise.
9213 * config/aarch64/thunderx.md: Likewise.
9214 * config/aarch64/thunderx2t99.md: Likewise.
9215 * config/aarch64/falkor.md: Likewise.
9217 2017-09-12 Martin Liska <mliska@suse.cz>
9219 * attribs.c (private_lookup_attribute): New function.
9220 * attribs.h (private_lookup_attribute): Declared here.
9221 (lookup_attribute): Called from this place.
9223 2017-09-12 Richard Biener <rguenther@suse.de>
9225 PR tree-optimization/82157
9226 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9227 stmts with side-effects.
9229 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9230 Alan Hayward <alan.hayward@arm.com>
9231 David Sherwood <david.sherwood@arm.com>
9233 * target.def (hard_regno_nregs): New hook.
9234 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9235 * targhooks.h (default_hard_regno_nregs): Declare.
9236 * targhooks.c (default_hard_regno_nregs): New function.
9237 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9238 (TARGET_HARD_REGNO_NREGS): ...this hook.
9239 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9240 (CLASS_MAX_NREGS): Likewise.
9241 * doc/tm.texi: Regenerate.
9242 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9243 instead of HARD_REGNO_NREGS.
9244 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9245 HARD_REGNO_NREGS in the comment.
9246 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9247 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9248 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9249 Return an unsigned int.
9250 (TARGET_HARD_REGNO_NREGS): Redefine.
9251 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9252 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9253 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9254 (arc_hard_regno_nregs): New function.
9255 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9256 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9257 (arm_hard_regno_nregs): New function.
9258 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9259 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9260 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9261 (TARGET_HARD_REGNO_NREGS): Redefine.
9262 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9263 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9264 (HARD_REGNO_NREGS): Delete.
9265 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9266 (cr16_hard_regno_nregs): New function.
9267 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9268 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9269 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9270 (cris_hard_regno_nregs): New function.
9271 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9272 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9273 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9274 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9275 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9276 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9277 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9278 (frv_hard_regno_nregs): Make static. Take and return an
9280 (frv_class_max_nregs): Remove outdated copy of documentation.
9281 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9282 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9283 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9284 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9285 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9286 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9287 (TARGET_HARD_REGNO_NREGS): Redefine.
9288 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9289 (CLASS_MAX_NREGS): Update comment.
9290 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9291 (ia64_hard_regno_nregs): New function.
9292 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9293 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9294 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9295 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9296 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9298 (m32c_hard_regno_nregs): Likewise. Make static.
9299 (TARGET_HARD_REGNO_NREGS): Redefine.
9300 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9301 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9302 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9303 (m68k_hard_regno_nregs): New function.
9304 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9305 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9306 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9307 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9308 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9309 Take and return an unsigned int.
9310 (TARGET_HARD_REGNO_NREGS): Redefine.
9311 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9312 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9313 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9314 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9315 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9316 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9317 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9318 (msp430_hard_regno_nregs): Make static. Take and return an
9320 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9321 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9322 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9323 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9324 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9325 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9326 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9327 (TARGET_HARD_REGNO_NREGS): Redefine.
9328 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9329 (PA_HARD_REGNO_NREGS): ...this.
9330 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9331 (PA_HARD_REGNO_NREGS): ...this.
9332 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9333 (pa_hard_regno_nregs): New function.
9334 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9335 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9336 (pdp11_hard_regno_nregs): New function.
9337 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9338 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9339 (rs6000_hard_regno_nregs_hook): New function.
9340 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9341 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9342 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9343 Take and return an unsigned int. Move earlier in file.
9344 (TARGET_HARD_REGNO_NREGS): Redefine.
9345 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9346 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9347 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9348 (rl78_hard_regno_nregs): Make static. Take and return an
9350 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9351 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9352 (rs6000_hard_regno_nregs_hook): New function.
9353 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9354 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9355 (TARGET_HARD_REGNO_NREGS): Redefine.
9356 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9357 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9358 instead of HARD_REGNO_NREGS.
9359 (s390_hard_regno_nregs): New function.
9360 (s390_hard_regno_mode_ok): Add comment from s390.h.
9361 (TARGET_HARD_REGNO_NREGS): Redefine.
9362 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9363 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9364 (sh_hard_regno_nregs): New function.
9365 (sh_pass_in_reg_p): Use it.
9366 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9367 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9368 (sparc_hard_regno_nregs): New function.
9369 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9370 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9371 (spu_function_arg_advance): Use it, supplying a valid register number.
9372 (TARGET_HARD_REGNO_NREGS): Redefine.
9373 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9374 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9375 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9376 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9377 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9378 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9379 (CLASS_MAX_NREGS): Remove copy of old documentation.
9380 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9381 (visium_hard_regno_nregs): New function.
9382 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9383 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9384 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9385 xtensa_hard_regno_nregs): New function.
9386 * system.h (HARD_REGNO_NREGS): Poison.
9388 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9390 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9391 hard_regno_nregs instead of HARD_REGNO_NREGS.
9392 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9393 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9394 (c6x_expand_epilogue): Likewise.
9395 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9396 (frv_read_iacc_argument): Likewise.
9397 * config/sh/sh.c: Include regs.h.
9398 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9399 (regs_used): Likewise.
9400 (output_stack_adjust): Likewise.
9401 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9402 * expmed.c: Include regs.h.
9403 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9404 * ree.c: Include regs.h.
9405 (combine_reaching_defs): Use hard_regno_nregs instead of
9407 (add_removable_extension): Likewise.
9409 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9411 * regs.h (hard_regno_nregs): Turn into a function.
9412 (end_hard_regno): Update accordingly.
9413 * caller-save.c (setup_save_areas): Likewise.
9414 (save_call_clobbered_regs): Likewise.
9415 (replace_reg_with_saved_mem): Likewise.
9416 (insert_restore): Likewise.
9417 (insert_save): Likewise.
9418 * combine.c (can_change_dest_mode): Likewise.
9419 (move_deaths): Likewise.
9420 (distribute_notes): Likewise.
9421 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9422 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9423 (rs6000_split_multireg_move): Likewise.
9424 (rs6000_register_move_cost): Likewise.
9425 (rs6000_memory_move_cost): Likewise.
9426 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9427 (rs6000_split_multireg_move): Likewise.
9428 (rs6000_register_move_cost): Likewise.
9429 (rs6000_memory_move_cost): Likewise.
9430 * cselib.c (cselib_reset_table): Likewise.
9431 (cselib_lookup_1): Likewise.
9432 * emit-rtl.c (set_mode_and_regno): Likewise.
9433 * function.c (aggregate_value_p): Likewise.
9434 * ira-color.c (setup_profitable_hard_regs): Likewise.
9435 (check_hard_reg_p): Likewise.
9436 (calculate_saved_nregs): Likewise.
9437 (assign_hard_reg): Likewise.
9438 (improve_allocation): Likewise.
9439 (calculate_spill_cost): Likewise.
9440 * ira-emit.c (modify_move_list): Likewise.
9441 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9442 (ira_hard_reg_in_set_p): Likewise.
9443 * ira.c (setup_reg_mode_hard_regset): Likewise.
9444 (clarify_prohibited_class_mode_regs): Likewise.
9445 (check_allocation): Likewise.
9446 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9447 (lra_setup_reg_renumber): Likewise.
9448 (setup_try_hard_regno_pseudos): Likewise.
9449 (spill_for): Likewise.
9450 (assign_hard_regno): Likewise.
9451 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9452 * lra-constraints.c (in_class_p): Likewise.
9453 (lra_constraint_offset): Likewise.
9454 (simplify_operand_subreg): Likewise.
9455 (lra_constraints): Likewise.
9456 (split_reg): Likewise.
9457 (split_if_necessary): Likewise.
9458 (invariant_p): Likewise.
9459 (inherit_in_ebb): Likewise.
9460 * lra-lives.c (process_bb_lives): Likewise.
9461 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9462 (get_hard_regs): Likewise.
9463 (do_remat): Likewise.
9464 * lra-spills.c (assign_spill_hard_regs): Likewise.
9465 * mode-switching.c (create_pre_exit): Likewise.
9466 * postreload.c (reload_combine_recognize_pattern): Likewise.
9467 * recog.c (peep2_find_free_register): Likewise.
9468 * regcprop.c (kill_value_regno): Likewise.
9469 (set_value_regno): Likewise.
9470 (copy_value): Likewise.
9471 (maybe_mode_change): Likewise.
9472 (find_oldest_value_reg): Likewise.
9473 (copyprop_hardreg_forward_1): Likewise.
9474 * regrename.c (check_new_reg_p): Likewise.
9475 (regrename_do_replace): Likewise.
9476 * reload.c (push_reload): Likewise.
9477 (combine_reloads): Likewise.
9478 (find_dummy_reload): Likewise.
9479 (operands_match_p): Likewise.
9480 (find_reloads): Likewise.
9481 (find_equiv_reg): Likewise.
9482 (reload_adjust_reg_for_mode): Likewise.
9483 * reload1.c (count_pseudo): Likewise.
9484 (count_spilled_pseudo): Likewise.
9485 (find_reg): Likewise.
9486 (clear_reload_reg_in_use): Likewise.
9487 (free_for_value_p): Likewise.
9488 (allocate_reload_reg): Likewise.
9489 (choose_reload_regs): Likewise.
9490 (reload_adjust_reg_for_temp): Likewise.
9491 (emit_reload_insns): Likewise.
9492 (delete_output_reload): Likewise.
9493 * rtlanal.c (subreg_get_info): Likewise.
9494 * sched-deps.c (sched_analyze_reg): Likewise.
9495 * sel-sched.c (init_regs_for_mode): Likewise.
9496 (mark_unavailable_hard_regs): Likewise.
9497 (choose_best_reg_1): Likewise.
9498 (verify_target_availability): Likewise.
9499 * valtrack.c (dead_debug_insert_temp): Likewise.
9500 * var-tracking.c (track_loc_p): Likewise.
9501 (emit_note_insn_var_location): Likewise.
9502 * varasm.c (make_decl_rtl): Likewise.
9503 * reginfo.c (choose_hard_reg_mode): Likewise.
9504 (init_reg_modes_target): Refer directly to
9505 this_target_regs->x_hard_regno_nregs.
9507 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9509 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9510 instead of hard_regno_nregs.
9512 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9514 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9515 end_hard_regno instead of hard_regno_nregs.
9516 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9517 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9518 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9519 * ira-color.c (improve_allocation): Likewise.
9520 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9521 * lra-lives.c (mark_regno_live): Likewise.
9522 (mark_regno_dead): Likewise.
9523 * lra-remat.c (operand_to_remat): Likewise.
9524 * lra.c (collect_non_operand_hard_regs): Likewise.
9525 * postreload.c (reload_combine_note_store): Likewise.
9526 (move2add_valid_value_p): Likewise.
9527 * reload.c (regno_clobbered_p): Likewise.
9529 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9531 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9533 * config/v850/v850.c (v850_reorg): Likewise.
9534 * reload.c (refers_to_regno_for_reload_p): Likewise.
9535 (find_equiv_reg): Likewise.
9536 * reload1.c (reload_reg_reaches_end_p): Likewise.
9538 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9540 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9542 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9543 * config/arm/arm.c (output_move_neon): Likewise.
9544 (arm_attr_length_move_neon): Likewise.
9545 (neon_split_vcombine): Likewise.
9546 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9547 (c6x_mark_reg_written): Likewise.
9548 (c6x_dwarf_register_span): Likewise.
9549 * config/i386/i386.c (ix86_save_reg): Likewise.
9550 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9551 (rws_access_reg): Likewise.
9552 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9553 * mode-switching.c (create_pre_exit): Likewise.
9554 * ree.c (combine_reaching_defs): Likewise.
9555 (add_removable_extension): Likewise.
9556 * regcprop.c (find_oldest_value_reg): Likewise.
9557 (copyprop_hardreg_forward_1): Likewise.
9558 * reload.c (reload_inner_reg_of_subreg): Likewise.
9559 (push_reload): Likewise.
9560 (combine_reloads): Likewise.
9561 (find_dummy_reload): Likewise.
9562 (reload_adjust_reg_for_mode): Likewise.
9563 * reload1.c (find_reload_regs): Likewise.
9564 (forget_old_reloads_1): Likewise.
9565 (reload_reg_free_for_value_p): Likewise.
9566 (reload_adjust_reg_for_temp): Likewise.
9567 (emit_reload_insns): Likewise.
9568 (delete_output_reload): Likewise.
9569 * sel-sched.c (choose_best_reg_1): Likewise.
9570 (choose_best_pseudo_reg): Likewise.
9572 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9573 Alan Hayward <alan.hayward@arm.com>
9574 David Sherwood <david.sherwood@arm.com>
9576 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9577 * target.def (slow_unaligned_access): New hook.
9578 * targhooks.h (default_slow_unaligned_access): Declare.
9579 * targhooks.c (default_slow_unaligned_access): New function.
9580 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9581 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9582 * doc/tm.texi: Regenerate.
9583 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9584 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9585 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9587 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9588 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9590 (rs6000_slow_unaligned_access): New function.
9591 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9592 (expand_block_compare): Likewise.
9593 (expand_strn_compare): Likewise.
9594 (rs6000_rtx_costs): Likewise.
9595 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9596 (riscv_slow_unaligned_access): Likewise.
9597 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9598 (riscv_slow_unaligned_access_p): ...this and make static.
9599 (riscv_option_override): Update accordingly.
9600 (riscv_slow_unaligned_access): New function.
9601 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9602 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9603 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9604 (rs6000_slow_unaligned_access): New function.
9605 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9606 (rs6000_rtx_costs): Likewise.
9607 * config/rs6000/rs6000-string.c (expand_block_compare)
9608 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9609 of SLOW_UNALIGNED_ACCESS.
9610 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9611 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9612 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9613 of SLOW_UNALIGNED_ACCESS.
9614 * expmed.c (simple_mem_bitfield_p): Likewise.
9615 * expr.c (alignment_for_piecewise_move): Likewise.
9616 (emit_group_load_1): Likewise.
9617 (emit_group_store): Likewise.
9618 (copy_blkmode_from_reg): Likewise.
9619 (emit_push_insn): Likewise.
9620 (expand_assignment): Likewise.
9621 (store_field): Likewise.
9622 (expand_expr_real_1): Likewise.
9623 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9624 * lra-constraints.c (simplify_operand_subreg): Likewise.
9625 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9626 * gimple-ssa-store-merging.c: Likewise in block comment at start
9628 * tree-ssa-strlen.c: Include target.h.
9629 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9630 of SLOW_UNALIGNED_ACCESS.
9631 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9633 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9635 PR rtl-optimization/82185
9636 * expmed.c (emit_store_flag_int): Only test tem if it has been
9639 2017-09-12 Richard Biener <rguenther@suse.de>
9642 * match.pd ((FTYPE) N CMP CST): Fix typo.
9644 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9646 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9648 (mips_near_type_p): Add 'short_call' attribute as a synonym
9650 * doc/extend.texi (short_call): Document new function attribute.
9652 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9655 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9656 assertion check that in the condition.
9657 (get_atomic_generic_size): Likewise. Before testing if parameter
9658 has pointer type, if it has array type, call for C++
9659 default_conversion to perform array-to-pointer conversion.
9661 2017-09-12 Richard Biener <rguenther@suse.de>
9663 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9664 for operations we cannot scalarize.
9666 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9668 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9669 vectors heap vectors. Clean up comments.
9670 Make visited_bbs a reference.
9671 (profitable_jump_thread_path): Make GC
9672 vectors heap vectors. Clean up comments.
9674 (convert_and_register_jump_thread_path): Make GC vectors heap
9676 (check_subpath_and_update_thread_path): Same. Clean up comments.
9677 Make visited_bbs a reference.
9678 (handle_phi): Abstract common code to to
9679 register_jump_thread_path_if_profitable.
9680 Rename VAR_BB to DEF_BB.
9682 Make GC vectors heap vectors.
9683 Make visited_bbs a reference.
9684 (handle_assignment): Same.
9685 (register_jump_thread_path_if_profitable): New.
9686 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9688 Make GC vectors heap vectors. Clean up comments.
9689 Make visited_bbs a reference.
9690 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9691 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9694 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9697 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9698 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9700 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9702 Revert r251800 and r251799.
9704 2017-09-11 Martin Jambor <mjambor@suse.cz>
9707 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9708 arguments in advance.
9709 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9710 use it to find predecessor edges.
9711 (naive_outof_ssa): Collect vector of predecessors.
9713 2017-09-08 Jason Merrill <jason@redhat.com>
9715 PR c++/70029 - ICE with ref-qualifier and -flto
9716 * langhooks.h (struct lang_hooks_for_types): Add
9717 copy_lang_qualifiers.
9718 * attribs.c (build_type_attribute_qual_variant): Use it.
9719 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9721 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9722 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9724 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9727 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9728 (*mulsi3_sp64): New instruction.
9729 (mulsi3): New expander.
9731 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9733 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9735 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9737 * sancov.c: Include memmodel.h.
9739 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9742 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9745 2017-09-07 Carl Love <cel@us.ibm.com>
9747 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9748 the sldi instruction.
9750 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9752 * sancov.c: Include tm_p.h.
9754 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9757 * output.h (switch_to_other_text_partition): New declaration.
9758 * varasm.c (switch_to_other_text_partition): New function.
9759 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9760 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9761 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9762 to the other text partition before emitting LCL label and switch back
9763 after emitting the word after it.
9765 2017-09-07 Richard Biener <rguenther@suse.de>
9767 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9768 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9769 critical edges here, after loop init.
9770 (pass_data_pre): Remove PROP_no_crit_edges flags.
9771 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9772 for valueization of call args to avoid leaking VN_TOP.
9773 (visit_use): Assert we do not visit default defs.
9774 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9775 Use error_mark_node to more easily detect leaking VN_TOP.
9776 All default-defs are varying, not VN_TOP. Mark them visited.
9777 (run_scc_vn): Make code match comment.
9779 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9781 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9782 OPTION_MASK_FLOAT128_KEYWORD.
9783 (POWERPC_MASKS): Likewise.
9784 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9785 support for the -mfloat128-type option, and make -mfloat128
9786 default on PowerPC Linux systems. Define or undefine
9787 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9788 Define __float128 to be __ieee128 if IEEE 128-bit support is
9789 enabled, or undefine it.
9790 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9791 Delete defining __FLOAT128_TYPE__.
9792 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9793 -mfloat128-type option and make -mfloat128 default on PowerPC
9795 (TARGET_FLOAT128_TYPE): Likewise.
9796 (-mfloat128-type): Likewise.
9797 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9798 Delete the -mfloat128-type option and make -mfloat128 default on
9799 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9800 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9801 128-bit floating point is enabled. Change tests from using
9802 -mfloat128-type to -mfloat128.
9803 (rs6000_mangle_type): Use the correct mangling for the __float128
9804 type even if normal long double is restricted to 64-bits.
9805 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9807 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9808 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9809 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9811 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9812 documentation for -mfloat128.
9814 2017-09-06 Olivier Hainque <hainque@adacore.com>
9816 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9818 2017-09-06 Wish Wu <wishwu007@gmail.com>
9819 Jakub Jelinek <jakub@redhat.com>
9821 * asan.c (initialize_sanitizer_builtins): Add
9822 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9823 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9824 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9825 BT_FN_VOID_UINT64_PTR variables.
9826 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9827 (BT_FN_VOID_UINT16_UINT16): Likewise.
9828 (BT_FN_VOID_UINT32_UINT32): Likewise.
9829 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9830 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9831 (BT_FN_VOID_UINT64_PTR): Likewise.
9832 * common.opt (flag_sanitize_coverage): New variable.
9833 (fsanitize-coverage=trace-pc): Remove.
9834 (fsanitize-coverage=): Add.
9835 * flag-types.h (enum sanitize_coverage_code): New enum.
9836 * fold-const.c (fold_range_test): Disable non-short-circuit
9837 optimization if flag_sanitize_coverage.
9838 (fold_truth_andor): Likewise.
9839 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9840 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9841 (coverage_sanitizer_opts): New array.
9842 (get_closest_sanitizer_option): Add OPTS argument, handle also
9843 OPT_fsanitize_coverage_.
9844 (parse_sanitizer_options): Adjusted to also handle
9845 OPT_fsanitize_coverage_.
9846 (common_handle_option): Add OPT_fsanitize_coverage_.
9847 * sancov.c (instrument_comparison, instrument_switch): New function.
9848 (sancov_pass): Add trace-cmp support.
9849 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9850 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9851 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9852 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9853 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9854 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9855 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9856 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9857 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9858 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9860 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9862 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9863 error. Only quit immediately if parsing is complete.
9864 (BEGIN): Initialize fatal_err and parse_done.
9865 (begin fpu, end fpu): Check number of arguments.
9866 (begin arch, end arch): Likewise.
9867 (begin cpu, end cpu): Likewise.
9868 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9869 (optalias): Likewise.
9871 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9873 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9874 * config/arm/arm-isa.h: Delete. Move definitions to ...
9875 * arm-cpus.in: ... here. Use new feature and fgroup values.
9876 * config/arm/arm.c (arm_option_override): Use lower case for feature
9878 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9879 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9880 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9882 (print_isa_bits_for): New function.
9883 (gen_isa): New function.
9884 (gen_comm_data): Use print_isa_bits_for.
9885 (define feature): New keyword.
9886 (define fgroup): New keyword.
9887 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9888 (arm-isa.h): Add rule to generate file.
9889 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9890 case for feature bit names.
9892 2017-09-06 Richard Biener <rguenther@suse.de>
9894 * tree-ssa-pre.c (NECESSARY): Remove.
9895 (create_expression_by_pieces): Do not touch pass-local flags.
9896 (insert_into_preds_of_block): Likewise.
9897 (do_pre_regular_insertion): Likewise.
9898 (eliminate_insert): Likewise.
9899 (eliminate_dom_walker::before_dom_children): Likewise.
9900 (fini_eliminate): Do not look at inserted_exprs.
9901 (mark_operand_necessary): Remove.
9902 (remove_dead_inserted_code): Replace with simple work-list
9903 algorithm based on inserted_exprs and SSA uses.
9904 (pass_pre::execute): Re-order fini_eliminate and
9905 remove_dead_inserted_code.
9907 2017-09-06 Olivier Hainque <hainque@adacore.com>
9909 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9910 for VxWorks 7. Adjust surrounding comments.
9912 2017-09-06 Richard Biener <rguenther@suse.de>
9914 * gimple-ssa-strength-reduction.c
9915 (find_candidates_dom_walker::before_dom_children): Also allow
9918 2017-09-06 Richard Biener <rguenther@suse.de>
9920 PR tree-optimization/82108
9921 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9922 for gap in the non-permutation SLP case.
9924 2017-09-06 Martin Jambor <mjambor@suse.cz>
9926 PR tree-optimization/82078
9927 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9928 add_access_to_work_queue...
9929 (build_accesses_from_assign): ...here.
9930 (propagate_all_subaccesses): Make sure racc is the group
9931 representative, if there is one.
9933 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9936 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9939 2017-09-06 Richard Biener <rguenther@suse.de>
9941 * gimple-ssa-strength-reduction.c
9942 (find_candidates_dom_walker::before_doom_children): Use a
9943 type and not a mode check.
9945 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9948 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9950 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9952 2017-09-05 Jeff Law <law@redhat.com>
9954 PR tree-optimization/64910
9955 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9956 cases where we have 3 or more operands.
9958 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9961 * omp-low.c (lower_omp_for): Recompute tree invariant if
9962 gimple_omp_for_initial/final is ADDR_EXPR.
9965 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9966 into gimple val before gimplification fo the COND_EXPR.
9968 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9970 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9971 REGION_COPY argument.
9972 (thread_through_all_blocks): Remove unused argument to
9973 duplicate_thread_path.
9975 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9976 Alan Hayward <alan.hayward@arm.com>
9977 David Sherwood <david.sherwood@arm.com>
9979 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9980 Take a scalar_mode rather than a machine_mode.
9981 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9982 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9983 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9984 (aarch64_gen_adjusted_ldpstp): Likewise.
9985 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9987 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9988 Alan Hayward <alan.hayward@arm.com>
9989 David Sherwood <david.sherwood@arm.com>
9991 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9992 Take a scalar_int_mode instead of a machine_mode.
9993 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9994 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9995 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9996 (aarch64_simd_attr_length_rglist): Delete.
9997 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9998 a scalar_int_mode instead of a machine_mode.
9999 (aarch64_add_offset): Likewise.
10000 (aarch64_internal_mov_immediate): Likewise
10001 (aarch64_add_constant_internal): Likewise.
10002 (aarch64_add_constant): Likewise.
10003 (aarch64_movw_imm): Likewise.
10004 (aarch64_rtx_arith_op_extract_p): Likewise.
10005 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10006 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10007 Remove assert that the mode isn't a vector.
10008 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10009 (aarch64_expand_mov_immediate): Update calls after above changes.
10010 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10011 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10012 (aarch64_move_imm): Likewise.
10013 (aarch64_can_const_movi_rtx_p): Likewise.
10014 (aarch64_strip_extend): Likewise.
10015 (aarch64_extr_rtx_p): Likewise.
10016 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10017 a CONST_INT when the mode parameter is VOIDmode.
10018 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10020 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10022 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10023 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10024 (bitwise_type_for_mode): Update accordingly.
10026 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10028 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10029 * stor-layout.c (mode_for_size_tree): Likewise.
10030 (mode_for_array): Update accordingly.
10031 (layout_decl): Likewise.
10032 (compute_record_mode): Likewise. Only set the mode once.
10034 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10036 * target.def (get_mask_mode): Change return type to opt_mode.
10038 * doc/tm.texi: Regenerate.
10039 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10040 * targhooks.c (default_get_mask_mode): Likewise.
10041 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10042 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10043 targetm.get_mask_mode.
10044 * tree.c (build_truth_vector_type): Likewise.
10046 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10048 * machmode.h (mode_for_vector): Return an opt_mode.
10049 * stor-layout.c (mode_for_vector): Likewise.
10050 (mode_for_int_vector): Update accordingly.
10051 (layout_type): Likewise.
10052 * config/i386/i386.c (emit_memmov): Likewise.
10053 (ix86_expand_set_or_movmem): Likewise.
10054 (ix86_expand_vector_init): Likewise.
10055 (ix86_get_mask_mode): Likewise.
10056 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10058 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10059 * expmed.c (extract_bit_field_1): Likewise.
10060 * expr.c (expand_expr_real_2): Likewise.
10061 * optabs-query.c (can_vec_perm_p): Likewise.
10062 (can_vec_mask_load_store_p): Likewise.
10063 * optabs.c (expand_vec_perm): Likewise.
10064 * targhooks.c (default_get_mask_mode): Likewise.
10065 * tree-vect-stmts.c (vectorizable_store): Likewise.
10066 (vectorizable_load): Likewise.
10067 (get_vectype_for_scalar_type_and_size): Likewise.
10069 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10071 * machmode.h (mode_for_int_vector): New function.
10072 * stor-layout.c (mode_for_int_vector): Likewise.
10073 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10074 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10075 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10076 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10077 (s390_expand_vcond): Likewise.
10079 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10081 * machmode.h (opt_machine_mode): New type.
10082 (opt_mode<T>): Allow construction from anything that can be
10084 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10085 (mode_for_size): Return an opt_machine_mode.
10086 * stor-layout.c (mode_for_size): Likewise.
10087 (mode_for_size_tree): Update call accordingly.
10088 (bitwise_mode_for_mode): Likewise.
10089 (make_fract_type): Likewise.
10090 (make_accum_type): Likewise.
10091 * caller-save.c (replace_reg_with_saved_mem): Update call
10093 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10094 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10095 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10096 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10097 * expmed.c (extract_bit_field_1): Likewise.
10098 * reload.c (get_secondary_mem): Likewise.
10099 * varasm.c (assemble_integer): Likewise.
10100 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10103 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10105 * machmode.h (decimal_float_mode_for_size): New function.
10106 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10107 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10108 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10109 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10110 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10111 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10113 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10115 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10116 (get_builtin_sync_mode): Likewise.
10117 (expand_ifn_atomic_compare_exchange): Likewise.
10118 (expand_builtin_atomic_clear): Likewise.
10119 (expand_builtin_atomic_test_and_set): Likewise.
10120 (fold_builtin_atomic_always_lock_free): Likewise.
10121 * calls.c (compute_argument_addresses): Likewise.
10122 (emit_library_call_value_1): Likewise.
10123 (store_one_arg): Likewise.
10124 * combine.c (combine_instructions): Likewise.
10125 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10126 * config/arm/arm.c (arm_function_value): Likewise.
10127 (aapcs_allocate_return_reg): Likewise.
10128 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10129 * config/i386/i386.c (construct_container): Likewise.
10130 (ix86_gimplify_va_arg): Likewise.
10131 (ix86_expand_sse_cmp): Likewise.
10132 (emit_memmov): Likewise.
10133 (emit_memset): Likewise.
10134 (expand_small_movmem_or_setmem): Likewise.
10135 (ix86_expand_pextr): Likewise.
10136 (ix86_expand_pinsr): Likewise.
10137 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10138 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10140 * config/mips/mips.c (mips_function_value_1) Likewise.
10141 (mips_block_move_straight): Likewise.
10142 (mips_expand_ins_as_unaligned_store): Likewise.
10143 * config/powerpcspe/powerpcspe.c
10144 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10145 (rs6000_darwin64_record_arg_flush): Likewise.
10146 * config/rs6000/rs6000.c
10147 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10148 (rs6000_darwin64_record_arg_flush): Likewise.
10149 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10150 (sparc_function_value_1): Likewise.
10151 * config/spu/spu.c (adjust_operand): Likewise.
10152 (spu_emit_branch_or_set): Likewise.
10153 (arith_immediate_p): Likewise.
10154 * emit-rtl.c (gen_lowpart_common): Likewise.
10155 * expr.c (expand_expr_real_1): Likewise.
10156 * function.c (assign_parm_setup_block): Likewise.
10157 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10158 * reload1.c (alter_reg): Likewise.
10159 * stor-layout.c (mode_for_vector): Likewise.
10160 (layout_type): Likewise.
10162 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10164 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10165 (spu_convert_move): Likewise.
10166 * lower-subreg.c (resolve_simple_move): Likewise.
10168 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10171 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10172 define_insn to a define_expand.
10173 (altivec_vsum2sws_direct): New define_insn.
10174 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10176 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10178 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10181 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10186 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10187 realign stack if stack alignment needed is less than incoming
10190 2017-09-05 Marek Polacek <polacek@redhat.com>
10193 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10196 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10198 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10200 2017-09-05 Richard Biener <rguenther@suse.de>
10202 PR tree-optimization/82084
10203 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10205 2017-09-05 Richard Biener <rguenther@suse.de>
10207 PR tree-optimization/82102
10208 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10210 2017-09-05 Martin Liska <mliska@suse.cz>
10212 PR tree-optimization/82032
10213 * tree-cfg.c (generate_range_test): New function.
10214 * tree-cfg.h (generate_range_test): Declared here.
10215 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10216 (cleanup_control_expr_graph): Use it.
10217 * tree-switch-conversion.c (try_switch_expansion): Remove
10219 (emit_case_nodes): Use generate_range_test.
10221 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10224 * config/i386/i386.md (*<btsc><mode>_mask): Add
10225 TARGET_USE_BT to insn constraint.
10226 (*btr<mode>_mask): Ditto.
10228 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10230 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10231 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10233 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10236 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10237 TARGET_NEON and TARGET_IWMMXT.
10238 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10239 TARGET_NEON and TARGET_IWMMXT.
10240 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10242 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10244 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10245 (ix86_rewrite_tls_address): Ditto.
10246 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10247 (ix86_rewrite_tls_address_1): Ditto.
10248 (ix86_rewrite_tls_address): Ditto.
10249 * config/i386/predicates.md (tls_address_pattern): New predicate.
10250 * config/i386/i386.md (TLS address splitter): New splitter.
10252 2017-09-04 Richard Biener <rguenther@suse.de>
10254 PR tree-optimization/82084
10255 * fold-const.h (can_native_encode_string_p): Declare.
10256 * fold-const.c (can_native_encode_string_p): Factor out from ...
10257 (native_encode_string): ... here.
10258 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10259 vectorizing stores from constants we later cannot handle.
10261 2017-09-04 Marek Polacek <polacek@redhat.com>
10264 * doc/invoke.texi: Update -Wtautological-compare documentation.
10266 2017-09-04 Jeff Law <law@redhat.com>
10268 PR tree-optimization/64910
10269 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10270 swap the first and last operand if the last is a constant.
10272 2017-09-04 Marek Polacek <polacek@redhat.com>
10275 * convert.c (do_narrow): When sanitizing signed integer overflows,
10276 bail out for signed types.
10277 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10279 2017-09-04 Richard Biener <rguenther@suse.de>
10281 PR tree-optimization/82060
10282 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10283 Move devirtualization after stmt folding and before EH/AB/noreturn
10284 cleanup to get the stmt refs canonicalized. Use a bool instead
10285 of gimple_modified_p since that doesn't work for NOPs. Schedule
10286 NOPs generated by folding for removal.
10288 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10289 Alan Hayward <alan.hayward@arm.com>
10290 David Sherwood <david.sherwood@arm.com>
10292 * coretypes.h (pad_direction): New enum.
10293 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10294 (FUNCTION_ARG_PADDING): Likewise.
10295 * target.def (function_arg_padding): New hook.
10296 * targhooks.h (default_function_arg_padding): Declare.
10297 * targhooks.c (default_function_arg_padding): New function.
10298 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10299 (TARGET_FUNCTION_ARG_PADDING): ...this.
10300 * doc/tm.texi: Regenerate.
10301 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10302 instead of direction.
10303 (compute_argument_addresses): Likewise.
10304 (load_register_parameters): Likewise.
10305 (emit_library_call_value_1): Likewise.
10306 (store_one_arg): Use targetm.calls.function_arg_padding instead
10307 of FUNCTION_ARG_PADDING.
10308 (must_pass_in_stack_var_size_or_pad): Likewise.
10309 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10310 (emit_group_store): Likewise.
10311 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10312 instead of FUNCTION_ARG_PADDING.
10313 (emit_push_insn): Likewise, and propagate enum change throughout
10315 * function.h (direction): Delete.
10316 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10318 * function.c (assign_parm_find_stack_rtl): Likewise.
10319 (assign_parm_setup_block_p): Likewise.
10320 (assign_parm_setup_block): Likewise.
10321 (gimplify_parameters): Likewise.
10322 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10323 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10325 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10326 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10327 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10328 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10329 (aarch64_function_arg_padding): ...this new function.
10330 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10331 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10332 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10333 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10334 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10335 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10336 (arm_pad_arg_upward): Replace with...
10337 (arm_function_arg_padding): ...this new function.
10338 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10340 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10341 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10342 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10343 (ia64_hpux_function_arg_padding): Replace with...
10344 (ia64_function_arg_padding): ...this new function. Use pad_direction
10345 instead of direction. Check for TARGET_HPUX.
10346 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10347 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10348 (iq2000_function_arg_padding): New function.
10349 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10350 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10351 (mips_function_arg_padding): ...this new function.
10352 (mips_pad_reg_upward): Update accordingly.
10353 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10354 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10355 targetm.calls.function_arg_padding.
10356 (FUNCTION_ARG_PADDING): Delete.
10357 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10358 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10359 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10360 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10361 (nios2_block_reg_padding): Return pad_direction instead of direction.
10362 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10363 instead of direction.
10364 (nios2_function_arg_padding): Likewise. Make static.
10365 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10366 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10367 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10368 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10369 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10370 (pa_function_arg_padding): Make static. Return pad_direction instead
10372 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10373 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10374 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10375 instead of direction. Use targetm.calls.function_arg_padding.
10376 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10377 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10378 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10379 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10380 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10382 (function_arg_padding): Rename to...
10383 (rs6000_function_arg_padding): ...this. Make static. Return
10384 pad_direction instead of direction.
10385 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10386 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10387 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10388 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10389 instead of direction. Use targetm.calls.function_arg_padding.
10390 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10391 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10392 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10393 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10394 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10395 (function_arg_padding): Rename to...
10396 (rs6000_function_arg_padding): ...this. Make static. Return
10397 pad_direction instead of direction.
10398 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10399 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10400 * config/s390/s390.c (s390_function_arg_padding): New function.
10401 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10402 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10403 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10404 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10405 (function_arg_padding): Rename to...
10406 (sparc_function_arg_padding): ...this. Make static. Return
10407 pad_direction instead of direction.
10408 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10409 * config/spu/spu.c (spu_function_arg_padding): New function.
10410 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10411 * system.h (FUNCTION_ARG_PADDING): Poison.
10413 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10414 Alan Hayward <alan.hayward@arm.com>
10415 David Sherwood <david.sherwood@arm.com>
10417 * target.def (modes_tieable_p): New hook.
10418 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10419 (TARGET_MODES_TIEABLE_P): ...this.
10420 * doc/tm.texi.in: Regenerate.
10421 * hooks.h (hook_bool_mode_mode_true): Declare.
10422 * hooks.c (hook_bool_mode_mode_true): New function.
10423 * combine.c (subst): Use targetm.modes_tieable_p instead of
10425 * dse.c (find_shift_sequence): Likewise.
10426 * expmed.c (extract_low_bits): Likewise.
10427 * lower-subreg.c: Include target.h.
10428 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10430 * rtlanal.c (rtx_cost): Likewise.
10431 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10432 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10433 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10434 (TARGET_MODES_TIEABLE_P): Redefine.
10435 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10436 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10437 (TARGET_MODES_TIEABLE_P): Redefine.
10438 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10439 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10440 (arc_modes_tieable_p): New function.
10441 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10442 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10443 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10444 (arm_modes_tieable_p): Make static.
10445 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10446 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10447 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10448 (TARGET_MODES_TIEABLE_P): Redefine.
10449 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10450 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10451 (TARGET_MODES_TIEABLE_P): Redefine.
10452 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10453 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10454 (cr16_modes_tieable_p): New function.
10455 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10456 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10457 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10458 (TRULY_NOOP_TRUNCATION): Update comment.
10459 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10460 (TRULY_NOOP_TRUNCATION): Update comment.
10461 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10462 (frv_modes_tieable_p): New function.
10463 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10464 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10465 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10466 (TARGET_MODES_TIEABLE_P): Redefine.
10467 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10468 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10469 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10470 (TARGET_MODES_TIEABLE_P): Redefine.
10471 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10472 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10473 (ia64_modes_tieable_p): New function.
10474 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10475 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10476 (iq2000_modes_tieable_p): New function.
10477 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10478 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10479 (lm32_modes_tieable_p): New function.
10480 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10481 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10482 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10483 (TARGET_MODES_TIEABLE_P): Redefine.
10484 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10485 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10486 (m32r_modes_tieable_p): New function.
10487 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10488 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10489 (m68k_modes_tieable_p): New function.
10490 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10491 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10492 (mcore_modes_tieable_p): New function.
10493 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10494 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10496 (TARGET_MODES_TIEABLE_P): Redefine.
10497 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10498 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10499 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10500 (TARGET_MODES_TIEABLE_P): Redefine.
10501 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10502 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10503 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10504 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10505 (mn10300_modes_tieable_p): ...this and make static.
10506 (TARGET_MODES_TIEABLE_P): Redefine.
10507 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10508 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10509 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10510 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10511 (msp430_modes_tieable_p): Make static.
10512 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10513 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10514 (TARGET_MODES_TIEABLE_P): Redefine.
10515 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10516 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10517 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10518 (TARGET_MODES_TIEABLE_P): Redefine.
10519 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10520 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10521 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10522 (TARGET_MODES_TIEABLE_P): Redefine.
10523 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10524 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10525 (pdp11_modes_tieable_p): New function.
10526 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10527 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10528 (rs6000_modes_tieable_p): New function.
10529 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10530 * config/powerpcspe/powerpcspe.md: Update comment.
10531 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10532 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10533 (TARGET_MODES_TIEABLE_P): Redefine.
10534 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10535 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10536 (rl78_modes_tieable_p): New function.
10537 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10538 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10539 (rs6000_modes_tieable_p): New function.
10540 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10541 * config/rs6000/rs6000.md: Update comment.
10542 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10543 * config/rx/rx.c (rx_modes_tieable_p): New function.
10544 (TARGET_MODES_TIEABLE_P): Redefine.
10545 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10546 * config/s390/s390.c (s390_modes_tieable_p): New function.
10547 (TARGET_MODES_TIEABLE_P): Redefine.
10548 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10549 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10550 (sh_modes_tieable_p): New function.
10551 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10552 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10553 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10554 (sparc_modes_tieable_p): Make static.
10555 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10556 * config/spu/spu.c (spu_modes_tieable_p): New function.
10557 (TARGET_MODES_TIEABLE_P): Redefine.
10558 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10559 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10560 (TARGET_MODES_TIEABLE_P): Redefine.
10561 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10562 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10563 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10564 * config/v850/v850.c (v850_modes_tieable_p): New function.
10565 (TARGET_MODES_TIEABLE_P): Redefine.
10566 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10567 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10568 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10569 (visium_modes_tieable_p): New function.
10570 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10571 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10572 (xtensa_modes_tieable_p): New function.
10573 * system.h (MODES_TIEABLE_P): Poison.
10575 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10576 Alan Hayward <alan.hayward@arm.com>
10577 David Sherwood <david.sherwood@arm.com>
10579 * target.def (hard_regno_mode_ok): New hook.
10580 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10581 (TARGET_HARD_REGNO_MODE_OK): ...this.
10582 * doc/tm.texi.in: Regenerate.
10583 * hooks.h (hook_bool_uint_mode_true): Declare.
10584 * hooks.c (hook_bool_uint_mode_true): New function.
10585 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10586 HARD_REGNO_MODE_OK.
10587 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10588 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10589 instead of HARD_REGNO_MODE_OK.
10590 * caller-save.c: Include target.h.
10591 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10592 HARD_REGNO_MODE_OK.
10593 * combine.c (can_combine_p): Likewise.
10594 (combinable_i3pat): Likewise.
10595 (can_change_dest_mode): Likewise.
10596 * expr.c (init_expr_target): Likewise.
10597 (convert_move): Likewise.
10598 (convert_modes): Likewise.
10599 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10600 (setup_prohibited_mode_move_regs): Likewise.
10601 * ira.h (target_ira): Likewise.
10602 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10603 * lra-constraints.c (process_alt_operands): Likewise.
10604 (split_reg): Likewise.
10605 * recog.c (peep2_find_free_register): Likewise.
10606 * ree.c (combine_reaching_defs): Likewise.
10607 * regcprop.c (maybe_mode_change): Likewise.
10608 * reginfo.c (init_reg_sets_1): Likewise.
10609 (choose_hard_reg_mode): Likewise.
10610 (simplifiable_subregs): Likewise.
10611 * regrename.c (check_new_reg_p): Likewise.
10612 * reload.c (find_valid_class): Likewise.
10613 (find_valid_class_1): Likewise.
10614 (reload_inner_reg_of_subreg): Likewise.
10615 (push_reload): Likewise.
10616 (combine_reloads): Likewise.
10617 (find_dummy_reload): Likewise.
10618 (find_reloads): Likewise.
10619 * reload1.c (find_reg): Likewise.
10620 (set_reload_reg): Likewise.
10621 (allocate_reload_reg): Likewise.
10622 (choose_reload_regs): Likewise.
10623 (reload_adjust_reg_for_temp): Likewise.
10624 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10625 (simplify_subreg_regno): Likewise.
10626 * sel-sched.c (init_regs_for_mode): Likewise.
10627 * varasm.c (make_decl_rtl): Likewise.
10628 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10629 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10630 HARD_REGNO_MODE_OK.
10631 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10632 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10633 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10634 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10635 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10636 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10637 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10638 (arc_mode_class): Delete.
10639 (HARD_REGNO_MODE_OK): Delete.
10640 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10641 (arc_hard_regno_mode_ok): Rename old array to...
10642 (arc_hard_regno_mode_ok_modes): ...this.
10643 (arc_conditional_register_usage): Update accordingly.
10644 (arc_mode_class): Make static.
10645 (arc_hard_regno_mode_ok): New function.
10646 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10647 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10648 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10649 (arm_hard_regno_mode_ok): Make static.
10650 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10651 HARD_REGNO_MODE_OK.
10652 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10653 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10654 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10656 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10657 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10658 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10659 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10660 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10661 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10662 * config/bfin/predicates.md (valid_reg_operand): Use
10663 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10664 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10665 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10666 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10667 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10668 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10669 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10670 (cr16_hard_regno_mode_ok): Make static and return a bool.
10671 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10672 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10673 (cris_hard_regno_mode_ok): New function.
10674 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10675 (epiphany_mode_class): Delete.
10676 (HARD_REGNO_MODE_OK): Delete.
10677 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10678 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10679 (hard_regno_mode_ok): Rename to...
10680 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10681 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10682 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10683 HARD_REGNO_MODE_OK.
10684 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10685 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10686 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10687 (frv_hard_regno_mode_ok): Make static and return a bool.
10688 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10689 HARD_REGNO_MODE_OK.
10690 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10691 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10692 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10693 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10695 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10696 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10697 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10698 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10700 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10701 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10702 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10703 (ia64_hard_regno_mode_ok): New function.
10704 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10705 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10706 (iq2000_hard_regno_mode_ok): New function.
10707 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10708 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10709 (lm32_hard_regno_mode_ok): New function.
10710 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10711 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10712 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10713 instead of HARD_REGNO_MODE_OK.
10714 (m32c_hard_regno_ok): Rename to...
10715 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10716 (m32c_cannot_change_mode_class): Update accordingly.
10717 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10718 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10719 (m32r_mode_class): Delete.
10720 (HARD_REGNO_MODE_OK): Delete.
10721 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10722 (m32r_hard_regno_mode_ok): Rename to...
10723 (m32r_hard_regno_modes): ...this.
10724 (m32r_mode_class): Make static.
10725 (m32r_hard_regno_mode_ok): New function.
10726 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10727 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10728 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10729 (m68k_hard_regno_mode_ok): Make static.
10730 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10731 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10732 (mcore_hard_regno_mode_ok): New function.
10733 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10734 (HARD_REGNO_MODE_OK): Delete.
10735 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10737 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10738 (microblaze_hard_regno_mode_ok): New function.
10739 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10740 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10741 (mips_hard_regno_mode_ok): Delete.
10742 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10743 (mips_hard_regno_mode_ok_p): ...this and make static.
10744 (mips_hard_regno_mode_ok_p): Rename to...
10745 (mips_hard_regno_mode_ok_uncached): ...this.
10746 (mips_hard_regno_mode_ok): New function.
10747 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10748 of HARD_REGNO_MODE_OK.
10749 (mips_option_override): Update after above name changes.
10750 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10751 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10752 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10753 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10754 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10755 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10756 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10757 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10758 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10759 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10760 (msp430_hard_regno_mode_ok): Make static and return a bool.
10761 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10762 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10763 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10765 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10766 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10767 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10768 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10769 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10770 (PA_HARD_REGNO_MODE_OK): ...this
10771 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10772 (PA_HARD_REGNO_MODE_OK): ...this.
10773 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10774 (pa_hard_regno_mode_ok): New function.
10775 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10776 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10777 (pdp11_hard_regno_mode_ok): New function.
10778 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10779 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10781 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10783 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10784 (rs6000_hard_regno_mode_ok): Rename to...
10785 (rs6000_hard_regno_mode_ok_uncached): ...this.
10786 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10787 (rs6000_hard_regno_mode_ok): New function.
10788 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10789 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10790 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10791 (riscv_hard_regno_mode_ok): ...this and make static.
10792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10793 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10794 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10795 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10796 (rl78_hard_regno_mode_ok): Make static and return bool.
10797 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10798 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10800 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10801 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10802 (rs6000_hard_regno_mode_ok): Rename to...
10803 (rs6000_hard_regno_mode_ok_uncached): ...this.
10804 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10805 (rs6000_hard_regno_mode_ok): New function.
10806 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10807 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10808 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10809 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10810 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10811 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10812 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10813 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10814 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10815 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10816 (sh_hard_regno_mode_ok): Make static.
10817 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10818 instead of HARD_REGNO_MODE_OK.
10819 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10820 (sparc_mode_class): Delete.
10821 (HARD_REGNO_MODE_OK): Delete.
10822 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10823 (hard_regno_mode_classes): Make static.
10824 (sparc_mode_class): Likewise.
10825 (sparc_hard_regno_mode_ok): New function.
10826 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10827 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10828 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10830 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10831 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10832 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10833 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10834 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10835 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10836 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10837 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10838 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10839 (visium_hard_regno_mode_ok): New function.
10840 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10841 instead of HARD_REGNO_MODE_OK.
10842 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10843 (HARD_REGNO_MODE_OK): Delete.
10844 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10845 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10846 (xtensa_option_override): Update accordingly.
10847 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10848 (xtensa_hard_regno_mode_ok): New function.
10849 * system.h (HARD_REGNO_MODE_OK): Poison.
10851 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10852 Alan Hayward <alan.hayward@arm.com>
10853 David Sherwood <david.sherwood@arm.com>
10855 * target.def (hard_regno_call_part_clobbered): New hook.
10856 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10857 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10858 * doc/tm.texi: Regenerate.
10859 * hooks.h (hook_bool_uint_mode_false): Declare.
10860 * hooks.c (hook_bool_uint_mode_false): New function.
10861 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10862 * cselib.c (cselib_process_insn): Use
10863 targetm.hard_regno_call_part_clobbered instead of
10864 HARD_REGNO_CALL_PART_CLOBBERED.
10865 * ira-conflicts.c (ira_build_conflicts): Likewise.
10866 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10867 * lra-constraints.c (need_for_call_save_p): Likewise.
10868 * lra-lives.c: Include target.h.
10869 (check_pseudos_live_through_calls): Use
10870 targetm.hard_regno_call_part_clobbered instead of
10871 HARD_REGNO_CALL_PART_CLOBBERED.
10872 * regcprop.c: Include target.h.
10873 (copyprop_hardreg_forward_1): Use
10874 targetm.hard_regno_call_part_clobbered instead of
10875 HARD_REGNO_CALL_PART_CLOBBERED.
10876 * reginfo.c (choose_hard_reg_mode): Likewise.
10877 * regrename.c (check_new_reg_p): Likewise.
10878 * reload.c (find_equiv_reg): Likewise.
10879 * reload1.c (emit_reload_insns): Likewise.
10880 * sched-deps.c (deps_analyze_insn): Likewise.
10881 * sel-sched.c (init_regs_for_mode): Likewise.
10882 (mark_unavailable_hard_regs): Likewise.
10883 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10884 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10885 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10887 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10888 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10889 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10891 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10893 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10894 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10895 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10897 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10898 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10899 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10901 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10902 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10904 * config/powerpcspe/powerpcspe.c
10905 (rs6000_hard_regno_call_part_clobbered): New function.
10906 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10907 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10908 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10910 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10911 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10912 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10914 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10915 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10916 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10918 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10919 Alan Hayward <alan.hayward@arm.com>
10920 David Sherwood <david.sherwood@arm.com>
10922 * rtl.h (subreg_memory_offset): Declare.
10923 * emit-rtl.c (subreg_memory_offset): New function.
10924 * expmed.c (store_bit_field_1): Use it.
10925 * expr.c (undefined_operand_subword_p): Likewise.
10926 * simplify-rtx.c (simplify_subreg): Likewise.
10928 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10930 PR rtl-optimization/57448
10933 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10934 using atomic_load pattern.
10935 (expand_atomic_store): Likewise.
10937 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10940 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10941 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10942 handling. Use replace_call_with_value with NULL instead of
10943 gsi_replace, unlink_stmt_vdef and release_defs.
10945 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10948 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10950 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10953 * rtl.h (emit_library_call_value_1): Declare.
10954 (emit_library_call): Replace declaration with a series of overloads.
10955 Remove the parameter count argument.
10956 (emit_library_call_value): Likewise.
10957 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10958 with an "rtx_mode_t *".
10959 (emit_library_call_value): Delete.
10960 (emit_library_call): Likewise.
10961 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10962 (asan_emit_allocas_unpoison): Likewise.
10963 * builtins.c (expand_builtin_powi): Likewise.
10964 (expand_asan_emit_allocas_unpoison): Likewise.
10965 * cfgexpand.c (expand_main_function): Likewise.
10966 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10967 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10968 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10969 * config/arm/arm.c (arm_trampoline_init): Likewise.
10970 (arm_call_tls_get_addr): Likewise.
10971 (arm_expand_divmod_libfunc): Likewise.
10972 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10973 (smulsi3_highpart): Likewise.
10974 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10975 (c6x_expand_compare): Likewise.
10976 (c6x_expand_movmem): Likewise.
10977 * config/frv/frv.c (frv_trampoline_init): Likewise.
10978 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10979 (ix86_expand_divmod_libfunc): Likewise.
10980 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10981 (ia64_expand_compare): Likewise.
10982 (ia64_profile_hook): Likewise.
10983 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10984 (nonlocal_goto): Likewise.
10985 (restore_stack_nonlocal): Likewise.
10986 * config/m32r/m32r.c (block_move_call): Likewise.
10987 (m32r_trampoline_init): Likewise.
10988 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10989 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10990 (m68k_call_m68k_read_tp): Likewise.
10991 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10992 (microblaze_expand_divide): Likewise.
10993 * config/mips/mips.h (mips_args): Likewise.
10994 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10995 (MIPS_ICACHE_SYNC): Likewise.
10996 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10997 (nios2_trampoline_init): Likewise.
10998 * config/pa/pa.c (hppa_tls_call): Likewise.
10999 (pa_trampoline_init): Likewise.
11000 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11001 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11002 (expand_strn_compare): Likewise.
11003 (rs6000_generate_compare): Likewise.
11004 (rs6000_expand_float128_convert): Likewise.
11005 (output_profile_hook): Likewise.
11006 (rs6000_trampoline_init): Likewise.
11007 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11008 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11009 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11010 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11011 (rs6000_generate_compare): Likewise.
11012 (rs6000_expand_float128_convert): Likewise.
11013 (output_profile_hook): Likewise.
11014 (rs6000_trampoline_init): Likewise.
11015 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11016 * config/sh/sh.c (sh_trampoline_init): Likewise.
11017 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11018 (sparc_emit_float_lib_cmp): Likewise.
11019 (sparc32_initialize_trampoline): Likewise.
11020 (sparc64_initialize_trampoline): Likewise.
11021 (sparc_profile_hook): Likewise.
11022 * config/spu/spu.c (ea_load_store): Likewise.
11023 * config/spu/spu.md (floatunssidf2): Likewise.
11024 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11025 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11026 * config/visium/visium.c (expand_block_move_4): Likewise.
11027 (expand_block_move_2): Likewise.
11028 (expand_block_move_1): Likewise.
11029 (expand_block_set_4): Likewise.
11030 (expand_block_set_2): Likewise.
11031 (expand_block_set_1): Likewise.
11032 (visium_trampoline_init): Likewise.
11033 (visium_profile_hook): Likewise.
11034 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11035 (xtensa_setup_frame_addresses): Likewise.
11036 (xtensa_trampoline_init): Likewise.
11037 * except.c (sjlj_emit_function_enter): Likewise.
11038 (sjlj_emit_function_exit): Likewise.
11039 * explow.c (allocate_dynamic_stack_space): Likewise.
11040 (probe_stack_range): Likewise.
11041 * expr.c (convert_mode_scalar): Likewise.
11042 * optabs.c (expand_binop): Likewise.
11043 (expand_twoval_binop_libfunc): Likewise.
11044 (expand_unop): Likewise.
11045 (prepare_cmp_insn): Likewise.
11046 (prepare_float_lib_cmp): Likewise.
11047 (expand_float): Likewise.
11048 (expand_fix): Likewise.
11049 (expand_fixed_convert): Likewise.
11050 (maybe_emit_sync_lock_test_and_set): Likewise.
11051 (expand_atomic_compare_and_swap): Likewise.
11052 (expand_mem_thread_fence): Likewise.
11053 (expand_atomic_fetch_op): Likewise.
11055 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11057 * doc/generic.texi (OpenACC): Adjust URL.
11058 * doc/invoke.texi (C Dialect Options): Ditto.
11060 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11062 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11063 predicate for operand 1. Add (m,<S>) constraint.
11064 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11065 Prevent memory operand 1 with register operand 2.
11067 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11069 PR rtl-optimization/82024
11070 * combine.c (try_combine): If the combination result is a PARALLEL,
11071 and we only need to retain the SET in there that would be placed
11072 at I2, check that we can place that at I3 instead, before doing so.
11074 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11077 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11079 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11080 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11083 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11084 Jeff Law <law@redhat.com>
11086 * varasm.c (bss_initializer_p): Do not put constants into .bss
11087 (categorize_decl_for_section): Handle bss_initializer_p returning
11088 false when DECL_INITIAL is NULL.
11090 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11093 * config/s390/s390.c (s390_can_inline_p): New function.
11095 2017-09-01 Jeff Law <law@redhat.com>
11097 PR tree-optimization/82052
11098 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11099 Always initialize the returned slot after a hash table miss
11100 when INSERT is true.
11102 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11104 * config/s390/s390.md (mem_signal_fence): Remove.
11105 * doc/md.texi (mem_signal_fence): Remove.
11106 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11108 * target-insns.def (mem_signal_fence): Remove.
11110 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11113 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11116 * asan.c (create_odr_indicator): Strip name encoding from assembler
11117 name before appending it after __odr_asan_.
11119 2017-09-01 Martin Liska <mliska@suse.cz>
11121 PR tree-optimization/82059
11122 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11123 frequency only when an edge is redirected.
11125 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11127 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11128 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11129 (arc_conditional_register_usage): Remove ARC600 lp_count
11131 (arc_file_start): Emit Tag_ARC_CPU_variation.
11132 (arc_can_use_doloop_p): New conditions to use ZOLs.
11133 (hwloop_fail): New function.
11134 (hwloop_optimize): Likewise.
11135 (hwloop_pattern_reg): Likewise.
11136 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11137 (arc_reorg_loops): New function, calls reorg_loops.
11138 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11139 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11141 (arc_loop_hazard): Remove function, functionality moved into
11143 (arc_hazard): Remove arc_loop_hazard call.
11144 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11145 into hwloop_optimize.
11146 (arc_label_align): Remove ZOL handling.
11147 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11148 * config/arc/arc.md (doloop_begin): Remove pattern.
11149 (doloop_begin_i): Likewise.
11150 (doloop_end_i): Likewise.
11151 (doloop_fallback): Likewise.
11152 (doloop_fallback_m): Likewise.
11153 (doloop_end): Reimplement expand.
11154 (arc_lp): New pattern for LP instruction.
11155 (loop_end): New pattern.
11156 (loop_fail): Likewise.
11157 (decrement_and_branch_until_zero): Likewise.
11158 * config/arc/arc.opt (mlpc-width): New option.
11159 * doc/invoke.texi (mlpc-width): Document option.
11161 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11163 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11164 (arc_ccfsm_advance): Fix checking for delay slots.
11165 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11167 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11169 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11171 (movhi_insn): Update store instruction constraint which are saving
11172 6-bit short immediates.
11173 (movsi_insn): Consider also short scaled load operations.
11174 (zero_extendhisi2_i): Use Usd constraint instead of T.
11175 (extendhisi2_i): Add q constraint.
11176 (arc_clzsi2): Add type and length attributes.
11177 (arc_ctzsi2): Likewise.
11178 * config/arc/constraints.md (Usc): Update constraint, the
11179 assembler can parse two relocations for a single instruction.
11181 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11183 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11184 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11186 2017-08-31 Olivier Hainque <hainque@adacore.com>
11188 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11189 match as powerpc-wrs-vxworks*.
11191 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11193 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11194 register constraint for by-element operand.
11195 (aarch64_mls_elt_merge<mode>): Likewise.
11197 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11199 * config/arc/arc.c (arc_can_follow_jump): Check for short
11202 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11204 * config.gcc: Use g.opt for arc.
11205 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11206 functionality moved to ...
11207 (legitimate_scaled_address_p): New function, ...here.
11208 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11209 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11210 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11212 (arc_override_options): Handle G option.
11213 (arc_output_pic_addr_const): Correct function definition.
11214 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11215 (arc_decl_anon_ns_mem_p): Delete.
11216 (arc_in_small_data_p): Overhaul this function to take into
11217 consideration the value given via G option.
11218 (arc_rewrite_small_data_1): Renamed and corrected old
11219 arc_rewrite_small_data function.
11220 (arc_rewrite_small_data): New function.
11221 (small_data_pattern): Don't use pic_offset_table_rtx.
11222 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11223 * config/arc/simdext.md (movmisalignv2hi): Use
11224 prepare_move_operands function.
11226 (movmisalign*): Likewise.
11227 * doc/invoke.texi (ARC options): Document -G option.
11229 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11231 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11233 * config/arc/arc.c (arc_print_operand): Output scalled address for
11234 sdata whenever is possible.
11235 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11236 load/stores are available.
11237 (compact_sda_memory_operand): Check for the alignment required by
11238 code density instructions.
11239 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11241 * config/arc/constraints.md (Usd): Update constraint.
11242 (Us0): New constraint.
11243 (Usc): Update constraint.
11245 2017-08-31 Richard Biener <rguenther@suse.de>
11247 PR middle-end/82054
11248 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11249 function only once.
11251 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11253 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11254 Resize type_signature.
11256 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11257 Alan Hayward <alan.hayward@arm.com>
11258 David Sherwood <david.sherwood@arm.com>
11260 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11261 subregs whose inner modes can be stored in GPRs.
11262 (aarch64_classify_index): Likewise.
11264 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11265 Alan Hayward <alan.hayward@arm.com>
11266 David Sherwood <david.sherwood@arm.com>
11268 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11269 (V_INT_EQUIV): ...this.
11270 (v_cmp_result): Rename to...
11271 (v_int_equiv): ...this.
11272 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11273 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11274 (copysign<mode>3): Likewise.
11275 (aarch64_simd_bsl<mode>_internal): Likewise.
11276 (aarch64_simd_bsl<mode>): Likewise.
11277 (vec_cmp<mode><mode>): Likewise.
11278 (vcond<mode><mode>): Likewise.
11279 (vcond<v_cmp_mixed><mode>): Likewise.
11280 (vcondu<mode><v_cmp_mixed>): Likewise.
11281 (aarch64_cm<optab><mode>): Likewise.
11282 (aarch64_cmtst<mode>): Likewise.
11283 (aarch64_fac<optab><mode>): Likewise.
11284 (vec_perm_const<mode>): Likewise.
11285 (vcond_mask_<mode><v_cmp_result>): Rename to...
11286 (vcond_mask_<mode><v_int_equiv>): ...this.
11287 (vec_cmp<mode><v_cmp_result>): Rename to...
11288 (vec_cmp<mode><v_int_equiv>): ...this.
11290 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11291 Alan Hayward <alan.hayward@arm.com>
11292 David Sherwood <david.sherwood@arm.com>
11294 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11296 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11297 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11298 (UNSPEC_LD4_DREG): New unspecs.
11299 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11300 (aarch64_ld2<mode>_dreg_be): Replace with...
11301 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11303 (aarch64_ld3<mode>_dreg_le)
11304 (aarch64_ld3<mode>_dreg_be): Replace with...
11305 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11307 (aarch64_ld4<mode>_dreg_le)
11308 (aarch64_ld4<mode>_dreg_be): Replace with...
11309 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11312 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11314 PR tree-optimization/81987
11315 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11316 insert an initializer in a location not dominated by the stride
11319 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11321 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11322 on the entire header of the finally block in the fallthru case.
11324 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11326 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11328 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11330 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11331 rs6000_emit_move_from_cr and call renamed function.
11332 (rs6000_emit_prologue): Call renamed functions.
11333 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11334 movesi_from_cr, remove volatile CRs.
11336 2017-08-30 Jon Beniston <jon@beniston.com>
11337 Richard Biener <rguenther@suse.de>
11339 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11340 of VECTOR_MODE_P check.
11341 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11342 element vector types.
11344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11346 * df.h (df_read_modify_subreg_p): Remove in favor of...
11347 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11348 const_rtx instead of an rtx.
11349 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11350 * df-problems.c (df_word_lr_mark_ref): Likewise.
11351 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11352 (mark_pseudo_reg_dead): Likewise.
11353 (mark_ref_dead): Likewise.
11354 * reginfo.c (init_subregs_of_mode): Likewise.
11355 * sched-deps.c (sched_analyze_1): Likewise.
11356 * df-scan.c (df_def_record_1): Likewise.
11357 (df_uses_record): Likewise.
11358 (df_read_modify_subreg_p): Remove in favor of...
11359 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11360 const_rtx instead of an rtx.
11362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11363 Alan Hayward <alan.hayward@arm.com>
11364 David Sherwood <david.sherwood@arm.com>
11366 * rtl.h (partial_subreg_p): New function.
11367 * caller-save.c (save_call_clobbered_regs): Use it.
11368 * calls.c (expand_call): Likewise.
11369 * combine.c (combinable_i3pat): Likewise.
11370 (simplify_set): Likewise.
11371 (make_extraction): Likewise.
11372 (make_compound_operation_int): Likewise.
11373 (gen_lowpart_or_truncate): Likewise.
11374 (force_to_mode): Likewise.
11375 (make_field_assignment): Likewise.
11376 (reg_truncated_to_mode): Likewise.
11377 (record_truncated_value): Likewise.
11378 (move_deaths): Likewise.
11379 * cse.c (record_jump_cond): Likewise.
11380 (cse_insn): Likewise.
11381 * cselib.c (cselib_lookup_1): Likewise.
11382 * expmed.c (extract_bit_field_using_extv): Likewise.
11383 * function.c (assign_parm_setup_reg): Likewise.
11384 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11385 * ira-build.c (create_insn_allocnos): Likewise.
11386 * lra-coalesce.c (merge_pseudos): Likewise.
11387 * lra-constraints.c (match_reload): Likewise.
11388 (simplify_operand_subreg): Likewise.
11389 (curr_insn_transform): Likewise.
11390 * lra-lives.c (process_bb_lives): Likewise.
11391 * lra.c (new_insn_reg): Likewise.
11392 (lra_substitute_pseudo): Likewise.
11393 * regcprop.c (mode_change_ok): Likewise.
11394 (maybe_mode_change): Likewise.
11395 (copyprop_hardreg_forward_1): Likewise.
11396 * reload.c (push_reload): Likewise.
11397 (find_reloads): Likewise.
11398 (find_reloads_subreg_address): Likewise.
11399 * reload1.c (alter_reg): Likewise.
11400 (eliminate_regs_1): Likewise.
11401 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11403 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11405 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11406 back to if statements, including unpack.
11408 2017-08-30 Martin Liska <mliska@suse.cz>
11410 PR inline-asm/82001
11411 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11413 (func_checker::compare_asm_inputs_outputs): ... this function.
11414 (func_checker::compare_gimple_asm): Use the function to compare
11415 also ASM constrains.
11416 * ipa-icf-gimple.h: Rename the function.
11418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11419 Alan Hayward <alan.hayward@arm.com>
11420 David Sherwood <david.sherwood@arm.com>
11422 * coretypes.h (complex_mode): New type.
11423 * gdbhooks.py (build_pretty_printer): Handle it.
11424 * machmode.h (complex_mode): New class.
11425 (complex_mode::includes_p): New function.
11426 (is_complex_int_mode): Likewise.
11427 (is_complex_float_mode): Likewise.
11428 * genmodes.c (get_mode_class): Handle complex mode classes.
11429 * function.c (expand_function_end): Use is_complex_int_mode.
11431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11432 Alan Hayward <alan.hayward@arm.com>
11433 David Sherwood <david.sherwood@arm.com>
11435 * coretypes.h (scalar_mode_pod): New typedef.
11436 * gdbhooks.py (build_pretty_printer): Handle it.
11437 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11438 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11439 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11440 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11441 as_a <scalar_mode>.
11443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11444 Alan Hayward <alan.hayward@arm.com>
11445 David Sherwood <david.sherwood@arm.com>
11447 * machmode.h (mode_for_vector): Take a scalar_mode instead
11449 * stor-layout.c (mode_for_vector): Likewise.
11450 * explow.c (promote_mode): Use as_a <scalar_mode>.
11451 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11454 Alan Hayward <alan.hayward@arm.com>
11455 David Sherwood <david.sherwood@arm.com>
11457 * target.def (preferred_simd_mode): Take a scalar_mode
11458 instead of a machine_mode.
11459 * targhooks.h (default_preferred_simd_mode): Likewise.
11460 * targhooks.c (default_preferred_simd_mode): Likewise.
11461 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11462 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11463 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11464 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11465 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11466 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11467 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11468 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11470 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11471 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11472 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11473 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11474 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11475 * doc/tm.texi: Regenerate.
11476 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11480 Alan Hayward <alan.hayward@arm.com>
11481 David Sherwood <david.sherwood@arm.com>
11483 * target.def (scalar_mode_supported_p): Take a scalar_mode
11484 instead of a machine_mode.
11485 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11486 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11487 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11488 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11489 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11490 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11491 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11492 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11493 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11494 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11495 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11496 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11497 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11498 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11500 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11501 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11502 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11503 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11504 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11506 * doc/tm.texi: Regenerate.
11508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11509 Alan Hayward <alan.hayward@arm.com>
11510 David Sherwood <david.sherwood@arm.com>
11512 * coretypes.h (opt_scalar_mode): New typedef.
11513 * gdbhooks.py (build_pretty_printers): Handle it.
11514 * machmode.h (mode_iterator::get_2xwider): Add overload for
11516 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11518 * expr.c (convert_mode_scalar): Likewise.
11519 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11520 * optabs.c (expand_float): Likewise.
11521 (expand_fix): Likewise.
11522 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11525 Alan Hayward <alan.hayward@arm.com>
11526 David Sherwood <david.sherwood@arm.com>
11528 * optabs.c (expand_float): Explicitly check for scalars before
11529 using a branching expansion.
11530 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
11537 (convert_mode_scalar): ...this new function. Treat the modes
11540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11541 Alan Hayward <alan.hayward@arm.com>
11542 David Sherwood <david.sherwood@arm.com>
11544 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11546 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11549 Alan Hayward <alan.hayward@arm.com>
11550 David Sherwood <david.sherwood@arm.com>
11552 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11553 rather than a machine_mode.
11554 (fixed_from_string): Likewise.
11555 (fixed_convert): Likewise.
11556 (fixed_convert_from_int): Likewise.
11557 (fixed_convert_from_real): Likewise.
11558 (real_convert_from_fixed): Likewise.
11559 * fixed-value.c (fixed_from_double_int): Likewise.
11560 (fixed_from_string): Likewise.
11561 (fixed_convert): Likewise.
11562 (fixed_convert_from_int): Likewise.
11563 (fixed_convert_from_real): Likewise.
11564 (real_convert_from_fixed): Likewise.
11565 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11568 Alan Hayward <alan.hayward@arm.com>
11569 David Sherwood <david.sherwood@arm.com>
11571 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11572 of separate mode class checks. Do not allow vector modes here.
11573 (immed_wide_int_const): Use as_a <scalar_mode>.
11574 * explow.c (trunc_int_for_mode): Likewise.
11575 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11576 (wi::shwi): Likewise.
11577 (wi::min_value): Likewise.
11578 (wi::max_value): Likewise.
11579 * dwarf2out.c (loc_descriptor): Likewise.
11580 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11581 for CONST_WIDE_INT.
11583 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11584 Alan Hayward <alan.hayward@arm.com>
11585 David Sherwood <david.sherwood@arm.com>
11587 * tree.h (SCALAR_TYPE_MODE): New macro.
11588 * expr.c (expand_expr_addr_expr_1): Use it.
11589 (expand_expr_real_2): Likewise.
11590 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11591 (fold_convert_const_fixed_from_int): Likewise.
11592 (fold_convert_const_fixed_from_real): Likewise.
11593 (native_encode_fixed): Likewise
11594 (native_encode_complex): Likewise
11595 (native_encode_vector): Likewise.
11596 (native_interpret_fixed): Likewise.
11597 (native_interpret_real): Likewise.
11598 (native_interpret_complex): Likewise.
11599 (native_interpret_vector): Likewise.
11600 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11601 (simd_clone_adjust_argument_types): Likewise.
11602 (simd_clone_init_simd_arrays): Likewise.
11603 (simd_clone_adjust): Likewise.
11604 * stor-layout.c (layout_type): Likewise.
11605 * tree.c (build_minus_one_cst): Likewise.
11606 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11607 * tree-inline.c (estimate_move_cost): Likewise.
11608 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11609 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11610 (vectorizable_reduction): Likewise.
11611 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11612 (vect_recog_mixed_size_cond_pattern): Likewise.
11613 (check_bool_pattern): Likewise.
11614 (adjust_bool_pattern): Likewise.
11615 (search_type_for_mask_1): Likewise.
11616 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11617 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11618 (vectorizable_load): Likewise.
11619 (vectorizable_store): Likewise.
11620 * ubsan.c (ubsan_encode_value): Likewise.
11621 * varasm.c (output_constant): Likewise.
11623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11624 Alan Hayward <alan.hayward@arm.com>
11625 David Sherwood <david.sherwood@arm.com>
11627 * coretypes.h (scalar_mode): New class.
11628 * machmode.h (scalar_mode): Likewise.
11629 (scalar_mode::includes_p): New function.
11630 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11631 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11632 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11633 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11634 * expmed.c (store_bit_field_1): Likewise.
11635 (extract_bit_field_1): Likewise.
11636 * expr.c (write_complex_part): Likewise.
11637 (read_complex_part): Likewise.
11638 (emit_move_complex_push): Likewise.
11639 (expand_expr_real_2): Likewise.
11640 * function.c (assign_parm_setup_reg): Likewise.
11641 (assign_parms_unsplit_complex): Likewise.
11642 * optabs.c (expand_binop): Likewise.
11643 * rtlanal.c (subreg_get_info): Likewise.
11644 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11645 * varasm.c (output_constant_pool_2): Likewise.
11647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11648 Alan Hayward <alan.hayward@arm.com>
11649 David Sherwood <david.sherwood@arm.com>
11651 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11653 (expmed_mult_highpart_optab): Likewise.
11654 (expmed_mult_highpart): Likewise.
11656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11657 Alan Hayward <alan.hayward@arm.com>
11658 David Sherwood <david.sherwood@arm.com>
11660 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11661 instead of a machine_mode.
11662 (builtin_memset_read_str): Likewise.
11663 * builtins.c (c_readstr): Likewise.
11664 (builtin_memcpy_read_str): Likewise.
11665 (builtin_strncpy_read_str): Likewise.
11666 (builtin_memset_read_str): Likewise.
11667 (builtin_memset_gen_str): Likewise.
11668 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11669 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11670 instead of a machine_mode.
11671 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11673 (make_extraction): Likewise.
11674 (try_widen_shift_mode): Take and return scalar_int_modes instead
11676 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11677 a scalar_int_mode instead of a machine_mode.
11678 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11679 (avr_addr_space_pointer_mode): Likewise.
11680 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11681 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11682 (msp430_unwind_word_mode): Likewise.
11683 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11684 (spu_addr_space_pointer_mode): Likewise.
11685 (spu_addr_space_address_mode): Likewise.
11686 (spu_libgcc_cmp_return_mode): Likewise.
11687 (spu_libgcc_shift_count_mode): Likewise.
11688 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11689 (rl78_addr_space_pointer_mode): Likewise.
11690 (fl78_unwind_word_mode): Likewise.
11691 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11693 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11694 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11695 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11696 (mips_valid_pointer_mode): Likewise.
11697 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11698 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11699 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11701 (ft32_addr_space_address_mode): Likewise.
11702 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11703 scalar_int_mode instead of a machine_mode.
11704 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11706 (m32c_addr_space_address_mode): Likewise.
11707 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11708 (rs6000_eh_return_filter_mode): Likewise.
11709 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11710 (rs6000_eh_return_filter_mode): Likewise.
11711 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11712 (s390_libgcc_shift_count_mode): Likewise.
11713 (s390_unwind_word_mode): Likewise.
11714 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11716 * target.def (mode_rep_extended): Likewise.
11717 (valid_pointer_mode): Likewise.
11718 (addr_space.valid_pointer_mode): Likewise.
11719 (eh_return_filter_mode): Return a scalar_int_mode rather than
11721 (libgcc_cmp_return_mode): Likewise.
11722 (libgcc_shift_count_mode): Likewise.
11723 (unwind_word_mode): Likewise.
11724 (addr_space.pointer_mode): Likewise.
11725 (addr_space.address_mode): Likewise.
11726 * doc/tm.texi: Regenerate.
11727 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11729 (do_jump): Use scalar_int_mode for local variables.
11730 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11731 rather than a machine_mode.
11732 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11733 (scompare_loc_descriptor_wide): Likewise.
11734 (scompare_loc_descriptor_narrow): Likewise.
11735 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11737 * except.c (sjlj_emit_dispatch_table): Likewise.
11738 (expand_builtin_eh_copy_values): Likewise.
11739 * explow.c (convert_memory_address_addr_space_1): Likewise.
11740 Take a scalar_int_mode rather than a machine_mode.
11741 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11742 than a machine_mode.
11743 (memory_address_addr_space): Use scalar_int_mode for local variables.
11744 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11745 rather than a machine_mode.
11746 * expmed.c (mask_rtx): Likewise.
11747 (init_expmed_one_conv): Likewise.
11748 (expand_mult_highpart_adjust): Likewise.
11749 (extract_high_half): Likewise.
11750 (expmed_mult_highpart_optab): Likewise.
11751 (expmed_mult_highpart): Likewise.
11752 (expand_smod_pow2): Likewise.
11753 (expand_sdiv_pow2): Likewise.
11754 (emit_store_flag_int): Likewise.
11755 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11757 (extract_low_bits): Likewise.
11758 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11760 * expr.c (pieces_addr::adjust): Likewise.
11761 (can_store_by_pieces): Likewise.
11762 (store_by_pieces): Likewise.
11763 (clear_by_pieces_1): Likewise.
11764 (expand_expr_addr_expr_1): Likewise.
11765 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11766 (expand_expr_real_1): Likewise.
11767 (try_casesi): Likewise.
11768 * final.c (shorten_branches): Likewise.
11769 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11770 type of "mode" to machine_mode.
11771 * internal-fn.c (expand_arith_overflow_result_store): Take a
11772 scalar_int_mode rather than a machine_mode.
11773 (expand_mul_overflow): Use scalar_int_mode for local variables.
11774 * loop-doloop.c (doloop_modify): Likewise.
11775 (doloop_optimize): Likewise.
11776 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11777 than a machine_mode.
11778 (expand_doubleword_shift_condmove): Likewise.
11779 (expand_doubleword_shift): Likewise.
11780 (expand_doubleword_clz): Likewise.
11781 (expand_doubleword_popcount): Likewise.
11782 (expand_doubleword_parity): Likewise.
11783 (expand_absneg_bit): Use scalar_int_mode for local variables.
11784 (prepare_float_lib_cmp): Likewise.
11785 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11786 rather than a machine_mode.
11787 (convert_memory_address_addr_space): Likewise.
11788 (get_mode_bounds): Likewise.
11789 (get_address_mode): Return a scalar_int_mode rather than a
11791 * rtlanal.c (get_address_mode): Likewise.
11792 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11793 than a machine_mode.
11794 * targhooks.c (default_mode_rep_extended): Likewise.
11795 (default_valid_pointer_mode): Likewise.
11796 (default_addr_space_valid_pointer_mode): Likewise.
11797 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11798 than a machine_mode.
11799 (default_libgcc_cmp_return_mode): Likewise.
11800 (default_libgcc_shift_count_mode): Likewise.
11801 (default_unwind_word_mode): Likewise.
11802 (default_addr_space_pointer_mode): Likewise.
11803 (default_addr_space_address_mode): Likewise.
11804 * targhooks.h (default_eh_return_filter_mode): Likewise.
11805 (default_libgcc_cmp_return_mode): Likewise.
11806 (default_libgcc_shift_count_mode): Likewise.
11807 (default_unwind_word_mode): Likewise.
11808 (default_addr_space_pointer_mode): Likewise.
11809 (default_addr_space_address_mode): Likewise.
11810 (default_mode_rep_extended): Take a scalar_int_mode rather than
11812 (default_valid_pointer_mode): Likewise.
11813 (default_addr_space_valid_pointer_mode): Likewise.
11814 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11816 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11817 rather than a machine_mode.
11818 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11819 for local variables.
11820 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11821 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11822 than a machine_mode.
11824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11825 Alan Hayward <alan.hayward@arm.com>
11826 David Sherwood <david.sherwood@arm.com>
11828 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11829 the mode argument to scalar_int_mode.
11830 (do_jump_by_parts_zero_rtx): Likewise.
11831 (do_jump_by_parts_equality_rtx): Likewise.
11832 (do_jump_by_parts_greater): Take a mode argument.
11833 (do_jump_by_parts_equality): Likewise.
11834 (do_jump_1): Update calls accordingly.
11836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11837 Alan Hayward <alan.hayward@arm.com>
11838 David Sherwood <david.sherwood@arm.com>
11840 * is-a.h (safe_dyn_cast): New function.
11841 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11842 (jump_table_for_label): Likewise.
11843 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11844 instead of an rtx_insn *.
11845 (shorten_branches): Use dyn_cast instead of LABEL_P and
11846 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11847 rtx_jump_table_data::get_data_mode.
11848 (final_scan_insn): Likewise.
11850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11851 Alan Hayward <alan.hayward@arm.com>
11852 David Sherwood <david.sherwood@arm.com>
11854 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11855 trying to combine a full-register integer set with a subreg
11858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11859 Alan Hayward <alan.hayward@arm.com>
11860 David Sherwood <david.sherwood@arm.com>
11862 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11863 that is always either address_mode or pointer_mode.
11865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11866 Alan Hayward <alan.hayward@arm.com>
11867 David Sherwood <david.sherwood@arm.com>
11869 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11870 when the two are known to be equal.
11872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11873 Alan Hayward <alan.hayward@arm.com>
11874 David Sherwood <david.sherwood@arm.com>
11876 * simplify-rtx.c (simplify_const_unary_operation): Use
11877 is_a <scalar_int_mode> instead of checking for a nonzero
11878 precision. Forcibly convert op_mode to a scalar_int_mode
11879 in that case. More clearly differentiate the operand and
11880 result modes and use the former when deciding what the value
11881 of a count-bits operation should be. Use is_int_mode instead
11882 of checking for a MODE_INT. Remove redundant check for whether
11883 this mode has a zero precision.
11885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11886 Alan Hayward <alan.hayward@arm.com>
11887 David Sherwood <david.sherwood@arm.com>
11889 * optabs.c (widen_leading): Change the type of the mode argument
11890 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11891 (widen_bswap): Likewise.
11892 (expand_parity): Likewise.
11893 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11894 (expand_ffs): Likewise.
11895 (epand_unop): Check for scalar integer modes before calling the
11898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11899 Alan Hayward <alan.hayward@arm.com>
11900 David Sherwood <david.sherwood@arm.com>
11902 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11904 (expand_expr_real_1): Update call accordingly.
11906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11907 Alan Hayward <alan.hayward@arm.com>
11908 David Sherwood <david.sherwood@arm.com>
11910 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11911 value_mode arguments. Use scalar_int_mode internally.
11912 (store_bit_field_1): Rename the new integer mode from imode
11913 to op0_mode and use it instead of GET_MODE (op0). Update calls
11914 to store_split_bit_field, store_bit_field_using_insv and
11915 store_fixed_bit_field.
11916 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11917 Use scalar_int_mode internally. Use a bit count rather than a mode
11918 when calculating the largest bit size for get_best_mode.
11919 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11920 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11921 Remove assertion that OP0 has a scalar integer mode.
11922 (store_split_bit_field): Add op0_mode and value_mode arguments.
11923 Update calls to extract_fixed_bit_field.
11924 (extract_bit_field_using_extv): Add an op0_mode argument.
11925 Use scalar_int_mode internally.
11926 (extract_bit_field_1): Rename the new integer mode from imode to
11927 op0_mode and use it instead of GET_MODE (op0). Update calls to
11928 extract_split_bit_field, extract_bit_field_using_extv and
11929 extract_fixed_bit_field.
11930 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11931 to extract_split_bit_field and extract_fixed_bit_field_1.
11932 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11933 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11934 on the target mode.
11935 (extract_split_bit_field): Add an op0_mode argument. Update call
11936 to extract_fixed_bit_field.
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 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11943 * explow.c (hard_function_value): Likewise.
11944 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11945 convert_to_mode call outside the loop.
11946 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11947 for the mode iterator. Require the mode specified by max_pieces
11949 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11951 (copy_blkmode_to_reg): Likewise.
11952 (set_storage_via_setmem): Likewise.
11953 * optabs.c (prepare_cmp_insn): Likewise.
11954 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11955 * stor-layout.c (finish_bitfield_representative): Likewise.
11957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11958 Alan Hayward <alan.hayward@arm.com>
11959 David Sherwood <david.sherwood@arm.com>
11961 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11962 * expr.c (convert_move): Use them.
11963 (convert_modes): Likewise.
11964 (store_expr_with_bounds): Likewise.
11966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11967 Alan Hayward <alan.hayward@arm.com>
11968 David Sherwood <david.sherwood@arm.com>
11970 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11971 parameter for the mode of "x". Remove the "known_x", "known_mode"
11972 and "known_ret" arguments. Change the type of the mode argument
11973 to scalar_int_mode.
11974 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11975 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11976 (reg_num_sign_bit_copies_for_combine): Likewise.
11977 * rtlanal.c (nonzero_bits1): Likewise.
11978 (num_sign_bit_copies1): Likewise.
11979 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11980 (reg_num_sign_bit_copies_general): Likewise.
11981 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11982 (reg_nonzero_bits_general): Likewise.
11984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11985 Alan Hayward <alan.hayward@arm.com>
11986 David Sherwood <david.sherwood@arm.com>
11988 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11989 than in subroutines. Return 1 for non-integer modes.
11990 (cached_num_sign_bit_copies): Change the type of the mode parameter
11991 to scalar_int_mode.
11992 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11993 classes. Handle CONST_INT_P first and then check whether X also
11994 has a scalar integer mode. Check the same thing for inner registers
11995 of a SUBREG and for values that are being extended or truncated.
11997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11998 Alan Hayward <alan.hayward@arm.com>
11999 David Sherwood <david.sherwood@arm.com>
12001 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12002 in subroutines. Return the mode mask for non-integer modes.
12003 (cached_nonzero_bits): Change the type of the mode parameter
12004 to scalar_int_mode.
12005 (nonzero_bits1): Likewise. Remove early exit for other mode
12006 classes. Handle CONST_INT_P first and then check whether X
12007 also has a scalar integer mode.
12009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12010 Alan Hayward <alan.hayward@arm.com>
12011 David Sherwood <david.sherwood@arm.com>
12013 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12014 Return a scalar_int_mode and assert that the size is greater than
12016 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12017 (op_by_pieces_d::op_by_pieces_d): Likewise.
12018 (op_by_pieces_d::run): Likewise.
12019 (can_store_by_pieces): Likewise.
12021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12022 Alan Hayward <alan.hayward@arm.com>
12023 David Sherwood <david.sherwood@arm.com>
12025 * combine.c (extract_left_shift): Add a mode argument and update
12027 (make_compound_operation_int): Change the type of the mode parameter
12028 to scalar_int_mode and update the call to extract_left_shift.
12030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12031 Alan Hayward <alan.hayward@arm.com>
12032 David Sherwood <david.sherwood@arm.com>
12034 * combine.c (simplify_and_const_int): Change the type of the mode
12035 parameter to scalar_int_mode.
12036 (simplify_and_const_int_1): Likewise. Update recursive call.
12038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12039 Alan Hayward <alan.hayward@arm.com>
12040 David Sherwood <david.sherwood@arm.com>
12042 * combine.c (simplify_compare_const): Check that the mode is a
12043 scalar_int_mode (rather than VOIDmode) before testing its
12045 (simplify_comparison): Move COMPARISON_P handling out of the
12046 loop and restrict the latter part of the loop to scalar_int_modes.
12047 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12048 and when considering SUBREG_REGs. Use is_int_mode instead of
12049 checking GET_MODE_CLASS against MODE_INT.
12051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12052 Alan Hayward <alan.hayward@arm.com>
12053 David Sherwood <david.sherwood@arm.com>
12055 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12056 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12057 shift_unit_mode and for modes involved in scalar shifts.
12059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12060 Alan Hayward <alan.hayward@arm.com>
12061 David Sherwood <david.sherwood@arm.com>
12063 * combine.c (force_int_to_mode): New function, split out from...
12064 (force_to_mode): ...here. Keep xmode up-to-date and use it
12065 instead of GET_MODE (x).
12067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12068 Alan Hayward <alan.hayward@arm.com>
12069 David Sherwood <david.sherwood@arm.com>
12071 * optabs-query.h (extraction_insn::struct_mode): Change type to
12072 opt_scalar_int_mode and update comment.
12073 (extraction_insn::field_mode): Change type to scalar_int_mode.
12074 (extraction_insn::pos_mode): Likewise.
12075 * combine.c (make_extraction): Update accordingly.
12076 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12077 (get_optab_extraction_insn): Likewise.
12078 * recog.c (simplify_while_replacing): Likewise.
12079 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12080 parameter to opt_scalar_int_mode.
12082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12083 Alan Hayward <alan.hayward@arm.com>
12084 David Sherwood <david.sherwood@arm.com>
12086 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12087 to a scalar_int_mode instead of a machine_mode.
12088 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12089 (get_best_mode): Return a boolean and use a pointer argument to store
12090 the selected mode. Replace the limit mode parameter with a bit limit.
12091 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12092 for the values returned by bit_field_mode_iterator::next_mode.
12093 (store_bit_field): Update call to get_best_mode.
12094 (store_fixed_bit_field): Likewise.
12095 (extract_fixed_bit_field): Likewise.
12096 * expr.c (optimize_bitfield_assignment_op): Likewise.
12097 * fold-const.c (optimize_bit_field_compare): Likewise.
12098 (fold_truth_andor_1): Likewise.
12099 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12100 Update for new type of m_mode.
12101 (get_best_mode): As above.
12103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12104 Alan Hayward <alan.hayward@arm.com>
12105 David Sherwood <david.sherwood@arm.com>
12107 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12108 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12109 (store_bit_field): Check is_a <scalar_int_mode> before calling
12110 strict_volatile_bitfield_p.
12111 (extract_bit_field): Likewise.
12113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12114 Alan Hayward <alan.hayward@arm.com>
12115 David Sherwood <david.sherwood@arm.com>
12117 * target.def (cstore_mode): Return a scalar_int_mode.
12118 * doc/tm.texi: Regenerate.
12119 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12120 * targhooks.h (default_cstore_mode): Likewise.
12121 * targhooks.c (default_cstore_mode): Likewise, using a forced
12123 * expmed.c (emit_cstore): Expect the target of the cstore to be
12126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12127 Alan Hayward <alan.hayward@arm.com>
12128 David Sherwood <david.sherwood@arm.com>
12130 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12132 (niter_desc): Likewise mode.
12133 (iv_analyze): Add a mode parameter.
12135 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12136 and change its type to scalar_int_mode.
12137 * loop-iv.c: Update commentary at head of file.
12138 (iv_constant): Pass the mode paraeter before the rtx it describes
12139 and change its type to scalar_int_mode. Remove VOIDmode handling.
12140 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12141 (iv_extend): Likewise.
12142 (shorten_into_mode): Likewise.
12143 (iv_add): Use scalar_int_mode.
12144 (iv_mult): Likewise.
12145 (iv_shift): Likewise.
12146 (canonicalize_iv_subregs): Likewise.
12147 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12148 it describes and change its mode to scalar_int_mode. Also change
12149 the type of the returned inner_mode to scalar_int_mode.
12150 (get_biv_step): Likewise, turning outer_mode from a pointer
12151 into a direct parameter. Update call to get_biv_step_1.
12152 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12153 iv_constant and get_biv_step.
12154 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12155 and change its type to scalar_int_mode. Don't initialise iv->mode
12156 to VOIDmode and remove later checks for its still being VOIDmode.
12157 Update calls to iv_analyze_op and iv_analyze_expr. Check
12158 is_a <scalar_int_mode> when changing the mode under consideration.
12159 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12160 Update call to iv_analyze_expr.
12161 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12162 inner register is not also a scalar_int_mode. Update call to
12164 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12165 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12166 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12167 separate mode class checks. Update calls to iv_analyze. Remove
12168 fix-up of VOIDmodes after iv_analyze_biv.
12169 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12170 don't have a scalar_int_mode. Update call to biv_p.
12172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12173 Alan Hayward <alan.hayward@arm.com>
12174 David Sherwood <david.sherwood@arm.com>
12176 * cfgexpand.c (convert_debug_memory_address): Use
12177 as_a <scalar_int_mode>.
12178 * combine.c (expand_compound_operation): Likewise.
12179 (make_extraction): Likewise.
12180 (change_zero_ext): Likewise.
12181 (simplify_comparison): Likewise.
12182 * cse.c (cse_insn): Likewise.
12183 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12184 (mem_loc_descriptor): Likewise.
12185 (loc_descriptor): Likewise.
12186 * expmed.c (init_expmed_one_mode): Likewise.
12187 (synth_mult): Likewise.
12188 (emit_store_flag_1): Likewise.
12189 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12190 of a comparison with size.
12191 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12192 (reduce_to_bit_field_precision): Likewise.
12193 * function.c (expand_function_end): Likewise.
12194 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12195 * loop-doloop.c (doloop_modify): Likewise.
12196 * optabs.c (expand_binop): Likewise.
12197 (expand_unop): Likewise.
12198 (expand_copysign_absneg): Likewise.
12199 (prepare_cmp_insn): Likewise.
12200 (maybe_legitimize_operand): Likewise.
12201 * recog.c (const_scalar_int_operand): Likewise.
12202 * rtlanal.c (get_address_mode): Likewise.
12203 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12204 (simplify_cond_clz_ctz): Likewise.
12205 * tree-nested.c (get_nl_goto_field): Likewise.
12206 * tree.c (build_vector_type_for_mode): Likewise.
12207 * var-tracking.c (use_narrower_mode): Likewise.
12209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12210 Alan Hayward <alan.hayward@arm.com>
12211 David Sherwood <david.sherwood@arm.com>
12213 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12214 * builtins.c (expand_builtin_signbit): Use it.
12215 * cfgexpand.c (expand_debug_expr): Likewise.
12216 * dojump.c (do_jump): Likewise.
12217 (do_compare_and_jump): Likewise.
12218 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12219 * expmed.c (make_tree): Likewise.
12220 * expr.c (expand_expr_real_2): Likewise.
12221 (expand_expr_real_1): Likewise.
12222 (try_casesi): Likewise.
12223 * fold-const-call.c (fold_const_call_ss): Likewise.
12224 * fold-const.c (unextend): Likewise.
12225 (extract_muldiv_1): Likewise.
12226 (fold_single_bit_test): Likewise.
12227 (native_encode_int): Likewise.
12228 (native_encode_string): Likewise.
12229 (native_interpret_int): Likewise.
12230 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12231 * internal-fn.c (expand_addsub_overflow): Likewise.
12232 (expand_neg_overflow): Likewise.
12233 (expand_mul_overflow): Likewise.
12234 (expand_arith_overflow): Likewise.
12235 * match.pd: Likewise.
12236 * stor-layout.c (layout_type): Likewise.
12237 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12238 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12239 * tree-ssanames.c (get_range_info): Likewise.
12240 * tree-switch-conversion.c (array_value_type) Likewise.
12241 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12242 (vect_recog_divmod_pattern): Likewise.
12243 (vect_recog_mixed_size_cond_pattern): Likewise.
12244 * tree-vrp.c (extract_range_basic): Likewise.
12245 (simplify_float_conversion_using_ranges): Likewise.
12246 * tree.c (int_fits_type_p): Likewise.
12247 * ubsan.c (instrument_bool_enum_load): Likewise.
12248 * varasm.c (mergeable_string_section): Likewise.
12249 (narrowing_initializer_constant_valid_p): Likewise.
12250 (output_constant): Likewise.
12252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12253 Alan Hayward <alan.hayward@arm.com>
12254 David Sherwood <david.sherwood@arm.com>
12256 * machmode.h (NARROWEST_INT_MODE): New macro.
12257 * expr.c (alignment_for_piecewise_move): Use it instead of
12258 GET_CLASS_NARROWEST_MODE (MODE_INT).
12259 (push_block): Likewise.
12260 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12262 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12265 Alan Hayward <alan.hayward@arm.com>
12266 David Sherwood <david.sherwood@arm.com>
12268 * postreload.c (move2add_valid_value_p): Change the type of the
12269 mode parameter to scalar_int_mode.
12270 (move2add_use_add2_insn): Add a mode parameter and use it instead
12272 (move2add_use_add3_insn): Likewise.
12273 (reload_cse_move2add): Update accordingly.
12275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12276 Alan Hayward <alan.hayward@arm.com>
12277 David Sherwood <david.sherwood@arm.com>
12279 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12281 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12282 * optabs.c (expand_unop): Likewise.
12284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12285 Alan Hayward <alan.hayward@arm.com>
12286 David Sherwood <david.sherwood@arm.com>
12288 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12289 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12290 (popcount_loc_descriptor): Likewise.
12291 (bswap_loc_descriptor): Likewise.
12292 (rotate_loc_descriptor): Likewise.
12293 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12294 calling the functions above.
12296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12297 Alan Hayward <alan.hayward@arm.com>
12298 David Sherwood <david.sherwood@arm.com>
12300 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12302 (try_combine): Likewise.
12303 (simplify_if_then_else): Likewise.
12304 * cse.c (cse_insn): Likewise.
12305 * dwarf2out.c (mem_loc_descriptor): Likewise.
12306 * emit-rtl.c (gen_lowpart_common): Likewise.
12307 * simplify-rtx.c (simplify_truncation): Likewise.
12308 (simplify_binary_operation_1): Likewise.
12309 (simplify_const_relational_operation): Likewise.
12310 (simplify_ternary_operation): Likewise.
12311 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12314 Alan Hayward <alan.hayward@arm.com>
12315 David Sherwood <david.sherwood@arm.com>
12317 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12318 * reload.c (push_reload): Likewise.
12319 (find_reloads): Likewise.
12321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12322 Alan Hayward <alan.hayward@arm.com>
12323 David Sherwood <david.sherwood@arm.com>
12325 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12326 (make_compound_operation_int): Likewise.
12327 (change_zero_ext): Likewise.
12328 * expr.c (convert_move): Likewise.
12329 (convert_modes): Likewise.
12330 * fwprop.c (forward_propagate_subreg): Likewise.
12331 * loop-iv.c (get_biv_step_1): Likewise.
12332 * optabs.c (widen_operand): Likewise.
12333 * postreload.c (move2add_valid_value_p): Likewise.
12334 * recog.c (simplify_while_replacing): Likewise.
12335 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12336 (simplify_binary_operation_1): Likewise. Remove redundant
12337 mode equality check.
12339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12340 Alan Hayward <alan.hayward@arm.com>
12341 David Sherwood <david.sherwood@arm.com>
12343 * combine.c (combine_simplify_rtx): Add checks for
12344 is_a <scalar_int_mode>.
12345 (simplify_if_then_else): Likewise.
12346 (make_field_assignment): Likewise.
12347 (simplify_comparison): Likewise.
12348 * ifcvt.c (noce_try_bitop): Likewise.
12349 * loop-invariant.c (canonicalize_address_mult): Likewise.
12350 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12353 Alan Hayward <alan.hayward@arm.com>
12354 David Sherwood <david.sherwood@arm.com>
12356 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12357 is_a <scalar_int_mode> instead of != BLKmode.
12359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12360 Alan Hayward <alan.hayward@arm.com>
12361 David Sherwood <david.sherwood@arm.com>
12363 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12364 instead of != VOIDmode.
12365 * combine.c (if_then_else_cond): Likewise.
12366 (change_zero_ext): Likewise.
12367 * dwarf2out.c (mem_loc_descriptor): Likewise.
12368 (loc_descriptor): Likewise.
12369 * rtlanal.c (canonicalize_condition): Likewise.
12370 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12373 Alan Hayward <alan.hayward@arm.com>
12374 David Sherwood <david.sherwood@arm.com>
12376 * simplify-rtx.c (simplify_binary_operation_1): Use
12377 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12380 Alan Hayward <alan.hayward@arm.com>
12381 David Sherwood <david.sherwood@arm.com>
12383 * wide-int.h (int_traits<unsigned char>) New class.
12384 (int_traits<unsigned short>) Likewise.
12385 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12386 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12388 * combine.c (set_nonzero_bits_and_sign_copies): Use
12389 is_a <scalar_int_mode>.
12390 (find_split_point): Likewise.
12391 (combine_simplify_rtx): Likewise.
12392 (simplify_logical): Likewise.
12393 (expand_compound_operation): Likewise.
12394 (expand_field_assignment): Likewise.
12395 (make_compound_operation): Likewise.
12396 (extended_count): Likewise.
12397 (change_zero_ext): Likewise.
12398 (simplify_comparison): Likewise.
12399 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12400 (ucompare_loc_descriptor): Likewise.
12401 (minmax_loc_descriptor): Likewise.
12402 (mem_loc_descriptor): Likewise.
12403 (loc_descriptor): Likewise.
12404 * expmed.c (init_expmed_one_mode): Likewise.
12405 * lra-constraints.c (lra_constraint_offset): Likewise.
12406 * optabs.c (prepare_libcall_arg): Likewise.
12407 * postreload.c (move2add_note_store): Likewise.
12408 * reload.c (operands_match_p): Likewise.
12409 * rtl.h (load_extend_op): Likewise.
12410 * rtlhooks.c (gen_lowpart_general): Likewise.
12411 * simplify-rtx.c (simplify_truncation): Likewise.
12412 (simplify_unary_operation_1): Likewise.
12413 (simplify_binary_operation_1): Likewise.
12414 (simplify_const_binary_operation): Likewise.
12415 (simplify_const_relational_operation): Likewise.
12416 (simplify_subreg): Likewise.
12417 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12418 * var-tracking.c (adjust_mems): Likewise.
12419 (prepare_call_arguments): Likewise.
12421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12422 Alan Hayward <alan.hayward@arm.com>
12423 David Sherwood <david.sherwood@arm.com>
12425 * machmode.h (is_int_mode): New fuction.
12426 * combine.c (find_split_point): Use it.
12427 (combine_simplify_rtx): Likewise.
12428 (simplify_if_then_else): Likewise.
12429 (simplify_set): Likewise.
12430 (simplify_shift_const_1): Likewise.
12431 (simplify_comparison): Likewise.
12432 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12433 * cse.c (notreg_cost): Likewise.
12434 (cse_insn): Likewise.
12435 * cselib.c (cselib_lookup_1): Likewise.
12436 * dojump.c (do_jump_1): Likewise.
12437 (do_compare_rtx_and_jump): Likewise.
12438 * dse.c (get_call_args): Likewise.
12439 * dwarf2out.c (rtl_for_decl_init): Likewise.
12440 (native_encode_initializer): Likewise.
12441 * expmed.c (emit_store_flag_1): Likewise.
12442 (emit_store_flag): Likewise.
12443 * expr.c (convert_modes): Likewise.
12444 (store_field): Likewise.
12445 (expand_expr_real_1): Likewise.
12446 * fold-const.c (fold_read_from_constant_string): Likewise.
12447 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12448 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12449 * optabs.c (expand_binop): Likewise.
12450 (expand_unop): Likewise.
12451 (expand_abs_nojump): Likewise.
12452 (expand_one_cmpl_abs_nojump): Likewise.
12453 * simplify-rtx.c (mode_signbit_p): Likewise.
12454 (val_signbit_p): Likewise.
12455 (val_signbit_known_set_p): Likewise.
12456 (val_signbit_known_clear_p): Likewise.
12457 (simplify_relational_operation_1): Likewise.
12458 * tree.c (vector_type_mode): Likewise.
12460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12461 Alan Hayward <alan.hayward@arm.com>
12462 David Sherwood <david.sherwood@arm.com>
12464 * machmode.h (smallest_mode_for_size): Fix formatting.
12465 (smallest_int_mode_for_size): New function.
12466 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12467 instead of smallest_mode_for_size.
12468 * combine.c (make_extraction): Likewise.
12469 * config/arc/arc.c (arc_expand_movmem): Likewise.
12470 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12471 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12472 * config/s390/s390.c (s390_expand_insv): Likewise.
12473 * config/sparc/sparc.c (assign_int_registers): Likewise.
12474 * config/spu/spu.c (spu_function_value): Likewise.
12475 (spu_function_arg): Likewise.
12476 * coverage.c (get_gcov_type): Likewise.
12477 (get_gcov_unsigned_t): Likewise.
12478 * dse.c (find_shift_sequence): Likewise.
12479 * expmed.c (store_bit_field_1): Likewise.
12480 * expr.c (convert_move): Likewise.
12481 (store_field): Likewise.
12482 * internal-fn.c (expand_arith_overflow): Likewise.
12483 * optabs-query.c (get_best_extraction_insn): Likewise.
12484 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12485 * stor-layout.c (layout_type): Likewise.
12486 (initialize_sizetypes): Likewise.
12487 * targhooks.c (default_get_mask_mode): Likewise.
12488 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12491 Alan Hayward <alan.hayward@arm.com>
12492 David Sherwood <david.sherwood@arm.com>
12494 * machmode.h (opt_mode::else_blk): New function.
12495 (int_mode_for_mode): Declare.
12496 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12497 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12499 * cfgexpand.c (expand_debug_expr): Likewise.
12500 * combine.c (gen_lowpart_or_truncate): Likewise.
12501 (gen_lowpart_for_combine): Likewise.
12502 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12503 * config/avr/avr.c (avr_to_int_mode): Likewise.
12504 (avr_out_plus_1): Likewise.
12505 (avr_out_plus): Likewise.
12506 (avr_out_round): Likewise.
12507 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12508 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12509 (s390_expand_vcond): Likewise.
12510 * config/spu/spu.c (spu_split_immediate): Likewise.
12511 (spu_expand_mov): Likewise.
12512 * dse.c (get_stored_val): Likewise.
12513 * expmed.c (store_bit_field_1): Likewise.
12514 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12516 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12517 (extract_low_bits): Likewise.
12518 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12519 handling rather than repeating the check.
12520 (emit_group_store): Likewise.
12521 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12522 * optabs.c (expand_absneg_bit): Likewise.
12523 (expand_copysign_absneg): Likewise.
12524 (expand_copysign_bit): Likewise.
12525 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12526 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12527 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12528 * var-tracking.c (prepare_call_arguments): Likewise.
12529 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12530 int_mode_for_mode instead of mode_for_size.
12531 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12534 Alan Hayward <alan.hayward@arm.com>
12535 David Sherwood <david.sherwood@arm.com>
12537 * machmode.h (int_mode_for_size): New function.
12538 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12539 instead of mode_for_size.
12540 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12542 * combine.c (expand_field_assignment): Use int_mode_for_size
12543 instead of mode_for_size.
12544 (make_extraction): Likewise.
12545 (simplify_shift_const_1): Likewise.
12546 (simplify_comparison): Likewise.
12547 * dojump.c (do_jump): Likewise.
12548 * dwarf2out.c (mem_loc_descriptor): Likewise.
12549 * emit-rtl.c (init_derived_machine_modes): Likewise.
12550 * expmed.c (flip_storage_order): Likewise.
12551 (convert_extracted_bit_field): Likewise.
12552 * expr.c (copy_blkmode_from_reg): Likewise.
12553 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12554 * internal-fn.c (expand_mul_overflow): Likewise.
12555 * lower-subreg.c (simple_move): Likewise.
12556 * optabs-libfuncs.c (init_optabs): Likewise.
12557 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12558 * tree.c (vector_type_mode): Likewise.
12559 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12560 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12561 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12562 * tree-vect-stmts.c (vectorizable_load): Likewise.
12563 (vectorizable_store): Likewise.
12565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12566 Alan Hayward <alan.hayward@arm.com>
12567 David Sherwood <david.sherwood@arm.com>
12569 * coretypes.h (pod_mode): New type.
12570 (scalar_int_mode_pod): New typedef.
12571 * machmode.h (pod_mode): New class.
12572 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12573 * genmodes.c (emit_mode_int_n): Update accordingly.
12574 * lower-subreg.h (target_lower_subreg): Change type to
12575 scalar_int_mode_pod.
12576 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12577 scalar_int_mode_pod.
12579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12580 Alan Hayward <alan.hayward@arm.com>
12581 David Sherwood <david.sherwood@arm.com>
12583 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12584 machine_mode to scalar_int_mode.
12585 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12586 (rs6000_option_override_internal): Remove cast to int.
12587 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12588 machine_mode to scalar_int_mode.
12589 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12590 (rs6000_option_override_internal): Remove cast to int.
12591 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12592 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12594 * config/s390/s390.c (s390_expand_builtin): Likewise.
12595 * coretypes.h (scalar_int_mode): New type.
12596 (opt_scalar_int_mode): New typedef.
12597 * machmode.h (scalar_int_mode): New class.
12598 (scalar_int_mode::includes_p): New function.
12599 (byte_mode): Change type to scalar_int_mode.
12600 (word_mode): Likewise.
12601 (ptr_mode): Likewise.
12602 * emit-rtl.c (byte_mode): Likewise.
12603 (word_mode): Likewise.
12604 (ptr_mode): Likewise.
12605 (init_derived_machine_modes): Update accordingly.
12606 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12607 and MODE_PARTIAL_INT.
12608 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12609 opt_scalar_int_mode.
12611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12612 Alan Hayward <alan.hayward@arm.com>
12613 David Sherwood <david.sherwood@arm.com>
12615 * target.def (libgcc_floating_mode_supported_p): Take a
12617 * doc/tm.texi: Regenerate.
12618 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12620 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12621 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12625 Alan Hayward <alan.hayward@arm.com>
12626 David Sherwood <david.sherwood@arm.com>
12628 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12629 * doc/tm.texi: Regenerate.
12630 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12631 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12632 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12633 * targhooks.h (default_floatn_mode): Likewise.
12634 * targhooks.c (default_floatn_mode): Likewise.
12635 * tree.c (build_common_tree_nodes): Update accordingly.
12637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12638 Alan Hayward <alan.hayward@arm.com>
12639 David Sherwood <david.sherwood@arm.com>
12641 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12642 (mode_iterator::iterate_p): Likewise.
12643 (mode_iterator::get_wider): Likewise.
12644 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12647 Alan Hayward <alan.hayward@arm.com>
12648 David Sherwood <david.sherwood@arm.com>
12650 * coretypes.h (opt_scalar_float_mode): New typedef.
12651 * machmode.h (float_mode_for_size): New function.
12652 * emit-rtl.c (double_mode): Delete.
12653 (init_emit_once): Use float_mode_for_size.
12654 * stor-layout.c (layout_type): Likewise.
12655 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12658 Alan Hayward <alan.hayward@arm.com>
12659 David Sherwood <david.sherwood@arm.com>
12661 * output.h (assemble_real): Take a scalar_float_mode.
12662 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12663 * config/arm/arm.md (consttable_4): Likewise.
12664 (consttable_8): Likewise.
12665 (consttable_16): Likewise.
12666 * config/mips/mips.md (consttable_float): Likewise.
12667 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12668 * varasm.c (assemble_real): Take a scalar_float_mode.
12669 (output_constant_pool_2): Update accordingly.
12670 (output_constant): Likewise.
12672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12673 Alan Hayward <alan.hayward@arm.com>
12674 David Sherwood <david.sherwood@arm.com>
12676 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12677 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12678 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12679 (native_encode_real): Likewise.
12680 (native_interpret_real): Likewise.
12681 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12682 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12685 Alan Hayward <alan.hayward@arm.com>
12686 David Sherwood <david.sherwood@arm.com>
12688 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12689 <scalar_float_mode>. Simplify.
12690 (gen_extend_conv_libfunc): Likewise.
12692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12693 Alan Hayward <alan.hayward@arm.com>
12694 David Sherwood <david.sherwood@arm.com>
12696 * coretypes.h (scalar_float_mode): New type.
12697 * machmode.h (mode_traits::from_int): Use machine_mode if
12698 USE_ENUM_MODES is defined.
12699 (is_a): New function.
12701 (dyn_cast): Likewise.
12702 (scalar_float_mode): New class.
12703 (scalar_float_mode::includes_p): New function.
12704 (is_float_mode): Likewise.
12705 * gdbhooks.py (MachineModePrinter): New class.
12706 (build_pretty_printer): Use it for scalar_float_mode.
12707 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12708 (format_helper::format_helper): Turn into a template.
12709 * genmodes.c (get_mode_class): New function.
12710 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12711 or machine_mode if none.
12712 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12713 as_a <scalar_float_mode>.
12714 * dwarf2out.c (mem_loc_descriptor): Likewise.
12715 (insert_float): Likewise.
12716 (add_const_value_attribute): Likewise.
12717 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12718 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12719 (expand_unop): Update accordingly.
12720 (expand_abs_nojump): Likewise.
12721 (expand_copysign_absneg): Take a scalar_float_mode.
12722 (expand_copysign_bit): Likewise.
12723 (expand_copysign): Update accordingly.
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 * coretypes.h (opt_mode): New class.
12730 * machmode.h (opt_mode): Likewise.
12731 (opt_mode::else_void): New function.
12732 (opt_mode::require): Likewise.
12733 (opt_mode::exists): Likewise.
12734 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12735 (GET_MODE_2XWIDER_MODE): Likewise.
12736 (mode_iterator::get_wider): Update accordingly.
12737 (mode_iterator::get_2xwider): Likewise.
12738 (mode_iterator::get_known_wider): Likewise, turning into a template.
12739 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12740 forcing a wider mode to exist.
12741 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12742 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12743 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12744 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12745 * lower-subreg.c (init_lower_subreg): Likewise.
12746 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12747 on the final iteration.
12748 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12749 a wider mode exists before asking for a move pattern.
12750 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12751 forcing a wider mode to exist.
12752 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12753 returning false if no such mode exists.
12754 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12755 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12756 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12757 Avoid checking for a MODE_INT if we already know the mode is not a
12759 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12760 forcing a wider mode to exist.
12761 (expmed_mult_highpart_optab): Likewise.
12762 (expmed_mult_highpart): Likewise.
12763 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12765 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12766 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12767 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12768 * internal-fn.c (expand_mul_overflow): Update use of
12769 GET_MODE_2XWIDER_MODE.
12770 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12771 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12772 GET_MODE_WIDER_MODE.
12773 (convert_plusminus_to_widen): Likewise.
12774 * tree-switch-conversion.c (array_value_type): Likewise.
12775 * var-tracking.c (emit_note_insn_var_location): Likewise.
12776 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12777 Return false inside rather than outside the loop if no wider mode
12779 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12780 and GET_MODE_2XWIDER_MODE
12781 (can_compare_p): Use else_void.
12782 * gdbhooks.py (OptMachineModePrinter): New class.
12783 (build_pretty_printer): Use it for opt_mode.
12785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12786 Alan Hayward <alan.hayward@arm.com>
12787 David Sherwood <david.sherwood@arm.com>
12789 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12790 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12793 Alan Hayward <alan.hayward@arm.com>
12794 David Sherwood <david.sherwood@arm.com>
12796 * machmode.h (mode_traits): New structure.
12797 (get_narrowest_mode): New function.
12798 (mode_iterator::start): Likewise.
12799 (mode_iterator::iterate_p): Likewise.
12800 (mode_iterator::get_wider): Likewise.
12801 (mode_iterator::get_known_wider): Likewise.
12802 (mode_iterator::get_2xwider): Likewise.
12803 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12804 (FOR_EACH_MODE): Likewise.
12805 (FOR_EACH_MODE_FROM): Likewise.
12806 (FOR_EACH_MODE_UNTIL): Likewise.
12807 (FOR_EACH_WIDER_MODE): Likewise.
12808 (FOR_EACH_2XWIDER_MODE): Likewise.
12809 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12810 * combine.c (simplify_comparison): Likewise
12811 * config/i386/i386.c (type_natural_mode): Likewise.
12812 * cse.c (cse_insn): Likewise.
12813 * dse.c (find_shift_sequence): Likewise.
12814 * emit-rtl.c (init_derived_machine_modes): Likewise.
12815 (init_emit_once): Likewise.
12816 * explow.c (hard_function_value): Likewise.
12817 * expmed.c (extract_fixed_bit_field_1): Likewise.
12818 (extract_bit_field_1): Likewise.
12819 (expand_divmod): Likewise.
12820 (emit_store_flag_1): Likewise.
12821 * expr.c (init_expr_target): Likewise.
12822 (convert_move): Likewise.
12823 (alignment_for_piecewise_move): Likewise.
12824 (widest_int_mode_for_size): Likewise.
12825 (emit_block_move_via_movmem): Likewise.
12826 (copy_blkmode_to_reg): Likewise.
12827 (set_storage_via_setmem): Likewise.
12828 (compress_float_constant): Likewise.
12829 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12830 * optabs-query.c (get_best_extraction_insn): Likewise.
12831 * optabs.c (expand_binop): Likewise.
12832 (expand_twoval_unop): Likewise.
12833 (expand_twoval_binop): Likewise.
12834 (widen_leading): Likewise.
12835 (widen_bswap): Likewise.
12836 (expand_parity): Likewise.
12837 (expand_unop): Likewise.
12838 (prepare_cmp_insn): Likewise.
12839 (prepare_float_lib_cmp): Likewise.
12840 (expand_float): Likewise.
12841 (expand_fix): Likewise.
12842 (expand_sfix_optab): Likewise.
12843 * postreload.c (move2add_use_add2_insn): Likewise.
12844 * reg-stack.c (reg_to_stack): Likewise.
12845 * reginfo.c (choose_hard_reg_mode): Likewise.
12846 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12847 * stor-layout.c (mode_for_size): Likewise.
12848 (smallest_mode_for_size): Likewise.
12849 (mode_for_vector): Likewise.
12850 (finish_bitfield_representative): Likewise.
12851 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12852 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12853 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12854 * var-tracking.c (prepare_call_arguments): Likewise.
12856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12857 Alan Hayward <alan.hayward@arm.com>
12858 David Sherwood <david.sherwood@arm.com>
12860 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12861 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12862 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12863 * machmode.h (mode_size): Move earlier in file.
12864 (mode_precision): Likewise.
12865 (mode_inner): Likewise.
12866 (mode_nunits): Likewise.
12867 (mode_unit_size): Likewise.
12868 (unit_unit_precision): Likewise.
12869 (mode_wider): Likewise.
12870 (mode_2xwider): Likewise.
12871 (machine_mode): New class.
12872 (mode_to_bytes): New function.
12873 (mode_to_bits): Likewise.
12874 (mode_to_precision): Likewise.
12875 (mode_to_inner): Likewise.
12876 (mode_to_unit_size): Likewise.
12877 (mode_to_unit_precision): Likewise.
12878 (mode_to_nunits): Likewise.
12879 (GET_MODE_SIZE): Use mode_to_bytes.
12880 (GET_MODE_BITSIZE): Use mode_to_bits.
12881 (GET_MODE_PRECISION): Use mode_to_precision.
12882 (GET_MODE_INNER): Use mode_to_inner.
12883 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12884 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12885 (GET_MODE_NUNITS): Use mode_to_nunits.
12886 * system.h (ALWAYS_INLINE): New macro.
12887 * config/powerpcspe/powerpcspe-c.c
12888 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12889 int for arg1_mode and arg2_mode.
12891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12892 Alan Hayward <alan.hayward@arm.com>
12893 David Sherwood <david.sherwood@arm.com>
12895 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12896 Prefix mode names with E_ in case statements.
12897 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12898 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12899 (aarch64_split_simd_move): Likewise.
12900 (aarch64_gen_storewb_pair): Likewise.
12901 (aarch64_gen_loadwb_pair): Likewise.
12902 (aarch64_gen_store_pair): Likewise.
12903 (aarch64_gen_load_pair): Likewise.
12904 (aarch64_get_condition_code_1): Likewise.
12905 (aarch64_constant_pool_reload_icode): Likewise.
12906 (get_rsqrte_type): Likewise.
12907 (get_rsqrts_type): Likewise.
12908 (get_recpe_type): Likewise.
12909 (get_recps_type): Likewise.
12910 (aarch64_gimplify_va_arg_expr): Likewise.
12911 (aarch64_simd_container_mode): Likewise.
12912 (aarch64_emit_load_exclusive): Likewise.
12913 (aarch64_emit_store_exclusive): Likewise.
12914 (aarch64_expand_compare_and_swap): Likewise.
12915 (aarch64_gen_atomic_cas): Likewise.
12916 (aarch64_emit_bic): Likewise.
12917 (aarch64_emit_atomic_swap): Likewise.
12918 (aarch64_emit_atomic_load_op): Likewise.
12919 (aarch64_evpc_trn): Likewise.
12920 (aarch64_evpc_uzp): Likewise.
12921 (aarch64_evpc_zip): Likewise.
12922 (aarch64_evpc_ext): Likewise.
12923 (aarch64_evpc_rev): Likewise.
12924 (aarch64_evpc_dup): Likewise.
12925 (aarch64_gen_ccmp_first): Likewise.
12926 (aarch64_gen_ccmp_next): Likewise.
12927 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12928 (alpha_emit_xfloating_libcall): Likewise.
12929 (emit_insxl): Likewise.
12930 (alpha_arg_type): Likewise.
12931 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12932 (arc_preferred_simd_mode): Likewise.
12933 (arc_secondary_reload): Likewise.
12934 (get_arc_condition_code): Likewise.
12935 (arc_print_operand): Likewise.
12936 (arc_legitimate_constant_p): Likewise.
12937 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12938 * config/arc/arc.md (casesi_load): Likewise.
12939 (casesi_compact_jump): Likewise.
12940 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12941 (cc_use_register): Likewise.
12942 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12943 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12944 (arm_init_iwmmxt_builtins): Likewise.
12945 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12946 (neon_expand_vector_init): Likewise.
12947 (arm_attr_length_move_neon): Likewise.
12948 (maybe_get_arm_condition_code): Likewise.
12949 (arm_emit_vector_const): Likewise.
12950 (arm_preferred_simd_mode): Likewise.
12951 (arm_output_iwmmxt_tinsr): Likewise.
12952 (thumb1_output_casesi): Likewise.
12953 (thumb2_output_casesi): Likewise.
12954 (arm_emit_load_exclusive): Likewise.
12955 (arm_emit_store_exclusive): Likewise.
12956 (arm_expand_compare_and_swap): Likewise.
12957 (arm_evpc_neon_vuzp): Likewise.
12958 (arm_evpc_neon_vzip): Likewise.
12959 (arm_evpc_neon_vrev): Likewise.
12960 (arm_evpc_neon_vtrn): Likewise.
12961 (arm_evpc_neon_vext): Likewise.
12962 (arm_validize_comparison): Likewise.
12963 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12964 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12965 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12966 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12967 (c6x_preferred_simd_mode): Likewise.
12968 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12969 (epiphany_rtx_costs): Likewise.
12970 * config/epiphany/predicates.md (proper_comparison_operator):
12972 * config/frv/frv.c (condexec_memory_operand): Likewise.
12973 (frv_emit_move): Likewise.
12974 (output_move_single): Likewise.
12975 (output_condmove_single): Likewise.
12976 (frv_hard_regno_mode_ok): Likewise.
12977 (frv_matching_accg_mode): Likewise.
12978 * config/h8300/h8300.c (split_adds_subs): Likewise.
12979 (h8300_rtx_costs): Likewise.
12980 (h8300_print_operand): Likewise.
12981 (compute_mov_length): Likewise.
12982 (output_logical_op): Likewise.
12983 (compute_logical_op_length): Likewise.
12984 (compute_logical_op_cc): Likewise.
12985 (h8300_shift_needs_scratch_p): Likewise.
12986 (output_a_shift): Likewise.
12987 (compute_a_shift_length): Likewise.
12988 (compute_a_shift_cc): Likewise.
12989 (expand_a_rotate): Likewise.
12990 (output_a_rotate): Likewise.
12991 * config/i386/i386.c (classify_argument): Likewise.
12992 (function_arg_advance_32): Likewise.
12993 (function_arg_32): Likewise.
12994 (function_arg_64): Likewise.
12995 (function_value_64): Likewise.
12996 (ix86_gimplify_va_arg): Likewise.
12997 (ix86_legitimate_constant_p): Likewise.
12998 (put_condition_code): Likewise.
12999 (split_double_mode): Likewise.
13000 (ix86_avx256_split_vector_move_misalign): Likewise.
13001 (ix86_expand_vector_logical_operator): Likewise.
13002 (ix86_split_idivmod): Likewise.
13003 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13004 (ix86_build_const_vector): Likewise.
13005 (ix86_build_signbit_mask): Likewise.
13006 (ix86_match_ccmode): Likewise.
13007 (ix86_cc_modes_compatible): Likewise.
13008 (ix86_expand_branch): Likewise.
13009 (ix86_expand_sse_cmp): Likewise.
13010 (ix86_expand_sse_movcc): Likewise.
13011 (ix86_expand_int_sse_cmp): Likewise.
13012 (ix86_expand_vec_perm_vpermi2): Likewise.
13013 (ix86_expand_vec_perm): Likewise.
13014 (ix86_expand_sse_unpack): Likewise.
13015 (ix86_expand_int_addcc): Likewise.
13016 (ix86_split_to_parts): Likewise.
13017 (ix86_vectorize_builtin_gather): Likewise.
13018 (ix86_vectorize_builtin_scatter): Likewise.
13019 (avx_vpermilp_parallel): Likewise.
13020 (inline_memory_move_cost): Likewise.
13021 (ix86_tieable_integer_mode_p): Likewise.
13022 (x86_maybe_negate_const_int): Likewise.
13023 (ix86_expand_vector_init_duplicate): Likewise.
13024 (ix86_expand_vector_init_one_nonzero): Likewise.
13025 (ix86_expand_vector_init_one_var): Likewise.
13026 (ix86_expand_vector_init_concat): Likewise.
13027 (ix86_expand_vector_init_interleave): Likewise.
13028 (ix86_expand_vector_init_general): Likewise.
13029 (ix86_expand_vector_set): Likewise.
13030 (ix86_expand_vector_extract): Likewise.
13031 (emit_reduc_half): Likewise.
13032 (ix86_emit_i387_round): Likewise.
13033 (ix86_mangle_type): Likewise.
13034 (ix86_expand_round_sse4): Likewise.
13035 (expand_vec_perm_blend): Likewise.
13036 (canonicalize_vector_int_perm): Likewise.
13037 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13038 (expand_vec_perm_1): Likewise.
13039 (expand_vec_perm_interleave3): Likewise.
13040 (expand_vec_perm_even_odd_pack): Likewise.
13041 (expand_vec_perm_even_odd_1): Likewise.
13042 (expand_vec_perm_broadcast_1): Likewise.
13043 (ix86_vectorize_vec_perm_const_ok): Likewise.
13044 (ix86_expand_vecop_qihi): Likewise.
13045 (ix86_expand_mul_widen_hilo): Likewise.
13046 (ix86_expand_sse2_abs): Likewise.
13047 (ix86_expand_pextr): Likewise.
13048 (ix86_expand_pinsr): Likewise.
13049 (ix86_preferred_simd_mode): Likewise.
13050 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13051 * config/i386/sse.md (*andnot<mode>3): Likewise.
13052 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13053 (*<code><mode>3): Likewise.
13054 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13055 (ia64_expand_atomic_op): Likewise.
13056 (ia64_arg_type): Likewise.
13057 (ia64_mode_to_int): Likewise.
13058 (ia64_scalar_mode_supported_p): Likewise.
13059 (ia64_vector_mode_supported_p): Likewise.
13060 (expand_vec_perm_broadcast): Likewise.
13061 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13062 (iq2000_function_arg_advance): Likewise.
13063 (iq2000_function_arg): Likewise.
13064 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13065 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13066 (m68k_libcall_value): Likewise.
13067 (m68k_function_value): Likewise.
13068 (sched_attr_op_type): Likewise.
13069 * config/mcore/mcore.c (mcore_output_move): Likewise.
13070 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13072 (microblaze_function_arg): Likewise.
13073 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13074 (mips_print_operand): Likewise.
13075 (mips_mode_ok_for_mov_fmt_p): Likewise.
13076 (mips_vector_mode_supported_p): Likewise.
13077 (mips_preferred_simd_mode): Likewise.
13078 (mips_expand_vpc_loongson_even_odd): Likewise.
13079 (mips_expand_vec_unpack): Likewise.
13080 (mips_expand_vi_broadcast): Likewise.
13081 (mips_expand_vector_init): Likewise.
13082 (mips_expand_vec_reduc): Likewise.
13083 (mips_expand_msa_cmp): Likewise.
13084 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13085 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13086 (cc_flags_for_mode): Likewise.
13087 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13088 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13089 (nds32_output_casesi_pc_relative): Likewise.
13090 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13091 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13092 (nvptx_gen_unpack): Likewise.
13093 (nvptx_gen_pack): Likewise.
13094 (nvptx_gen_shuffle): Likewise.
13095 (nvptx_gen_wcast): Likewise.
13096 (nvptx_preferred_simd_mode): Likewise.
13097 * config/pa/pa.c (pa_secondary_reload): Likewise.
13098 * config/pa/predicates.md (base14_operand): Likewise.
13099 * config/powerpcspe/powerpcspe-c.c
13100 (altivec_resolve_overloaded_builtin): Likewise.
13101 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13103 (rs6000_preferred_simd_mode): Likewise.
13104 (output_vec_const_move): Likewise.
13105 (rs6000_expand_vector_extract): Likewise.
13106 (rs6000_split_vec_extract_var): Likewise.
13107 (reg_offset_addressing_ok_p): Likewise.
13108 (rs6000_legitimate_offset_address_p): Likewise.
13109 (rs6000_legitimize_address): Likewise.
13110 (rs6000_emit_set_const): Likewise.
13111 (rs6000_const_vec): Likewise.
13112 (rs6000_emit_move): Likewise.
13113 (spe_build_register_parallel): Likewise.
13114 (rs6000_darwin64_record_arg_recurse): Likewise.
13115 (swap_selector_for_mode): Likewise.
13116 (spe_init_builtins): Likewise.
13117 (paired_init_builtins): Likewise.
13118 (altivec_init_builtins): Likewise.
13119 (do_load_for_compare): Likewise.
13120 (rs6000_generate_compare): Likewise.
13121 (rs6000_expand_float128_convert): Likewise.
13122 (emit_load_locked): Likewise.
13123 (emit_store_conditional): Likewise.
13124 (rs6000_output_function_epilogue): Likewise.
13125 (rs6000_handle_altivec_attribute): Likewise.
13126 (rs6000_function_value): Likewise.
13127 (emit_fusion_gpr_load): Likewise.
13128 (emit_fusion_p9_load): Likewise.
13129 (emit_fusion_p9_store): Likewise.
13130 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13131 (fusion_gpr_mem_load): Likewise.
13132 (fusion_addis_mem_combo_load): Likewise.
13133 (fusion_addis_mem_combo_store): Likewise.
13134 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13135 (fusion_gpr_mem_load): Likewise.
13136 (fusion_addis_mem_combo_load): Likewise.
13137 (fusion_addis_mem_combo_store): Likewise.
13138 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13140 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13141 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13142 (rs6000_preferred_simd_mode): Likewise.
13143 (output_vec_const_move): Likewise.
13144 (rs6000_expand_vector_extract): Likewise.
13145 (rs6000_split_vec_extract_var): Likewise.
13146 (reg_offset_addressing_ok_p): Likewise.
13147 (rs6000_legitimate_offset_address_p): Likewise.
13148 (rs6000_legitimize_address): Likewise.
13149 (rs6000_emit_set_const): Likewise.
13150 (rs6000_const_vec): Likewise.
13151 (rs6000_emit_move): Likewise.
13152 (rs6000_darwin64_record_arg_recurse): Likewise.
13153 (swap_selector_for_mode): Likewise.
13154 (paired_init_builtins): Likewise.
13155 (altivec_init_builtins): Likewise.
13156 (rs6000_expand_float128_convert): Likewise.
13157 (emit_load_locked): Likewise.
13158 (emit_store_conditional): Likewise.
13159 (rs6000_output_function_epilogue): Likewise.
13160 (rs6000_handle_altivec_attribute): Likewise.
13161 (rs6000_function_value): Likewise.
13162 (emit_fusion_gpr_load): Likewise.
13163 (emit_fusion_p9_load): Likewise.
13164 (emit_fusion_p9_store): Likewise.
13165 * config/rx/rx.c (rx_gen_move_template): Likewise.
13166 (flags_from_mode): Likewise.
13167 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13168 (s390_slb_comparison): Likewise.
13169 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13170 (s390_vector_mode_supported_p): Likewise.
13171 (s390_cc_modes_compatible): Likewise.
13172 (s390_match_ccmode_set): Likewise.
13173 (s390_canonicalize_comparison): Likewise.
13174 (s390_emit_compare_and_swap): Likewise.
13175 (s390_branch_condition_mask): Likewise.
13176 (s390_rtx_costs): Likewise.
13177 (s390_secondary_reload): Likewise.
13178 (__SECONDARY_RELOAD_CASE): Likewise.
13179 (s390_expand_cs): Likewise.
13180 (s390_preferred_simd_mode): Likewise.
13181 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13182 * config/sh/sh.c (sh_print_operand): Likewise.
13183 (dump_table): Likewise.
13184 (sh_secondary_reload): Likewise.
13185 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13186 * config/sh/sh.md (casesi_worker_1): Likewise.
13187 (casesi_worker_2): Likewise.
13188 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13189 (fcc_comparison_operator): Likewise.
13190 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13191 (emit_soft_tfmode_cvt): Likewise.
13192 (sparc_preferred_simd_mode): Likewise.
13193 (output_cbranch): Likewise.
13194 (sparc_print_operand): Likewise.
13195 (sparc_expand_vec_perm_bmask): Likewise.
13196 (vector_init_bshuffle): Likewise.
13197 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13198 (spu_vector_mode_supported_p): Likewise.
13199 (spu_expand_insv): Likewise.
13200 (spu_emit_branch_or_set): Likewise.
13201 (spu_handle_vector_attribute): Likewise.
13202 (spu_builtin_splats): Likewise.
13203 (spu_builtin_extract): Likewise.
13204 (spu_builtin_promote): Likewise.
13205 (spu_expand_sign_extend): Likewise.
13206 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13207 (tilegx_simd_int): Likewise.
13208 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13209 (tilepro_simd_int): Likewise.
13210 * config/v850/v850.c (const_double_split): Likewise.
13211 (v850_print_operand): Likewise.
13212 (ep_memory_offset): Likewise.
13213 * config/vax/vax.c (vax_rtx_costs): Likewise.
13214 (vax_output_int_move): Likewise.
13215 (vax_output_int_add): Likewise.
13216 (vax_output_int_subtract): Likewise.
13217 * config/visium/predicates.md (visium_branch_operator): Likewise.
13218 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13219 (visium_print_operand_address): Likewise.
13220 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13221 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13222 (xtensa_expand_conditional_branch): Likewise.
13223 (xtensa_copy_incoming_a7): Likewise.
13224 (xtensa_output_literal): Likewise.
13225 * dfp.c (decimal_real_maxval): Likewise.
13226 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13229 Alan Hayward <alan.hayward@arm.com>
13230 David Sherwood <david.sherwood@arm.com>
13232 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13233 (mode_nunits_inline): Likewise.
13234 (mode_inner_inline): Likewise.
13235 (mode_unit_size_inline): Likewise.
13236 (mode_unit_precision_inline): Likewise.
13237 (emit_insn_modes_h): Likewise. Also emit a #define of the
13239 (emit_mode_wider): Add an E_ prefix to mode names.
13240 (emit_mode_complex): Likewise.
13241 (emit_mode_inner): Likewise.
13242 (emit_mode_adjustments): Likewise.
13243 (emit_mode_int_n): Likewise.
13244 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13245 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13246 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13247 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13248 (CRC32_BUILTIN, ENTRY): Likewise.
13249 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13250 (aarch64_pop_regs): Likewise.
13251 (aarch64_process_components): Likewise.
13252 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13253 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13254 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13255 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13256 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13257 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13259 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13260 (pcmpistr): Likewise.
13261 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13262 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13263 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13265 * config/rl78/rl78.c (mduc_regs): Likewise.
13266 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13267 (htm_expand_builtin): Likewise.
13268 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13269 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13270 * config/xtensa/xtensa.c (print_operand): Likewise.
13271 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13272 (NUM_MODE_VECTOR_INT): Likewise.
13273 * genoutput.c (null_operand): Likewise.
13274 (output_operand_data): Likewise.
13275 * genrecog.c (print_parameter_value): Likewise.
13276 * lra.c (debug_operand_data): Likewise.
13278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13280 * dwarf2out.c (scompare_loc_descriptor_wide)
13281 (scompare_loc_descriptor_narrow): New functions, split out from...
13282 (scompare_loc_descriptor): ...here.
13283 * expmed.c (emit_store_flag_int): New function, split out from...
13284 (emit_store_flag): ...here.
13286 2017-08-30 Richard Biener <rguenther@suse.de>
13288 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13289 (dwarf2out_early_finish): Move setting of AT_pubnames from
13290 early debug output to early finish.
13292 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13294 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13295 and -mdata-region to the assembler.
13297 2017-08-30 Richard Biener <rguenther@suse.de>
13299 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13301 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13302 (dwarf2out_early_global_decl): Only generate a DIE for the
13303 abstract origin if it doesn't already exist or is a declaration DIE.
13304 (resolve_addr): Do not add the linkage name twice when
13305 generating a stub DIE for the DW_TAG_GNU_call_site target.
13307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13309 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13310 Use machine_mode rather than int for arg1_mode.
13312 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13315 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13316 that the second argument of the built-in functions to unpack
13317 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13318 switch statement instead a lot of if statements.
13319 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13320 Allow 64-bit values to be in Altivec registers as well as
13321 traditional floating point registers.
13322 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13324 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13326 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13327 MAX_REGS_PER_ADDRESS == 1.
13329 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13331 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13332 * config/i386/i386.c (ix86_option_override_internal): Simplify
13333 setting of opts->x_flag_entry.
13335 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13336 Jakub Jelinek <jakub@redhat.com>
13337 Richard Biener <rguenther@suse.de>
13339 PR tree-optimization/81503
13340 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13341 folded constant fits in the target type; reorder tests for clarity.
13343 2017-08-29 Martin Liska <mliska@suse.cz>
13345 * passes.def: Include pass_lower_switch.
13346 * stmt.c (dump_case_nodes): Remove and move to
13347 tree-switch-conversion.
13348 (case_values_threshold): Likewise.
13349 (expand_switch_as_decision_tree_p): Likewise.
13350 (emit_case_decision_tree): Likewise.
13351 (expand_case): Likewise.
13352 (balance_case_nodes): Likewise.
13353 (node_has_low_bound): Likewise.
13354 (node_has_high_bound): Likewise.
13355 (node_is_bounded): Likewise.
13356 (emit_case_nodes): Likewise.
13357 (struct simple_case_node): New struct.
13358 (add_case_node): Remove.
13359 (emit_case_dispatch_table): Use vector instead of case_list.
13360 (reset_out_edges_aux): Remove.
13361 (compute_cases_per_edge): Likewise.
13362 (expand_case): Build list of simple_case_node.
13363 (expand_sjlj_dispatch_table): Use it.
13364 * tree-switch-conversion.c (struct case_node): Moved from
13365 stmt.c and adjusted.
13366 (emit_case_nodes): Likewise.
13367 (node_has_low_bound): Likewise.
13368 (node_has_high_bound): Likewise.
13369 (node_is_bounded): Likewise.
13370 (case_values_threshold): Likewise.
13371 (reset_out_edges_aux): Likewise.
13372 (compute_cases_per_edge): Likewise.
13373 (add_case_node): Likewise.
13374 (dump_case_nodes): Likewise.
13375 (balance_case_nodes): Likewise.
13376 (expand_switch_as_decision_tree_p): Likewise.
13377 (emit_jump): Likewise.
13378 (emit_case_decision_tree): Likewise.
13379 (try_switch_expansion): Likewise.
13380 (do_jump_if_equal): Likewise.
13381 (emit_cmp_and_jump_insns): Likewise.
13382 (fix_phi_operands_for_edge): New function.
13383 (record_phi_operand_mapping): Likewise.
13384 (class pass_lower_switch): New pass.
13385 (pass_lower_switch::execute): New function.
13386 (make_pass_lower_switch): Likewise.
13387 (conditional_probability):
13388 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13389 * tree-pass.h: Add make_pass_lower_switch.
13391 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13394 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13397 2017-08-29 Richard Biener <rguenther@suse.de>
13399 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13400 we do not add a DW_AT_inline attribute twice.
13401 (gen_subprogram_die): Remove code setting DW_AT_inline on
13402 DECL_ABSTRACT_P nodes.
13404 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13406 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13407 calls to internal functions.
13408 (gimplify_modify_expr): Likewise.
13409 * tree-call-cdce.c (use_internal_fn): Likewise.
13410 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13411 (convert_to_divmod): Set the nothrow flag.
13412 * tree-if-conv.c (predicate_mem_writes): Likewise.
13413 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13414 (vectorizable_call): Likewise.
13415 (vectorizable_store): Likewise.
13416 (vectorizable_load): Likewise.
13417 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13418 (vect_recog_mask_conversion_pattern): Likewise.
13420 2017-08-29 Martin Liska <mliska@suse.cz>
13423 * gcc.c (driver_handle_option): Add new argument.
13424 * opts-common.c (handle_option): Pass
13425 target_option_override_hook.
13426 * opts-global.c (lang_handle_option): Add new option.
13427 (set_default_handlers): Add new argument.
13428 (decode_options): Likewise.
13429 * opts.c (target_handle_option): Likewise.
13430 (common_handle_option): Call target_option_override_hook.
13431 * opts.h (struct cl_option_handler_func): Add hook for
13432 target option override.
13433 (struct cl_option_handlers): Likewise.
13434 (set_default_handlers): Add new argument.
13435 (decode_options): Likewise.
13436 (common_handle_option): Likewise.
13437 (target_handle_option): Likewise.
13438 * toplev.c (toplev::main): Pass targetm.target_option.override
13441 2017-08-29 Richard Biener <rguenther@suse.de>
13442 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13444 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13445 life to the active subtree.
13447 2017-08-28 Jeff Law <law@redhat.com>
13449 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13450 derive_equivalences.
13451 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13452 Code moved into....
13453 (edge_info::derive_equivalences): New private member function
13455 * tree-ssa-dom.c (class edge_info): Changed from a struct
13456 to a class. Add ctor/dtor, methods and data members.
13457 (edge_info::edge_info): Renamed from allocate_edge_info.
13458 Initialize additional members.
13459 (edge_info::~edge_info): New.
13460 (free_dom_edge_info): Delete the edge info.
13461 (record_edge_info): Use new class & associated member functions.
13462 Tighten forms for testing for edge equivalences.
13463 (record_temporary_equivalences): Iterate over the simple
13464 equivalences rather than assuming there's only one per edge.
13465 (cprop_into_successor_phis): Iterate over the simple
13466 equivalences rather than assuming there's only one per edge.
13467 (optimize_stmt): Use operand_equal_p rather than pointer
13468 equality for mini-DSE code.
13470 2017-08-28 Nathan Sidwell <nathan@acm.org>
13472 * gcc.c (execute): Fold SIGPIPE handling into switch
13473 statement. Adjust internal error message.
13475 2017-08-28 Richard Biener <rguenther@suse.de>
13478 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13479 Do nothing for removed DIEs.
13481 2017-08-28 Richard Biener <rguenther@suse.de>
13483 PR tree-optimization/81977
13484 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13487 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13490 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13491 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13492 using targetm.gen_mem_thread_fence.
13494 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13497 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13498 predicate to register_operand. Reorder operands.
13499 (*btr<mode>): Ditto.
13500 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13501 (*btr<mode>_mask): Ditto.
13503 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13505 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13506 * config/rs6000/xmmintrin.h: New file.
13507 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13509 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13512 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13513 parameter and_insn and return it.
13514 (recombine_lvx_pattern): Insert a copy to ensure availability of
13515 the base register of the copied masking operation at the point of
13516 the instruction replacement.
13517 (recombine_stvx_pattern): Likewise.
13519 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13521 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13522 undocumented switches.
13523 (-mpower9-dform-vector): Likewise.
13524 (-mpower9-dform): Likewise.
13525 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13526 comments to delete references to -mpower9-dform* switches.
13527 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13528 Delete reference to -mpower9-dform* switches, test for
13529 -mpower9-vector instead.
13530 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13531 (OTHER_P9_VECTOR_MASKS): Likewise.
13532 (POWERPC_MASKS): Likewise.
13533 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13534 tests against -mpower9-dform* to -mpower9-vector. Delete code
13535 that checked for -mpower9-dform* consistancy with other options.
13536 Add test for -mpower9-misc to enable other power9 switches.
13537 (rs6000_init_hard_regno_mode_ok): Likewise.
13538 (rs6000_option_override_internal): Likewise.
13539 (rs6000_emit_prologue): Likewise.
13540 (rs6000_emit_epilogue): Likewise.
13541 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13542 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13543 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13545 (emit_fusion_p9_store): Likewise.
13546 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13547 resetting these macros if the assembler does not support ISA 3.0
13549 (TARGET_P9_DFORM_VECTOR): Likewise.
13550 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13551 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13553 2017-08-25 Alan Modra <amodra@gmail.com>
13555 PR rtl-optimization/81747
13556 * cse.c (cse_extended_basic_block): Don't attempt to record
13557 equivalences for degenerate conditional branches that branch
13558 to their fall-through.
13560 2017-08-24 Martin Sebor <msebor@redhat.com>
13562 PR middle-end/81908
13563 * gimple-fold.c (size_must_be_zero_p): New function.
13564 (gimple_fold_builtin_memory_op): Call it.
13566 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13568 * config/rs6000/mm_malloc.h: New file.
13570 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13572 PR tree-optimization/81913
13573 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13574 analysis when either IVs in condition can wrap.
13576 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13578 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13579 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13581 2017-08-24 Richard Biener <rguenther@suse.de>
13584 * targhooks.c (default_target_can_inline_p): Properly
13585 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13586 is present and always compare.
13587 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13588 infer -mfpmath=sse from TARGET_SSE_P.
13589 (ix86_can_inline_p): Properly use target_option_default_node when
13590 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13592 2017-08-24 Richard Biener <rguenther@suse.de>
13595 * dwarf2out.c (output_die): Handle flag_generate_offload like
13597 (output_comp_unit): Likewise.
13598 (gen_array_type_die): Likewise.
13599 (dwarf2out_early_finish): Likewise.
13600 (note_variable_value_in_expr): Likewise.
13601 (dwarf2out_finish): Likewise. Adjust assert.
13602 * cgraphunit.c (symbol_table::compile): Move setting of
13603 flag_generate_offload earlier ...
13604 (symbol_table::finalize_compilation_unit): ... here, before
13605 early debug finalization.
13607 2017-08-24 Richard Biener <rguenther@suse.de>
13609 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13610 and ipa-fnsummary.h.
13611 (ix86_can_inline_p): When ix86_fpmath flags do not match
13612 check whether the callee uses FP math at all.
13614 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13616 PR middle-end/81931
13617 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13618 instead of TYPE_PRECISION.
13620 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13621 Alan Hayward <alan.hayward@arm.com>
13622 David Sherwood <david.sherwood@arm.com>
13624 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13626 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13628 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13630 (ix86_frame::stack_realign_allocate): New field.
13631 (struct machine_frame_state): Modify comments.
13632 (machine_frame_state::sp_realigned_fp_end): New field.
13633 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13634 layout calculation.
13635 (sp_valid_at): Add assertion to assure no attempt to access invalid
13636 offset of a realigned stack.
13637 (fp_valid_at): Likewise.
13638 (choose_baseaddr): Modify comments.
13639 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13640 ix86_expand_prologue.
13641 (ix86_expand_prologue): Modify stack realignment and allocation.
13642 (ix86_expand_epilogue): Modify comments.
13643 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13644 avx2_runtime, avx512f, and avx512f_runtime.
13646 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13648 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13649 (mstackrealign): Do not init to -1.
13650 * config/i386/i386.c (ix86_option_override_internal):
13651 Check opts_set, not opts when setting default value of
13652 opts->x_ix86_force_align_arg_pointer.
13654 2017-08-23 Richard Biener <rguenther@suse.de>
13656 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13657 lang_hooks.decl_printable_name.
13658 * print-rtl-function.c (print_rtx_function): Likewise.
13659 * tree-pretty-print.c (dump_function_header): Likewise.
13661 2017-08-23 Richard Biener <rguenther@suse.de>
13664 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13665 -g0 at compile-time.
13667 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13669 PR middle-end/19706
13670 * doc/sourcebuild.texi (Other hardware attributes):
13673 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13675 PR middle-end/19706
13676 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13677 Fix single-use check.
13679 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13681 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13683 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13685 * doc/install.texi: Modify to add more details on running selected
13688 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13690 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13691 is combined with -mabi=ms.
13692 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13695 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13697 PR tree-optimization/81488
13698 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13699 and cached_basis fields.
13700 (MAX_SPREAD): New constant.
13701 (alloc_cand_and_find_basis): Initialize new fields.
13702 (clear_visited): New function.
13703 (create_phi_basis_1): Rename from create_phi_basis, set visited
13704 and cached_basis fields.
13705 (create_phi_basis): New wrapper function.
13706 (phi_add_costs_1): Rename from phi_add_costs, add spread
13707 parameter, set visited field, short-circuit when limits reached.
13708 (phi_add_costs): New wrapper function.
13709 (record_phi_increments_1): Rename from record_phi_increments, set
13711 (record_phi_increments): New wrapper function.
13712 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13713 (phi_incr_cost): New wrapper function.
13714 (all_phi_incrs_profitable_1): Rename from
13715 all_phi_incrs_profitable, set visited field.
13716 (all_phi_incrs_profitable): New wrapper function.
13718 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13719 Alan Hayward <alan.hayward@arm.com>
13720 David Sherwood <david.sherwood@arm.com>
13722 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13723 that takes the outer and inner modes.
13724 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13725 comparison as the canonical test for a paradoxical subreg.
13726 * combine.c (simplify_set): Use paradoxical_subreg_p.
13727 (make_extraction): Likewise.
13728 (force_to_mode): Likewise.
13729 (rtx_equal_for_field_assignment_p): Likewise.
13730 (gen_lowpart_for_combine): Likewise.
13731 (simplify_comparison): Likewise.
13732 * cse.c (equiv_constant): Likewise.
13733 * expmed.c (store_bit_field_1): Likewise.
13734 * final.c (alter_subreg): Likewise.
13735 * fwprop.c (propagate_rtx): Likewise.
13736 (forward_propagate_subreg): Likewise.
13737 * ira-conflicts.c (ira_build_conflicts): Likewise.
13738 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13739 * lra-constraints.c (curr_insn_transform): Likewise.
13740 (split_reg): Likewise.
13741 * lra-eliminations.c (move_plus_up): Likewise.
13742 (lra_eliminate_regs_1): Likewise.
13743 * recog.c (general_operand): Likewise.
13744 * ree.c (combine_reaching_defs): Likewise.
13745 * reload.c (push_reload): Likewise.
13746 (find_reloads): Likewise.
13747 * reload1.c (elimination_effects): Likewise.
13748 (compute_reload_subreg_offset): Likewise.
13749 (choose_reload_regs): Likewise.
13750 * rtlanal.c (subreg_lsb_1): Likewise.
13751 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13752 (simplify_subreg): Likewise.
13753 * var-tracking.c (track_loc_p): Likewise.
13754 * emit-rtl.c (byte_lowpart_offset): Likewise.
13755 (paradoxical_subreg_p): Delete out-of-line definition.
13757 2017-08-22 Jeff Law <law@redhat.com>
13759 PR tree-optimization/81741
13760 PR tree-optimization/71947
13761 * tree-ssa-dom.c: Include tree-inline.h.
13762 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13763 equivalences if one is more expensive to compute than the other.
13764 * tree-ssa-scopedtables.h (class const_or_copies): Make
13765 record_const_or_copy_raw method private.
13766 (class avail_exprs_stack): New method simplify_binary_operation.
13767 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13768 avail_exprs_stack::simplify_binary_operation as needed.
13769 (avail_exprs_stack::simplify_binary_operation): New function.
13771 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13773 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13774 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13775 (DOT_SYMBOLS): Likewise.
13776 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13777 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13778 (RS6000_ABI_NAME): Likewise.
13779 (TARGET_CMODEL): Likewise.
13780 (TOC_SECTION_ASM_OP): Likewise.
13781 (SET_CMODEL): New macro.
13782 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13784 2017-08-22 Richard Biener <rguenther@suse.de>
13786 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13787 to allow for free-lang-data replacements similar to verify_type_variant.
13789 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13790 Alan Hayward <alan.hayward@arm.com>
13791 David Sherwood <david.sherwood@arm.com>
13793 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13794 VOIDmode for the LABEL_REF.
13796 2017-08-22 Richard Biener <rguenther@suse.de>
13798 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13800 2017-08-22 Richard Biener <rguenther@suse.de>
13802 * common.opt (feliminate-dwarf2-dups): Ignore.
13803 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13804 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13805 same_die_p_wrap, compute_section_prefix,
13806 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13807 (comdat_symbol_id, comdat_symbol_number): Likewise.
13808 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13810 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13811 (output_die): Mark unreachable path unreachable.
13812 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13813 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13814 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13815 (dwarf2out_early_finish): Likewise.
13817 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13819 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13821 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13824 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13825 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13826 (avr_attribute_table) <io, io_low, address>: Initialize
13827 .decl_required with true.
13829 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13831 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13832 undocumented debugging options.
13833 (-mvsx-scalar-double): Likewise.
13834 (-mallow-df-permute): Likewise.
13835 (-mvectorize-builtins): Likewise.
13836 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13837 (rs6000_builtin_vectorized_function): Likewise.
13838 (rs6000_builtin_md_vectorized_function): Likewise.
13839 (rs6000_opt_vars): Likewise.
13841 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13844 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13845 (*btrq_imm): Rename from *btrq.
13846 (*btcq_imm): Rename from *btcq.
13847 (btsc): New code attribute.
13848 (*<btsc><mode>): New insn pattern.
13849 (*btr<mode>): Ditto.
13850 (*<btsc><mode>_mask): New insn_and_split pattern.
13851 (*btr<mode>_mask): Ditto.
13853 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13854 Alan Hayward <alan.hayward@arm.com>
13855 David Sherwood <david.sherwood@arm.com>
13857 * function.c (pad_below): Simplify padding calculation.
13859 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13860 Alan Hayward <alan.hayward@arm.com>
13861 David Sherwood <david.sherwood@arm.com>
13863 * target.def (function_prologue): Remove frame size argument.
13864 (function_epilogue): Likewise.
13865 * doc/tm.texi: Regenerate.
13866 * final.c (final_start_function): Update call to function_prologue.
13867 (final_end_function): Update call to function_epilogue.
13868 (default_function_pro_epilogue): Remove frame size argument.
13869 * output.h (default_function_pro_epilogue): Likewise.
13870 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13871 (arm_output_function_prologue): Likewise.
13872 * config/frv/frv.c (frv_function_prologue): Likewise.
13873 (frv_function_epilogue): Likewise.
13874 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13875 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13876 (ia64_output_function_epilogue): Likewise.
13877 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13878 (m32r_output_function_epilogue): Likewise.
13879 * config/microblaze/microblaze.c (microblaze_function_prologue)
13880 (microblaze_function_epilogue): Likewise.
13881 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13882 (mips_output_function_epilogue): Likewise.
13883 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13884 (mmix_target_asm_function_epilogue): Likewise.
13885 * config/msp430/msp430.c (msp430_start_function): Likewise.
13886 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13887 (nds32_asm_function_epilogue): Likewise.
13888 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13889 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13890 (pa_output_function_epilogue): Likewise.
13891 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13892 (rs6000_output_function_epilogue): Likewise.
13893 * config/rl78/rl78.c (rl78_start_function): Likewise.
13894 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13895 (rs6000_output_function_epilogue): Likewise.
13896 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13897 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13898 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13899 (sparc_asm_function_epilogue): Likewise.
13901 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13903 * tree.h (type_has_mode_precision_p): New function.
13904 * convert.c (convert_to_integer_1): Use it.
13905 * expr.c (expand_expr_real_2): Likewise.
13906 (expand_expr_real_1): Likewise.
13907 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13908 * match.pd: Likewise.
13909 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13910 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13911 * tree-tailcall.c (process_assignment): Likewise.
13912 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13913 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13914 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13915 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13916 (vectorizable_assignment): Likewise.
13917 (vectorizable_shift): Likewise.
13918 (vectorizable_operation): Likewise.
13919 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13921 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13923 * match.pd: Add pow (C, x) simplification.
13925 2017-08-21 Richard Biener <rguenther@suse.de>
13927 PR tree-optimization/81900
13928 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13929 for blocks with abnormal predecessors.
13930 (compute_antic): Do not set visited flag prematurely.
13932 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13935 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13937 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13939 * stor-layout.h (vector_type_mode): Move to...
13940 * tree.h (vector_type_mode): ...here.
13941 * stor-layout.c (vector_type_mode): Move to...
13942 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13944 2017-08-21 Richard Biener <rguenther@suse.de>
13946 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13947 register_external_die hooks.
13948 (debug_false_tree_charstarstar_uhwistar): Declare.
13949 (debug_nothing_tree_charstar_uhwi): Likewise.
13950 * debug.c (do_nothing_debug_hooks): Adjust.
13951 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13952 (debug_nothing_tree_charstar_uhwi): Likewise.
13953 * dbxout.c (dbx_debug_hooks): Adjust.
13954 (xcoff_debug_hooks): Likewise.
13955 * sdbout.c (sdb_debug_hooks): Likewise.
13956 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13957 * dwarf2out.c (macinfo_label_base): New global.
13958 (dwarf2out_register_external_die): New function for the
13959 register_external_die hook.
13960 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13961 (dwarf2_debug_hooks): Use them.
13962 (dwarf2_lineno_debug_hooks): Adjust.
13963 (struct die_struct): Add with_offset flag.
13964 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13965 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13966 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13967 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13968 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13969 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13970 defining section names for the early LTO debug variants.
13971 (reset_indirect_string): New helper.
13972 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13973 (print_dw_val): Add support for offsetted symbol references.
13974 (get_ultimate_context): Split out from is_cxx.
13975 (is_cxx): Use get_ultimate_context.
13976 (is_fortran): Add decl overload.
13977 (compute_comp_unit_symbol): Split out worker from
13978 compute_section_prefix.
13979 (compute_section_prefix): Call compute_comp_unit_symbol and
13980 set comdat_type_p here.
13981 (output_die): Skip DIE symbol output for the LTO added one.
13982 Handle DIE symbol references with offset.
13983 (output_comp_unit): Guard section name mangling properly.
13984 For LTO debug sections emit a symbol at the section beginning
13985 which we use to refer to its DIEs.
13986 (add_abstract_origin_attribute): For DIEs registered via
13987 dwarf2out_register_external_die directly refer to the early
13988 DIE rather than indirectly through the shadow one we created.
13989 Remove obsolete call to dwarf2out_abstract_function for
13990 non-function/block origins.
13991 (gen_array_type_die): When generating early LTO debug do
13992 not emit DW_AT_string_length.
13993 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13994 late when in LTO. As suggested place a gcc_unreachable for
13995 the DECL_ABSTRACT_P case.
13996 (gen_subprogram_die): Avoid another specification DIE
13997 for early built declarations/definitions for the late LTO case.
13998 (gen_variable_die): Add type references for late duplicated VLA dies
14000 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14001 we have the abstract instance already.
14002 (process_scope_var): Adjust decl DIE contexts in LTO which
14003 first puts them in limbo.
14004 (gen_decl_die): Do not generate type DIEs late apart from
14005 types for VLAs or for decls we do not yet have a DIE. Do not
14006 call dwarf2out_abstract_function late.
14007 (dwarf2out_early_global_decl): Make sure to create DIEs
14008 for abstract instances of a decl first.
14009 (dwarf2out_late_global_decl): Adjust comment.
14010 (output_macinfo_op): With multiple macro sections use
14011 macinfo_label_base to distinguish labels.
14012 (output_macinfo): Likewise. Update macinfo_label_base.
14013 Pass in the line info label.
14014 (note_variable_value_in_expr): When generating LTO resolve
14015 all variable values here by generating DIEs as needed.
14016 (init_sections_and_labels): Add early LTO debug flag parameter
14017 and generate different sections and names if set. Add generation
14018 counter for the labels so we can have multiple of them.
14019 (reset_dies): Helper to allow DIEs to be output multiple times.
14020 (dwarf2out_finish): When outputting DIEs to the fat part of an
14021 LTO object first reset DIEs.
14022 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14023 (modified_type_die): Check for decl_ultimate_origin being self
14025 (gen_type_die_with_usage): Likewise.
14026 (gen_typedef_die): Allow decl_ultimate_origin being self.
14027 (set_decl_abstract_flags): Remove.
14028 (set_block_abstract_flags): Likewise.
14029 (dwarf2out_abstract_function): Treat the early generated DIEs
14030 as the abstract copy and only add DW_AT_inline and
14031 DW_AT_artificial here and call set_decl_origin_self.
14032 If the DIE has an abstract origin don't do anything.
14033 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14034 if we have none yet (Go fails to build one, PR78628).
14035 (variably_modified_type_p): Prevent endless recursion for Ada
14036 cyclic pointer types.
14037 * lto-streamer-in.c: Include debug.h.
14038 (dref_queue): New global.
14039 (lto_read_tree_1): Stream in DIE references.
14040 (lto_input_tree): Register DIE references.
14041 (input_function): Stream DECL_DEBUG_ARGS.
14042 * lto-streamer-out.c: Include debug.h.
14043 (lto_write_tree_1): Output DIE references.
14044 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14045 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14046 (output_function): Stream DECL_DEBUG_ARGS.
14047 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14048 Stream DECL_ABSTRACT_ORIGIN.
14049 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14050 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14051 DECL_CONTEXT for file-scope decls.
14052 * lto-streamer.h (struct dref_entry): Declare.
14053 (dref_queue): Likewise.
14054 * cfgexpand.c (pass_expand::execute): Do not call the
14055 outlining_inline_function hook here.
14056 * lto-wrapper.c (debug_obj): New global.
14057 (tool_cleanup): Unlink it if required.
14058 (debug_objcopy): New function.
14059 (run_gcc): Handle early debug sections in the IL files by
14060 extracting them to separate files, partially linkin them and
14061 feeding the result back as result to the linker.
14062 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14063 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14064 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14065 sections into a separate segment.
14066 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14068 (darwin_asm_dwarf_section): Likewise.
14069 (darwin_asm_output_dwarf_offset): Likewise.
14070 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14072 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14073 Alan Hayward <alan.hayward@arm.com>
14074 David Sherwood <david.sherwood@arm.com>
14076 * read-md.h (md_reader::record_potential_iterator_use): Replace
14077 pointer argument with an rtx and an index.
14078 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14079 (apply_mode_iterator): Likewise.
14080 (apply_code_iterator): Likewise.
14081 (apply_int_iterator): Likewise.
14082 (apply_subst_iterator): Likewise.
14083 (record_iterator_use): Likewise.
14084 (record_attribute_use): Likewise.
14085 (md_reader::record_potential_iterator_use): Likewise. Update calls
14086 to record_iterator_use and apply_iterator.
14087 (iterator_use): Replace ptr with x and index.
14088 (attribute_use): Likewise.
14089 (apply_attribute_uses): Update calls to apply_iterator.
14090 (apply_iterators): Likewise. Update initialization of iterator_use.
14091 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14092 and record_potential_iterator_use.
14093 (rtx_reader::read_rtx_operand): Likewise.
14095 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14096 Alan Hayward <alan.hayward@arm.com>
14097 David Sherwood <david.sherwood@arm.com>
14099 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14102 2017-08-21 Richard Biener <rguenther@suse.de>
14104 PR middle-end/81884
14105 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14106 at struct end conservatively when comparing common bases.
14108 2017-08-21 Richard Biener <rguenther@suse.de>
14110 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14111 (mem_ref_in_stmt): Remove.
14112 (determine_max_movement): Use ref index to get at the reference.
14113 (invariantness_dom_walker::before_dom_children): Deal with
14114 lim data already initialized.
14115 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14117 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14119 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14120 (TARGET_ISA_ROUND): Ditto.
14121 (TARGET_ROUND): Ditto.
14122 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14123 * config/i386/i386.md: Ditto.
14124 * config/i386/sse.md: Ditto.
14125 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14126 with OPTION_MASK_ISA_SSE4_1.
14128 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14131 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14132 __builtin_ia32_lzcnt_u16.
14134 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14137 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14138 (rs6000_set_current_function): Rewrite function to use it.
14140 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14143 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14144 and TYPE_WARN_IF_NOT_ALIGN.
14145 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14146 (handle_warn_if_not_align): New.
14147 (place_union_field): Call handle_warn_if_not_align.
14148 (place_field): Call handle_warn_if_not_align.
14149 Copy TYPE_WARN_IF_NOT_ALIGN.
14150 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14151 (layout_type): Likewise.
14152 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14154 (tree_decl_common): Add warn_if_not_align.
14155 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14156 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14157 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14158 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14159 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14160 * doc/extend.texi: Document warn_if_not_aligned attribute.
14161 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14163 2017-08-17 Martin Liska <mliska@suse.cz>
14166 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14167 (get_data_dependence): Use it as pointer type.
14168 (distribute_loop): Likewise.
14170 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14172 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14173 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14174 (p8_vmrgow_<mode>_direct): New define_insn.
14175 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14176 handle endianness for vmrgew and vmrgow permute patterns.
14178 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14180 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14181 * config/rs6000/rs6000-cpus.def: Remove comment.
14182 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14183 (POWERPC_MASKS): Likewise.
14184 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14185 use of TARGET_VSX_TIMODE.
14186 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14187 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14188 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14189 (rs6000_option_override_internal): Remove dead code.
14190 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14191 (rs6000_legitimize_reload_address): Likewise.
14192 (rs6000_legitimate_address_p): Likewise.
14193 (rs6000_opt_masks): Delete "vsx-timode".
14194 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14195 from function comment.
14196 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14197 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14198 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14200 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14201 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14202 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14203 (VSX_TI): Likewise.
14205 (define_peephole2): Likewise.
14207 2017-08-17 Martin Sebor <msebor@redhat.com>
14210 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14211 past the end of an array.
14212 (test_pp_format): Add test cases.
14214 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14216 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14217 missing ECF_NOTHROW flags.
14219 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14222 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14223 operands residing in integer registers.
14224 (*vsx_le_perm_load_<mode>): Likewise.
14225 (*vsx_le_perm_store_<mode>): Likewise.
14226 (define_peephole2): Add peepholes to optimize the above.
14228 2017-08-17 Marek Polacek <polacek@redhat.com>
14230 PR middle-end/81814
14231 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14232 to mimic what shorten_compare did. Change the return type to bool.
14233 (fold_cond_expr_with_comparison): Update call to
14234 operand_equal_for_comparison_p.
14235 (fold_ternary_loc): Likewise.
14237 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14239 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14241 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14243 2017-08-17 Richard Biener <rguenther@suse.de>
14245 * tree-ssa-structalias.c (solve_graph): When propagating
14246 to successors update the graphs succ edges and avoid duplicate work.
14248 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14251 * config/i386/i386.c (ix86_option_override_internal): Save target
14252 specific options after ix86_stack_protector_guard_reg was changed.
14254 2017-08-17 Richard Biener <rguenther@suse.de>
14256 PR tree-optimization/81827
14257 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14258 (new_var_info): Initialize it conservatively.
14259 (get_call_vi): Mark register vars.
14260 (new_scalar_tmp_constraint_exp): Likewise.
14261 (handle_rhs_call): Likewise.
14262 (handle_const_call): Likewise.
14263 (create_function_info_for): Likewise.
14264 (solve_constraints): Sort varinfos to separate register from
14265 non-register vars to pack points-to solution bitmaps during
14268 2017-08-17 Marek Polacek <polacek@redhat.com>
14270 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14272 2017-08-17 Richard Biener <rguenther@suse.de>
14274 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14275 to true when overflow is undefined and we saturated the result.
14277 2017-08-17 Alan Modra <amodra@gmail.com>
14280 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14281 Don't use store multiple if only one reg needs saving.
14282 (interesting_frame_related_regno): New function.
14283 (rs6000_frame_related): Don't emit frame info for regs that
14285 (rs6000_emit_epilogue): Likewise.
14287 2017-08-16 Nathan Sidwell <nathan@acm.org>
14289 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14290 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14291 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14292 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14293 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14294 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14295 (verify_type): Adjust for TYPE_BINFO move.
14296 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14297 process TYPE_BINFO directly.
14298 (hash_tree): Likewise.
14299 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14301 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14304 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14306 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14308 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14311 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14312 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14313 Add ix86_binary_operator_ok to insn constraint.
14314 (*iordi_1_bts): Ditto.
14315 (*xordi_1_btc): Ditto.
14316 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14317 Update corresponding peephole2 pattern.
14321 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14323 PR tree-optimization/81832
14324 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14325 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14327 2017-08-16 Marek Polacek <polacek@redhat.com>
14330 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14331 regarding size_zero_node.
14333 2017-08-16 Martin Liska <mliska@suse.cz>
14336 * config.gcc: Respect previously set extra_objs in case
14339 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14341 PR tree-optimization/81835
14342 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14343 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14344 not depend on the phi.
14346 2017-08-16 Alan Modra <amodra@gmail.com>
14348 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14351 2017-08-16 Alan Modra <amodra@gmail.com>
14353 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14354 (save_reg_p): ..into this. Update all callers.
14355 (first_reg_to_save): Simplify.
14357 2017-08-16 Alan Modra <amodra@gmail.com>
14359 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14362 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14366 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14367 constant count if that count is less than 32.
14369 2017-08-15 Nathan Sidwell <nathan@acm.org>
14371 * gcc.c (execute): Emit friendlier message if inferior is killed
14372 by an external cause.
14374 2017-08-15 Richard Biener <rguenther@suse.de>
14376 PR tree-optimization/81790
14377 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14378 CONSTRUCTORs from simplifying and VN.
14380 2017-08-14 Martin Sebor <msebor@redhat.com>
14382 * builtin-attrs.def: Add comments.
14384 2017-08-14 Martin Sebor <msebor@redhat.com>
14387 * doc/extend.texi (attribute nonstring): Document new attribute.
14389 2017-08-14 Martin Sebor <msebor@redhat.com>
14392 * tree-diagnostic.c (default_tree_printer): Handle %G.
14393 * gimple-pretty-print.h (percent_G_format): Declare new function.
14394 * gimple-pretty-print.c (percent_G_format): Define.
14395 * tree-pretty-print.c (percent_K_format): Add argument.
14397 2017-08-14 Martin Sebor <msebor@redhat.com>
14399 PR translation/79998
14400 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14401 Remove a stray space.
14403 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14406 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14407 (*iordi_1_bts): Ditto.
14408 (*xordi_1_btc): Ditto.
14410 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14413 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14414 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14416 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14417 quoted strings, and make more translator-friendly.
14418 (darwin_rs6000_override_options): Likewise.
14419 (rs6000_option_override_internal): Likewise.
14420 (rs6000_return_in_memory): Fix overlong line.
14421 (init_cmulative_args): Use quoted strings, and make more
14422 translator-friendly.
14423 (rs6000_pass_by_reference): Fix overlong line.
14424 (def_builtin): Use quoted strings.
14425 (altivec_expand_predicate_builtin): Use quoted strings, and make
14426 more translator-friendly.
14427 (htm_expand_builtin): Use quoted strings.
14428 (cpu_expand_builtin): Use quoted strings, and make more
14429 translator-friendly.
14430 (altivec_expand_builtin): Likewise.
14431 (paired_expand_predicate_builtin): Likewise.
14432 (rs6000_invalid_builtin): Likewise.
14433 (builtin_function_type): Use quoted strings.
14434 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14435 more translator-friendly.
14436 (rs6000_trampoline_init): Likewise.
14437 (rs6000_handle_altivec_attribute): Likewise.
14438 (rs6000_inner_target_options): Use quoted strings.
14439 (rs6000_disable_incompatible_switches): Likewise.
14440 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14441 strings, and make more translator-friendly.
14442 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14444 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14446 PR tree-optimization/81799
14447 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14448 cond_expr to simple gimple operand.
14450 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14452 PR middle-end/46932
14453 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14455 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14459 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14460 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14461 TARGET_GASISR_PROLOGUES.
14462 * config/avr/avr.c (avr_option_override): Same.
14463 (avr_pass_pre_proep::execute): Same.
14465 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14468 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14469 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14471 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14473 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14474 define_insn_and_split. Split to a memory load from 0 in
14475 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14476 using PTR mode iterator.
14477 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14478 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14479 (*add_tp_<mode>): Redefine as define_insn_and_split.
14480 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14481 address space. Merge with *add_tp_x32 using PTR mode iterator.
14482 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14483 Split to an add with a memory load from 0 in
14484 DEFAULT_TLS_SEG_REG address space.
14486 2017-08-12 Andrew Pinski <apinski@cavium.com>
14488 * config/aarch64/aarch64-option-extensions.def (rdma):
14489 Fix feature string to what Linux prints out in /proc/cpuinfo.
14491 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14494 * dwarf2out.c (modified_type_die): For C typedef types that have
14495 an ultimate origin, process the ultimate origin instead of the
14497 (gen_typedef_die): Assert that input DECLs have no ultimate
14499 (gen_type_die_with_usage): For typedef variants that have an
14500 ultimate origin, just call gen_decl_die on the original DECL.
14501 (process_scope_var): Avoid creating DIEs for local typedefs and
14502 concrete static variables.
14504 2017-08-12 Alan Modra <amodra@gmail.com>
14508 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14509 match gnu-user.h startfile.
14510 (ENDFILE_LINUX_SPEC): Similarly.
14512 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14515 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14517 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14519 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14520 * config/aarch64/aarch64.md (mov<mode>): Change.
14521 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14522 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14523 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14525 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14527 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14528 for storage order barriers.
14530 2017-08-11 Martin Liska <mliska@suse.cz>
14533 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14534 variables of void type.
14536 2017-08-11 Martin Liska <mliska@suse.cz>
14538 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14539 TARGET_SUPPORTS_ALIASES.
14540 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14541 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14542 (optimize_weakref): Likewise.
14543 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14544 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14545 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14547 2017-08-11 Martin Liska <mliska@suse.cz>
14550 * config/i386/i386.c (make_resolver_func): Do complete
14551 refactoring of the function.
14553 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14556 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14557 * config/i386/i386.c (ix86_stack_protect_guard): Use
14558 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14559 * doc/invoke.texi (x86 Options): Document
14560 -mstack-protector-guard-symbol= option.
14562 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14564 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14565 * config/i386/i386.c (ix86_split_stack_guard): New function.
14566 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14567 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14568 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14569 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14570 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14571 (split_stack_space_check): Call ix86_split_stack_guard.
14573 2017-08-10 Martin Sebor <msebor@redhat.com>
14575 * print-tree.c (print_node): Print location using the established
14577 Replace spaces with colons.
14578 (debug_raw, debug): Ditto.
14580 2017-08-10 Martin Sebor <msebor@redhat.com>
14583 * pretty-print.c (pp_format): Correct the handling of %s precision.
14585 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14588 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14590 (ix86_finalize_stack_frame_flags): This. Also clear
14591 frame_pointer_needed if -fno-omit-frame-pointer is used without
14593 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14594 with ix86_finalize_stack_frame_flags.
14595 (ix86_expand_epilogue): Likewise.
14596 (ix86_expand_split_stack_prologue): Likewise.
14597 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14599 2017-08-10 Martin Liska <mliska@suse.cz>
14602 * c-attribs.c (handle_target_attribute):
14603 Report warning for an empty string argument of target attribute.
14605 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14608 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14610 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14611 or DECL_NONLOCAL labels.
14612 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14613 or DECL_NONLOCAL labels here.
14615 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14617 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14618 to indicate when early gimple folding has been disabled.
14619 (rs6000_gimple_fold_builtin): Add debug content.
14620 (rs6000_invalid_builtin): Fix whitespace.
14621 (rs6000_expand_builtin): Fix whitespace.
14622 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14624 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14627 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14628 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14630 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14632 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14633 (qdf24xx): Likewise.
14634 * config/aarch64/aarch64.md: Include falkor.md.
14635 * config/aarch64/falkor.md: New.
14637 2017-08-09 Marek Polacek <polacek@redhat.com>
14640 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14641 * diagnostic.c (emit_diagnostic): Add a comment.
14642 (emit_diagnostic_valist): New function.
14644 2017-08-09 Marek Polacek <polacek@redhat.com>
14647 * input.c (make_location): New overload.
14648 * input.h (make_location): Declare.
14650 2017-08-08 Alan Modra <amodra@gmail.com>
14651 H.J. Lu <hongjiu.lu@intel.com>
14654 * gcc.c (NO_PIE_SPEC): Delete.
14655 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14657 (LINK_PIE_SPEC): ..to here.
14658 (LINK_COMMAND_SPEC): Support -no-pie.
14659 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14660 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14662 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14663 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14664 (ENDFILE_CRTEND_SPEC): Similarly.
14666 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14669 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14670 (mstack-protector-guard-offset=): Ditto.
14671 * config/i386/i386.c (ix86_option_override): Handle
14672 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14674 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14675 ix86_stack_protect_guard_offset variables.
14676 (TARGET_STACK_PROTECT_GUARD): Always define.
14677 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14678 and -mstack-protector-guard-offset= options.
14680 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14682 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14683 boundary case for the last candidate.
14685 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14687 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14688 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14690 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14692 PR middle-end/19706
14693 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14694 * config/aarch64/aarch64-builtins.c
14695 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14696 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14697 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14699 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14700 Andrew Pinski <pinskia@gmail.com>
14702 PR middle-end/19706
14703 * internal-fn.def (XORSIGN): New.
14704 * optabs.def (xorsign_optab): New.
14705 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14706 (convert_expand_mult_copysign): New.
14707 (pass_optimize_widening_mul::execute): Call
14708 convert_expand_mult_copysign.
14710 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14712 PR tree-optimization/81354
14713 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14714 Insert on edges rather than explicitly creating landing pads.
14715 (analyze_candidates_and_replace): Commit edge inserts.
14717 2017-08-08 Richard Biener <rguenther@suse.de>
14719 PR middle-end/81719
14720 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14721 (expand_simple_operations): Also look through ADDR_EXPRs with
14722 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14724 2017-08-08 Richard Biener <rguenther@suse.de>
14726 PR tree-optimization/81723
14727 * tree-vect-slp.c (struct bst_traits): New hash traits.
14728 (bst_fail): New global.
14729 (vect_build_slp_tree_2): New worker, split out from ...
14730 (vect_build_slp_tree): ... this now wrapping it with using
14731 bst_fail set to cache SLP tree build fails. Properly handle
14733 (vect_analyze_slp_instance): Allocate and free bst_fail.
14735 2017-08-08 Martin Liska <mliska@suse.cz>
14738 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14739 LABEL_DECLs that can be from a different function.
14741 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14743 PR tree-optimization/81744
14744 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14745 loop's number of iterations.
14747 2017-08-08 Martin Liska <mliska@suse.cz>
14749 * asan.c: Include header files.
14750 * attribs.c (build_decl_attribute_variant): New function moved
14752 (build_type_attribute_qual_variant): Likewise.
14753 (cmp_attrib_identifiers): Likewise.
14754 (simple_cst_list_equal): Likewise.
14755 (omp_declare_simd_clauses_equal): Likewise.
14756 (attribute_value_equal): Likewise.
14757 (comp_type_attributes): Likewise.
14758 (build_type_attribute_variant): Likewise.
14759 (lookup_ident_attribute): Likewise.
14760 (remove_attribute): Likewise.
14761 (merge_attributes): Likewise.
14762 (merge_type_attributes): Likewise.
14763 (merge_decl_attributes): Likewise.
14764 (merge_dllimport_decl_attributes): Likewise.
14765 (handle_dll_attribute): Likewise.
14766 (attribute_list_equal): Likewise.
14767 (attribute_list_contained): Likewise.
14768 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14769 (lookup_attribute_by_prefix): Likewise.
14770 * bb-reorder.c: Include header files.
14771 * builtins.c: Likewise.
14772 * calls.c: Likewise.
14773 * cfgexpand.c: Likewise.
14774 * cgraph.c: Likewise.
14775 * cgraphunit.c: Likewise.
14776 * convert.c: Likewise.
14777 * dwarf2out.c: Likewise.
14778 * final.c: Likewise.
14779 * fold-const.c: Likewise.
14780 * function.c: Likewise.
14781 * gimple-expr.c: Likewise.
14782 * gimple-fold.c: Likewise.
14783 * gimple-pretty-print.c: Likewise.
14784 * gimple.c: Likewise.
14785 * gimplify.c: Likewise.
14786 * hsa-common.c: Likewise.
14787 * hsa-gen.c: Likewise.
14788 * internal-fn.c: Likewise.
14789 * ipa-chkp.c: Likewise.
14790 * ipa-cp.c: Likewise.
14791 * ipa-devirt.c: Likewise.
14792 * ipa-fnsummary.c: Likewise.
14793 * ipa-inline.c: Likewise.
14794 * ipa-visibility.c: Likewise.
14796 * lto-cgraph.c: Likewise.
14797 * omp-expand.c: Likewise.
14798 * omp-general.c: Likewise.
14799 * omp-low.c: Likewise.
14800 * omp-offload.c: Likewise.
14801 * omp-simd-clone.c: Likewise.
14802 * opts-global.c: Likewise.
14803 * passes.c: Likewise.
14804 * predict.c: Likewise.
14805 * sancov.c: Likewise.
14806 * sanopt.c: Likewise.
14807 * symtab.c: Likewise.
14808 * toplev.c: Likewise.
14809 * trans-mem.c: Likewise.
14810 * tree-chkp.c: Likewise.
14811 * tree-eh.c: Likewise.
14812 * tree-into-ssa.c: Likewise.
14813 * tree-object-size.c: Likewise.
14814 * tree-parloops.c: Likewise.
14815 * tree-profile.c: Likewise.
14816 * tree-ssa-ccp.c: Likewise.
14817 * tree-ssa-live.c: Likewise.
14818 * tree-ssa-loop.c: Likewise.
14819 * tree-ssa-sccvn.c: Likewise.
14820 * tree-ssa-structalias.c: Likewise.
14821 * tree-ssa.c: Likewise.
14822 * tree-streamer-in.c: Likewise.
14823 * tree-vectorizer.c: Likewise.
14824 * tree-vrp.c: Likewise.
14825 * tsan.c: Likewise.
14826 * ubsan.c: Likewise.
14827 * varasm.c: Likewise.
14828 * varpool.c: Likewise.
14829 * tree.c: Remove functions moved to attribs.[ch].
14830 * tree.h: Likewise.
14831 * config/aarch64/aarch64.c: Add attrs.h header file.
14832 * config/alpha/alpha.c: Likewise.
14833 * config/arc/arc.c: Likewise.
14834 * config/arm/arm.c: Likewise.
14835 * config/avr/avr.c: Likewise.
14836 * config/bfin/bfin.c: Likewise.
14837 * config/c6x/c6x.c: Likewise.
14838 * config/cr16/cr16.c: Likewise.
14839 * config/cris/cris.c: Likewise.
14840 * config/darwin.c: Likewise.
14841 * config/epiphany/epiphany.c: Likewise.
14842 * config/fr30/fr30.c: Likewise.
14843 * config/frv/frv.c: Likewise.
14844 * config/ft32/ft32.c: Likewise.
14845 * config/h8300/h8300.c: Likewise.
14846 * config/i386/winnt.c: Likewise.
14847 * config/ia64/ia64.c: Likewise.
14848 * config/iq2000/iq2000.c: Likewise.
14849 * config/lm32/lm32.c: Likewise.
14850 * config/m32c/m32c.c: Likewise.
14851 * config/m32r/m32r.c: Likewise.
14852 * config/m68k/m68k.c: Likewise.
14853 * config/mcore/mcore.c: Likewise.
14854 * config/microblaze/microblaze.c: Likewise.
14855 * config/mips/mips.c: Likewise.
14856 * config/mmix/mmix.c: Likewise.
14857 * config/mn10300/mn10300.c: Likewise.
14858 * config/moxie/moxie.c: Likewise.
14859 * config/msp430/msp430.c: Likewise.
14860 * config/nds32/nds32-isr.c: Likewise.
14861 * config/nds32/nds32.c: Likewise.
14862 * config/nios2/nios2.c: Likewise.
14863 * config/nvptx/nvptx.c: Likewise.
14864 * config/pa/pa.c: Likewise.
14865 * config/pdp11/pdp11.c: Likewise.
14866 * config/powerpcspe/powerpcspe.c: Likewise.
14867 * config/riscv/riscv.c: Likewise.
14868 * config/rl78/rl78.c: Likewise.
14869 * config/rx/rx.c: Likewise.
14870 * config/s390/s390.c: Likewise.
14871 * config/sh/sh.c: Likewise.
14872 * config/sol2.c: Likewise.
14873 * config/sparc/sparc.c: Likewise.
14874 * config/spu/spu.c: Likewise.
14875 * config/stormy16/stormy16.c: Likewise.
14876 * config/tilegx/tilegx.c: Likewise.
14877 * config/tilepro/tilepro.c: Likewise.
14878 * config/v850/v850.c: Likewise.
14879 * config/vax/vax.c: Likewise.
14880 * config/visium/visium.c: Likewise.
14881 * config/xtensa/xtensa.c: Likewise.
14883 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14886 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14887 constraints since the -mupper-regs-* switches have been
14889 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14890 into a vector from a double word element that was extracted from
14891 another vector, and eliminate extra XXPERMDI instructions.
14892 (vsx_concat_<mode>_2): Likewise.
14893 (vsx_concat_<mode>_3): Likewise.
14894 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14895 concat to allow optimizing inserts from previous extracts.
14897 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14899 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14900 memory reference to a SSP offset in TLS address space.
14901 (ix86_print_operand) <case '@'>: Remove.
14902 (ix86_print_operand_punct_valid_p): Remove '@' code.
14903 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14904 UNSPEC_SP_TLS_TEST.
14905 (stack_tls_protect_set_<mode>): Remove.
14906 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14907 (stack_tls_protect_test_<mode>): Remove.
14908 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14910 2017-08-07 Olivier Hainque <hainque@adacore.com>
14913 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14915 2017-08-07 Douglas Rupp <rupp@adacore.com>
14917 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14918 variable was referenced as multidir in command.
14920 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14923 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14926 2017-08-07 Martin Liska <mliska@suse.cz>
14928 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14929 * config/rl78/rl78.c: Add include of attribs.h.
14930 * config/sh/sh.c: Likewise.
14931 * config/v850/v850.c: Likewise.
14933 2017-08-07 Tom de Vries <tom@codesourcery.com>
14935 PR middle-end/78266
14936 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14938 2017-08-07 Martin Liska <mliska@suse.cz>
14940 * config/mips/mips.c: Include attribs.h.
14942 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14945 * doc/invoke.texi: Document change in behvaior for -Ofast for
14948 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14950 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14952 (aarch64_pop_regs): Likewise.
14953 (aarch64_gen_load_pair): Likewise.
14954 (aarch64_save_callee_saves): Likewise.
14955 (aarch64_restore_callee_saves): Likewise.
14957 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14959 * config/i386/i386.c: Revert the last change.
14961 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14964 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14966 (ix86_finalize_stack_frame_flags): This. Also clear
14967 frame_pointer_needed if -fno-omit-frame-pointer is used without
14969 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14970 with ix86_finalize_stack_frame_flags.
14971 (ix86_expand_epilogue): Likewise.
14972 (ix86_expand_split_stack_prologue): Likewise.
14974 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14977 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14978 to P_AES for Westmere.
14980 2017-08-07 Jonathan Yong <10walls@gmail.com>
14982 * config/i386/mingw.opt (fset-stack-executable): Removed.
14983 * config/i386/cygming.opt (fset-stack-executable): Moved
14985 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14987 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14989 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14991 2017-08-07 Marek Polacek <polacek@redhat.com>
14993 PR middle-end/81737
14994 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14996 2017-08-07 Martin Liska <mliska@suse.cz>
14998 * attribs.h (canonicalize_attr_name): New function.
14999 (cmp_attribs): Move from c-format.c and adjusted.
15000 (is_attribute_p): Moved from tree.h.
15001 * tree-inline.c: Add new includes.
15002 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15003 (private_is_attribute_p): Remove.
15004 (private_lookup_attribute): Likewise.
15005 (private_lookup_attribute_by_prefix): Simplify.
15006 (remove_attribute): Use is_attribute_p.
15007 * tree.h: Remove removed declarations.
15009 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15011 PR middle-end/81698
15012 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15013 instead of computing it in the function. Formatting fix.
15014 (expand_case): Don't rely on default_edge being the first edge,
15015 clear it if removing it, pass default_edge to
15016 emit_case_dispatch_table.
15017 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15020 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15022 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15023 insn in the function, emit NOP after the insn.
15025 2017-08-06 Tom de Vries <tom@codesourcery.com>
15027 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15030 2017-08-06 Tom de Vries <tom@codesourcery.com>
15032 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15035 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15037 PR tree-optimization/57371
15038 * match.pd: New pattern.
15040 2017-08-04 Marek Polacek <polacek@redhat.com>
15042 PR middle-end/81695
15043 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15044 perform the computation in offset_int.
15046 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15048 PR tree-optimization/81136
15049 * tree-vectorizer.h: Include tree-hash-traits.h.
15050 (vec_base_alignments): New typedef.
15051 (vec_info): Add a base_alignments field.
15052 (vect_record_base_alignments): Declare.
15053 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15055 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15056 (create_data_ref): Add an is_conditional_in_stmt argument.
15057 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15058 the is_conditional_in_stmt field.
15059 (data_ref_loc): Add an is_conditional_in_stmt field.
15060 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15061 (find_data_references_in_stmt): Update call to create_data_ref.
15062 (graphite_find_data_references_in_stmt): Likewise.
15063 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15064 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15065 (vect_record_base_alignment): New function.
15066 (vect_record_base_alignments): Likewise.
15067 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15068 for nested statements even if we fail to compute a misalignment.
15069 Use pooled base alignments for unconditional references.
15070 (vect_find_same_alignment_drs): Compare base addresses instead
15072 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15073 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15075 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15077 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15078 Add an explicit name for the enum. Use auto_vec for slp_instances
15079 and grouped_stores.
15080 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15082 (_bb_vec_info): Add a constructor and destructor.
15083 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15084 (destroy_loop_vec_info): Delete.
15085 (vect_destroy_datarefs): Likewise.
15086 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15087 (vec_info::vec_info): New function.
15088 (vec_info::~vec_info): Likewise.
15089 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15090 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15091 destroy_loop_vec_info.
15092 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15093 (_loop_vec_info::_loop_vec_info): ...this.
15094 (destroy_loop_vec_info): Replace with...
15095 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15096 the stmt_vec_infos. Leave handling of vec_info information to its
15097 destructor. Remove explicit vector releases.
15098 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15099 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15100 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15101 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15102 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15103 (destroy_bb_vec_info): Replace with...
15104 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15105 information to its destructor.
15106 (vect_slp_analyze_bb_1): Use new and delete instead of
15107 new_bb_vec_info and destroy_bb_vec_info.
15108 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15111 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15113 * tree-data-ref.h (subscript): Add access_fn field.
15114 (data_dependence_relation): Add could_be_independent_p.
15115 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15116 (same_access_functions): Move to tree-data-ref.c.
15117 * tree-data-ref.c (ref_contains_union_access_p): New function.
15118 (access_fn_component_p): Likewise.
15119 (access_fn_components_comparable_p): Likewise.
15120 (dr_analyze_indices): Add a reference to access_fn_component_p.
15121 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15123 (constant_access_functions): Likewise.
15124 (add_other_self_distances): Likewise.
15125 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15126 (initialize_data_dependence_relation): Use XCNEW and remove
15127 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15128 of access functions that have the same type. Allow the
15129 subsequence to end with different bases in some circumstances.
15130 Record the chosen access functions in SUB_ACCESS_FN.
15131 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15132 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15133 (subscript_dependence_tester_1): Likewise dra and drb.
15134 (build_classic_dist_vector): Update calls accordingly.
15135 (subscript_dependence_tester): Likewise.
15136 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15137 DDR_COULD_BE_INDEPENDENT_P.
15138 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15139 comp_alias_ddrs instead of may_alias_ddrs.
15140 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15142 (vect_analyze_data_ref_dependence): Use it if
15143 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15144 distance vectors if that fails.
15145 (dependence_distance_ge_vf): New function.
15146 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15147 LOOP_VINFO_MAY_ALIAS_DDRS.
15149 2017-08-04 Richard Biener <rguenther@suse.de>
15151 PR middle-end/81705
15152 * fold-const.c (fold_binary_loc): Properly restrict
15153 minus_var0 && minus_var1 case when associating undefined overflow
15156 2017-08-04 Tom de Vries <tom@codesourcery.com>
15158 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15160 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15162 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15163 Don't start diagnostic messages with a capital letter.
15164 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15166 (rs6000_invalid_builtin): Likewise.
15167 (rs6000_trampoline_init): Likewise.
15169 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15172 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15173 after setting changeable df flags.
15175 2017-08-03 Richard Biener <rguenther@suse.de>
15177 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15178 up if the use is in USE - X.
15180 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15182 * toplev.c (dumpfile.h): New include.
15183 (internal_error_reentered): New static function. Use it...
15184 (internal_error_function): ...here to handle reentered internal_error.
15186 2017-08-03 Richard Biener <rguenther@suse.de>
15188 PR middle-end/81148
15189 * fold-const.c (split_tree): Add minus_var and minus_con
15190 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15191 here but always use minus_*.
15192 (associate_trees): Assert we never associate with MINUS_EXPR
15193 and NULL first operand. Do not recurse for PLUS_EXPR operands
15194 when associating as MINUS_EXPR either.
15195 (fold_binary_loc): Track minus_var and minus_con.
15197 2017-08-03 Tom de Vries <tom@codesourcery.com>
15200 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15201 ACCEL_COMPILER, apply finish_options on
15202 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15204 2017-08-03 Tom de Vries <tom@codesourcery.com>
15207 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15208 function_entry_patch_area_size > 0.
15210 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15213 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15214 instead of 10??LU, perform unit multiplication in wide_int,
15215 don't change alloc_object_size_limit if the limit is larger
15218 PR tree-optimization/81655
15219 PR tree-optimization/81588
15220 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15221 the case when ranges[i].low and high are 1 for unsigned type with
15224 PR middle-end/81052
15225 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15226 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15228 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15230 * tree-vrp.h: Add include guard.
15232 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15235 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15236 (ud2): New insn pattern.
15237 * config/i386/i386.c (ix86_expand_epilogue):
15238 For naked functions, generate ud2 instead of trap insn.
15240 2017-08-02 Marek Polacek <polacek@redhat.com>
15243 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15245 2017-08-02 Tom de Vries <tom@codesourcery.com>
15246 Cesar Philippidis <cesar@codesourcery.com>
15248 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15249 Add missing edge probabilities.
15251 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15253 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15254 Correct endianness.
15256 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15258 PR middle-end/79499
15259 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15260 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15261 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15262 prologue_seq sequences - if any.
15264 2017-08-02 Richard Biener <rguenther@suse.de>
15266 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15267 via vectors if supported, integer extracts via punning if supported
15268 or otherwise vector extracts.
15270 2017-08-02 Richard Biener <rguenther@suse.de>
15272 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15274 (bitmap_insert_into_set): ... this.
15276 2017-08-02 Richard Biener <rguenther@suse.de>
15278 PR tree-optimization/81633
15280 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15282 PR tree-optimization/71752
15283 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15285 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15287 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15288 (machine_function::call_ms2sysv_pad_out): Remove field.
15289 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15290 (ix86_compute_frame_layout): Likewise.
15292 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15295 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15296 attribute with interrupt attribute.
15298 2017-08-01 Andrew Pinski <apinski@cavium.com>
15300 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15301 BIT_INSERT_EXPR's operand 1
15302 to see if the types precision matches.
15304 2017-08-01 Martin Liska <mliska@suse.cz>
15306 PR middle-end/70140
15307 * builtins.c (expand_builtin_memcpy_args): Remove.
15308 (expand_builtin_memcpy): Call newly added function
15309 expand_builtin_memory_copy_args.
15310 (expand_builtin_memcpy_with_bounds): Likewise.
15311 (expand_builtin_mempcpy): Remove last argument.
15312 (expand_builtin_mempcpy_with_bounds): Likewise.
15313 (expand_builtin_memory_copy_args): New function created from
15314 expand_builtin_mempcpy_args with small modifications.
15315 (expand_builtin_mempcpy_args): Remove.
15316 (expand_builtin_stpcpy): Remove unused argument.
15317 (expand_builtin): Likewise.
15318 (expand_builtin_with_bounds): Likewise.
15320 2017-08-01 Martin Liska <mliska@suse.cz>
15323 Make mempcpy more optimal (PR middle-end/70140).
15325 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15328 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15329 __builtin_vec_cmpne verify both arguments are compatible vectors
15330 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15331 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15332 move computation of aligned to after checking the argument types.
15336 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15337 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15340 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15341 Doug Rupp <rupp@adacore.com>
15342 Olivier Hainque <hainque@adacore.com>
15344 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15345 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15347 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15348 for TARGET_OS_CPP_BUILTIN.
15349 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15350 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15352 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15353 passing required abi options to the assembler for EABI configurations.
15354 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15355 of .text.hot and .text.unlikely sections for kernel modules when
15356 using ARM style exceptions.
15357 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15358 options. Add EXTRA_CC1_SPEC.
15359 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15361 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15363 (ARM_TARGET2_DWARF_FORMAT): Define.
15364 * config/arm/t-vxworks: Adjust multilib control to removal of the
15365 Diab command line options.
15367 2017-08-01 Martin Liska <mliska@suse.cz>
15369 PR gcov-profile/81561
15370 * gcov.c (unblock): Make unblocking safe as we need to preserve
15371 index correspondence of blocks and block_lists.
15373 2017-08-01 Richard Biener <rguenther@suse.de>
15375 PR tree-optimization/81181
15376 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15377 (compute_antic): ... end of iteration here.
15379 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15381 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15382 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15383 (ftree-slp-vectorize): Likewise.
15384 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15385 can no longer be set independent of flag_tree_loop_vectorize.
15386 * omp-general.c (emp_max_vf): Likewise.
15387 * opts.c (enable_fdo_optimizations): Remove references to
15388 flag_tree_vectorize, these are now implicit.
15389 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15390 and leave it for the options machinery.
15392 2017-08-01 Martin Liska <mliska@suse.cz>
15394 PR middle-end/70140
15395 * builtins.c (expand_builtin_memcpy_args): Remove.
15396 (expand_builtin_memcpy): Call newly added function
15397 expand_builtin_memory_copy_args.
15398 (expand_builtin_memcpy_with_bounds): Likewise.
15399 (expand_builtin_mempcpy): Remove last argument.
15400 (expand_builtin_mempcpy_with_bounds): Likewise.
15401 (expand_builtin_memory_copy_args): New function created from
15402 expand_builtin_mempcpy_args with small modifications.
15403 (expand_builtin_mempcpy_args): Remove.
15404 (expand_builtin_stpcpy): Remove unused argument.
15405 (expand_builtin): Likewise.
15406 (expand_builtin_with_bounds): Likewise.
15408 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15411 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15412 print "ds:" only for immediates in generic address space.
15414 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15417 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15418 (ix86_function_ok_for_sibcall): Return false for naked functions.
15420 2017-08-01 Richard Biener <rguenther@suse.de>
15422 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15423 (compute_antic): Seed worklist with exit block predecessors.
15424 * cfganal.c (dfs_find_deadend): For a cycle return the source
15425 of the edge closing it.
15427 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15429 * config/aarch64/aarch64.c
15430 (aarch64_can_const_movi_rtx_p): Move 0 check.
15432 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15434 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15435 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15437 * match.pd: Ditto in address comparison pattern.
15439 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15441 PR tree-optimization/81627
15442 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15443 closed ssa form for store-store chain.
15445 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15447 PR tree-optimization/81620
15448 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15449 for store-store chain.
15451 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15453 PR tree-optimization/81588
15454 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15455 ranges[i].in_p, invert comparison code ccode. For >/>=,
15456 swap rhs1 and rhs2 and comparison code unconditionally,
15457 for </<= don't do that. Don't swap rhs1/rhs2 again if
15458 ranges[i].in_p, instead invert comparison code ccode if
15459 opcode or oe->rank is BIT_IOR_EXPR.
15462 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15463 a direct optab to conversion optab.
15464 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15465 with GET_MODE_INNER as last argument instead of optab_handler.
15466 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15467 vector extraction if possible and optab is available.
15468 * expr.c (store_constructor): Use convert_optab_handler instead
15469 of optab_handler. Use vector initialization from smaller
15470 vectors if possible and optab is available.
15471 * tree-vect-stmts.c (vectorizable_load): Likewise.
15472 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15473 now have two modes.
15474 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15475 of vec_init from half-sized vectors with the same element mode.
15476 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15477 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15478 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15479 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15480 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15481 after mode in gen_vec_extract* calls.
15482 (vec_extract<mode>): Renamed to ...
15483 (vec_extract<mode><ssescalarmodelower>): ... this.
15484 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15485 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15486 element mode after mode in gen_vec_init* calls.
15487 (VEC_INIT_HALF_MODE): New mode iterator.
15488 (vec_init<mode>): Renamed to ...
15489 (vec_init<mode><ssescalarmodelower>): ... this.
15490 (vec_init<mode><ssehalfvecmodelower>): New expander.
15491 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15492 (vec_extractv2sfsf): ... this.
15493 (vec_initv2sf): Renamed to ...
15494 (vec_initv2sfsf): ... this.
15495 (vec_extractv2si): Renamed to ...
15496 (vec_extractv2sisi): ... this.
15497 (vec_initv2si): Renamed to ...
15498 (vec_initv2sisi): ... this.
15499 (vec_extractv4hi): Renamed to ...
15500 (vec_extractv4hihi): ... this.
15501 (vec_initv4hi): Renamed to ...
15502 (vec_initv4hihi): ... this.
15503 (vec_extractv8qi): Renamed to ...
15504 (vec_extractv8qiqi): ... this.
15505 (vec_initv8qi): Renamed to ...
15506 (vec_initv8qiqi): ... this.
15507 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15508 (vec_init<mode>): Renamed to ...
15509 (vec_init<mode><VEC_base_l>): ... this.
15510 (vec_extract<mode>): Renamed to ...
15511 (vec_extract<mode><VEC_base_l>): ... this.
15512 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15513 (vec_initv2sfsf): ... this.
15514 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15515 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15516 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15517 element mode after mode in gen_vec_init* calls.
15518 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15519 (vec_init<mode><Vel>): ... this.
15520 (vec_extract<mode>): Renamed to ...
15521 (vec_extract<mode><Vel>): ... this.
15522 * config/aarch64/iterators.md (Vel): New mode attribute.
15523 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15524 Add element mode after mode in gen_vec_extract* calls.
15525 * config/s390/vector.md (non_vec_l): New mode attribute.
15526 (vec_extract<mode>): Renamed to ...
15527 (vec_extract<mode><non_vec_l>): ... this.
15528 (vec_init<mode>): Renamed to ...
15529 (vec_init<mode><non_vec_l>): ... this.
15530 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15531 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15533 * config/arm/iterators.md (V_elem_l): New mode attribute.
15534 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15535 (vec_extract<mode><V_elem_l>): ... this.
15536 (vec_extractv2di): Renamed to ...
15537 (vec_extractv2didi): ... this.
15538 (vec_init<mode>): Renamed to ...
15539 (vec_init<mode><V_elem_l>): ... this.
15540 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15541 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15542 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15543 Add element mode after gen_vec_extract* calls.
15544 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15545 (vec_init<mode><unitmode>): ... this.
15546 (vec_extract<mode>): Renamed to ...
15547 (vec_extract<mode><unitmode>): ... this.
15548 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15549 (vec_init<mode><unitmode>): ... this.
15550 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15551 (vec_initv2sfsf): ... this.
15552 (vec_extractv2sf): Renamed to ...
15553 (vec_extractv2sfsf): ... this.
15554 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15555 Add element mode after gen_vec_extract* calls.
15556 * config/mips/mips.md (unitmode): New mode iterator.
15557 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15558 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15559 * config/spu/spu.md (inner_l): New mode attribute.
15560 (vec_init<mode>): Renamed to ...
15561 (vec_init<mode><inner_l>): ... this.
15562 (vec_extract<mode>): Renamed to ...
15563 (vec_extract<mode><inner_l>): ... this.
15564 * config/sparc/sparc.md (veltmode): New mode iterator.
15565 (vec_init<VMALL:mode>): Renamed to ...
15566 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15567 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15568 (vec_initv2sisi): ... this.
15569 (vec_initv2sf): Renamed to ...
15570 (vec_initv2sfsf): ... this.
15571 (vec_extractv2sf): Renamed to ...
15572 (vec_extractv2sfsf): ... this.
15573 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15574 (vec_init<mode>): Renamed to ...
15575 (vec_init<mode><VEC_base_l>): ... this.
15576 (vec_extract<mode>): Renamed to ...
15577 (vec_extract<mode><VEC_base_l>): ... this.
15578 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15579 (vec_initv2sfsf): ... this.
15580 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15581 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15582 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15583 gen_vec_init* calls.
15585 2017-08-01 Richard Biener <rguenther@suse.de>
15587 PR tree-optimization/81297
15588 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15589 TREE_OVERFLOW from INTEGER_CSTs.
15591 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15593 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15595 2017-07-31 Carl Love <cel@us.ibm.com>
15597 * config/rs6000/rs6000-c: Add support for built-in functions
15598 vector signed char vec_xl_be (signed long long, signed char *);
15599 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15600 vector signed int vec_xl_be (signed long long, signed int *);
15601 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15602 vector signed long long vec_xl_be (signed long long, signed long long *);
15603 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15604 vector signed short vec_xl_be (signed long long, signed short *);
15605 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15606 vector double vec_xl_be (signed long long, double *);
15607 vector float vec_xl_be (signed long long, float *);
15608 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15609 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15610 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15612 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15613 (altivec_expand_builtin): Add switch statement to call
15614 altivec_expand_xl_be for each builtin.
15615 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15616 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15617 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15618 __builtin_vsx_le_be_v16qi.
15619 * doc/extend.texi: Update the built-in documentation file for the
15620 new built-in functions.
15622 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15625 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15627 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15629 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15631 * config.gcc: Add z14.
15632 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15633 CPU model numbers for z13s and z14.
15634 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15636 * config/s390/s390-opts.h (enum processor_type): Rename
15637 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15638 * config/s390/s390.c (processor_table): Add field for CPU name to
15639 be passed to Binutils.
15640 (s390_asm_output_machine_for_arch): Use the new field in
15641 processor_table for Binutils.
15642 (s390_expand_builtin): Replace arch12 with z14.
15643 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15644 (s390_get_sched_attrmask): Likewise.
15645 (s390_get_unit_mask): Likewise.
15646 * config/s390/s390.opt: Add z14 to processor_type enum.
15648 2017-07-31 Martin Jambor <mjambor@suse.cz>
15651 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15652 regardless of optimization level.
15654 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15655 Martin Liska <mliska@suse.cz>
15657 * predict.def: Remove old comment and adjust probability.
15658 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15659 PREDICT statements.
15661 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15664 * config/i386/i386.c (ix86_function_naked): New function.
15665 (ix86_can_use_return_insn_p): Return false for naked functions.
15666 (ix86_expand_prologue): Skip prologue for naked functions.
15667 (ix86_expand_epilogue): Skip epilogue for naked functions
15668 and emit trap instruction.
15669 (ix86_warn_func_return): New function.
15670 (ix86_attribute_table): Add "naked" attribute specification.
15671 (TARGET_WARN_FUNC_RETURN): Define.
15672 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15674 2017-07-31 Martin Liska <mliska@suse.cz>
15676 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15677 (dump_gimple_bb_header): Always dump BB info.
15678 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15680 2017-07-31 Martin Liska <mliska@suse.cz>
15683 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15684 also with current_function_decl non-null equality.
15686 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15689 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15690 change type to the element type, instead add eltype variable and
15691 use it where we are interested in the element type.
15693 PR tree-optimization/81603
15694 * ipa-polymorphic-call.c
15695 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15696 offset arithmetic in offset_int, bail out if the resulting bit offset
15697 doesn't fit into shwi.
15699 2017-07-31 Martin Liska <mliska@suse.cz>
15701 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15702 (gimplify_save_expr): Fix comment.
15704 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15707 * config/i386/i386.c (ix86_function_arg): Update arguments for
15709 (ix86_compute_frame_layout): Set the initial stack offset to
15710 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15711 INCOMING_FRAME_SP_OFFSET.
15712 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15713 stack before exception handler returns.
15714 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15715 the 'ERROR_CODE' for exception handler.
15717 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15719 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15720 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15721 (ASM_OUTPUT_REG_POP): Ditto.
15722 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15723 instead of asm_fprintf to output pure string.
15725 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15727 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15728 to imported_module_or_decl hook.
15729 (debug_nothing_tree_tree_tree_bool): Remove.
15730 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15731 * debug.c (do_nothing_debug_hooks): Use
15732 debug_nothing_tree_tree_tree_bool_bool instead of
15733 debug_nothing_tree_tree_tree_bool.
15734 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15735 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15736 * sdbout.c (sdb_debug_hooks): Likewise.
15737 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15738 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15740 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15741 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15742 attribute, don't add anything.
15744 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15746 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15747 (fold_build2_stat_loc): Likewise.
15748 (fold_build3_stat_loc): Likewise.
15749 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15750 (fold_build1_loc): Remove macro.
15751 (fold_build2_loc): Likewise.
15752 (fold_build3_loc): Likewise.
15754 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15756 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15757 (gimple_build_debug_bind_source_stat): Likewise.
15758 * gimple.h (gimple_build_debug_bind): Remove macro.
15759 (gimple_build_debug_bind_source): Likewise.
15761 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15763 * bitmap.c (bitmap_alloc): Adjust.
15764 (bitmap_gc_alloc): Likewise.
15765 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15767 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15769 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15770 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15771 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15772 (bitmap_gc_alloc_stat): Likewise.
15773 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15775 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15777 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15778 * rtl.h (shallow_copy_rtx): Remove macro.
15780 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15782 * emit-rtl.c (gen_raw_REG): Adjust.
15783 * gengenrtl.c (gendef): Likewise.
15784 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15785 * rtl.h (rtx_alloc): Remove macro.
15787 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15789 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15790 (build_tree_list_stat): Likewise.
15791 * tree.h (build_tree_list): Remove macro.
15792 (build_tree_list_vec): Likewise.
15794 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15796 * tree.c (make_vector_stat): Remove _stat from name.
15797 (build_vector_stat): Likewise.
15798 * tree.h (make_vector_stat): Remove macro.
15799 (build_vector_stat): Likewise.
15801 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15803 * tree.h (build_var_debug_value): Remove prototype.
15805 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15807 * tree.c (tree_cons_stat): Remove _stat from name.
15808 * tree.h (tree_cons): Remove macro.
15810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15812 * tree.c (build_vl_exp_stat): Remove _stat from name.
15813 * tree.h (build_vl_exp): Remove macro.
15815 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15817 * tree.c (build_decl_stat): Remove _stat from name.
15818 * tree.h (build_decl): Remove macro.
15820 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15822 * gimple.c (gimple_build_with_ops_stat): Adjust.
15823 (gimple_alloc_stat): Remove _stat from name.
15824 * gimple.h (gimple_alloc): Remove macro.
15826 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15828 * tree.c (make_tree_vec_stat): Remove _stat from name.
15829 (grow_tree_vec_stat): Likewise.
15830 * tree.h (make_tree_vec_stat): Adjust prototype.
15831 (grow_tree_vec_stat): Likewise.
15832 (make_tree_vec): Remove macro.
15833 (grow_tree_vec): Likewise.
15835 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15837 * fold-const.c (fold_build1_stat_loc): Adjust.
15838 (fold_build2_stat_loc): Likewise.
15839 (fold_build3_stat_loc): Likewise.
15840 * tree.c (build0_stat): Remove _stat from name.
15841 (build1_stat): Likewise.
15842 (build2_stat): Likewise.
15843 (build3_stat): Likewise.
15844 (build4_stat): Likewise.
15845 (build5_stat): Likewise.
15846 * tree.h (build1_loc): Remove macro, and rename _stat function
15848 (build2_loc): Likewise.
15849 (build3_loc): Likewise.
15850 (build4_loc): Likewise.
15851 (build5_loc): Likewise.
15853 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15855 * tree.c (make_int_cst_stat): Remove _stat from name.
15856 * tree.h (make_int_cst_stat): Adjust prototype.
15857 (make_int_cst): Remove macro.
15859 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15861 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15862 * tree.h (make_tree_binfo_stat): Adjust prototype.
15863 (make_tree_binfo): Remove.
15865 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15867 * tree.c (copy_node_stat): Rename to copy_node.
15868 (build_distinct_type_copy): Adjust.
15869 * tree.h (copy_node_stat): Adjust prototype.
15870 (copy_node): Remove macro.
15872 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15874 * tree.c (make_node_stat): rename to make_node.
15875 (build_tree_list_stat): Adjust.
15876 (build0_stat): Likewise.
15877 (build2_stat): Likewise.
15878 (build3_stat): Likewise.
15879 (build4_stat): Likewise.
15880 (build5_stat): Likewise.
15881 (build_decl_stat): Likewise.
15882 * tree.h (make_node_stat): Adjust prototype.
15883 (make_node): remove macro.
15885 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15887 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15888 (PPC_FEATURE2_SCV): Likewise.
15889 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15891 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15893 * config/aarch64/aarch64.c
15894 (aarch64_internal_mov_immediate): Add new special pattern.
15895 * config/aarch64/aarch64.md (*movdi_aarch64):
15896 Add reg/32bit const mov case.
15898 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15899 Richard Sandiford <richard.sandiford@linaro.org>
15901 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15902 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15903 Add integer and movi cases.
15904 (movi-split-hf-df-sf split, fp16): New.
15905 (enabled): Added TARGET_FP_F16INST.
15906 * config/aarch64/iterators.md (GPF_HF): New.
15907 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15909 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15911 * config/aarch64/aarch64.c
15912 (aarch64_simd_container_mode): Add prototype.
15913 (aarch64_expand_mov_immediate): Add HI support.
15914 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15915 (aarch64_can_const_movi_rtx_p): New.
15916 (aarch64_preferred_reload_class):
15917 Remove restrictions of using FP registers for certain SIMD operations.
15918 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15919 (aarch64_valid_floating_const): Add integer move validation.
15920 (aarch64_simd_imm_scalar_p): Remove.
15921 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15922 (aarch64_legitimate_constant_p): Expand list of supported cases.
15923 * config/aarch64/aarch64-protos.h
15924 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15925 (aarch64_reinterpret_float_as_int): New.
15926 (aarch64_simd_imm_scalar_p): Remove.
15927 * config/aarch64/constraints.md (Uvi): New.
15928 (Dd): Split into Ds and new Dd.
15929 * config/aarch64/aarch64.md (*movsi_aarch64):
15931 (*movdi_aarch64): Add SIMD mov case.
15933 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15935 * tree-predcom.c: (struct chain): Handle store-store chain in which
15936 stores for elimination only store loop invariant values.
15937 (execute_pred_commoning_chain): Ditto.
15938 (prepare_initializers_chain_store_elim): Ditto.
15939 (prepare_finalizers): Ditto.
15940 (is_inv_store_elimination_chain): New function.
15941 (initialize_root_vars_store_elim_1): New function.
15943 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15945 * tree-predcom.c: Revise general description of the pass.
15946 (enum chain_type): New enum type for store elimination.
15947 (struct chain): New field supporting store elimination.
15948 (struct component): Ditto.
15949 (dump_chain): Dump store-stores chain.
15950 (release_chain): Release resources.
15951 (split_data_refs_to_components): Compute and create component
15952 contains only stores for elimination.
15953 (get_chain_last_ref_at): New function.
15954 (make_invariant_chain): Initialization.
15955 (make_rooted_chain): Specify chain type in parameter and record it.
15956 (add_looparound_copies): Skip for store-stores chain.
15957 (determine_roots_comp): Compute type of chain and pass it to
15959 (initialize_root_vars_store_elim_2): New function.
15960 (finalize_eliminated_stores): New function.
15961 (remove_stmt): Handle store for elimination.
15962 (execute_pred_commoning_chain): Execute predictive commoning on
15963 store-store chains.
15964 (determine_unroll_factor): Skip unroll for store-stores chain.
15965 (prepare_initializers_chain_store_elim): New function.
15966 (prepare_initializers_chain): Hanlde store-store chain.
15967 (prepare_finalizers_chain, prepare_finalizers): New function.
15968 (tree_predictive_commoning_loop): Return integer value indicating
15969 if loop is unrolled or lcssa form is corrupted.
15970 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15972 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15974 * tree-predcom.c (initialize_root): Delete.
15975 (execute_pred_commoning_chain): Initialize root vars and replace
15976 reference of non-combined chain directly, rather than call above
15979 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15981 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15982 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15984 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15986 * tree-predcom.c (struct chain): New field init_seq.
15987 (release_chain): Release init_seq.
15988 (prepare_initializers_chain): Record intialization stmts in above
15990 (insert_init_seqs): New function.
15991 (tree_predictive_commoning_loop): Call insert_init_seqs.
15993 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15995 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15997 2017-07-28 Richard Biener <rguenther@suse.de>
15999 * match.pd: Remove superfluous :c.
16000 * genmatch.c (simplify::id): Add member.
16001 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16003 (current_id): New global.
16004 (dt_node::parent): Move from ...
16005 (dt_operand::parent): ... here. Add for_id member.
16006 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16007 (decision_tree::find_node): Relax order requirement when
16008 merging DT_TRUE nodes to ones inbetween the current simplify
16009 and the one we try to merge with. Add diagnostic whenever
16010 we need to enforce pattern order by not merging.
16011 (decision_tree::insert): Set current_id.
16012 (decision_tree::print_node): Dump parent node and for_id.
16013 (parser::last_id): Add member.
16014 (parser::push_simplify): Assign unique id.
16015 (parser::parser): Initialize last_id.
16017 2017-07-28 Martin Liska <mliska@suse.cz>
16020 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16021 gimple_build_debug_bind.
16023 2017-07-28 Richard Biener <rguenther@suse.de>
16025 PR tree-optimization/81502
16026 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16028 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16030 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16031 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16032 for BIT_FIELD_REF args.
16033 * fold-const.c (make_bit_field_ref): Likewise.
16034 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16036 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16039 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16040 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16041 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16042 Or it into SANITIZER_UNDEFINED.
16043 * ubsan.c: Include gimple-fold.h and varasm.h.
16044 (ubsan_expand_ptr_ifn): New function.
16045 (instrument_pointer_overflow): New function.
16046 (maybe_instrument_pointer_overflow): New function.
16047 (instrument_object_size): Formatting fix.
16048 (pass_ubsan::execute): Call instrument_pointer_overflow
16049 and maybe_instrument_pointer_overflow.
16050 * internal-fn.c (expand_UBSAN_PTR): New function.
16051 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16052 * sanitizer.def (__ubsan_handle_pointer_overflow,
16053 __ubsan_handle_pointer_overflow_abort): New builtins.
16054 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16055 * internal-fn.def (UBSAN_PTR): New internal function.
16056 * opts.c (sanitizer_opts): Add pointer-overflow.
16057 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16058 * fold-const.c (build_range_check): Compute pointer range check in
16059 integral type if pointer arithmetics would be needed. Formatting
16062 2017-07-28 Martin Liska <mliska@suse.cz>
16065 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16066 parameters that are of a variable-length.
16068 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16070 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16072 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16073 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16074 (CRT_CALL_STATIC_FUNCTION): Likewise.
16075 (ASM_DEFAULT_SPEC): New define.
16076 (ASM_SPEC32): Likewise.
16077 (ASM_SPEC64): Likewise.
16078 (ASM_SPEC_COMMON): Likewise.
16079 (ASM_SPEC): Likewise.
16080 (INVALID_64BIT): Likewise.
16081 (LINK_OS_DEFAULT_SPEC): Likewise.
16082 (LINK_OS_SPEC32): Likewise.
16083 (LINK_OS_SPEC64): Likewise.
16084 (POWERPC_LINUX): Likewise.
16085 (PTRDIFF_TYPE): Likewise.
16086 (RESTORE_FP_PREFIX): Likewise.
16087 (RESTORE_FP_SUFFIX): Likewise.
16088 (SAVE_FP_PREFIX): Likewise.
16089 (SAVE_FP_SUFFIX): Likewise.
16090 (SIZE_TYPE): Likewise.
16091 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16092 (TARGET_64BIT): Likewise.
16093 (TARGET_64BIT): Likewise.
16094 (TARGET_AIX): Likewise.
16095 (WCHAR_TYPE_SIZE): Likewise.
16096 (WCHAR_TYPE): Undefine.
16097 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16098 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16099 (CPP_OS_RTEMS_SPEC): Delete.
16100 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16101 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16103 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16105 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16107 PR tree-optimization/81578
16108 * tree-parloops.c (build_new_reduction): Bail out if
16109 reduction_code isn't one of the standard OpenMP reductions.
16110 Move the details printing after that decision.
16112 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16114 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16115 related to reload_in_progress.
16116 (splat_input_operand): Likewise.
16117 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16119 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16121 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16122 (TARGET_INSTANTIATE_DECLS): Likewise.
16123 (legitimate_indexed_address_p): Delete reload_in_progress code.
16124 (rs6000_debug_legitimate_address_p): Likewise.
16125 (rs6000_eliminate_indexed_memrefs): Likewise.
16126 (rs6000_emit_le_vsx_store): Likewise.
16127 (rs6000_emit_move_si_sf_subreg): Likewise.
16128 (rs6000_emit_move): Likewise.
16129 (register_to_reg_type): Likewise.
16130 (rs6000_pre_atomic_barrier): Likewise.
16131 (rs6000_machopic_legitimize_pic_address): Likewise.
16132 (rs6000_allocate_stack_temp): Likewise.
16133 (rs6000_address_for_fpconvert): Likewise.
16134 (rs6000_address_for_altivec): Likewise.
16135 (rs6000_secondary_memory_needed_rtx): Delete function.
16136 (rs6000_check_sdmode): Likewise.
16137 (rs6000_alloc_sdmode_stack_slot): Likewise.
16138 (rs6000_instantiate_decls): Likewise.
16139 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16140 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16141 Delete reload_in_progress.
16142 (*vec_reload_and_plus_<mptrsize>): Likewise.
16143 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16144 (vsx_div_v2di): Likewise.
16145 (vsx_udiv_v2di): Likewise.
16147 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16149 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16150 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16151 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16152 (rs6000_debug_reg_global): Delete print of LRA status.
16153 (rs6000_option_override_internal): Delete dead LRA related code.
16154 (rs6000_lra_p): Delete function.
16155 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16157 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16159 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16160 * config/riscv/rtems.h: New file.
16162 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16163 Sudakshina Das <sudi.das@arm.com>
16165 * config/aarch64/aarch64.md
16166 (define_split for and<mode>3nr_compare): Move
16167 non aarch64_logical_operand to a register.
16168 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16169 register immediate operand to a register.
16170 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16172 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16174 PR middle-end/81564
16175 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16177 2017-07-27 Richard Biener <rguenther@suse.de>
16179 PR tree-optimization/81573
16180 PR tree-optimization/81494
16181 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16182 multi defuse cycle case.
16184 2017-07-27 Richard Biener <rguenther@suse.de>
16186 PR tree-optimization/81571
16187 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16190 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16192 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16193 earlier and only if MASK_FPU is set. Adjust formatting.
16195 2017-07-27 Martin Liska <mliska@suse.cz>
16197 * opt-functions.awk: Add validation of value of Init.
16198 * optc-gen.awk: Pass new argument.
16200 2017-07-27 Martin Liska <mliska@suse.cz>
16202 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16203 Fix wrong condition.
16205 2017-07-27 Martin Liska <mliska@suse.cz>
16207 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16208 BBs and edges seen by autoFDO.
16210 2017-07-27 Richard Biener <rguenther@suse.de>
16212 PR tree-optimization/81502
16213 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16214 with incompatible but same sized type.
16215 (execute_update_addresses_taken): Likewise.
16217 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16219 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16220 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16222 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16225 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16226 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16227 Change s_operand to memory_operand.
16229 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16231 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16232 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16233 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16234 Emit instructions rather than returning an expression. Handle TFmode
16235 and KFmode by casting to TImode.
16236 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16237 (rs6000_emit_le_vsx_store): Likewise.
16238 * config/rs6000/vsx.md (VSX_TI): New iterator.
16239 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16240 (*vsx_le_undo_permute_<mode>): Likewise.
16241 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16242 emit the split sequence.
16243 (*vsx_le_perm_store_<mode>): Likewise.
16245 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16247 PR tree-optimization/81555
16248 PR tree-optimization/81556
16249 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16250 if true, force CHANGED for the recursive invocation.
16251 (reassociate_bb): Remember original length of ops array, pass
16252 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16254 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16255 attributes for noipa attribute. For naked attribute use
16256 lookup_attribute first before lookup_attribute_spec.
16257 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16259 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16260 for functions with noipa attribute.
16261 (cgraph_externally_visible_p): Return true for functions with noipa
16263 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16264 for functions with noipa attribute.
16265 * doc/extend.texi: Document noipa function attribute.
16266 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16267 also for functions with noipa attribute.
16268 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16270 2017-07-26 Andrew Pinski <apinski@cavium.com>
16272 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16273 vec_unalign_load_cost and vec_unalign_store_cost.
16275 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16277 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16278 -mvsx-small-integer option.
16279 (ISA_3_0_MASKS_IEEE): Likewise.
16280 (OTHER_VSX_VECTOR_MASKS): Likewise.
16281 (POWERPC_MASKS): Likewise.
16282 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16283 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16284 code, only testing for DImode being allowed in non-VSX floating
16286 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16287 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16289 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16290 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16291 TARGET_P8_VECTOR test.
16292 (rs6000_secondary_reload_simple_move): Likewise.
16293 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16294 since TARGET_P9_VECTOR was already tested.
16295 (rs6000_opt_masks): Remove -mvsx-small-integer.
16296 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16297 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16299 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16300 test for TARGET_VEXTRACTUB was used, and that uses
16302 (p9 extract splitter): Likewise.
16303 (vsx_extract_<mode>_di_p9): Likewise.
16304 (vsx_extract_<mode>_store_p9): Likewise.
16305 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16306 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16307 the elimination of TARGET_VSX_SMALL_INTEGER.
16308 (vsx_extract_<mode>_p8): Likewise.
16309 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16310 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16311 (vsx_set_<mode>_p9): Likewise.
16312 (vsx_set_v4sf_p9): Likewise.
16313 (vsx_set_v4sf_p9_zero): Likewise.
16314 (vsx_insert_extract_v4sf_p9): Likewise.
16315 (vsx_insert_extract_v4sf_p9_2): Likewise.
16316 * config/rs6000/rs6000.md (sign extend splitter): Change
16317 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16318 (floatsi<mode>2_lfiwax_mem): Likewise.
16319 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16320 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16321 since a test for TARGET_P9_VECTOR was used.
16322 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16323 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16324 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16325 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16326 TARGET_P8_VECTOR test.
16327 (fix_trunc<mode>si2_stfiwx): Likewise.
16328 (fix_trunc<mode>si2_internal): Likewise.
16329 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16330 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16332 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16333 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16334 TARGET_P8_VECTOR test.
16335 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16336 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16337 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16339 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16340 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16341 since a test for TARGET_P9_VECTOR was used.
16342 (splitter for loading small constants): Likewise.
16344 2017-07-26 Andrew Pinski <apinski@cavium.com>
16346 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16349 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16352 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16353 (fp_valid_at): Likewise.
16355 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16357 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16358 (qdf24xx_addrcost_table): Likewise.
16359 (cortexa57_tunings): Update to use generic_branch_cost.
16360 (cortexa72_tunings): Likewise.
16361 (cortexa73_tunings): Likewise.
16362 (qdf24xx_tunings): Likewise.
16364 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16366 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16367 (thunderx2t99_branch_cost): Likewise.
16368 (cortexa35_tunings): Update to use generic_branch_cost.
16369 (cortexa53_tunings): Likewise.
16370 (cortexa57_tunings): Likewise.
16371 (cortexa72_tunings): Likewise.
16372 (cortexa73_tunings): Likewise.
16373 (thunderx2t99_tunings): Likewise.
16375 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16377 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16378 (sparc_option_override): Honour MASK_FSMULD.
16379 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16380 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16381 * config/sparc/sparc.opt (mfsmuld): New option.
16382 * doc/invoke.texi (mfsmuld): Document option.
16384 2017-07-26 Marek Polacek <polacek@redhat.com>
16386 PR middle-end/70992
16387 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16389 2017-07-26 Richard Biener <rguenther@suse.de>
16391 * gimple-match-head.c (do_valueize): Return OP if valueize
16393 (get_def): New helper to get at the def stmt of a SSA name
16394 if valueize allows.
16395 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16396 do_valueize to get at the def stmt.
16397 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16399 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16401 PR middle-end/46932
16402 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16404 2017-07-26 Martin Liska <mliska@suse.cz>
16407 * function.c (expand_function_start): Make expansion of
16408 nonlocal_goto_save_area after parm_birth_insn.
16410 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16412 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16413 from all CPU target flags enable members.
16415 2017-07-26 Richard Biener <rguenther@suse.de>
16417 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16418 (decision_tree::gen): Make 'type' const.
16419 (write_predicate): Likewise.
16421 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16423 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16424 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16425 (rs6000_option_override_internal): Likewise.
16426 (rs6000_expand_vector_set): Likewise.
16427 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16428 (TARGET_UPPER_REGS_SF): Likewise.
16429 (TARGET_UPPER_REGS_DI): Likewise.
16430 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16431 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16432 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16433 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16434 (Splitters for DI constants in Altivec registers): Likewise.
16435 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16436 (vsx_set_v4sf_p9): Likewise.
16437 (vsx_set_v4sf_p9_zero): Likewise.
16438 (vsx_insert_extract_v4sf_p9): Likewise.
16439 (vsx_insert_extract_v4sf_p9_2): Likewise.
16441 2017-07-25 Carl Love <cel@us.ibm.com>
16443 * doc/extend.texi: Update the built-in documentation file for the
16444 existing built-in functions
16445 vector signed char vec_cnttz (vector signed char);
16446 vector unsigned char vec_cnttz (vector unsigned char);
16447 vector signed short vec_cnttz (vector signed short);
16448 vector unsigned short vec_cnttz (vector unsigned short);
16449 vector signed int vec_cnttz (vector signed int);
16450 vector unsigned int vec_cnttz (vector unsigned int);
16451 vector signed long long vec_cnttz (vector signed long long);
16452 vector unsigned long long vec_cnttz (vector unsigned long long);
16454 2017-07-25 Andrew Pinski <apinski@cavium.com>
16456 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16457 accesses where the use is for the first operand of a BIT_INSERT.
16459 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16462 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16463 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16465 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16467 * config/i386/gstabs.h: Delete.
16468 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16470 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16472 * config/i386/i386.c (ix86_decompose_address): Do not check for
16473 register RTX when looking at index_reg or base_reg.
16474 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16476 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16478 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16479 to update EH info here.
16481 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16483 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16485 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16487 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16489 2017-07-25 Torsten Duwe <duwe@suse.de>
16491 * common.opt: Introduce -fpatchable-function-entry
16492 command line option, and its variables function_entry_patch_area_size
16493 and function_entry_patch_area_start.
16494 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16495 including a two-value parser.
16496 * target.def (print_patchable_function_entry): New target hook.
16497 * targhooks.h (default_print_patchable_function_entry): New function.
16498 * targhooks.c (default_print_patchable_function_entry): Likewise.
16499 * toplev.c (process_options): Switch off IPA-RA if
16500 patchable function entries are being generated.
16501 * varasm.c (assemble_start_function): Look at the
16502 patchable-function-entry command line switch and current
16503 function attributes and maybe generate NOP instructions by
16504 calling the print_patchable_function_entry hook.
16505 * doc/extend.texi: Document patchable_function_entry attribute.
16506 * doc/invoke.texi: Document -fpatchable_function_entry
16507 command line option.
16508 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16510 * doc/tm.texi: Re-generate.
16512 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16515 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16516 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16518 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16520 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16522 2017-07-25 Richard Biener <rguenther@suse.de>
16524 PR tree-optimization/81455
16525 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16526 not walk in cycles when looking for guards.
16528 2017-07-25 Richard Biener <rguenther@suse.de>
16530 PR tree-optimization/81529
16531 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16532 when optimizing backedge uses.
16534 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16536 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16538 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16539 to dl_section_ref. On AIX, append an expression to subtract
16540 the size of the section length to dl_section_ref.
16542 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16544 * configure.ac: If any of the config.* scripts fail, exit 1.
16545 * configure: Regenerate.
16547 2017-07-25 Richard Biener <rguenther@suse.de>
16549 PR middle-end/81546
16550 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16551 of immediate uses, be more verbose on errors.
16553 2017-07-25 Richard Biener <rguenther@suse.de>
16555 PR tree-optimization/81510
16556 * tree-vect-loop.c (vect_is_simple_reduction): When the
16557 reduction stmt is not inside the loop bail out.
16559 2017-07-25 Richard Biener <rguenther@suse.de>
16561 PR tree-optimization/81303
16562 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16563 profitability check against LOOP_VINFO_NITERSM1.
16565 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16567 * domwalk.c (cmp_bb_postorder): Simplify.
16568 (sort_bbs_postorder): New function. Use it...
16569 (dom_walker::walk): ...here to optimize common cases.
16571 2017-07-25 Martin Liska <mliska@suse.cz>
16574 * ipa-visibility.c (function_and_variable_visibility): Make the
16575 redirection just on target that supports aliasing.
16576 Fix GNU coding style.
16578 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16581 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16583 * config/arm/rtems.h: Likewise.
16584 * config/bfin/rtems.h: Likewise.
16585 * config/i386/rtemself.h: Likewise.
16586 * config/lm32/rtems.h: Likewise.
16587 * config/m32c/rtems.h: Likewise.
16588 * config/m68k/rtemself.h: Likewise.
16589 * config/microblaze/rtems.h: Likewise.
16590 * config/mips/rtems.h: Likewise.
16591 * config/moxie/rtems.h: Likewise.
16592 * config/nios2/rtems.h: Likewise.
16593 * config/powerpcspe/rtems.h: Likewise.
16594 * config/rs6000/rtems.h: Likewise.
16595 * config/rtems.h: Likewise.
16596 * config/sh/rtems.h: Likewise.
16597 * config/sh/rtemself.h: Likewise.
16598 * config/sparc/rtemself.h: Likewise.
16600 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16603 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16604 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16605 * tree-ssa-structalias.c (alias_get_name): Same.
16607 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16610 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16611 instructions if no du chain is found.
16613 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16615 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16617 2017-07-25 Richard Biener <rguenther@suse.de>
16619 PR middle-end/81505
16620 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16623 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16625 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16626 upper-regs options.
16627 (ISA_2_7_MASKS_SERVER): Likewise.
16628 (ISA_3_0_MASKS_IEEE): Likewise.
16629 (OTHER_P8_VECTOR_MASKS): Likewise.
16630 (OTHER_VSX_VECTOR_MASKS): Likewise.
16631 (POWERPC_MASKS): Likewise.
16632 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16633 duplicate list of options.
16634 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16635 explicit -mupper-regs options.
16636 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16637 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16638 alias for -mupper-regs-df.
16639 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16640 (rs6000_init_hard_regno_mode_ok): Likewise.
16641 (rs6000_option_override_internal): Likewise.
16642 (rs6000_opt_masks): Likewise.
16643 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16644 options in terms of whether -mvsx or -mpower8-vector was used.
16645 (TARGET_UPPER_REGS_DI): Likewise.
16646 (TARGET_UPPER_REGS_SF): Likewise.
16647 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16648 -mupper-regs-* options.
16650 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16652 * passes.c (emergency_dump_function): Print some empty lines and a
16653 header before the RTL dump.
16655 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16657 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16659 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16662 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16663 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16665 2017-07-24 Carl Love <cel@us.ibm.com>
16667 * config/rs6000/rs6000-c.c: Add support for built-in functions
16668 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16669 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16670 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16671 vec_extract_fp_from_shortl): Add defines for the two builtins.
16672 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16673 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16675 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16676 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16677 * doc/extend.texi: Update the built-in documentation file for the
16678 new built-in function.
16680 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16683 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16685 * doc/generic.texi: Likewise.
16686 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16687 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16689 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16691 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16692 (aarch64_mls_elt_merge<mode>): Likewise.
16694 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16696 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16697 having __cxa_atexit.
16699 2017-07-23 Michael Collison <michael.collison@arm.com>
16701 * config/arm/arm.c (arm_option_override): Deprecate
16702 use of -mstructure-size-boundary.
16703 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16704 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16706 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16709 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16710 Reduce cost estimate for direct moves.
16712 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16715 * config/i386/i386.c (ix86_option_override_internal): Disable
16716 BMI, BMI2 and TBM instructions for -m16.
16718 2017-07-21 Carl Love <cel@us.ibm.com>
16720 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16721 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16722 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16723 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16724 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16725 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16726 VMULOSW): New enum "unspec" values.
16727 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16728 altivec_vmulosw): New patterns.
16729 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16730 VMULOSW): Add definitions.
16732 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16734 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16735 (qdf24xx): Likewise.
16736 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16737 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16738 (AARCH64_FL_V8_1): Renumber.
16739 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16740 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16741 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16742 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16743 rdma to feature modifiers list.
16745 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16747 PR middle-end/56727
16748 * ipa-visibility (function_and_variable_visibility): Convert
16749 recursive PLT call to direct call if appropriate.
16751 2017-07-21 Andrew Pinski <apinski@cavium.com>
16753 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16754 operand 1 to see if the types precision matches.
16755 * fold-const.c (operand_equal_p): Likewise.
16757 2017-07-21 Richard Biener <rguenther@suse.de>
16759 PR tree-optimization/81303
16760 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16761 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16762 (vect_peeling_hash_get_lowest_cost): Adjust.
16763 (vect_enhance_data_refs_alignment): Likewise. Use
16764 vect_get_peeling_costs_all_drs to compute the penalty for no
16765 peeling to match up costs.
16767 2017-07-21 Richard Biener <rguenther@suse.de>
16769 PR tree-optimization/81500
16770 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16771 we didn't identify a reduction path.
16773 2017-07-21 Tom de Vries <tom@codesourcery.com>
16774 Cesar Philippidis <cesar@codesourcery.com>
16776 PR gcov-profile/81442
16777 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16780 2017-07-21 Tom de Vries <tom@codesourcery.com>
16783 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16785 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16786 nvptx_override_options_after_change.
16788 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16790 * dwarf2out.c (output_file_names): Avoid double testing for
16791 dwarf_version >= 5.
16793 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16795 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16797 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16799 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16801 (try_crossjump_to_edge): Do not punt on partitioned functions.
16803 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16805 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16806 Put all BBs reachable only via paths crossing cold region to cold
16808 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16810 2016-07-21 Richard Biener <rguenther@suse.de>
16812 PR tree-optimization/81303
16813 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16814 into account prologue and epilogue iterations when raising
16815 min_profitable_iters to sth at least covering one vector iteration.
16817 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16819 * config/arm/arm.c (arm_test_cpu_arch_dat):
16822 2017-07-20 Nathan Sidwell <nathan@acm.org>
16824 Remove TYPE_METHODS.
16825 * tree.h (TYPE_METHODS): Delete.
16826 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16827 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16828 (dbxout_type_methods): Scan TYPE_FIELDS.
16829 (dbxout_type): Don't check TYPE_METHODS here.
16830 * function.c (use_register_for_decl): Always ignore register for
16831 class types when not optimizing.
16832 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16833 * tree.c (free_lang_data_in_type): Stitch out member functions and
16834 templates from TYPE_FIELDS.
16835 (build_distinct_type_copy, verify_type_variant,
16836 verify_type): Member fns are on TYPE_FIELDS.
16837 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16838 * tree-pretty-print.c (dump_generic_node): Likewise.
16840 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16843 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16844 V2TImode and V4TImode.
16845 (ix86_expand_vector_extract): Likewise.
16846 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16847 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16848 (ssescalarmode): Handle V4TImode and V2TImode.
16849 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16850 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16851 (VEXTRACTI128_MODE): New mode iterator.
16852 (splitter for *vec_extractv?ti first element): New.
16853 (VEC_INIT_MODE): New mode iterator.
16854 (vec_init<mode>): Consolidate 3 expanders into one using
16855 VEC_INIT_MODE mode iterator.
16857 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16859 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16860 non_spilled_static_chain_regno_p.
16862 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16864 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16866 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16868 * bb-reorder.c (connect_traces): Allow copying of blocks within
16871 2017-07-20 Richard Biener <rguenther@suse.de>
16873 * gimple.h (gimple_phi_result): Add gphi * overload.
16874 (gimple_phi_result_ptr): Likewise.
16875 (gimple_phi_arg): Likewise. Adjust index assert to only
16876 allow actual argument accesses rather than all slots available
16878 (gimple_phi_arg_def): Add gphi * overload.
16879 * tree-phinodes.c (make_phi_node): Initialize only actual
16881 (resize_phi_node): Clear memory not covered by old node,
16882 do not initialize excess argument slots.
16883 (reserve_phi_args_for_new_edge): Initialize new argument slot
16886 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16888 PR tree-optimization/81388
16890 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16892 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16893 by removing computation of may_be_zero.
16895 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16896 Tom de Vries <tom@codesourcery.com>
16898 PR middle-end/81030
16899 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16900 when gimple level profile disagrees with what RTL expander did.
16902 2017-07-20 Richard Biener <rguenther@suse.de>
16904 PR tree-optimization/61171
16905 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16906 (vect_analyze_stmt): Add slp instance parameter.
16907 (vectorizable_reduction): Likewise.
16908 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16909 (vect_is_simple_reduction): Deal with chains not detected
16910 as SLP reduction chain, specifically not properly associated
16911 chains containing a mix of plus/minus.
16912 (get_reduction_op): Remove.
16913 (get_initial_defs_for_reduction): Simplify, pass in whether
16914 this is a reduction chain, pass in the SLP node for the PHIs.
16915 (vect_create_epilog_for_reduction): Get the SLP instance as
16917 (vectorizable_reduction): Get the SLP instance as arg.
16918 During analysis remember the SLP node with the PHIs in the
16919 instance. Simplify getting at the vectorized reduction PHIs.
16920 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16921 through SLP instance.
16922 (vect_slp_analyze_operations): Likewise.
16923 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16924 (vect_transform_stmt): Likewise.
16926 2017-07-20 Tom de Vries <tom@codesourcery.com>
16928 PR tree-optimization/81489
16929 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16930 read of phi arg location to before loop that modifies phi.
16932 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16934 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16937 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16939 PR middle-end/81331
16940 * except.c (execute): Fix ordering issue.
16942 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16944 PR rtl-optimization/81423
16945 * combine.c (make_compound_operation_int): Don't try to optimize
16946 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16948 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16950 PR rtl-optimization/81423
16951 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16952 with a constant that is -1 in the truncated to mode.
16954 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16956 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16957 (determine_unlikely_bbs): ... here.
16958 * predict.h (propagate_unlikely_bbs_forward): Declare.
16959 * cfgexpand.c (pass_expand::execute): Use it.
16960 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16962 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16963 propagate_unlikely_bbs_forward.
16965 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16967 PR middle-end/81331
16968 * except.c (maybe_add_nop_after_section_switch): New function.
16971 2017-07-19 Tom de Vries <tom@codesourcery.com>
16973 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16975 2017-07-19 Tom de Vries <tom@codesourcery.com>
16977 * gimple.h (gimple_phi_arg): Make assert more strict.
16979 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16981 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16982 * config/rs6000/mmintrin.h: New file.
16983 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16985 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16987 PR tree-optimization/81346
16988 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16990 2017-07-19 Tom de Vries <tom@codesourcery.com>
16992 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16994 2017-07-19 Tom de Vries <tom@codesourcery.com>
16996 * config/nvptx/nvptx-modes.def: Add V2DImode.
16997 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16998 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16999 (nvptx_output_mov_insn): Handle lack of mov.b128.
17000 (nvptx_print_operand): Handle 'H' and 'L' codes.
17001 (nvptx_vector_mode_supported): Allow V2DImode.
17002 (nvptx_preferred_simd_mode): New function.
17003 (nvptx_data_alignment): New function.
17004 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17005 nvptx_preferred_simd_mode.
17006 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17008 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17010 2017-07-19 Tom de Vries <tom@codesourcery.com>
17012 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17013 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17014 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17015 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17016 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17017 (mov<VECIM>_insn): New define_insn.
17018 (define_expand "mov<VECIM>): New define_expand.
17020 2017-07-19 Tom de Vries <tom@codesourcery.com>
17022 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17024 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17026 PR tree-optimization/81346
17027 * fold-const.h (fold_div_compare, range_check_type): Declare.
17028 * fold-const.c (range_check_type): New function.
17029 (build_range_check): Use range_check_type.
17030 (fold_div_compare): No longer static, rewritten into
17031 a match.pd helper function.
17032 (fold_comparison): Don't call fold_div_compare here.
17033 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17034 as helper function.
17036 2017-07-19 Nathan Sidwell <nathan@acm.org>
17038 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17039 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17040 * tree.c (find_decls_types_r, verify_type): Use
17041 TYPE_{MIN,MAX}_VALUE_RAW.
17042 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17043 (hash_tree): Likewise.
17044 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17046 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17049 2017-07-18 Tom de Vries <tom@codesourcery.com>
17051 PR middle-end/81464
17052 * omp-expand.c (expand_omp_for_static_chunk): Handle
17053 equal-argument loop exit phi.
17055 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17058 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17059 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17060 operand 2 predicate.
17061 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17062 operand 2 predicate.
17063 (ror,rol -> rorx splitters): Use const_int_operand as
17064 operand 2 predicate.
17066 2017-06-18 Richard Biener <rguenther@suse.de>
17068 PR tree-optimization/81410
17069 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17070 the gap in the ! slp_perm SLP case after each group.
17072 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17074 PR middle-end/81463
17075 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17078 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17080 PR middle-end/81462
17081 * predict.c (set_even_probabilities): Cleanup; do not affect
17082 probabilities that are already known.
17083 (combine_predictions_for_bb): Call even when count is set.
17085 2017-07-18 Nathan Sidwell <nathan@acm.org>
17087 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17090 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17093 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17094 optimization for loop niter analysis.
17096 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17099 * config/avr/avr.c (avr_optimize_casesi): Don't use
17100 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17102 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17104 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17105 body_cost_vec from _vect_peel_extended_info.
17106 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17107 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17110 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17112 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17114 2017-07-18 Richard Biener <rguenther@suse.de>
17116 PR tree-optimization/80620
17117 PR tree-optimization/81403
17118 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17119 info when re-using a VN table entry.
17121 2017-07-18 Richard Biener <rguenther@suse.de>
17123 PR tree-optimization/81418
17124 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17125 vectype_in. Verify that with lane-reducing reduction operations
17126 we have a single def-use cycle.
17128 2017-07-17 Carl Love <cel@us.ibm.com>
17130 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17132 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17133 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17134 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17135 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17136 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17137 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17138 VMULOSW): New enum "unspec" values.
17139 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17140 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17141 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17142 altivec_vmulosw): New patterns.
17143 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17144 VMULOSW): Add definitions.
17145 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17147 * config/alpha/alpha.c: Include predict.h.
17149 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17151 * tree-vrp.c (compare_assert_loc): Fix comparison function
17152 to return predictable results.
17154 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17156 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17158 (subdi3): Likewise.
17159 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17160 * doc/invoke.texi (mexpand-adddi): Update text.
17162 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17164 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17165 that also clobbers the CC register. The old expand code is moved
17167 (*arc_clzsi2): ... here.
17168 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17169 the CC register. The old expand code is moved to ...
17170 (arc_ctzsi2): ... here.
17172 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17174 * config/arc/arc.opt (mindexed-loads): Use initial value
17175 TARGET_INDEXED_LOADS_DEFAULT.
17176 (mauto-modify-reg): Use initial value
17177 TARGET_AUTO_MODIFY_REG_DEFAULT.
17178 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17179 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17180 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17181 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17183 2017-07-17 Martin Liska <mliska@suse.cz>
17186 * opts.c (finish_options): Do not allow -fgnu-tm
17187 w/ -fsanitize={kernel-,}address. Say sorry.
17189 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17192 * tree-loop-distribution.c (classify_partition): Only assert on
17193 numer of iterations.
17194 (merge_dep_scc_partitions): Delete prameter. Update function call.
17195 (distribute_loop): Remove code handling loop with unknown niters.
17196 (pass_loop_distribution::execute): Skip loop with unknown niters.
17198 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17201 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17202 function sort_partitions_by_post_order.
17204 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17206 PR tree-optimization/81374
17207 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17208 the max index of basic blocks, rather than number of basic blocks.
17210 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17212 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17214 (arc_legitimate_pic_operand_p): Likewise.
17215 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17217 (arc_needs_pcl_p): Likewise.
17218 (arc_legitimate_pc_offset_p): Likewise.
17219 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17220 function is also used in constrains.md.
17221 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17222 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17223 PLUS. Only return true/false in known cases, otherwise assert.
17224 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17225 is already called in arc_legitimate_constant_p.
17226 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17228 (LEGITIMATE_PIC_OPERAND_P): Use
17229 arc_raw_symbolic_reference_mentioned_p function.
17230 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17235 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17236 Andrew Burgess <andrew.burgess@embecosm.com>
17238 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17239 (arc_return_address_register): New function.
17240 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17241 (arc_handle_fndecl_attribute): Add naked attribute.
17242 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17243 (TARGET_WARN_FUNC_RETURN): Likewise.
17244 (arc_allocate_stack_slots_for_args): New function.
17245 (arc_warn_func_return): Likewise.
17246 (machine_function): Change type fn_type.
17247 (arc_compute_function_type): Consider new naked function type,
17248 change function return type.
17249 (arc_must_save_register): Adapt to handle new
17250 arc_compute_function_type's return type.
17251 (arc_expand_prologue): Likewise.
17252 (arc_expand_epilogue): Likewise.
17253 (arc_return_address_regs): Delete.
17254 (arc_return_address_register): New function.
17255 (arc_epilogue_uses): Use above function.
17256 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17257 (arc_function_type): Change encoding, add naked type.
17258 (ARC_INTERRUPT_P): Change to handle the new encoding.
17259 (ARC_FAST_INTERRUPT_P): Likewise.
17260 (ARC_NORMAL_P): Define.
17261 (ARC_NAKED_P): Likewise.
17262 (arc_compute_function_type): Delete prototype.
17263 * config/arc/arc.md (in_ret_delay_slot): Use
17264 arc_return_address_register function.
17265 (simple_return): Likewise.
17266 (p_return_i): Likewise.
17268 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17270 PR tree-optimization/81428
17271 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17272 can't be built for those types.
17274 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17276 Remove stuff dead since r239246.
17278 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17279 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17280 (avr_inform_devices): Remove dead stuff.
17282 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17284 * config/arm/arm_neon.h: Fix softp typo.
17286 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17288 PR tree-optimization/81365
17289 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17290 aggregate moves onto bb predecessor edges, make sure there are no
17291 loads that could alias the lhs in between the start of bb and the
17294 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17297 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17298 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17299 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17301 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17303 PR tree-optimization/81396
17304 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17305 (init_symbolic_number): Initialize it to 1.
17306 (perform_symbolic_merge): Add n_ops from both operands into the new
17308 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17309 without base_addr as useless if they need more than one operation.
17310 (bswap_replace): Handle !bswap case for NULL base_addr.
17312 2017-07-17 Tom de Vries <tom@codesourcery.com>
17315 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17318 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17320 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17321 conditional builtin define __FIX_LEON3FT_B2BST.
17323 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17325 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17326 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17329 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17331 PR rtl-optimization/81424
17332 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17333 to remove potential trapping from operands if -fnon-call-exceptions.
17335 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17337 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17338 profile_proability for scalling.
17339 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17341 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17343 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17345 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17347 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17348 fixpoint arithmetics.
17350 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17352 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17353 fixpoint arithmetics.
17355 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17357 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17358 fixpoint arithmetics.
17360 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17362 * profile-count.h (profile_probability::from_reg_br_prob_note,
17363 profile_probability::to_reg_br_prob_note): New functions.
17364 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17365 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17366 * predict.c (probability_reliable_p): Update.
17367 (edge_probability_reliable_p): Update.
17368 (br_prob_note_reliable_p): Update.
17369 (invert_br_probabilities): Update.
17370 (add_reg_br_prob_note): New function.
17371 (combine_predictions_for_insn): Update.
17372 * asan.c (asan_clear_shadow): Update.
17373 * cfgbuild.c (compute_outgoing_frequencies): Update.
17374 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17375 (update_br_prob_note): Update.
17376 (rtl_verify_edges): Update.
17377 (purge_dead_edges): Update.
17378 (fixup_reorder_chain): Update.
17379 * emit-rtl.c (try_split): Update.
17380 * ifcvt.c (cond_exec_process_insns): Update.
17381 (cond_exec_process_if_block): Update.
17382 (dead_or_predicable): Update.
17383 * internal-fn.c (expand_addsub_overflow): Update.
17384 (expand_neg_overflow): Update.
17385 (expand_mul_overflow): Update.
17386 * loop-doloop.c (doloop_modify): Update.
17387 * loop-unroll.c (compare_and_jump_seq): Update.
17388 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17389 * predict.h: Update.
17390 * reorg.c (mostly_true_jump): Update.
17392 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17393 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17394 * config/arc/arc.c: (emit_unlikely_jump): Update.
17395 * config/arm/arm.c: (emit_unlikely_jump): Update.
17396 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17397 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17398 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17399 (ix86_print_operand): Update.
17400 (ix86_split_fp_branch): Update.
17401 (predict_jump): Update.
17402 * config/ia64/ia64.c (ia64_print_operand): Update.
17403 * config/mmix/mmix.c (mmix_print_operand): Update.
17404 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17405 (rs6000_expand_split_stack_prologue): Update.
17406 * config/rs6000/rs6000.c: Update.
17407 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17408 (s390_expand_vec_movstr): Update.
17409 (s390_expand_cs_tdsi): Update.
17410 (s390_expand_split_stack_prologue): Update.
17411 * config/sh/sh.c (sh_print_operand): Update.
17412 (expand_cbranchsi4): Update.
17413 (expand_cbranchdi4): Update.
17414 * config/sparc/sparc.c (output_v9branch): Update.
17415 * config/spu/spu.c (get_branch_target): Update.
17416 (ea_load_store_inline): Update.
17417 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17418 * config/tilepro/tilepro.c: Update.
17420 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17422 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17423 (gimplify_save_expr): Likewise.
17425 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17427 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17429 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17431 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17433 * timevar.def (TV_IPA_FNSUMMARY): Define.
17435 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17437 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17438 to back store errata sensitive sequence from being generated.
17439 (sqrtdf2_fix): Likewise.
17441 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17443 * tree-ssa-threadupdate.c (compute_path_counts,
17444 update_joiner_offpath_counts): Use profile_probability.
17446 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17449 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17451 * config/arm/arm-c.c (arm_cpu_builtins): Define
17452 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17454 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17456 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17457 array entries to represent __ieee128 versions of the
17458 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17459 scalar_extract_sig, and scalar_insert_exp built-in functions.
17460 (altivec_resolve_overloaded_builtin): Add special case handling
17461 for the __builtin_scalar_insert_exp function, as represented by
17462 the P9V_BUILTIN_VEC_VSIEDP constant.
17463 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17464 exponent support for __ieee128 argument.
17465 (VSESQP): Add scalar extract signature support for __ieee128
17467 (VSTDCNQP): Add scalar test negative support for __ieee128
17469 (VSIEQP): Add scalar insert exponent support for __int128 argument
17470 with __ieee128 result.
17471 (VSIEQPF): Add scalar insert exponent support for __ieee128
17472 argument with __ieee128 result.
17473 (VSTDCQP): Add scalar test data class support for __ieee128
17475 (VSTDCNQP): Add overload support for scalar test negative with
17476 __ieee128 argument.
17477 (VSTDCQP): Add overload support for scalar test data class
17478 __ieee128 argument.
17479 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17480 UNSPEC_VSX_SXSIGDP.
17481 (UNSPEC_VSX_SIEXPQP): New constant.
17482 (xsxexpqp): New insn for VSX scalar extract exponent quad
17484 (xsxsigqp): New insn for VSX scalar extract significand quad
17486 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17487 precision with floating point argument.
17488 (xststdcqp): New expand for VSX scalar test data class quad
17490 (xststdcnegqp): New expand for VSX scalar test negative quad
17492 (xststdcqp): New insn to match expansions for VSX scalar test data
17493 class quad precision and VSX scalar test negative quad precision.
17494 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17495 special case operand checking to enforce that second operand of
17496 VSX scalar test data class with quad precision argument is a 7-bit
17498 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17499 prototypes and descriptions of __ieee128 versions of
17500 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17501 scalar_test_data_class, and scalar_test_neg built-in functions.
17503 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17505 PR tree-optimization/81162
17506 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17507 replace a negate with an add.
17509 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17511 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17513 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17515 * config/arm/arm-c.c (arm_cpu_builtins): Define
17516 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17518 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17520 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17521 (armv8-r): Set ARM Cortex-R52 as default CPU.
17522 * config/arm/arm-tables.opt: Regenerate.
17523 * config/arm/arm-tune.md: Regenerate.
17524 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17526 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17527 extension for -mcpu=cortex-r52.
17529 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17531 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17532 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17533 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17534 (fp-armv8): Define it as FP_ARMv8 only.
17535 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17536 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17538 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17539 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17540 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17541 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17542 than TARGET_FPU_ARMV8.
17543 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17544 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17545 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17547 * config/arm/neon.md (neon_vrint): Likewise.
17548 (neon_vcvt): Likewise.
17549 (neon_<fmaxmin_op><mode>): Likewise.
17550 (<fmaxmin><mode>3): Likewise.
17551 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17552 * config/arm/predicates.md (arm_cond_move_operator): Check against
17553 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17555 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17557 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17558 to top of function.
17560 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17562 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17563 loop in comment with memset.
17565 2017-07-14 Martin Liska <mliska@suse.cz>
17567 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17568 * dwarf2out.c (is_java): Remove the function.
17569 (output_pubname): Remove usage of the function.
17570 (lower_bound_default): Remove usage of DW_LANG_Java.
17571 (gen_compile_unit_die): Likewise.
17572 * gcc.c: Remove compiler defaults for .java and .zip files.
17573 * gimple-expr.c (remove_suffix): Change as there's no longer
17574 extension than 4-letter one.
17575 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17576 (gimplify_save_expr): Likewise.
17577 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17578 as it's possible even for other languages than Java.
17579 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17580 * lto-opts.c (lto_write_options): Remove reference to Java.
17581 * opts.c (strip_off_ending): Update file extension handling.
17582 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17583 * tree-eh.c (lower_resx): Likewise.
17584 * tree.c (free_lang_data_in_type): Remove dead code.
17585 (find_decls_types_r): Likewise.
17586 (build_common_builtin_nodes): Remove Java from a comment.
17587 (verify_type): Remove dead code.
17588 * varasm.c (assemble_external): Remove Java from a comment.
17590 2017-07-14 Martin Liska <mliska@suse.cz>
17592 * opts.c (finish_options): Add quotes.
17593 (common_handle_option): Likewise.
17595 2017-07-14 Martin Liska <mliska@suse.cz>
17597 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17598 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17599 Remove N_SO_PASCAL.
17600 * dwarf2out.c (lower_bound_default): Do not handle
17602 (gen_compile_unit_die): Likewise.
17603 * gcc.c: Remove default extension binding for GNU Pascal.
17604 * stmt.c: Remove Pascal language from a comment.
17605 * xcoffout.c: Likewise.
17607 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17610 * diagnostic-show-locus.c (fixit_cmp): New function.
17611 (layout::layout): Sort m_fixit_hints.
17612 (column_range::column_range): Assert that the values are valid.
17613 (struct char_span): New struct.
17614 (correction::overwrite): New method.
17615 (struct source_line): New struct.
17616 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17617 calls in terms of classes source_line and char_span, and
17618 correction::overwrite.
17619 (selftest::test_overlapped_fixit_printing_2): New function.
17620 (selftest::diagnostic_show_locus_c_tests): Call it.
17622 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17624 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17625 early if there is no lhs.
17627 2017-07-13 Martin Liska <mliska@suse.cz>
17629 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17630 (gen_reference_type_die): Likewise.
17631 * stor-layout.c: Remove Pascal-related comment.
17633 2017-07-13 Martin Liska <mliska@suse.cz>
17635 * opts.c (finish_options): Add quotes to error messages.
17636 (parse_sanitizer_options): Likewise.
17638 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17640 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17642 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17644 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17646 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17648 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17650 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17652 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17655 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17656 provides the hardware capability bits, define the macro
17657 __BUILTIN_CPU_SUPPORTS__.
17658 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17659 if GLIBC does not provide the hardware capability bits. Add a
17660 gcc_unreachable call if the built-in cpu function is neither
17661 __builtin_cpu_is nor __builtin_cpu_supports.
17662 (rs6000_get_function_versions_dispatcher): Change the warning
17663 that an old GLIBC is used which does not export the capability
17664 bits to be an error.
17665 * doc/extend.texi (target_clones attribute): Document the
17666 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17667 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17668 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17669 the macros defined by GCC if the newer GLIBC is available.
17671 2017-07-12 Jeff Law <law@redhat.com>
17673 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17674 remaining includes slightly.
17675 * config/riscv/riscv-builtins.c: Include profile-count.h.
17677 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17680 * config/avr/avr.c (avr_set_current_function): In diagnostic
17681 messages: Quote keywords and (parts of) identifiers.
17682 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17685 2017-07-12 Carl Love <cel@us.ibm.com>
17687 * config/rs6000/rs6000-c.c: Add support for built-in functions
17688 vector bool char vec_revb (vector bool char);
17689 vector bool short vec_revb (vector short char);
17690 vector bool int vec_revb (vector bool int);
17691 vector bool long long vec_revb (vector bool long long);
17692 * doc/extend.texi: Update the built-in documentation file for the
17693 new built-in functions.
17695 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17697 * config/s390/s390.md: Remove movcc splitter.
17699 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17701 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17702 load/store on condition.
17704 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17707 * config/avr/avr.c (avr_encode_section_info)
17708 [progmem && !TREE_READONLY]: Error if progmem object needs
17711 2017-07-11 Michael Collison <michael.collison@arm.com>
17713 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17716 2017-07-11 Carl Love <cel@us.ibm.com>
17718 * config/rs6000/rs6000-c.c: Add support for builtins
17719 vector unsigned int vec_parity_lsbb (vector signed int);
17720 vector unsigned int vec_parity_lsbb (vector unsigned int);
17721 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17722 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17723 vector unsigned long long vec_parity_lsbb (vector signed long long);
17724 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17725 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17726 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17727 * doc/extend.texi: Update the built-in documentation file for the
17728 new built-in functions.
17730 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17732 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17733 (layout::m_primary_loc): New field.
17734 (layout::layout): Initialize new field. Move location filtering
17735 logic from here to...
17736 (layout::maybe_add_location_range): ...this new method. Add
17737 support for filtering to just the lines already specified by other
17739 (layout::will_show_line_p): New method.
17740 (gcc_rich_location::add_location_if_nearby): New method.
17741 (selftest::test_add_location_if_nearby): New test function.
17742 (selftest::diagnostic_show_locus_c_tests): Call it.
17743 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17746 2017-07-11 Tom de Vries <tom@codesourcery.com>
17748 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17749 (bb_first_real_insn): New function.
17750 (nvptx_single): Add extra initialization of broadcasted condition
17753 2017-07-11 Nathan Sidwell <nathan@acm.org>
17755 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17757 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17759 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17760 Binutils doc as TEXI will mess them up.
17761 * doc/invoke.texi (AVR Options): Same here.
17763 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17765 * config/sparc/sparc.opt (mfix-ut700): New option.
17766 (mfix-gr712rc): Likewise.
17767 (sparc_fix_b2bst): New variable.
17768 * doc/invoke.texi (SPARC options): Document them.
17769 (ARM options): Fix warnings.
17770 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17771 instructions to prevent sequences that can trigger the store-store
17772 errata for certain LEON3FT processors.
17773 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17774 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17775 * config/sparc/sparc.md (fix_b2bst): New attribute.
17776 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17778 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17781 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17783 (*rsqrtsf2_sse): Ditto.
17785 (div<mode>3): Macroize insn from divdf3 and divsf3
17786 using MODEF mode iterator.
17788 2017-07-10 Martin Sebor <msebor@redhat.com>
17790 PR tree-optimization/80397
17791 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17792 instead of testing for equality to INTEGER_TYPE.
17794 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17796 * config.gcc: Remove uclibc from arc target spec.
17798 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17800 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17802 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17805 * lto-wrapper.c (remove_option): New function.
17806 (merge_and_complain): Merge PIC/PIE options more realistically.
17808 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17810 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17814 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17815 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17816 * config.in: Regenerate.
17817 * configure: Regenerate.
17818 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17819 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17820 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17821 (TARGET_GASISR_PROLOGUES): ...target mask.
17822 * common/config/avr/avr-common.c
17823 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17824 Set -mgas-isr-prologues.
17825 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17826 INSERT_PASS_BEFORE for it.
17827 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17828 * config/avr/avr.c (avr_option_override)
17829 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17830 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17831 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17832 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17833 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17835 (make_avr_pass_pre_proep): New function.
17836 (emit_push_sfr) <treg>: Add argument to function and use it
17837 instead of TMP_REG.
17838 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17839 and set machine->gasisr.yes.
17840 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17841 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17842 __gcc_isr.n_pushed to .L__stack_usage.
17843 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17844 (avr_asm_final_postscan_insn): ...this new static function.
17845 * config/avr/avr.h (machine_function)
17846 <is_no_gccisr, use_L__stack_usage>: New fields.
17847 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17848 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17849 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17850 (gasisr, *gasisr): New expander and insn.
17851 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17852 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17853 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17855 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17857 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17860 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17862 Move jump-tables out of .text again.
17865 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17866 (ASM_OUTPUT_ADDR_VEC): New function.
17867 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17868 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17869 INSN_ADDRESSes as asm comment.
17870 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17871 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17872 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17873 * config/avr/avr.md (*tablejump): Adjust comment.
17874 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17875 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17877 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17878 (avr_output_addr_vec): New proto.
17879 (avr_log_t) <insn_addresses>: New field.
17881 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17884 * config/i386/i386.c (ix86_function_arg_advance): Set
17885 outgoing_args_on_stack to true if there are outgoing arguments
17887 (ix86_function_arg): Likewise.
17888 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17889 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17890 * config/i386/i386.h (machine_function): Add
17891 outgoing_args_on_stack.
17893 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17895 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17896 supporting pthreds.
17897 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17899 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17901 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17902 (REAL_H): Remove $(MACHMODE_H).
17903 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17905 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17906 $(MACHMODE_H) and double-int.h.
17907 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17909 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17912 2017-07-07 Andrew Pinski <apinski@cavium.com>
17914 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17915 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17917 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17919 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17920 Add warning if GCC was not configured to link against a GLIBC that
17921 exports the hardware capability bits.
17922 (make_resolver_func): Make resolver function private and not a
17923 COMDAT function. Create the name with clone_function_name instead
17924 of make_unique_name.
17927 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17928 correct operand in doing the split.
17930 2017-07-07 Carl Love <cel@us.ibm.com>
17932 * config/rs6000/rs6000-c: Add support for built-in function
17933 vector unsigned short vec_pack_to_short_fp32 (vector float,
17935 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17936 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17937 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17938 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17939 (convert_4f32_8i16): Add define_expand.
17940 * doc/extend.texi: Update the built-in documentation file for the
17941 new built-in function.
17943 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17945 * config/sparc/m8.md: New file.
17946 * config/sparc/sparc.md: Include m8.md.
17948 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17950 * config/sparc/sparc.opt: New option -mvis4b.
17951 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17952 (sparc_option_override): Handle VIS4B.
17953 (enum sparc_builtins): Define
17954 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17955 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17956 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17957 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17958 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17959 (check_constant_argument): New function.
17960 (sparc_vis_init_builtins): Define builtins
17961 __builtin_vis_dictunpack{8,16,32},
17962 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17963 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17964 __builtin_vis_fpcmpde{8,16,32}shl and
17965 __builtin_vis_fpcmpur{8,16,32}shl.
17966 (sparc_expand_builtin): Check that the constant operands to
17967 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17968 constant and in range.
17969 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17971 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17972 (SPARC_IMM5_P): Likewise.
17973 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17974 (enabled): Handle vis4b.
17975 (UNSPEC_DICTUNPACK): New unspec.
17976 (UNSPEC_FPCMPSHL): Likewise.
17977 (UNSPEC_FPUCMPSHL): Likewise.
17978 (UNSPEC_FPCMPDESHL): Likewise.
17979 (UNSPEC_FPCMPURSHL): Likewise.
17980 (cpu_feature): New CPU feature `vis4b'.
17981 (dictunpack{8,16,32}): New insns.
17982 (FPCSMODE): New mode iterator.
17983 (fpcscond): New code iterator.
17984 (fpcsucond): Likewise.
17985 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17986 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17987 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17988 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17989 * config/sparc/constraints.md: Define constraints `q' for unsigned
17990 2-bit integer constants and `t' for unsigned 5-bit integer
17992 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17994 (imm5_operand_dictunpack16): Likewise.
17995 (imm5_operand_dictunpack32): Likewise.
17996 (imm2_operand): Likewise.
17997 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17998 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17999 ditunpack* and fpcmp*shl builtins.
18001 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18003 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18004 * config.in: Add HAVE_AS_SPARC6 define.
18005 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18007 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18009 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18010 (CPP_CPU_SPEC): Handle m8.
18011 (ASM_CPU_SPEC): Likewise.
18012 * config/sparc/sparc-opts.h (enum processor_type): Add
18014 * config/sparc/sparc.c (m8_costs): New struct.
18015 (sparc_option_override): Handle TARGET_CPU_m8.
18016 (sparc32_initialize_trampoline): Likewise.
18017 (sparc64_initialize_trampoline): Likewise.
18018 (sparc_issue_rate): Likewise.
18019 (sparc_register_move_cost): Likewise.
18020 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18021 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18022 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18023 (CPP_CPU_SPEC): Handle M8.
18024 (ASM_CPU_SPEC): Likewise.
18025 (AS_M8_FLAG): Define.
18026 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18027 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18028 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18030 * configure: Regenerate.
18031 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18034 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18036 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18038 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18039 ("*movdi_insn_sp32"): Do not set v3pipe.
18040 ("*movsi_insn"): Likewise.
18041 ("*movdi_insn_sp64"): Likewise.
18042 ("*movsf_insn"): Likewise.
18043 ("*movdf_insn_sp32"): Likewise.
18044 ("*movdf_insn_sp64"): Likewise.
18045 ("*zero_extendsidi2_insn_sp64"): Likewise.
18046 ("*sign_extendsidi2_insn"): Likewise.
18047 ("*mov<VM32:mode>_insn"): Likewise.
18048 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18049 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18050 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18051 ("<vlop:code><VL:mode>3"): Likewise.
18052 ("*not_<vlop:code><VL:mode>3"): Likewise.
18053 ("*nand<VL:mode>_vis"): Likewise.
18054 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18055 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18056 ("one_cmpl<VL:mode>2"): Likewise.
18057 ("faligndata<VM64:mode>_vis"): Likewise.
18058 ("alignaddrsi_vis"): Likewise.
18059 ("alignaddrdi_vis"): Likweise.
18060 ("alignaddrlsi_vis"): Likewise.
18061 ("alignaddrldi_vis"): Likewise.
18062 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18063 ("bmaskdi_vis"): Likewise.
18064 ("bmasksi_vis"): Likewise.
18065 ("bshuffle<VM64:mode>_vis"): Likewise.
18066 ("cmask8<P:mode>_vis"): Likewise.
18067 ("cmask16<P:mode>_vis"): Likewise.
18068 ("cmask32<P:mode>_vis"): Likewise.
18069 ("pdistn<P:mode>_vis"): Likewise.
18070 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18072 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18074 * config/sparc/sparc.md ("subtype"): New insn attribute.
18075 ("*wrgsr_sp64"): Set insn subtype.
18076 ("*rdgsr_sp64"): Likewise.
18077 ("alignaddrsi_vis"): Likewise.
18078 ("alignaddrdi_vis"): Likewise.
18079 ("alignaddrlsi_vis"): Likewise.
18080 ("alignaddrldi_vis"): Likewise.
18081 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18082 ("fexpand_vis"): Likewise.
18083 ("fpmerge_vis"): Likewise.
18084 ("faligndata<VM64:mode>_vis"): Likewise.
18085 ("bshuffle<VM64:mode>_vis"): Likewise.
18086 ("cmask8<P:mode>_vis"): Likewise.
18087 ("cmask16<P:mode>_vis"): Likewise.
18088 ("cmask32<P:mode>_vis"): Likewise.
18089 ("fchksm16_vis"): Likewise.
18090 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18091 ("fmean16_vis"): Likewise.
18092 ("fp<plusminus_insn>64_vis"): Likewise.
18093 ("<plusminus_insn>v8qi3"): Likewise.
18094 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18095 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18096 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18097 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18098 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18099 ("*movqi_insn"): Likewise.
18100 ("*movhi_insn"): Likewise.
18101 ("*movsi_insn"): Likewise.
18102 ("movsi_pic_gotdata_op"): Likewise.
18103 ("*movdi_insn_sp32"): Likewise.
18104 ("*movdi_insn_sp64"): Likewise.
18105 ("movdi_pic_gotdata_op"): Likewise.
18106 ("*movsf_insn"): Likewise.
18107 ("*movdf_insn_sp32"): Likewise.
18108 ("*movdf_insn_sp64"): Likewise.
18109 ("*zero_extendhisi2_insn"): Likewise.
18110 ("*zero_extendqihi2_insn"): Likewise.
18111 ("*zero_extendqisi2_insn"): Likewise.
18112 ("*zero_extendqidi2_insn"): Likewise.
18113 ("*zero_extendhidi2_insn"): Likewise.
18114 ("*zero_extendsidi2_insn_sp64"): Likewise.
18115 ("ldfsr"): Likewise.
18116 ("prefetch_64"): Likewise.
18117 ("prefetch_32"): Likewise.
18118 ("tie_ld32"): Likewise.
18119 ("tie_ld64"): Likewise.
18120 ("*tldo_ldub_sp32"): Likewise.
18121 ("*tldo_ldub1_sp32"): Likewise.
18122 ("*tldo_ldub2_sp32"): Likewise.
18123 ("*tldo_ldub_sp64"): Likewise.
18124 ("*tldo_ldub1_sp64"): Likewise.
18125 ("*tldo_ldub2_sp64"): Likewise.
18126 ("*tldo_ldub3_sp64"): Likewise.
18127 ("*tldo_lduh_sp32"): Likewise.
18128 ("*tldo_lduh1_sp32"): Likewise.
18129 ("*tldo_lduh_sp64"): Likewise.
18130 ("*tldo_lduh1_sp64"): Likewise.
18131 ("*tldo_lduh2_sp64"): Likewise.
18132 ("*tldo_lduw_sp32"): Likewise.
18133 ("*tldo_lduw_sp64"): Likewise.
18134 ("*tldo_lduw1_sp64"): Likewise.
18135 ("*tldo_ldx_sp64"): Likewise.
18136 ("*mov<VM32:mode>_insn"): Likewise.
18137 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18138 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18140 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18142 * config/sparc/sparc.md ("type"): New insn type viscmp.
18143 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18145 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18146 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18147 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18148 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18150 ("n7_vis_logical_11cycle"): Likewise.
18151 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18152 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18153 * config/sparc/niagara.md ("niag_vis"): Likewise.
18154 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18155 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18157 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18159 * config/sparc/sparc.md: New instruction type `bmask'.
18160 (bmaskdi_vis): Use the `bmask' type.
18161 (bmasksi_vis): Likewise.
18162 * config/sparc/ultra3.md (us3_array): Likewise.
18163 * config/sparc/niagara7.md (n7_array): Likewise.
18164 * config/sparc/niagara4.md (n4_array): Likewise.
18165 * config/sparc/niagara2.md (niag2_vis): Likewise.
18166 (niag3_vis): Likewise.
18167 * config/sparc/niagara.md (niag_vis): Likewise.
18169 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18171 * ipa-comdats.c: Remove optimize check from gate.
18172 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18173 for functions not optimized.
18174 (ipa_fn_summary_read): Skip optimize check.
18175 (ipa_fn_summary_write): Likewise.
18176 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18178 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18180 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18181 (check_callers): Give up on caller which is not optimized.
18182 (inline_small_functions): Likewise.
18183 (ipa_inline): Do not give up when not optimizing.
18184 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18185 away unoptimizes cdtors.
18186 (whole_program_function_and_variable_visibility): Do
18187 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18188 * ipa.c (process_references): Do not check optimize.
18189 (symbol_table::remove_unreachable_nodes): Update optimize check.
18190 (set_writeonly_bit): Update optimize check.
18191 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18192 (pass_ipa_single_use::gate): Remove.
18194 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18196 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18197 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18198 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18199 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18200 permute_load, permute_store, adjust_extract, adjust_splat,
18201 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18202 replace_swap_with_copy, dump_swap_insn_table,
18203 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18204 recombine_lvx_pattern, recombine_stvx_pattern,
18205 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18206 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18207 to file rs6000-p8swap.c.
18208 * config/rs6000/rs6000-p8swap.c: New file.
18209 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18210 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18211 and rs6000*-*-* targets.
18213 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18215 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18217 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18219 * lto-wrapper.c (merge_and_complain): Do not merge
18220 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18221 fsigned_zeros, ftrapping_math, fwrapv.
18222 (append_compiler_options): Do not track these options.
18223 (append_linker_options): Likewie
18225 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18227 * cgraphunit.c (cgraph_node::finalize_function): When
18228 !flag_toplevel_reorde set no_reorder flag.
18229 (varpool_node::finalize_decl): Likewise.
18230 (symbol_table::compile): Drop no toplevel reorder path.
18232 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18234 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18235 edges; zero probability is not better than uninitialized.
18237 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18239 * asan.h (asan_sanitize_allocas_p): Declare.
18240 * asan.c (asan_sanitize_allocas_p): New function.
18241 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18242 (handle_builtin_alloca): Likewise.
18243 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18244 if !asan_sanitize_allocas_p.
18245 * params.def (asan-instrument-allocas): Add new option.
18246 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18247 * opts.c (common_handle_option): Disable allocas sanitization for
18250 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18252 * asan.c: Include gimple-fold.h.
18253 (get_last_alloca_addr): New function.
18254 (handle_builtin_stackrestore): Likewise.
18255 (handle_builtin_alloca): Likewise.
18256 (asan_emit_allocas_unpoison): Likewise.
18257 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18258 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18259 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18260 (instrument_builtin_call): Pass gimple iterator to
18261 get_mem_refs_of_builtin_call.
18262 (last_alloca_addr): New global.
18263 * asan.h (asan_emit_allocas_unpoison): Declare.
18264 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18265 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18266 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18267 if function calls alloca.
18268 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18269 * gimple-fold.h (replace_call_with_value): Declare.
18270 * internal-fn.c: Include asan.h.
18271 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18272 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18274 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18276 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18277 (C_SELFTEST_FLAGS): New.
18278 (CPP_SELFTEST_FLAGS): New.
18279 (SELFTEST_DEPS): New, from deps of s-selftest.
18280 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18281 (CPP_SELFTEST_DEPS): New.
18282 (selftest): Add dependency on s-selftest-c++.
18283 (s-selftest): Rename to...
18284 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18285 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18286 than SELFTEST_FLAGS.
18287 (selftest-gdb): Rename to...
18288 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18290 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18291 (selftest-valgrind): Rename to...
18292 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18294 (selftest-valgrind): Reintroduce as an alias for
18295 selftest-c-valgrind.
18296 (s-selftest-c++): New.
18297 (selftest-c++-gdb): New.
18298 (selftest-c++-valgrind): New.
18300 2017-07-06 Olivier Hainque <hainque@adacore.com>
18302 * gcc.c (process_command): When deciding if undefined variables
18303 should be ignored when processing specs, accept "gcc -v" as well.
18305 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18307 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18308 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18310 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18312 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18313 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18314 * config/arm/arm-tables.opt: Regenerate.
18315 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18317 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18319 2017-07-06 Carl Love <cel@us.ibm.com>
18321 * ChangeLog: Clean up from mid air collision
18323 2017-07-06 Carl Love <cel@us.ibm.com>
18325 * config/rs6000/rs6000-c.c: Add support for built-in functions
18326 vector signed int vec_subc (vector signed int, vector signed int);
18327 vector signed __int128 vec_subc (vector signed __int128,
18328 vector signed __int128);
18329 vector unsigned __int128 vec_subc (vector unsigned __int128,
18330 vector unsigned __int128);
18331 vector signed int vec_sube (vector signed int, vector signed int,
18332 vector signed int);
18333 vector unsigned int vec_sube (vector unsigned int,
18334 vector unsigned int,
18335 vector unsigned int);
18336 vector signed __int128 vec_sube (vector signed __int128,
18337 vector signed __int128,
18338 vector signed__int128);
18339 vector unsigned __int128 vec_sube (vector unsigned __int128,
18340 vector unsigned __int128,
18341 vector unsigned __int128);
18342 vector signed int vec_subec (vector signed int, vector signed int,
18343 vector signed int);
18344 vector unsigned int vec_subec (vector unsigned int,
18345 vector unsigned int,
18346 vector unsigned int);
18347 vector signed __int128 vec_subec (vector signed __int128,
18348 vector signed __int128,
18349 vector signed__int128);
18350 vector unsigned __int128 vec_subec (vector unsigned __int128,
18351 vector unsigned __int128,
18352 vector unsigned __int128);
18353 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18354 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18355 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18356 BU_ALTIVEC_OVERLOAD_X definitions.
18357 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18358 * doc/extend.texi: Update the built-in documentation file for the new
18359 built-in functions.
18361 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18364 * diagnostic-show-locus.c (layout::layout): Use start and finish
18365 spelling location for the start and finish of each range.
18366 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18367 Add unused aspect param.
18368 * input.c (expand_location_1): Add "aspect" param, and use it
18369 to access the correct part of the location.
18370 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18372 (expand_location_to_spelling_point): Likewise.
18373 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18374 param, and pass it to expand_location_1.
18376 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18378 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18379 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18380 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18381 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18382 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18383 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18384 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18385 _mm_maskz_getmant_ss): New intrinsics.
18386 (__builtin_ia32_getexpss128_mask): Changed to ...
18387 __builtin_ia32_getexpss128_round ... this.
18388 (__builtin_ia32_getexpsd128_mask): Changed to ...
18389 __builtin_ia32_getexpsd128_round ... this.
18390 * config/i386/i386-builtin-types.def
18391 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18392 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18393 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18394 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18395 __builtin_ia32_getmantss_mask_round): New builtins.
18396 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18397 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18398 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18399 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18400 * config/i386/sse.md
18401 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18402 avx512f_sgetexp<mode><mask_scalar_name>
18403 <round_saeonly_scalar_name> ... this.
18404 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18405 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18406 vgetexp<ssescalarmodesuffix>
18407 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18408 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18409 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18410 avx512f_vgetmant<mode><mask_scalar_name>
18411 <round_saeonly_scalar_name> ... this.
18412 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18413 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18414 vgetmant<ssescalarmodesuffix>
18415 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18416 %0<mask_scalar_operand4>, %1, %2
18417 <round_saeonly_scalar_mask_op4>, %3} ... this.
18418 * config/i386/subst.md (mask_scalar_operand4,
18419 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18420 round_saeonly_scalar_nimm_predicate): New subst attributes.
18422 2017-07-06 Julia Koval <julia.koval@intel.com>
18424 * config/i386/i386.c (ix86_erase_embedded_rounding):
18425 Remove code for old rounding pattern.
18427 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18429 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18431 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18433 * doc/sourcebuild.texi (Test Directives, Variants of
18434 dg-require-support): Add documentation for dg-require-stack-check.
18436 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18438 * config/i386/subst.md (mask_scalar, round_scalar,
18439 round_saeonly_scalar): New meta-templates.
18440 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18441 round_scalar_mask_operand3, round_scalar_mask_op3,
18442 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18443 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18444 round_saeonly_scalar_constraint,
18445 round_saeonly_scalar_prefix): New subst attribute.
18446 * config/i386/sse.md
18447 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18448 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18449 <round_scalar_name> ... this.
18450 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18451 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18452 <round_scalar_name> ... this.
18453 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18454 <sse>_vm<code><mode>3<mask_scalar_name>
18455 <round_saeonly_scalar_name> ... this.
18456 (v<plusminus_mnemonic><ssescalarmodesuffix>
18457 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18458 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18459 v<plusminus_mnemonic><ssescalarmodesuffix>
18460 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18461 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18462 (v<multdiv_mnemonic><ssescalarmodesuffix>
18463 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18464 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18465 v<multdiv_mnemonic><ssescalarmodesuffix>
18466 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18467 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18468 (v<maxmin_float><ssescalarmodesuffix>
18469 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18470 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18471 v<maxmin_float><ssescalarmodesuffix>
18472 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18473 %0<mask_scalar_operand3>, %1, %<iptr>2
18474 <round_saeonly_scalar_mask_op3>} ... this.
18476 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18478 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18479 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18481 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18482 Alan Hayward <alan.hayward@arm.com>
18483 David Sherwood <david.sherwood@arm.com>
18485 * combine.c (simplify_if_then_else): Remove "enum" before
18487 * compare-elim.c (can_eliminate_compare): Likewise.
18488 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18490 (aarch64_lookup_simd_builtin_type): Likewise.
18491 (aarch64_simd_builtin_type): Likewise.
18492 (aarch64_init_simd_builtin_types): Likewise.
18493 (aarch64_simd_expand_args): Likewise.
18494 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18496 (aarch64_reverse_mask): Likewise.
18497 (aarch64_simd_emit_reg_reg_move): Likewise.
18498 (aarch64_gen_adjusted_ldpstp): Likewise.
18499 (aarch64_ccmp_mode_to_code): Likewise.
18500 (aarch64_operands_ok_for_ldpstp): Likewise.
18501 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18502 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18504 (aarch64_min_divisions_for_recip_mul): Likewise.
18505 (aarch64_reassociation_width): Likewise.
18506 (aarch64_get_condition_code_1): Likewise.
18507 (aarch64_simd_emit_reg_reg_move): Likewise.
18508 (aarch64_simd_attr_length_rglist): Likewise.
18509 (aarch64_reverse_mask): Likewise.
18510 (aarch64_operands_ok_for_ldpstp): Likewise.
18511 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18512 (aarch64_gen_adjusted_ldpstp): Likewise.
18513 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18515 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18516 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18517 (arm_lookup_simd_builtin_type): Likewise.
18518 (arm_simd_builtin_type): Likewise.
18519 (arm_init_simd_builtin_types): Likewise.
18520 (arm_expand_builtin_args): Likewise.
18521 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18522 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18523 (ft32_setup_incoming_varargs): Likewise.
18524 (ft32_function_arg): Likewise.
18525 (ft32_function_arg_advance): Likewise.
18526 (ft32_pass_by_reference): Likewise.
18527 (ft32_arg_partial_bytes): Likewise.
18528 (ft32_valid_pointer_mode): Likewise.
18529 (ft32_addr_space_pointer_mode): Likewise.
18530 (ft32_addr_space_legitimate_address_p): Likewise.
18531 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18533 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18534 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18535 (iamcu_alignment): Likewise.
18536 (canonicalize_vector_int_perm): Likewise.
18537 (ix86_noce_conversion_profitable_p): Likewise.
18538 (ix86_mpx_bound_mode): Likewise.
18539 (ix86_operands_ok_for_move_multiple): Likewise.
18540 * config/microblaze/microblaze-protos.h
18541 (microblaze_expand_conditional_branch_reg): Likewise.
18542 * config/microblaze/microblaze.c
18543 (microblaze_expand_conditional_branch_reg): Likewise.
18544 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18546 (rs6000_reassociation_width): Likewise.
18547 (rs6000_invalid_binary_op): Likewise.
18548 (fusion_p9_p): Likewise.
18549 (emit_fusion_p9_load): Likewise.
18550 (emit_fusion_p9_store): Likewise.
18551 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18553 (riscv_hard_regno_mode_ok_p): Likewise.
18554 (riscv_address_insns): Likewise.
18555 (riscv_split_symbol): Likewise.
18556 (riscv_legitimize_move): Likewise.
18557 (riscv_function_value): Likewise.
18558 (riscv_hard_regno_nregs): Likewise.
18559 (riscv_expand_builtin): Likewise.
18560 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18561 (riscv_build_integer): Likewise.
18562 (riscv_split_integer): Likewise.
18563 (riscv_legitimate_constant_p): Likewise.
18564 (riscv_cannot_force_const_mem): Likewise.
18565 (riscv_regno_mode_ok_for_base_p): Likewise.
18566 (riscv_valid_base_register_p): Likewise.
18567 (riscv_valid_offset_p): Likewise.
18568 (riscv_valid_lo_sum_p): Likewise.
18569 (riscv_classify_address): Likewise.
18570 (riscv_legitimate_address_p): Likewise.
18571 (riscv_address_insns): Likewise.
18572 (riscv_load_store_insns): Likewise.
18573 (riscv_force_binary): Likewise.
18574 (riscv_split_symbol): Likewise.
18575 (riscv_force_address): Likewise.
18576 (riscv_legitimize_address): Likewise.
18577 (riscv_move_integer): Likewise.
18578 (riscv_legitimize_const_move): Likewise.
18579 (riscv_legitimize_move): Likewise.
18580 (riscv_address_cost): Likewise.
18581 (riscv_subword): Likewise.
18582 (riscv_output_move): Likewise.
18583 (riscv_canonicalize_int_order_test): Likewise.
18584 (riscv_emit_int_order_test): Likewise.
18585 (riscv_function_arg_boundary): Likewise.
18586 (riscv_pass_mode_in_fpr_p): Likewise.
18587 (riscv_pass_fpr_single): Likewise.
18588 (riscv_pass_fpr_pair): Likewise.
18589 (riscv_get_arg_info): Likewise.
18590 (riscv_function_arg): Likewise.
18591 (riscv_function_arg_advance): Likewise.
18592 (riscv_arg_partial_bytes): Likewise.
18593 (riscv_function_value): Likewise.
18594 (riscv_pass_by_reference): Likewise.
18595 (riscv_setup_incoming_varargs): Likewise.
18596 (riscv_print_operand): Likewise.
18597 (riscv_elf_select_rtx_section): Likewise.
18598 (riscv_save_restore_reg): Likewise.
18599 (riscv_for_each_saved_reg): Likewise.
18600 (riscv_register_move_cost): Likewise.
18601 (riscv_hard_regno_mode_ok_p): Likewise.
18602 (riscv_hard_regno_nregs): Likewise.
18603 (riscv_class_max_nregs): Likewise.
18604 (riscv_memory_move_cost): Likewise.
18605 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18606 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18607 (rl78_addr_space_address_mode): Likewise.
18608 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18610 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18611 (rs6000_reassociation_width): Likewise.
18612 (rs6000_invalid_binary_op): Likewise.
18613 (fusion_p9_p): Likewise.
18614 (emit_fusion_p9_load): Likewise.
18615 (emit_fusion_p9_store): Likewise.
18616 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18617 (ok_for_simple_move_operands): Likewise.
18618 (ok_for_simple_move_strict_operands): Likewise.
18619 (ok_for_simple_arith_logic_operands): Likewise.
18620 (visium_legitimize_reload_address): Likewise.
18621 (visium_select_cc_mode): Likewise.
18622 (output_cbranch): Likewise.
18623 (visium_split_double_move): Likewise.
18624 (visium_expand_copysign): Likewise.
18625 (visium_expand_int_cstore): Likewise.
18626 (visium_expand_fp_cstore): Likewise.
18627 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18628 (visium_function_arg): Likewise.
18629 (visium_function_arg_advance): Likewise.
18630 (visium_libcall_value): Likewise.
18631 (visium_setup_incoming_varargs): Likewise.
18632 (visium_legitimate_constant_p): Likewise.
18633 (visium_legitimate_address_p): Likewise.
18634 (visium_legitimize_address): Likewise.
18635 (visium_secondary_reload): Likewise.
18636 (visium_register_move_cost): Likewise.
18637 (visium_memory_move_cost): Likewise.
18638 (prepare_move_operands): Likewise.
18639 (ok_for_simple_move_operands): Likewise.
18640 (ok_for_simple_move_strict_operands): Likewise.
18641 (ok_for_simple_arith_logic_operands): Likewise.
18642 (visium_function_value_1): Likewise.
18643 (rtx_ok_for_offset_p): Likewise.
18644 (visium_legitimize_reload_address): Likewise.
18645 (visium_split_double_move): Likewise.
18646 (visium_expand_copysign): Likewise.
18647 (visium_expand_int_cstore): Likewise.
18648 (visium_expand_fp_cstore): Likewise.
18649 (visium_split_cstore): Likewise.
18650 (visium_select_cc_mode): Likewise.
18651 (visium_split_cbranch): Likewise.
18652 (output_cbranch): Likewise.
18653 (visium_print_operand_address): Likewise.
18654 * expmed.c (flip_storage_order): Likewise.
18655 * expmed.h (emit_cstore): Likewise.
18656 (flip_storage_order): Likewise.
18657 * genrecog.c (validate_pattern): Likewise.
18658 * hsa-gen.c (gen_hsa_addr): Likewise.
18659 * internal-fn.c (expand_arith_overflow): Likewise.
18660 * ira-color.c (allocno_copy_cost_saving): Likewise.
18661 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18662 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18663 (process_invariant_for_inheritance): Likewise.
18664 * lra-eliminations.c (move_plus_up): Likewise.
18665 * omp-low.c (lower_oacc_reductions): Likewise.
18666 * simplify-rtx.c (simplify_subreg): Likewise.
18667 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18668 (TARGET_CHKP_BOUND_MODE): Likewise..
18669 * targhooks.c (default_chkp_bound_mode): Likewise.
18670 (default_setup_incoming_vararg_bounds): Likewise.
18671 * targhooks.h (default_chkp_bound_mode): Likewise.
18672 (default_setup_incoming_vararg_bounds): Likewise.
18673 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18674 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18675 (have_whole_vector_shift): Likewise.
18676 * tree-vect-stmts.c (vectorizable_load): Likewise.
18677 * doc/tm.texi: Regenerate.
18679 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18681 Graceful degrade if Binutils PR21472 is not available.
18684 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18685 .rodata in flash test fails.
18686 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18687 * confgure: Regenerate.
18688 * config.in: Regenerate.
18689 * config/avr/avr.c (avr_asm_named_section)
18690 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18691 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18692 (avr_asm_init_sections): Same.
18694 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18696 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18697 (fma<VH:mode>4_intrinsic): Likewise.
18698 (*fmsub<VCVTF:mode>4): Likewise.
18699 (*fmsub<VH:mode>4_intrinsic): Likewise.
18701 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18704 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18705 Don't depend on "optimize > 0".
18706 (out_movhi_r_mr, out_movqi_mr_r): Same.
18707 (out_movhi_mr_r, out_movqi_r_mr): Same.
18708 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18709 io_address_operand on "optimize > 0".
18711 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18713 * tree-loop-distribution.c: Add general explanantion on the pass.
18714 (generate_loops_for_partition): Mark distributed loop.
18715 (pg_add_dependence_edges): New parameter. Handle alias data
18716 dependence specially and record it in the parameter if asked.
18717 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18718 (init_partition_graph_vertices, add_partition_graph_edge): New.
18719 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18720 (free_partition_graph_vdata, build_partition_graph): New.
18721 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18722 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18723 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18724 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18725 (version_for_distribution_p, finalize_partitions): New.
18726 (distribute_loop): Handle alias data dependence specially. Factor
18727 out loop fusion code as functions and call these functions.
18729 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18731 * tree-loop-distribution.c (classify_partition): New parameter and
18732 better handle reduction statement.
18733 (rdg_build_partitions): Revise comment.
18734 (distribute_loop): Compute statements in all partitions and pass it
18735 to classify_partition.
18737 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18739 * tree-loop-distribution.c (enum partition_type): New.
18740 (struct partition): New field type.
18741 (partition_merge_into): Add parameter. Update partition type.
18742 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18743 (build_rdg_partition_for_vertex): Compute partition type.
18744 (rdg_build_partitions): Dump partition type.
18745 (distribute_loop): Update calls to partition_merge_into.
18747 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18749 * tree-loop-distribution.c (struct ddr_hasher): New.
18750 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18752 (classify_partition): Call get_data_dependence.
18753 (pg_add_dependence_edges): Ditto.
18754 (distribute_loop): Release data dependence hash table.
18756 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18758 * tree-loop-distribution.c (ref_base_address): Delete.
18759 (similar_memory_accesses): Rename ...
18760 (share_memory_accesses): ... to this. Check if partitions access
18761 the same memory reference.
18762 (distribute_loop): Call share_memory_accesses.
18764 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18766 * tree-loop-distribution.c (struct partition): New field recording
18767 its data reference.
18768 (partition_alloc, partition_free): Init and release data refs.
18769 (partition_merge_into): Merge data refs.
18770 (build_rdg_partition_for_vertex): Collect data refs for partition.
18771 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18773 (distribute_loop): Remve data refs from vertice data of partition
18776 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18778 * tree-loop-distribution.c (params.h): Include header file.
18779 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18780 (datarefs_vec): New global var.
18781 (create_rdg_vertices): Use datarefs_vec directly.
18782 (free_rdg): Don't free data references.
18783 (build_rdg): Update use. Don't free data references.
18784 (distribute_loop): Compute global variable for data references.
18785 Bail out if there are too many data references.
18787 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18789 * tree-loop-distribution.c (loop_nest): New global var.
18790 (build_rdg): Use loop directly, rather than loop nest.
18791 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18793 (distribute_loop): Compute global variable loop nest. Update use.
18795 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18797 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18798 (partition_merge_into): New parameter. Dump reason for fusion.
18799 (distribute_loop): Update use of partition_merge_into.
18801 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18803 * tree-loop-distribution.c (bb_top_order_index): New.
18804 (bb_top_order_index_size, bb_top_order_cmp): New.
18805 (stmts_from_loop): Use topological order.
18806 (pass_loop_distribution::execute): Compute and release topological
18807 order for basic blocks.
18809 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18811 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18814 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18816 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18817 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18818 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18819 * internal-fn.def (LOOP_DIST_ALIAS): New.
18820 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18821 (fold_loop_internal_call): ... this.
18822 (vect_loop_dist_alias_call): New function.
18823 (set_uid_loop_bbs): Call fold_loop_internal_call.
18824 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18827 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18830 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18831 Require dead FLAGS_REG at the beginning of a peephole.
18833 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18836 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18837 arguments in the call to __builtin_ia32_sbb_u32.
18838 (_subborrow_u64): Swap _X and _Y arguments in the call to
18839 __builtin_ia32_sbb_u64.
18841 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18844 * tree-vrp.c (compare_assert_loc): Turn into a function template
18845 with stable template parameter. Only test if a->e is NULL,
18846 !a->e == !b->e has been verified already. Use e == NULL or
18847 e != NULL instead of e or ! e tests. If stable is true, don't use
18848 iterative_hash_expr, on the other side allow a or b or both NULL
18849 and sort the NULLs last.
18850 (process_assert_insertions): Sort using compare_assert_loc<false>
18851 instead of compare_assert_loc, later sort using
18852 compare_assert_loc<true> before calling process_assert_insertions_for
18853 in a loop. Use break instead of continue once seen NULL pointer.
18855 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18857 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18858 Cortex-R7 and Cortex-R8 processors.
18860 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18862 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18863 uninitialized while src is not.
18865 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18867 * common/config/arm/arm-common.c: Adjust include path for
18869 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18870 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18871 (arm-cpu-data.h): Likewise.
18872 (arm-cpu-cdata.h): Likewise.
18873 * config/arm/arm-cpu.h: Delete.
18874 * config/arm/arm-cpu-cdata.h: Delete.
18875 * config/arm/arm-cpu-data.h: Delete.
18877 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18879 * config/arm/arm-cpus.in (cortex-a55): New.
18880 (cortex-a75): Likewise.
18881 (cortex-a75.cortex-a55): Likewise.
18882 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18884 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18885 * config/arm/arm-cpu-cdata.h: Regenerate.
18886 * config/arm/arm-cpu-data.h: Regenerate.
18887 * config/arm/arm-cpu.h: Regenerate.
18888 * config/arm/arm-tables.opt: Regenerate.
18889 * config/arm/arm-tune.md: Regenerate.
18891 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18893 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18895 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18897 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18900 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18902 PR tree-optimization/81292
18903 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18904 full_string_p, also call adjust_related_strinfos if the adjustment
18905 is simple, otherwise invalidate related strinfos.
18907 2017-07-04 Martin Liska <mliska@suse.cz>
18910 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18911 newly created variable as DECL_IGNORED_P.
18913 2017-07-04 Martin Liska <mliska@suse.cz>
18916 * ipa-inline.c (inline_small_functions):
18917 Use xstrdup_for_dump.
18919 2017-07-04 Tom de Vries <tom@codesourcery.com>
18921 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18923 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18926 * config/darwin.c (darwin_function_switched_text_sections):
18929 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18931 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18933 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18935 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18937 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18939 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18940 min_profitable_iters, and th as inclusive lower bounds.
18941 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18942 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18943 for min_profitable_iters and min_profitable_estimate.
18944 (vect_transform_loop): Treat th as an inclusive lower bound.
18945 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18947 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18950 * config/darwin.c (darwin_function_switched_text_sections):
18951 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18952 in two pieces, and suppress the use of buf.
18954 2017-07-03 Nathan Sidwell <nathan@acm.org>
18956 * hash-table.h (hash_table_mod1): Fix indentation.
18958 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18960 PR middle-end/81290
18961 * predict.c (force_edge_cold): Be more careful about propagation
18963 * profile-count.h (profile_probability::guessed,
18964 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18966 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18968 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18970 * doc/invoke.texi (rcpc architecture extension): Document it.
18972 2017-07-03 Richard Biener <rguenther@suse.de>
18974 PR tree-optimization/60510
18975 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18976 the scalar reduction PHI and use it.
18977 (vectorizable_reduction): Properly guard the single_defuse_cycle
18978 path for non-SLP reduction chains where we cannot use it.
18979 Rework reduc_def/index and vector type deduction. Rework
18980 vector operand gathering during reduction op code-gen.
18981 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18982 chains dissolve the chain and leave it to non-SLP reduction
18985 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18987 * tree-data-ref.h (dr_alignment): Declare.
18988 * tree-data-ref.c (dr_alignment): New function.
18989 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18990 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18992 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18994 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18996 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18997 and base_misalignment fields.
18998 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18999 * tree-data-ref.c: Include builtins.h.
19000 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19001 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19002 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19003 * tree-vect-data-refs.c: Include tree-cfg.h.
19004 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19005 fields instead of calculating an alignment here.
19006 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19007 innermost_loop_behavior fields.
19009 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19011 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19013 (DR_STEP_ALIGNMENT): New macro.
19014 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19015 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19016 (create_data_ref): Print it.
19017 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19018 to tell whether the step preserves vector (mis)alignment.
19019 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19020 Move the check for an integer step and generalise to all INTEGER_CST.
19021 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19022 Print the outer step alignment.
19024 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19026 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19027 with offset_alignment.
19028 (DR_ALIGNED_TO): Delete.
19029 (DR_OFFSET_ALIGNMENT): New macro.
19030 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19031 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19032 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19033 (create_data_ref): Likewise.
19034 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19035 (vect_analyze_data_refs): Likewise.
19036 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19037 creating dummy innermost behavior.
19039 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19041 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19042 with a "innermost_loop_behavior *" and refeence tree.
19043 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19044 (create_data_ref): Update call accordingly.
19045 * tree-predcom.c (find_looparound_phi): Likewise.
19047 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19049 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19050 fields with dr_wrt_vec_loop.
19051 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19052 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19053 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19054 (vect_dr_behavior): New function.
19055 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19056 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19057 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19058 track whether the step preserves the misalignment.
19059 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19060 Use vect_dr_behavior.
19061 (vect_setup_realignment): Update call accordingly.
19062 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19063 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19064 call to vect_create_addr_base_for_vector_ref.
19065 (vect_create_cond_for_align_checks): Likewise.
19066 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19067 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19068 (vect_recog_mask_conversion_pattern): Likewise.
19069 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19070 (new_stmt_vec_info): Remove redundant zeroing.
19072 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19074 * common/config/arm/arm-common.c (arm_be8_option): New function.
19075 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19076 (ISA_ARMv6): Add isa_bit_be8.
19077 * config/arm/arm.h (arm_be8_option): Add prototype.
19078 (BE8_SPEC_FUNCTION): New define.
19079 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19080 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19081 (mlittle-endian): Similarly.
19082 (mbe8, mbe32): New options.
19083 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19084 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19086 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19088 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19090 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19092 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19093 (cleanup_tree_cfg_bb): Use it.
19094 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19096 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19098 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19101 * loop-doloop.c (add_test): Update profile.
19103 2017-07-03 Martin Liska <mliska@suse.cz>
19106 * sanopt.c (rewrite_usage_of_param): New function.
19107 (sanitize_rewrite_addressable_params): Likewise.
19108 (pass_sanopt::execute): Call rewrite_usage_of_param.
19110 2017-07-03 Richard Biener <rguenther@suse.de>
19112 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19113 back to using VIEW_CONVERT_EXPR.
19115 2017-07-03 Martin Liska <mliska@suse.cz>
19118 * doc/extend.texi: Document when a resolver function is
19119 generated for target_clones.
19121 2017-07-03 Martin Liska <mliska@suse.cz>
19123 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19124 and shadow memory of auto variables which are subject of
19125 use-after-scope sanitization.
19126 (asan_expand_mark_ifn): Add do set only when is_poison.
19128 2016-07-03 Richard Biener <rguenther@suse.de>
19130 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19132 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19133 (vectorizable_reduction): Perform reduction PHI creation when
19134 visiting a reduction PHI and adjust and simplify code generation
19135 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19136 (vect_transform_loop): Visit reduction PHIs.
19137 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19138 defs into the SLP tree.
19139 (vect_build_slp_tree): Reduction defs terminate the recursion.
19140 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19142 (vect_get_vec_defs_for_stmt_copy): Export.
19143 (vect_get_vec_defs): Likewise.
19144 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19146 (vect_get_vec_defs_for_stmt_copy): Declare.
19147 (vect_get_vec_defs): Likewise.
19149 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19151 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19152 parameter with a "loop" parameter and use it instead of the
19153 loop containing DR_STMT. Don't check simple_iv when doing
19154 BB analysis. Describe the two analysis modes in the comment.
19156 2017-07-03 Tom de Vries <tom@codesourcery.com>
19158 PR tree-optimization/69468
19159 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19160 (find_same_succ_bb): Handle ignore_edge_flags.
19162 2017-07-03 Tom de Vries <tom@codesourcery.com>
19164 PR tree-optimization/81192
19165 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19167 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19169 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19171 2017-07-03 Tom de Vries <tom@codesourcery.com>
19173 PR tree-optimization/81192
19174 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19175 BB_SAME_SUCC (bb) == NULL.
19177 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19179 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19182 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19184 * dumpfile.c: Include profile-count.h
19185 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19187 (insert_cond_bb): Update profile.
19188 * tree-cfg.h (insert_cond_bb): Update prototype.
19189 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19190 * tree-dump.c: Do not include tree-cfg.
19192 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19194 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19196 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19198 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19201 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19203 * tree-complex.c (expand_complex_div_wide): update profile.
19205 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19206 Alan Hayward <alan.hayward@arm.com>
19207 David Sherwood <david.sherwood@arm.com>
19209 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19210 (CORETYPES_H): New define.
19211 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19212 (insn-modes-inline.h, s-modes-inline-h): New rules.
19213 (generated_files): Add insn-modes-inline.h.
19214 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19215 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19216 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19217 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19218 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19219 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19220 (build/gencodes.o, build/genconditions.o): Likewise.
19221 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19222 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19223 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19224 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19225 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19226 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19227 * coretypes.h: Include everything up to real.h for generators.
19228 Include insn-modes.h first. Include wide-int-print.h after
19229 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19230 * machmode.h: Don't include insn-modes.h here.
19231 * function-tests.c: Remove includes of signop.h, machmode.h,
19232 double-int.h and wide-int.h.
19234 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19236 * optc-save-gen.awk: Likewise.
19237 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19238 * godump.c: Remove include of wide-int-print.h.
19239 * pretty-print.h: Likewise.
19240 * wide-int-print.cc: Likewise.
19241 * wide-int.cc: Likewise.
19242 * hash-map-tests.c: Remove include of signop.h.
19243 * hash-set-tests.c: Likewise.
19244 * rtl-tests.c: Likewise.
19245 * mkconfig.sh: Remove include of machmode.h.
19246 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19248 (emit_insn_modes_inline_h): ...this new function. Emit the code
19249 into an insn-modes-inline.h header file, adding appropriate
19250 include guards and end comments.
19251 (emit_insn_modes_c_header): Remove include of machmode.h.
19252 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19254 (main): Handle -i flag and call emit_insn_modes_inline_h when
19257 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19259 * tree-ssa-strlen.c (strinfo): Rename the length field to
19260 nonzero_chars. Add a full_string_p field.
19261 (compare_nonzero_chars, zero_length_string_p): New functions.
19262 (get_addr_stridx): Add an offset_out parameter.
19263 Use compare_nonzero_chars.
19264 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19265 (new_strinfo): Update after above changes to strinfo.
19266 (set_endptr_and_length): Set full_string_p.
19267 (get_string_length): Update after above changes to strinfo.
19268 (unshare_strinfo): Update call to new_strinfo.
19269 (maybe_invalidate): Likewise.
19270 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19271 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19272 as a uhwi instead of an shwi. Update after above changes to
19273 strinfo and new_strinfo.
19274 (zero_length_string): Assert that chainsi contains full strings.
19275 Use zero_length_string_p. Update call to new_strinfo.
19276 (adjust_related_strinfos): Update after above changes to strinfo.
19277 Copy full_string_p from origsi.
19278 (adjust_last_stmt): Use zero_length_string_p.
19279 (handle_builtin_strlen): Update after above changes to strinfo and
19280 new_strinfo. Install the lhs as the string length if the previous
19281 entry didn't describe a full string.
19282 (handle_builtin_strchr): Update after above changes to strinfo
19284 (handle_builtin_strcpy): Likewise.
19285 (handle_builtin_strcat): Likewise.
19286 (handle_builtin_malloc): Likewise.
19287 (handle_pointer_plus): Likewise.
19288 (handle_builtin_memcpy): Likewise. Track nonzero characters
19289 that aren't necessarily followed by a nul terminator.
19290 (handle_char_store): Likewise.
19292 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19294 PR tree-optimization/80769
19295 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19296 for malloc and calloc. Document the new invariant that all related
19297 strinfos have delayed lengths or none do.
19298 (verify_related_strinfos): Move earlier in file.
19299 (set_endptr_and_length): New function, split out from...
19300 (get_string_length): ...here. Also set the lengths of related
19302 (zero_length_string): Assert that chainsi has known (rather than
19304 (adjust_related_strinfos): Likewise.
19306 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19308 PR tree-optimization/81136
19309 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19310 assert that two references with the same misalignment have the same
19311 compile-time misalignment if those compile-time misalignments
19314 2017-07-01 Andi Kleen <ak@linux.intel.com>
19316 * print-tree.c (print_node): Print all attributes.
19318 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19320 * cfg.c (scale_bbs_frequencies): New function.
19321 * cfg.h (scale_bbs_frequencies): Declare it.
19322 * cfgloopanal.c (single_likely_exit): Cleanup.
19323 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19325 (scale_loop_profile): Likewise.
19326 (loop_version): Likewise.
19327 (create_empty_loop_on_edge): Update.
19328 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19329 scale_loop_frequencies, scale_loop_profile, loopify,
19330 loop_version): Update prototypes.
19331 * modulo-sched.c (sms_schedule): Update.
19332 * predict.c (unlikely_executed_edge_p): Also check probability.
19333 (probably_never_executed_edge_p): Fix typo.
19334 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19335 * tree-parloops.c (gen_parallel_loop): Update.
19336 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19337 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19338 * tree-ssa-loop-split.c (split_loop): Update.
19339 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19340 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19341 (vect_loop_versioning): Update.
19342 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19344 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19346 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19348 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19350 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19351 to keep profile consistent.
19353 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19355 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19356 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19357 * profile-count.h (max_safe_multiplier): Make unsigned.
19358 (profile_count::guessed_zero): New.
19360 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19362 * bb-reorder.c (fix_up_crossing_landing_pad,
19363 fix_crossing_conditional_branches): Use make_single_succ_edge
19364 to keep profile consistent.
19366 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19368 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19371 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19374 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19375 the right scopes, make sure cond_jump isn't preserved between multiple
19376 iterations. Search for fallthru edge whenever there are 3+ edges and
19377 use find_fallthru_edge for it.
19379 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19381 Patch by Alexander Monakov <amonakov@ispras.ru>
19382 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19383 probabilities consistently.
19385 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19387 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19388 emit_cmp_and_jump_insns.
19390 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19393 * tree-inline.c (expand_call_inline): Combine profile statuses.
19395 2017-06-30 Andrew Pinski <apinski@cavium.com>
19397 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19398 fold_stmt returned true.
19400 2017-06-30 Nathan Sidwell <nathan@acm.org>
19402 * ggc.h (empty_string): Delete.
19403 * cfgexpand.c (expand_asm_stmt): Use plain "".
19404 * optabs.c (expand_asm_memory_barrier): Likewise.
19405 * stringpool.c (empty_string): Delete.
19406 (digit_vector, digit_string): Delete.
19407 (ggc_alloc_string): Use plain "", don't optimize single digit
19408 strings. Use ggc_alloc_atomic.
19410 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19412 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19413 comparison set and one other set, use the cost of the non-comparison
19416 2017-06-30 Nathan Sidwell <nathan@acm.org>
19418 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19421 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19423 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19424 loops. Remove now unneeded calls to gimple_switch_set_label() that
19425 just set removed labels to NULL_TREE.
19427 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19429 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19430 (set_range_info): ...here. Only call set_range_info_raw if domain
19432 (set_nonzero_bits): Call set_range_info_raw.
19433 * tree-ssanames.h (set_range_info_raw): New.
19435 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19438 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19439 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19440 of nonimmediate_operand and <store_mask_constraint> instead of m
19441 for the input operand. For V8FI iterator, always split if input
19442 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19443 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19444 <store_mask_predicate> instead of register_operand and
19445 <store_mask_constraint> instead of v for the input operand. Make
19446 sure both operands aren't MEMs for if not <mask_applied>.
19448 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19450 * lto-wrapper.c (copy_file) Close both file descriptors before
19453 2017-06-30 Martin Liska <mliska@suse.cz>
19456 * multiple_target.c (create_dispatcher_calls): Make ifunc
19457 also for function that don't have calls or are not referenced.
19459 2017-06-30 Richard Biener <rguenther@suse.de>
19461 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19462 analyze the first scalar stmt. Move vector type computation
19463 for the BB case here from ...
19464 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19465 live operation processing in the SLP case properly.
19467 2017-06-30 Richard Biener <rguenther@suse.de>
19469 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19471 2017-06-30 Martin Liska <mliska@suse.cz>
19474 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19475 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19477 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19479 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19482 2017-06-30 Richard Biener <rguenther@suse.de>
19484 PR tree-optimization/81249
19485 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19486 condition reduction result to original scalar type.
19488 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19490 * profile-count.h (enum profile_quality): Fix typos and whitespace
19493 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19495 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19496 type for branch probabilities.
19498 2017-06-29 Julian Brown <julian@codesourcery.com>
19499 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19501 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19502 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19503 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19504 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19506 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19508 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19509 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19510 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19511 CC usage from generic code to here.
19512 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19513 CC usage into the target macros.
19515 2017-06-29 Maya Rashish <coypu@sdf.org>
19517 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19520 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19522 * arm/arm-builtins.c: Include profile-count.h
19523 * except.c (sjlj_emit_function_enter): Use
19524 profile_probability::unlikely.
19526 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19528 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19529 and tocrel_offset be pointer args rather than implicitly using
19531 (legitimate_constant_pool_address_p, rs6000_emit_move,
19532 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19533 tocrel_offset and use in toc_relative_expr_p call.
19534 (print_operand, print_operand_address): Use static tocrel_base_oac
19535 and tocrel_offset_oac.
19536 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19539 2017-06-29 Maya Rashish <coypu@sdf.org>
19541 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19543 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19545 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19546 objects, take into account only the alignment of 'op0' and 'mode1' if
19549 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19551 * ccmp.c (ccmp_tree_comparison_p): New function.
19552 (ccmp_candidate_p): Update to use above function.
19553 (get_compare_parts): New function.
19554 (expand_ccmp_next): Update to use new functions.
19555 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19557 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19558 take mode as argument.
19559 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19560 * expr.c (expand_expr_real_1): Pass mode as argument.
19562 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19564 * combine.c (combine_instructions): Print insns to dump_file, together
19567 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19569 * asan.c (asan_emit_stack_protection): Update.
19570 (create_cond_insert_point): Update.
19571 * auto-profile.c (afdo_propagate_circuit): Update.
19572 * basic-block.h (struct edge_def): Turn probability to
19573 profile_probability.
19574 (EDGE_FREQUENCY): Update.
19575 * bb-reorder.c (find_traces_1_round): Update.
19576 (better_edge_p): Update.
19577 (sanitize_hot_paths): Update.
19578 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19579 (make_single_succ_edge): Update.
19580 (check_bb_profile): Update.
19581 (dump_edge_info): Update.
19582 (update_bb_profile_for_threading): Update.
19583 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19584 probabilitycount to 0.
19585 * cfgbuild.c (compute_outgoing_frequencies): Update.
19586 * cfgcleanup.c (try_forward_edges): Update.
19587 (outgoing_edges_match): Update.
19588 (try_crossjump_to_edge): Update.
19589 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19590 (expand_gimple_tailcall): Update.
19591 (construct_init_block): Use make_single_succ_edge.
19592 (construct_exit_block): Use make_single_succ_edge.
19593 * cfghooks.c (verify_flow_info): Update.
19594 (redirect_edge_succ_nodup): Update.
19595 (split_edge): Update.
19596 (account_profile_record): Update.
19597 * cfgloopanal.c (single_likely_exit): Update.
19598 * cfgloopmanip.c (scale_loop_profile): Update.
19599 (set_zero_probability): Remove.
19600 (duplicate_loop_to_header_edge): Update.
19601 * cfgloopmanip.h (loop_version): Update prototype.
19602 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19603 (force_nonfallthru_and_redirect): Update.
19604 (update_br_prob_note): Update.
19605 (rtl_verify_edges): Update.
19606 (purge_dead_edges): Update.
19607 (rtl_lv_add_condition_to_bb): Update.
19608 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19609 * cgraphunit.c (init_lowered_empty_function): Update.
19610 (cgraph_node::expand_thunk): Update.
19611 * cilk-common.c: Include profile-count.h
19612 * dojump.c (inv): Remove.
19613 (jumpifnot): Update.
19614 (jumpifnot_1): Update.
19615 (do_jump_1): Update.
19617 (do_jump_by_parts_greater_rtx): Update.
19618 (do_compare_rtx_and_jump): Update.
19619 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19620 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19621 * dwarf2cfi.c: Include profile-count.h
19622 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19623 (sjlj_emit_dispatch_table): Likewise.
19624 * explow.c: Include profile-count.h
19625 * expmed.c (emit_store_flag_force): Update.
19626 (do_cmp_and_jump): Update.
19627 * expr.c (compare_by_pieces_d::generate): Update.
19628 (compare_by_pieces_d::finish_mode): Update.
19629 (emit_block_move_via_loop): Update.
19630 (store_expr_with_bounds): Update.
19631 (store_constructor): Update.
19632 (expand_expr_real_2): Update.
19633 (expand_expr_real_1): Update.
19634 * expr.h (try_casesi, try_tablejump): Update prototypes.
19635 * gimple-pretty-print.c (dump_probability): Update.
19636 (dump_profile): New.
19637 (dump_gimple_label): Update.
19638 (dump_gimple_bb_header): Update.
19639 * graph.c (draw_cfg_node_succ_edges): Update.
19640 * hsa-gen.c (convert_switch_statements): Update.
19641 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19642 (find_if_case_1): Update.
19643 (find_if_case_2): Update.
19644 * internal-fn.c (expand_arith_overflow_result_store): Update.
19645 (expand_addsub_overflow): Update.
19646 (expand_neg_overflow): Update.
19647 (expand_mul_overflow): Update.
19648 (expand_vector_ubsan_overflow): Update.
19649 * ipa-cp.c (good_cloning_opportunity_p): Update.
19650 * ipa-split.c (split_function): Use make_single_succ_edge.
19651 * ipa-utils.c (ipa_merge_profiles): Update.
19652 * loop-doloop.c (add_test): Update.
19653 (doloop_modify): Update.
19654 * loop-unroll.c (compare_and_jump_seq): Update.
19655 (unroll_loop_runtime_iterations): Update.
19656 * lra-constraints.c (lra_inheritance): Update.
19657 * lto-streamer-in.c (input_cfg): Update.
19658 * lto-streamer-out.c (output_cfg): Update.
19659 * mcf.c (adjust_cfg_counts): Update.
19660 * modulo-sched.c (sms_schedule): Update.
19661 * omp-expand.c (expand_omp_for_init_counts): Update.
19662 (extract_omp_for_update_vars): Update.
19663 (expand_omp_ordered_sink): Update.
19664 (expand_omp_for_ordered_loops): Update.
19665 (expand_omp_for_generic): Update.
19666 (expand_omp_for_static_nochunk): Update.
19667 (expand_omp_for_static_chunk): Update.
19668 (expand_cilk_for): Update.
19669 (expand_omp_simd): Update.
19670 (expand_omp_taskloop_for_outer): Update.
19671 (expand_omp_taskloop_for_inner): Update.
19672 * omp-simd-clone.c (simd_clone_adjust): Update.
19673 * optabs.c (expand_doubleword_shift): Update.
19674 (expand_abs): Update.
19675 (emit_cmp_and_jump_insn_1): Update.
19676 (expand_compare_and_swap_loop): Update.
19677 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19678 * predict.c (predictable_edge_p): Update.
19679 (edge_probability_reliable_p): Update.
19680 (set_even_probabilities): Update.
19681 (combine_predictions_for_insn): Update.
19682 (combine_predictions_for_bb): Update.
19683 (propagate_freq): Update.
19684 (estimate_bb_frequencies): Update.
19685 (force_edge_cold): Update.
19686 * profile-count.c (profile_count::dump): Add missing space into dump.
19687 (profile_count::debug): Add newline.
19688 (profile_count::differs_from_p): Explicitly convert to unsigned.
19689 (profile_count::stream_in): Update.
19690 (profile_probability::dump): New member function.
19691 (profile_probability::debug): New member function.
19692 (profile_probability::differs_from_p): New member function.
19693 (profile_probability::differs_lot_from_p): New member function.
19694 (profile_probability::stream_in): New member function.
19695 (profile_probability::stream_out): New member function.
19696 * profile-count.h (profile_count_quality): Rename to ...
19697 (profile_quality): ... this one.
19698 (profile_probability): New.
19699 (profile_count): Update.
19700 * profile.c (compute_branch_probabilities): Update.
19701 * recog.c (peep2_attempt): Update.
19702 * sched-ebb.c (schedule_ebbs): Update.
19703 * sched-rgn.c (find_single_block_region): Update.
19704 (compute_dom_prob_ps): Update.
19705 (schedule_region): Update.
19706 * sel-sched-ir.c (compute_succs_info): Update.
19707 * stmt.c (struct case_node): Update.
19708 (do_jump_if_equal): Update.
19709 (get_outgoing_edge_probs): Update.
19710 (conditional_probability): Update.
19711 (emit_case_dispatch_table): Update.
19712 (expand_case): Update.
19713 (expand_sjlj_dispatch_table): Update.
19714 (emit_case_nodes): Update.
19715 * targhooks.c: Update.
19716 * tracer.c (better_p): Update.
19717 (find_best_successor): Update.
19718 * trans-mem.c (expand_transaction): Update.
19719 * tree-call-cdce.c: Update.
19720 * tree-cfg.c (gimple_split_edge): Upate.
19721 (move_sese_region_to_fn): Upate.
19722 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19723 * tree-eh.c (lower_resx): Upate.
19724 (cleanup_empty_eh_move_lp): Upate.
19725 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19726 * tree-inline.c (copy_edges_for_bb): Update.
19727 (copy_cfg_body): Update.
19728 * tree-parloops.c (gen_parallel_loop): Update.
19729 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19730 (gimple_gen_time_profiler): Update.
19731 * tree-ssa-dce.c (remove_dead_stmt): Update.
19732 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19733 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19734 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19735 (unloop_loops): Update.
19736 (try_peel_loop): Update.
19737 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19738 * tree-ssa-loop-split.c (connect_loops): Update.
19739 (split_loop): Update.
19740 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19741 (hoist_guard): Update.
19742 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19743 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19744 (value_replacement): Update.
19745 * tree-ssa-reassoc.c (branch_fixup): Update.
19746 * tree-ssa-tail-merge.c (replace_block_by): Update.
19747 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19748 (create_edge_and_update_destination_phis): Update.
19749 (compute_path_counts): Update.
19750 (recompute_probabilities): Update.
19751 (update_joiner_offpath_counts): Update.
19752 (freqs_to_counts_path): Update.
19753 (duplicate_thread_path): Update.
19754 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19755 (struct switch_conv_info): Update.
19756 (gen_inbound_check): Update.
19757 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19758 (vect_do_peeling): Update.
19759 (vect_loop_versioning): Update.
19760 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19761 (optimize_mask_stores): Update.
19762 * ubsan.c (ubsan_expand_null_ifn): Update.
19763 * value-prof.c (gimple_divmod_fixed_value): Update.
19764 (gimple_divmod_fixed_value_transform): Update.
19765 (gimple_mod_pow2): Update.
19766 (gimple_mod_pow2_value_transform): Update.
19767 (gimple_mod_subtract): Update.
19768 (gimple_mod_subtract_transform): Update.
19769 (gimple_ic): Update.
19770 (gimple_stringop_fixed_value): Update.
19771 (gimple_stringops_transform): Update.
19772 * value-prof.h: Update.
19774 2017-06-29 Carl Love <cel@us.ibm.com>
19776 * config/rs6000/rs6000-c.c: Add support for built-in functions
19777 vector signed int vec_signed (vector float);
19778 vector signed long long vec_signed (vector double);
19779 vector signed int vec_signed2 (vector double, vector double);
19780 vector signed int vec_signede (vector double);
19781 vector signed int vec_signedo (vector double);
19782 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19783 instruction generator.
19784 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19785 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19786 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19788 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19789 vunsignede_v2df): Add define_expands.
19790 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19791 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19792 VEC_UNSIGNEDO): Add definitions.
19793 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19794 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19795 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19796 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19797 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19798 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19799 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19800 vec_unsignede, vec_unsignedo): Add builtin defines.
19801 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19803 * doc/extend.texi: Update the built-in documentation file for the
19804 new built-in functions.
19806 2017-06-29 Richard Biener <rguenther@suse.de>
19808 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19809 reduction chains to LOOP_VINFO_REDUCTIONS.
19810 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19811 SLP reductions after processing reduction chains.
19813 2017-06-29 Nathan Sidwell <nathan@acm.org>
19815 * builtins.c (fold_builtin_FUNCTION): Use
19816 lang_hooks.decl_printable_name.
19818 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19820 PR middle-end/81194
19821 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19822 with only one label.
19823 * stmt.c (expand_case): Assert NCASES is greater than one.
19825 2017-06-29 Richard Biener <rguenther@suse.de>
19827 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19829 (group_case_labels): Likewise.
19830 (find_taken_edge): Push sanity checking on val to workers...
19831 (find_taken_edge_cond_expr): ... here
19832 (find_taken_edge_switch_expr): ... and here, handle cases
19833 with just a default label.
19834 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19835 (group_case_labels): Likewise.
19836 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19837 group_case_labels does anything cleanup the CFG again.
19839 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19841 PR tree-optimization/81196
19842 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19843 exit condition comparing two IVs.
19845 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19847 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19848 profile to the dummy entry at the end of the list of architectures.
19849 * config/arm/arm-cpu-cdata.h: Regenerated.
19851 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19852 Michael Collison <michael.collison@arm.com>
19855 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19857 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19858 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19859 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19860 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19861 of shift when the shift amount is masked with constant equal to
19862 the size of the mode.
19863 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19866 2017-06-29 Martin Liska <mliska@suse.cz>
19868 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19870 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19873 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19876 (edge_growth_cache_entry::edge_growth_cache_entry): New
19878 (reset_edge_growth_cache): Update to use constructor.
19880 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19882 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19883 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19884 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19886 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19888 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19889 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19891 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19893 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19894 (*-linux-uclibc*): Add t-uclibc tmake_file.
19895 * config/t-musl: New.
19896 * config/t-uclibc: New.
19898 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19900 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19902 (gen_comm_data): Emit architectural setting of arch_prof.
19903 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19905 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19906 (armv8-m.base, armv8-m.main): Likewise.
19907 * arm-protos.h (arm_build_target): Add profile field.
19908 (arch_option): Likewise.
19909 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19911 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19912 arm_active_target.profile.
19914 2017-06-28 Richard Biener <rguenther@suse.de>
19916 PR middle-end/81227
19917 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19918 TYPE_OVERFLOW_WRAPS.
19919 * match.pd (negate_expr_p): Likewise.
19920 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19921 fold_build2, not fold_binary.
19923 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19925 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19926 Convert memory address to Pmode.
19927 (aarch64_print_operand): Assert MEM operands are always Pmode.
19929 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19932 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19933 Remove redundant if.
19934 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19935 * config/arm/aarch-common-protos.h
19936 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19937 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19939 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19942 * multiple_target.c (create_dispatcher_calls): Set the default
19943 clone to be static, not public.
19945 2017-06-28 Richard Biener <rguenther@suse.de>
19947 * tree-vect-loop.c (vectorizable_reduction): Move special
19948 cond reduction IV var creation ...
19949 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19950 parameter. Use STMT_VINFO_VECTYPE.
19951 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19954 2017-06-28 Martin Liska <mliska@suse.cz>
19957 * ipa-visibility.c (non_local_p): Handle visibility.
19959 2017-06-28 Martin Liska <mliska@suse.cz>
19962 * common.opt: Add IntegerRange to various options.
19963 * opt-functions.awk (integer_range_info): New function.
19964 * optc-gen.awk: Add integer_range_info to cl_options struct.
19965 * opts-common.c (decode_cmdline_option): Handle
19966 CL_ERR_INT_RANGE_ARG.
19967 (cmdline_handle_error): Likewise.
19968 * opts.c (print_filtered_help): Show valid interval in
19969 when --help is provided.
19970 * opts.h (struct cl_option): Add range_min and range_max fields.
19971 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19973 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19975 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19976 (x * C EQ/NE y * C): New transformation.
19978 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19980 * genmultilib (combination_space): Accept '+' in option names.
19982 2017-06-28 Martin Liska <mliska@suse.cz>
19985 * asan.c (instrument_derefs): Bail out inner references
19986 that are hard register variables.
19988 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19991 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19992 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19994 2017-06-28 Richard Biener <rguenther@suse.de>
19996 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19997 (vect_get_slp_defs): Adjust.
19998 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20000 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20002 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20003 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20004 (vectorizable_reduction): Adjust.
20005 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20007 (vect_get_slp_defs): Likewise.
20008 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20009 (vectorizable_bswap): Adjust.
20010 (vectorizable_call): Likewise.
20011 (vectorizable_conversion): Likewise.
20012 (vectorizable_assignment): Likewise.
20013 (vectorizable_shift): Likewise.
20014 (vectorizable_operation): Likewise.
20015 (vectorizable_store): Likewise.
20016 (vectorizable_condition): Likewise.
20017 (vectorizable_comparison): Likewise.
20019 2017-06-28 Michael Collison <michael.collison@arm.com>
20022 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20024 (arm_gen_movmemqi): Removed unused variable 'i'.
20025 Convert 'for' loop into 'while' loop.
20026 (arm_expand_prologue): Remove last unnecessary set of insn.
20027 (thumb_pop): Remove unused variable 'pushed_words'.
20028 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20030 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20032 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20033 * config/s390/s390-protos.h: Add prototype of
20034 s390_rel_address_ok_p.
20035 * config/s390/s390.c (s390_got_symbol): New function.
20036 (s390_rel_address_ok_p): New function.
20037 (legitimize_pic_address): Use s390_rel_address_ok_p.
20038 (s390_load_got): Use s390_got_symbol.
20039 (s390_option_override): Issue error if
20040 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20041 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20043 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20045 2017-06-27 Andrew Pinski <apinski@cavium.com>
20047 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20048 (X * copysign (1.0, X)): New pattern.
20049 (X * copysign (1.0, -X)): New pattern.
20050 (copysign (-1.0, CST)): New pattern.
20052 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20054 * genmultilib (combination_space): Remove variable.
20055 Validate reuse rules against regular expression for any sequence
20056 of multilib options in any order.
20058 2017-06-27 Michael Collison <michael.collison@arm.com>
20060 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20061 call aarch64_split_simd_combine.
20062 * (aarch64_combine_internal<mode>): Delete pattern.
20063 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20064 Allow register and subreg operands.
20066 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20068 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20069 specific need, just fallback on defaults.
20070 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20072 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20073 Olivier Hainque <hainque@adacore.com>
20075 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20077 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20078 targets. Pick a default if no particular attempt applied.
20079 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20082 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20084 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20085 (x86_64-wrs-vxworks7): Likewise.
20087 2017-06-27 Marek Polacek <polacek@redhat.com>
20090 * ubsan.c (instrument_null): Check get_base_address's result for null.
20092 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20094 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20096 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20098 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20099 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20100 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20101 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20102 New function types.
20103 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20104 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20105 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20106 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20107 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20108 BUILT_IN_FEUPDATEENV): New builtins.
20109 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20110 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20111 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20112 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20114 (builtin_structptr_types): Adjust size.
20115 * tree.c (builtin_structptr_types): Add four entries.
20117 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20118 Olivier Hainque <hainque@adacore.com>
20120 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20121 (TLS_SYM): New local macro, forcing reference to __tls__ on
20122 link command lines for VxWorks 7 RTPs, triggering initialization
20124 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20125 OS features TLS support, true for RTPs on VxWorks 7.
20126 * config/vxworks.c (vxworks_override_options): Setup emutls
20129 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20131 * predict.c (test_prediction_value_range): Use -1U instead of -1
20132 to avoid narrowing conversion warning.
20133 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20134 to avoid narrowing conversion warning.
20135 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20137 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20139 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20141 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20142 64bit configurations.
20143 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20144 (SIZE_TYPE): Likewise.
20145 * config/vxworks.c (vxworks_emutls_var_fields): Use
20146 long_unsigned_type_node instead of unsigned_type_node as the offset
20147 field type, which is "pointer" mode in emutls.c.
20149 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20152 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20154 PR middle-end/81207
20155 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20156 gimple_vuse copying separately from gimple_vdef copying.
20158 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20160 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20163 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20164 Olivier Hainque <hainque@adacore.com>
20166 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20167 for all vxworks7 targets.
20168 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20169 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20170 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20171 variations for VX6/VX7 and 32/64bits later on in ...
20172 (VXWORKS_LIB_SPEC): Leverage new macros.
20173 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20174 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20176 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20178 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20179 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20181 2017-06-26 Carl Love <cel@us.ibm.com>
20183 * config/rs6000/rs6000-c.c: Add support for built-in functions
20184 vector bool char vec_reve (vector bool char);
20185 vector signed char vec_reve (vector signed char);
20186 vector unsigned char vec_reve (vector unsigned char);
20187 vector bool int vec_reve (vector bool int);
20188 vector signed int vec_reve (vector signed int);
20189 vector unsigned int vec_reve (vector unsigned int);
20190 vector bool long long vec_reve (vector bool long long);
20191 vector signed long long vec_reve (vector signed long long);
20192 vector unsigned long long vec_reve (vector unsigned long long);
20193 vector bool short vec_reve (vector bool short);
20194 vector signed short vec_reve (vector signed short);
20195 vector double vec_reve (vector double);
20196 vector float vec_reve (vector float);
20197 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20198 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20199 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20200 (altivec_vreve): New pattern.
20201 * config/rs6000/altivec.h (vec_reve): New define.
20202 * doc/extend.texi (vec_rev): Update the built-in documentation file
20203 for the new built-in functions.
20205 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20207 PR tree-optimization/71815
20208 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20210 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20212 (slsr_process_phi): Likewise.
20213 (replace_uncond_cands_and_profitable_phis): Don't replace a
20214 multiply candidate with a stride of 1 (copy or cast).
20215 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20217 (lowest_cost_path): Likewise.
20218 (total_savings): Likewise.
20220 2017-06-26 Richard Biener <rguenther@suse.de>
20223 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20224 Use def_builtin_pure for all gather builtins.
20226 2017-06-26 Richard Biener <rguenther@suse.de>
20228 PR tree-optimization/81203
20229 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20230 non-dominating BBs.
20232 2017-06-26 Marek Polacek <polacek@redhat.com>
20235 * doc/invoke.texi: Document -Wmultistatement-macros.
20237 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20239 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20240 arm_neon_ok_no_float_abi effective target.
20242 2017-06-26 Richard Biener <rguenther@suse.de>
20244 PR tree-optimization/80928
20245 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20246 (copy_bbs): Set BB_DUPLICATED flag early.
20247 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20249 (execute_on_shrinking_pred): Likewise.
20250 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20251 BB_DUPLICATED blocks.
20252 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20253 iterate over all PHIs considering removal of *gsi.
20255 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20257 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20260 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20262 * config/rs6000/rs6000-string.c: (expand_block_clear,
20263 do_load_for_compare, select_block_compare_mode,
20264 compute_current_alignment, expand_block_compare,
20265 expand_strncmp_align_check, expand_strn_compare,
20266 expand_block_move, rs6000_output_load_multiple)
20267 Move functions related to string/block move/compare
20268 to a separate file.
20269 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20270 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20271 for this function which is now used in two files.
20272 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20273 * config.gcc: Add rs6000-string.o to extra_objs for
20274 targets powerpc*-*-* and rs6000*-*-*.
20276 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20279 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20280 32-bit, since indexed is not valid for DImode.
20281 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20282 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20283 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20284 (define_peephole2 for Altivec d-form store): Likewise.
20287 * multiple_target.c (create_dispatcher_calls): Only create the
20288 dispatcher call if the function is the default clone of a
20289 versioned function.
20291 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20293 PR middle-end/80902
20294 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20295 a call, force the call to not be a tail call.
20297 2017-06-23 Jeff Law <law@redhat.com>
20299 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20302 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20304 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20305 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20306 handling for early expansion of vector shifts (sl,sr,sra,rl).
20307 (builtin_function_type): Add vector shift right instructions
20308 to the unsigned argument list.
20310 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20312 rtl-optimizatoin/79286
20313 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20314 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20315 trap. PIC register plus a const unspec without offset can never trap.
20317 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20319 * tree.h (builtin_structptr_type): New type.
20320 (builtin_structptr_types): Declare new array.
20321 * tree.c (builtin_structptr_types): New array.
20322 (free_lang_data, build_common_tree_nodes): Use it.
20324 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20327 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20330 2017-06-22 Matt Turner <mattst88@gmail.com>
20332 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20333 Lake models to skylake case. Assume skylake for unknown
20334 models with clflushopt.
20336 2017-06-22 Jeff Law <law@redhat.com>
20338 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20339 frame sizes that do not satisfy aarch64_uimm12_shift.
20341 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20343 * profile-count.h (apply_probability,
20344 apply_scale, probability_in): Fix checks for zero.
20346 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20348 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20349 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20351 2016-06-22 Richard Biener <rguenther@suse.de>
20353 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20354 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20355 REDUC_MAX_EXPR support.
20356 (vectorizable_reduction): Likewise.
20357 (vect_create_epilog_for_reduction): Likewise.
20359 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20361 * match.pd (A / (1 << B) -> A >> B): New.
20362 * generic-match-head.c: Include optabs-tree.h.
20363 * gimple-match-head.c: Likewise.
20364 * optabs-tree.h (target_supports_op_p): New.
20365 * optabs-tree.c (target_supports_op_p): New.
20367 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20369 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20370 $gcc_cv_ld --help output.
20371 (gcc_cv_ld_demangle): Likewise.
20372 (gcc_cv_ld_eh_frame_hdr): Likewise.
20373 (gcc_cv_ld_pie): Likewise.
20374 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20375 (gcc_cv_ld_buildid): Likewise.
20376 (gcc_cv_ld_sysroot): Likewise.
20377 (ld_bndplt_support): Likewise.
20378 (ld_pushpopstate_support): Likewise.
20379 * configure: Regenerate.
20380 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20382 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20385 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20386 operands indexes to avoid gap between operands and match_dups.
20388 2017-06-21 Andrew Pinski <apinski@cavium.com>
20390 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20391 Increment Arith_shift and Arith_shift_reg by 1.
20392 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20394 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20395 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20396 (aarch64_strip_extend): Add new argument and test for it.
20397 (aarch64_cheap_mult_shift_p): New function.
20398 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20399 add a cost if it is true.
20400 Update calls to aarch64_strip_extend.
20401 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20403 2017-06-21 Andrew Pinski <apinski@cavium.com>
20405 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20407 (thunderxt88): Likewise.
20408 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20409 (thunderx_prefetch_tune): New variable.
20410 (thunderx2t99_prefetch_tune): Update for the correct values.
20411 (thunderxt88_tunings): New variable.
20412 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20413 generic_prefetch_tune.
20414 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20416 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20418 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20419 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20420 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20421 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20422 (aarch64_atomic_cas<mode>, GPI): Likewise.
20424 2017-06-21 Martin Liska <mliska@suse.cz>
20426 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20427 statements on cold and hot labels.
20428 * predict.c (tree_estimate_probability_bb): Remove the
20429 prediction from this place.
20431 2017-06-21 Martin Liska <mliska@suse.cz>
20433 PR tree-optimization/79489
20434 * gimplify.c (maybe_add_early_return_predict_stmt): New
20436 (gimplify_return_expr): Call the function.
20437 * predict.c (tree_estimate_probability_bb): Remove handling
20439 * predict.def: Update comment about early return predictor.
20440 * gimple-predict.h (is_gimple_predict): New function.
20441 * predict.def: Change default value of early return to 66.
20442 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20444 * passes.def: Put pass_strip_predict_hints to the beginning of
20447 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20449 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20450 FUNCTION_DECL declarations.
20451 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20453 (dwaf2out_decl): Likewise.
20454 * godump.c (go_early_global_decl): Skip call to the real debug hook
20455 for FUNCTION_DECL declarations.
20456 * passes.c (rest_of_decl_compilation): Skip call to the
20457 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20458 -fdump-go-spec is passed.
20460 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20462 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20463 Reorder for compactness.
20464 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20465 (def_builtin_pure, def_builtin_pure2): New functions.
20466 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20468 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20470 * match.pd (nop_convert): New predicate.
20471 ((A +- CST1) +- CST2): Allow some NOP conversions.
20473 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20476 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20477 with ctors/dtors if GOVD_SHARED is set.
20479 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20481 * config/aarch64/aarch64.md (movti_aarch64):
20482 Emit mov rather than orr.
20483 (movtf_aarch64): Likewise.
20484 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20485 Emit mov rather than orr.
20487 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20489 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20490 Swap alternatives, make integer dup more expensive.
20492 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20494 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20495 Return true for non-tls symbols.
20497 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20499 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20500 (cortex-a75): Likewise.
20501 (cortex-a75.cortex-a55): Likewise.
20502 * config/aarch64/aarch64-tune.md: Regenerate.
20503 * doc/invoke.texi (-mtune): Document new values for -mtune.
20505 2017-06-21 Tom de Vries <tom@codesourcery.com>
20507 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20508 stack_size feature.
20509 (Effective-Target Keywords, Other attributes): Suggest using
20510 dg-add-options stack_size feature to get stack limit in stack_size
20511 effective target documentation.
20513 2017-06-21 Julian Brown <julian@codesourcery.com>
20514 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20516 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20517 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20518 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20520 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20521 attribute type list for neon_multiply.
20522 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20523 attribute type list for neon_multiply.
20524 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20525 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20526 attribute type list for neon_multiply.
20527 * config/arm/types.md (crypto_pmull): Add.
20528 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20529 attribute type list.
20531 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20533 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20536 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20538 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20539 to make sure not to dereference a NULL cost_classes_ptr pointer.
20541 2017-06-20 Carl Love <cel@us.ibm.com>
20543 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20544 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20545 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20546 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20547 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20548 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20549 VMULOSW): New enum "unspec" values.
20550 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20551 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20552 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20553 altivec_vmulosw): New patterns.
20554 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20555 VMULOSW): Add definitions.
20557 2017-06-20 Julia Koval <julia.koval@intel.com>
20559 * config/i386/i386.c: Fix rounding expand for new pattern.
20560 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20562 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20564 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20565 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20567 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20569 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20572 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20574 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20575 architecture, then by implementer ID.
20576 * config/aarch64/aarch64-tune.md: Regenerate.
20578 2017-06-20 Richard Biener <rguenther@suse.de>
20580 PR middle-end/81097
20581 * fold-const.c (split_tree): Fold to type before negating.
20583 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20585 * diagnostic-show-locus.c
20586 (selftest::test_fixit_deletion_affecting_newline): New function.
20587 (selftest::diagnostic_show_locus_c_tests): Call it.
20589 2017-06-20 Andreas Schwab <schwab@suse.de>
20592 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20595 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20597 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20598 __ARM_FEATURE_COPROC according to support.
20600 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20602 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20603 Rewritten to avoid overflow for > 32-bit pointers.
20606 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20607 by removing enum keyword.
20608 (ubsan_type_descriptor): Likewise. Formatting fix.
20611 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20612 splitter): Require TARGET_SSE2 in the condition.
20614 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20617 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20618 for doing vector set of SFmode on ISA 3.0.
20619 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20620 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20622 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20623 SFmode value into a V4SF variable that was extracted from another
20624 V4SF variable without converting the element to double precision
20625 and back to single precision vector format.
20626 (vsx_insert_extract_v4sf_p9_2): Likewise.
20628 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20630 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20631 in UWHI to avoid undefined overflow.
20634 * ubsan.h (enum ubsan_encode_value_phase): New.
20635 (ubsan_encode_value): Change second argument to
20636 enum ubsan_encode_value_phase with default value of
20637 UBSAN_ENCODE_VALUE_GENERIC.
20638 * ubsan.c (ubsan_encode_value): Change second argument to
20639 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20640 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20641 create_tmp_var_raw instead of create_tmp_var and use a
20643 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20644 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20645 ubsan_encode_value callers.
20648 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20649 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20650 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20652 2017-06-19 Richard Biener <rguenther@suse.de>
20654 PR middle-end/81118
20655 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20656 estimates if we changed anything.
20658 2017-06-19 Richard Biener <rguenther@suse.de>
20660 PR tree-optimization/80887
20661 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20662 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20663 simplified lookups, then reset mprts_hook.
20664 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20666 (try_to_simplify): Likewise.
20668 2017-06-19 Martin Liska <mliska@suse.cz>
20671 * gimplify.c (gimplify_switch_expr):
20672 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20674 2017-06-19 Martin Liska <mliska@suse.cz>
20676 * doc/install.texi: Document that PGO runs in 4 stages.
20678 2017-06-19 Martin Liska <mliska@suse.cz>
20681 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20682 to dispatcher function name.
20683 * multiple_target.c (replace_function_decl): New function.
20684 (create_dispatcher_calls): Redirect both edges and references.
20686 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20688 * profile-count.c (profile_count::dump): Dump quality.
20689 (profile_count::differs_from_p): Update for unsigned val.
20690 * profile-count.h (profile_count_quality): New enum.
20691 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20693 2017-06-19 Richard Biener <rguenther@suse.de>
20695 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20696 struct function as arg.
20697 (estimate_numbers_of_iterations): Export overload with loop arg.
20698 (free_numbers_of_iterations_estimates_loop): Use an overload of
20699 free_numbers_of_iterations_estimates instead.
20700 * tree-cfg.c (remove_bb): Adjust.
20701 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20702 * tree-parloops.c (gen_parallel_loop): Likewise.
20703 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20705 (tree_unroll_loops_completely): Likewise.
20706 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20707 Use an overload instead and export.
20708 (estimated_loop_iterations): Adjust.
20709 (max_loop_iterations): Likewise.
20710 (likely_max_loop_iterations): Likewise.
20711 (estimate_numbers_of_iterations): Take struct function as arg
20713 (loop_exits_before_overflow): Adjust.
20714 (free_numbers_of_iterations_estimates_loop): Use an overload.
20715 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20716 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20718 2017-06-19 Richard Biener <rguenther@suse.de>
20721 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20722 RANGE_EXPR conservatively.
20724 2017-06-16 Carl Love <cel@us.ibm.com>
20726 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20727 definitions for vec_float, vec_float2, vec_floato,
20728 vec_floate built-ins.
20729 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20730 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20732 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20733 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20734 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20735 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20736 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20737 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20738 vec_floato): Add builtin defines.
20739 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20740 Update the built-in documentation file for the new built-in
20743 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20745 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20746 (mthumb): Mark as the negative of -marm.
20748 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20750 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20752 (ARM Options, -mtune): Document that this accepts the same
20753 extension options as -mcpu.
20754 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20758 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20759 permitted extensions.
20761 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20763 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20764 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20765 (armv8-m.main): Add option +nodsp.
20766 * config/arm/arm-cpu-cdata.h: Regenerated.
20768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20770 * config/arm/t-fuchsia: New file.
20771 * config.gcc (arm*-*-fuchsia*): Use it.
20773 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20775 * config/arm/t-symbian: Rewrite for new option infrastructure.
20777 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20779 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20780 (MULTILIB_REQUIRED): Likewise.
20782 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20784 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20785 (MULTILIB_RESUE): Likewise.
20786 (MULTILIB_MATCHES): Likewise.
20787 (MULTLIB_REQUIRED): Likewise.
20789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20791 * config/arm/t-rtems: Rewrite for new option framework.
20793 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20795 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20796 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20797 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20798 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20799 * config/arm/t-multilib: ... here.
20800 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20801 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20802 armv7-a and armv8*-a when A-profile libraries have not been built.
20803 * config/arm/t-rmprofile: Rewrite.
20805 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20807 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20808 with a backslash. Remove the backslash after substituting unescaped
20810 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20812 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20814 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20815 the driver to pass through the default setting of -mfloat-abi.
20816 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20818 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20819 (all_feat_combs): New rule.
20820 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20822 * config/arm/t-aprofile: Rewrite.
20824 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20826 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20827 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20828 fpu is not specified by the user/command-line.
20829 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20830 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20831 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20832 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20833 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20834 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20838 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20839 * config/arm/t-arm-elf: Rewritten.
20841 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20843 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20844 have some floating-point instructions.
20845 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20846 (TARGET_MAYBE_HARD_FLOAT): New macro.
20847 * config/arm/arm-builtins.c (arm_init_builtins): Use
20848 TARGET_MAYBE_HARD_FLOAT.
20849 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20853 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20854 (configargs.h): Include it.
20855 (arm_print_hint_for_fpu_option): New function.
20856 (arm_parse_fpu_option): New function.
20857 (candidate_extension): New class.
20858 (arm_canon_for_multilib): New function.
20859 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20860 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20861 (ARCH_CANONICAL_SPECS): New macro.
20862 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20866 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20867 are set after handling multilib fragments. Set target_cpu_default2
20870 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20872 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20874 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20875 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20877 (check_arch): Likewise.
20878 * config/arm/arm.c (arm_configure_build_target): Handle
20879 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20880 options in the default.
20882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20884 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20885 when an option is an alias of another.
20886 * config/arm/parsecpu.awk (optalias): New parser token.
20887 (gen_comm_data): Mark non-alias options as such. Emit entries
20888 for extension aliases.
20889 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20890 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20891 (armv6kz, armv6zk, armv6t2): Likewise.
20892 (armv7): Make vfpv3-d16 an alias.
20893 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20895 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20896 Sort in canonical order.
20897 (armv8-a): Sort in canonical order.
20898 (armv8.1-a, armv8.2-a): Likewise.
20899 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20901 (cortex-a9): Sort in canonical order.
20902 * config/arm/arm.c (selftests.h): Include it.
20903 (arm_test_cpu_arch_data): New function.
20904 (arm_run_self_tests): New function.
20905 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20906 (targetm): Move declaration to the end of the file.
20907 * arm-cpu-cdata.h: Regenerated.
20909 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20911 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20912 call to target_mode_check describing the type of option passed.
20913 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20914 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20915 arm_parse_cpu_option_name to match parameters against list of
20917 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20918 arm_arch_core_flags data structure.
20919 * config/arm/arm-cpu_cdata.h: Regenerated.
20921 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20923 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20925 (arm_print_hint_for_cpu_option): Likewise.
20926 (arm_print_hint_for_arch_option): Likewise.
20927 (arm_parse_cpu_option_name): Likewise.
20928 (arm_parse_arch_option_name): Likewise.
20929 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20930 of entries in the all_fpus list.
20931 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20932 (arm_parse_cpu_option_name): Declare.
20933 (arm_parse_arch_option_name): Declare.
20934 (arm_parse_option_features): Declare.
20935 (arm_intialize_isa): Declare.
20936 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20938 (gen_comm_data): ... here. Make definitions non-static.
20939 * config/arm/arm-cpu-data.h: Regenerated.
20940 * config/arm/arm-cpu-cdata.h: Regenerated.
20942 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20944 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20945 (cpu_arch_extension): New structure.
20946 (cpu_arch_option, arch_option, cpu_option): New structures.
20947 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20948 architecture types.
20949 (gen_data): Generate new format data tables.
20950 * config/arm/arm.c (cpu_tune): New structure.
20951 (cpu_option, processors): Delete.
20952 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20953 (arm_print_hint_for_cpu_option): ... this and ...
20954 (arm_print_hint_for_arch_option): ... this.
20955 (arm_parse_arch_cpu_name): Delete. Replace with ...
20956 (arm_parse_cpu_option_name): ... this and ...
20957 (arm_parse_arch_option_name): ... this.
20958 (arm_unrecognized_feature): Change type of target parameter to
20960 (arm_parse_arch_cpu_features): Delete. Replace with ...
20961 (arm_parse_option_features): ... this.
20962 (arm_configure_build_target): Rework to use new configuration data
20964 (arm_print_tune_info): Rework for new configuration data tables.
20965 * config/arm/arm-cpu-data.h: Regenerated.
20966 * config/arm/arm-cpu.h: Regenerated.
20968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20970 * Makefile.in (OBJS): Move sbitmap.o from here ...
20971 (OBJS-libcommon): ... to here.
20973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20975 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20976 (ISA_ALL_CRYPTO): New macro.
20977 (ISA_ALL_SIMD): New macro
20978 (ISA_ALL_FP): New macro.
20979 * config/arm/arm.c (fpu_bitlist): Update initializer.
20980 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20982 (arm9e): Add fpu. Add option for nofp
20983 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20984 (arm926ej-s, arm1026ej-s): Likewise.
20985 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20986 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20987 neon-fp16, neon-vfpv4, nofp and nosimd.
20988 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20989 (cortex-a8): Add fpu. Add option for nofp.
20990 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20991 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20992 (cortex-r4f): Add fpu.
20993 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20994 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20996 (cortex-r8): Likewise.
20997 (cortex-m4): Add fpu. Add option for nofp.
20998 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20999 (cortex-a17.cortex-a7): Likewise.
21000 (cortex-a32): Add fpu. Add options for crypto and nofp.
21001 (cortex-a35, cortex-a53): Likewise.
21002 (cortex-a57): Add fpu. Add option for crypto.
21003 (cortex-a72, cortex-a73): Likewise.
21004 (exynos-m1): Likewise.
21005 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21006 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21007 (cortex-m33): Add fpu. Add option for nofp.
21008 * config/arm/arm-cpu-cdata.h: Regenerated
21009 * config/arm/arm-cpu-data.h: Regenerated.
21011 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21013 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21014 (armv5te, armv5tej): Likewise.
21015 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21016 (armv7): Add options fp and vfpv3-d16.
21017 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21018 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21020 (armv7ve): Likewise.
21021 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21022 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21023 (armv8-a): Add nocrypto option.
21024 (armv8.1-a, armv8.2-a): Likewise.
21025 (armv8-m.main): add options fp, fp.dp and nofp.
21027 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21029 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21031 (armv8-a+crc): Delete.
21032 (armv8.1-a): Add options simd, crypto and nofp.
21033 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21034 (armv8.2-a+fp16): Delete.
21035 (armv8-m.main): Add option dsp.
21036 (armv8-m.main+dsp): Delete.
21037 (cortex-a8): Add fpu. Add nofp option.
21038 (cortex-a9): Add fpu. Add nofp and nosimd options.
21039 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21040 link to main cpu and architecture data structures.
21041 (gen_comm_data): Only put isa attributes from the main architecture
21043 (option): New statement for architecture and CPU entries.
21044 * arm.c (struct cpu_option): New structure.
21045 (struct processors): Add entry for options.
21046 (arm_unrecognized_feature): New function.
21047 (arm_parse_arch_cpu_name): Ignore any characters after the first
21049 (arm_parse_arch_cpu_feature): New function.
21050 (arm_configure_build_target): Separate out any CPU and architecture
21051 features and parse separately. Don't error out if -mfpu=auto is
21052 used with only an architecture string.
21053 (arm_print_asm_arch_directives): New function.
21054 (arm_file_start): Call it.
21055 * config/arm/arm-cpu-cdata.h: Regenerated.
21056 * config/arm/arm-cpu-data.h: Likewise.
21057 * config/arm/arm-tables.opt: Likewise.
21059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21061 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21062 assembler when it is not -mfpu=auto.
21064 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21066 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21067 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21068 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21069 (ASM_CPU_SPEC): Rewrite.
21070 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21071 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21072 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21073 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21074 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21075 copied string is NUL-terminated. Also strip any characters prefixed
21077 (arm_rewrite_selected_arch): New function.
21078 (arm_rewrite_march): New function.
21080 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21082 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21083 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21084 (march, mcpu, mtune): Convert to string-based options.
21085 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21086 (arm_parse_arch_cpu_name): New function.
21087 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21088 identify selected architecture or CPU.
21089 (arm_option_save): New function.
21090 (TARGET_OPTION_SAVE): Redefine.
21091 (arm_option_restore): Restore string options.
21092 (arm_option_print): Print string options.
21094 2017-06-16 Martin Sebor <msebor@redhat.com>
21096 PR tree-optimization/80933
21097 PR tree-optimization/80934
21098 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21099 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21100 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21101 (gimple_fold_builtin): Call them.
21103 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21105 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21106 as unlikely; update profile.
21108 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21110 * predict.c (force_edge_cold): Handle declaring edges impossible
21113 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21115 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21117 (try_unroll_loop_completely): Fix reporting.
21119 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21121 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21123 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21126 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21127 if given a non-constant argument for an intrinsic which requires a
21130 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21132 * profile.c (compare_freqs): New function.
21133 (branch_prob): Sort edge list.
21134 (find_spanning_tree): Assume that the list is priority sorted.
21136 2017-06-16 Richard Biener <rguenther@suse.de>
21138 PR tree-optimization/81090
21139 * passes.def (pass_record_bounds): Remove.
21140 * tree-pass.h (make_pass_record_bounds): Likewise.
21141 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21142 make_pass_record_bounds): Likewise.
21143 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21144 not free niter estimates at the beginning but at the end.
21145 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21147 2017-06-16 Richard Biener <rguenther@suse.de>
21149 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21150 initializer to workaround ICE in host GCC 4.8.
21152 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21154 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21156 (clone_inlined_nodes): Update.
21158 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21160 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21161 prefetch settings, and enable prefetching by default at -O3.
21163 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21165 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21166 Set flag_prefetch_loop_arrays according to tuning data.
21168 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21170 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21171 New tune structure.
21172 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21173 [Unrelated to main purpose of the patch] Place the pointer field last
21174 to enable type checking errors when tune structure are wrongly merged.
21175 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21176 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21177 (thunderx2t99_prefetch_tune): New tune constants.
21178 (tune_params *_tunings): Update all tunings (no functional change).
21179 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21180 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21181 from tunings structures.
21183 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21186 * ubsan.c (instrument_null): Add T argument, use it instead
21187 of computing it based on IS_LHS.
21188 (instrument_object_size): Likewise.
21189 (pass_ubsan::execute): Adjust instrument_null and
21190 instrument_object_size callers to pass gimple_get_lhs or
21191 gimple_assign_rhs1 result to it. Use instrument_null instead of
21192 calling get_base_address and instrument_mem_ref. Handle
21193 aggregate call arguments for object-size sanitization.
21195 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21197 PR tree-optimization/81089
21198 * tree-vrp.c (is_masked_range_test): Validate operands of
21201 2017-06-15 Martin Sebor <msebor@redhat.com>
21204 * dumpfile.c (dump_register): Avoid calling memset to initialize
21205 a class with a default ctor.
21206 * gcc.c (struct compiler): Remove const qualification.
21207 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21208 * hash-table.h: Ditto.
21209 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21211 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21212 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21214 * params.h (struct param_info): Make struct members non-const.
21215 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21216 with default initialization.
21217 * vec.h (vec_copy_construct, vec_default_construct): New helper
21219 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21220 with vec_copy_construct.
21221 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21222 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21223 * doc/invoke.texi (-Wclass-memaccess): Document.
21225 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21227 * emit-rtl.h (is_leaf): Update comment about local
21228 register allocator.
21230 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21233 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21234 for a variable to have a section before checking if the section has a
21236 Set section to.persistent if persistent attribute is set.
21237 Warn if .persistent attribute is used on an automatic variable.
21239 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21241 PR rtl-optimization/80474
21242 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21244 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21246 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21249 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21251 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21252 (casesi): Emit bounds checking as RTL.
21253 (casesi_internal_mips16_<mode>): Remove bounds checking.
21255 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21257 * config/xtensa/xtensa.c (xtensa_option_override): Append
21258 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21259 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21260 xtensa_doloop_hooks): Define unconditionally.
21261 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21263 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21264 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21265 for it in xtensa_option_override.
21266 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21267 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21269 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21271 * doc/cppopts.texi: Document '-' special value to -MF.
21273 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21275 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21276 (cortex_a53_fconst): Likewise.
21277 (cortex_a53_fpmul): Likewise.
21278 (cortex_a53_f_load_64): Likewise.
21279 (cortex_a53_f_load_many): Likewise.
21280 (cortex_a53_advsimd_alu): Likewise.
21281 (cortex_a53_advsimd_alu_q): Likewise.
21282 (cortex_a53_advsimd_mul): Likewise.
21283 (cortex_a53_advsimd_mul_q): Likewise.
21284 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21285 Add missing fmul, r2f_cvt and fconst cases.
21287 2017-06-14 Richard Biener <rguenther@suse.de>
21289 PR middle-end/81088
21290 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21292 (fold_binary_loc): When associating do not treat pre-existing
21293 TREE_OVERFLOW on literal constants as a reason to allow
21294 TREE_OVERFLOW on associated literal constants.
21296 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21298 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21299 (MASK_FEATURES): New macro.
21300 * config/sparc/sparc.c (sparc_option_override): Remove the special
21301 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21303 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21305 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21306 a division of 0 if non-call exceptions are enabled.
21308 2017-06-14 Andrew Pinski <apinski@cavium.com>
21309 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21312 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21313 Improve vector initialization code gen for only variable case.
21315 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21317 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21319 2017-06-14 Richard Biener <rguenther@suse.de>
21321 PR tree-optimization/81083
21322 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21325 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21327 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21328 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21329 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21330 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21331 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21332 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21334 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21336 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21337 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21339 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21341 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21343 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21345 * config/rs6000/t-rtems: Don't handle SPE.
21347 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21349 * config/rs6000/t-linux: Don't handle SPE.
21351 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21353 * config/rs6000/eabispe.h: Delete file.
21355 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21357 * config/rs6000/t-spe: Delete file.
21359 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21361 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21362 (rs6000_legitimate_offset_address_p): Return false for anything in
21363 V2SImode or V2SFmode.
21365 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21367 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21368 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21369 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21371 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21372 (rs6000_legitimate_offset_address_p): Ditto.
21373 (rs6000_emit_move): Ditto.
21374 (rs6000_init_builtins): Remove V4HI_type_node.
21376 2017-06-13 Martin Liska <mliska@suse.cz>
21379 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21380 (gate_asan): Likewise.
21381 * asan.h (asan_no_sanitize_address_p): Remove the function.
21382 (sanitize_flags_p): New function.
21383 * builtins.def: Fix coding style.
21384 * common.opt: Use renamed enum value.
21385 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21386 * doc/extend.texi: Document no_sanitize attribute.
21387 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21388 to SANITIZE_UNDEFINED_NONDEFAULT.
21389 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21390 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21391 Use sanitize_flags_p.
21392 * gimplify.c (gimplify_function_tree): Likewise.
21393 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21394 * opts.c (parse_no_sanitize_attribute): New function.
21395 (common_handle_option): Use renamed enum value.
21396 * opts.h (parse_no_sanitize_attribute): Declare.
21397 * tree.c (sanitize_flags_p): New function.
21398 * tree.h: Declared here.
21399 * tsan.c: Use sanitize_flags_p.
21400 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21401 (instrument_mem_ref): Likewise.
21402 (instrument_bool_enum_load): Likewise.
21403 (do_ubsan_in_current_function): Remove the function.
21404 (pass_ubsan::execute): Use sanitize_flags_p.
21405 * ubsan.h: Remove do_ubsan_in_current_function
21406 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21407 (dump_function_to_file): Use it here.
21409 2017-06-13 Martin Jambor <mjambor@suse.cz>
21411 PR tree-optimization/80803
21412 PR tree-optimization/81063
21413 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21414 (propagate_subaccesses_across_link): Enqueue subtree whenever
21415 necessary instead of relying on the caller.
21417 2017-06-13 Martin Jambor <mjambor@suse.cz>
21419 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21420 that have a first_link.
21421 (sort_and_splice_var_accesses): Do not check first_link before
21423 (subtree_mark_written_and_enqueue): Likewise.
21424 (propagate_all_subaccesses): Likewise and do not stop at first
21425 parent with a first_link.
21427 2017-06-13 Martin Jambor <mjambor@suse.cz>
21429 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21432 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21434 * match.pd: New pattern.
21436 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21438 * tree-vrp.c (is_masked_range_test): New function.
21439 (register_edge_assert_for): Determine ranges for
21442 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21444 PR tree-optimization/67328
21445 * fold-const.c (maskable_range_p): New function.
21446 (build_range_check): Generate bittests if possible.
21448 2017-06-13 Martin Liska <mliska@suse.cz>
21450 * gimple-pretty-print.c (dump_probability): Add new argument.
21451 (dump_edge_probability): Dump both probability and count.
21452 (dump_gimple_label): Likewise.
21453 (dump_gimple_bb_header): Likewise.
21455 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21458 * config/avr/avr-devices.c: Fix indentation.
21459 * config/avr/gen-avr-mmcu-specs.c: Dito.
21461 2017-06-13 Richard Biener <rguenther@suse.de>
21463 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21464 instead get vector type from stmt_info.
21465 (vectorizable_reduction): Adjust. Remove dead code.
21467 2017-06-13 Richard Biener <rguenther@suse.de>
21469 PR middle-end/81065
21470 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21471 case of C * (x * C2 + C3).
21472 (fold_addr_of_array_ref_difference): Properly fold index difference.
21474 2017-06-12 David S. Miller <davem@davemloft.net>
21477 * config/sparc/sparc.md (return expander): Emit frame blockage if
21478 function uses alloca.
21480 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21482 * combine.c (make_field_assignment): Check len rather than the mode
21483 precision when calling force_to_mode.
21485 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21487 Support multilibs and devices that see flash in RAM address range.
21490 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21491 (avr_mcu_t) <flash_pm_offset>: New field.
21492 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21493 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21494 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21495 (AVR_TINY_PM_OFFSET): Remove macro.
21496 * config/avr/avr.opt (-mshort-calls): New option.
21497 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21498 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21499 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21500 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21501 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21502 instead of avr_arch->have_jmp_call.
21503 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21504 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21505 avr_arch->flash_pm_offset to define.
21506 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21507 new field flash_pm_offset. Add entry for avrxmega3.
21508 (avr_texinfo): Add entry for avrxmega3.
21509 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21510 attiny212, attiny214,
21511 attiny412, attiny414, attiny416, attiny417,
21512 attiny814, attiny816, attiny817,
21513 attiny1614, attiny1616, attiny1617,
21514 attiny3214, attiny3216, attiny3217.
21515 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21516 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21517 (avr_print_operand_address) [AVR_TINY]: Same.
21518 (avr_asm_init_sections) <readonly_data_section>: Only patch
21519 callback if avr_arch->flash_pm_offset = 0.
21520 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21521 for rodata if avr_arch->flash_pm_offset != 0.
21522 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21523 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21524 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21525 (m_options): Append opt_rcall.
21526 (m_dirnames): Append dir_rcall.
21527 * config/avr/t-multilib: Regenerate.
21529 * configure.ac [target=avr]: Check whether avrxmega3 default
21530 linker description file works as needed.
21531 * configure: Regenerate.
21532 * doc/avr-mmcu.texi: Regenerate.
21533 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21534 <__AVR_ARCH__>: Document avrxmega3 and 103.
21535 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21536 <__AVR_SHORT_CALLS__>: Document it.
21537 <__AVR_PM_BASE_ADDRESS__>: Document it.
21538 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21539 (AVR Variable Attributes) <progmem>: Document this is
21540 not needed for avrxmega3.
21541 (AVR Named Address Spaces) <__flash>: Dito.
21543 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21545 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21547 2017-06-12 Doug Rupp <rupp@adacore.com>
21549 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21550 Append vxworks-stdint.h to the tm_file list.
21551 * config/vxworks-stdint.h: New file.
21553 2017-06-12 Martin Liska <mliska@suse.cz>
21555 PR tree-optimization/81041
21556 * tree-profile.c (gimple_gen_ic_func_profiler):
21557 Create an extra BB in profile-generate
21558 (gimple_gen_time_profiler): Likewise.
21560 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21562 PR tree-optimization/81003
21563 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21564 (update_range_test): Use it instead of force_gimple_operand_gsi.
21566 2017-06-12 Richard Biener <rguenther@suse.de>
21568 PR tree-optimization/81053
21569 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21570 with backedge value not defined in loop. Simplify def stmt
21573 2017-06-11 Tom de Vries <tom@codesourcery.com>
21576 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21578 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21579 nvptx_cannot_force_const_mem.
21581 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21583 * opts.c (finish_options): Move test for flag_split_stack after
21584 it has been initialized.
21586 2017-06-11 Jason Merrill <jason@redhat.com>
21588 * tree.h (id_equal): New.
21589 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21590 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21591 instead of strcmp of IDENTIFIER_POINTER.
21593 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21595 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21596 (mark_all_inlined_calls_cdtor): Fix formating.
21597 (inline_transform): Rescale profile before inlining.
21599 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21601 * cgraph.h (cgraph_edge::clone): Update prototype.
21602 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21603 (cgraph_node::create_clone): Update.
21604 (cgraph_node::create_version_clone): Update.
21605 * tree-inline.c (copy_bb): Update.
21606 (expand_call_inline): Update.
21608 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21610 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21611 factored out from ...
21612 (rs6000_emit_prologue): ... here.
21614 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21616 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21617 factored out from ...
21618 (rs6000_emit_prologue): ... here.
21620 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21622 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21624 (handle_missing_profiles): Fix computation of tp_first_run.
21625 (counts_to_freqs): Do not touch freqs when count is 0.
21627 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21629 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21632 2017-06-10 Tom de Vries <tom@codesourcery.com>
21634 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21635 attributes): Document signal effective target.
21637 2017-06-10 Tom de Vries <tom@codesourcery.com>
21639 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21640 Document effective target stack_size.
21642 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21644 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21645 to the edit_context if they can be auto-applied.
21647 2017-06-9 Ian Lance Taylor <iant@golang.org>
21649 * opts.c (finish_options): If -fsplit-stack, disable implicit
21650 -forder-blocks-and-partition.
21651 * doc/invoke.texi (Optimize Options): Document that when using
21652 -fsplit-stack -forder-blocks-and-partition is not implicitly
21655 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21657 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21658 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21659 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21660 * builtins.def (abort, trap, unreachable): Declare cold.
21661 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21662 * tree-core.h (ECF_COLD): New.
21663 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21664 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21666 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21668 * predict.c (unlikely_executed_stmt_p): Cleanup.
21670 2017-06-09 Richard Biener <rguenther@suse.de>
21672 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21673 model if the ref is always written to.
21675 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21677 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21679 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21681 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21684 2017-06-09 Tom de Vries <tom@codesourcery.com>
21687 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21688 "target cannot support label values" when encountering LABEL_REF.
21690 2017-06-09 Martin Liska <mliska@suse.cz>
21692 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21693 (gimple_gen_ic_func_profiler): Emit direct comparison
21694 of __gcov_indirect_call_callee with NULL.
21695 (gimple_gen_time_profiler): Change probability from
21696 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21698 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21700 * profile.c (edge_gcov_counts): Turn to pointer.
21701 (compute_branch_probabilities, compute_branch_probabilities): Update.
21702 (branch_prob): Do not clear edge_gcov_count.
21703 * profile.h (edge_gcov_counts): Turn to pointer.
21704 (edge_gcov_count): Update.
21706 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21708 * gimple.h (gimple_check_failed): Mark cold.
21710 2017-06-09 Richard Biener <rguenther@suse.de>
21712 PR tree-optimization/66623
21713 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21714 refactor check_reduction into two parts, properly computing
21715 whether we have to check reduction validity for outer loop
21718 2017-06-09 Richard Biener <rguenther@suse.de>
21720 PR tree-optimization/79483
21721 * graphite-scop-detection.c (order): New global.
21722 (get_order): Compute bb to order mapping that satisfies code
21723 generation constraints.
21724 (cmp_pbbs): New helper.
21725 (build_scops): Start domwalk at entry block, sort generated
21728 2017-06-09 Richard Biener <rguenther@suse.de>
21730 PR middle-end/81007
21731 * ipa-polymorphic-call.c
21732 (ipa_polymorphic_call_context::restrict_to_inner_class):
21733 Skip FIELD_DECLs with error_mark_node type.
21734 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21737 2017-06-09 Martin Liska <mliska@suse.cz>
21739 * predict.c (struct branch_predictor): New struct.
21740 (test_prediction_value_range): New test.
21741 (predict_c_tests): New function.
21742 * selftest-run-tests.c (selftest::run_tests): Run the function.
21743 * selftest.h: Declare new tests.
21745 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21748 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21749 gen_add3_insn did not fail.
21750 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21751 r0, construct that number in a temporary reg and add that reg to r0.
21752 If asked to put the result in r0 as well, fail.
21754 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21756 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21757 for early expansion of vec_eqv.
21759 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21761 PR middle-end/81005
21762 * ubsan.c (instrument_null): Avoid pointless code temporary.
21763 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21765 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21769 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21772 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21775 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21777 * predict.c (unlikely_executed_edge_p): Move ahead.
21778 (probably_never_executed_edge_p): Use it.
21780 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21782 PR middle-end/79988
21783 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21784 gimple_call_builtin_p call.
21786 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21788 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21789 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21790 rtl_check_failed_type2, rtl_check_failed_code1,
21791 rtl_check_failed_code2, rtl_check_failed_code_mode,
21792 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21793 rtvec_check_failed_bounds, rtl_check_failed_flag,
21794 _fatal_insn_not_found, _fatal_insn): Likewise.
21795 * tree.h (tree_contains_struct_check_failed,
21796 tree_check_failed, tree_not_check_failed,
21797 tree_class_check_failed, tree_range_check_failed,
21798 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21799 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21800 tree_operand_check_failed, omp_clause_check_failed,
21801 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21804 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21806 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21807 flag_branch_probabilities.
21808 * ipa-inline.c (edge_badness): Likewise.
21809 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21810 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21811 * predict.c (maybe_hot_frequency_p): Likewise.
21812 (probably_never_executed): Likewise.
21813 * sched-ebb.c (schedule_ebbs): Likewise.
21814 * sched-rgn.c (find_single_block_region): Likewise.
21815 * tracer.c (tail_duplicate): Likewise.
21817 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21819 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21820 longer requires x_flag_profile_use.
21822 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21824 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21825 instead of flag_reorder_blocks_and_partition.
21826 * dbxout.c (dbxout_function_end): Likewise.
21827 * dwarf2out.c (gen_subprogram_die): Likewise.
21828 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21829 * hw-doloop.c (reorg_loops): Likewise.
21830 * varasm.c (assemble_start_function,
21831 assemble_end_function): Likewise.
21832 (decide_function_section): Do not check for
21833 flag_reorder_blocks_and_partition.
21835 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21837 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21839 (chkp_get_hard_register_fake_addr_expr): Ditto.
21840 (chkp_build_addr_expr): Add check for hard reg case.
21841 (chkp_parse_array_and_component_ref): Ditto.
21842 (chkp_find_bounds_1): Ditto.
21843 (chkp_process_stmt): Don't generate bounds store for
21846 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21848 * predict.c (maybe_hot_bb_p): Do not check profile status.
21849 (maybe_hot_edge_p): Likewise.
21850 (probably_never_executed): Check for zero counts even if profile
21852 (unlikely_executed_edge_p): New function.
21853 (unlikely_executed_stmt_p): New function.
21854 (unlikely_executed_bb_p): New function.
21855 (set_even_probabilities): Use unlikely predicates.
21856 (combine_predictions_for_bb): Likewise.
21857 (predict_paths_for_bb): Likewise.
21858 (predict_paths_leading_to_edge): Likewise.
21859 (determine_unlikely_bbs): New function.
21860 (estimate_bb_frequencies): Use it.
21861 (compute_function_frequency): Use zero counts even if profile is
21863 * profile-count.h: Fix typo.
21865 2017-08-08 Julia Koval <julia.koval@intel.com>
21867 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21868 _mm512_mask_cvtsepi16_storeu_epi8,
21869 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21870 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21871 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21872 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21873 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21874 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21875 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21876 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21877 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21878 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21879 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21880 __builtin_ia32_pmovuswb256mem_mask,
21881 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21882 __builtin_ia32_pmovwb512mem_mask): New builtins.
21884 2017-08-08 Julia Koval <julia.koval@intel.com>
21886 PR target/73350,80862
21887 * config/i386/subst.md (round): Fix round pattern.
21888 * config/i386/i386.c (ix86_erase_embedded_rounding):
21889 Fix erasing rounding for the fixed pattern.
21891 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21893 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21895 2017-06-08 Martin Liska <mliska@suse.cz>
21897 PR gcov-profile/80911
21898 * gcov.c (block_info::block_info): New constructor.
21900 2017-06-07 Carl Love <cel@us.ibm.com>
21902 * config/rs6000/rs6000-c: The return type of the following
21903 built-in functions was implemented as int not long long. Fix sign
21904 of return value for the unsigned version of vec_mulo and vec_mule.
21905 vector unsigned long long vec_bperm (vector unsigned long long,
21906 vector unsigned char)
21907 vector signed long long vec_mule (vector signed int,
21909 vector unsigned long long vec_mule (vector unsigned int,
21910 vector unsigned int)
21911 vector signed long long vec_mulo (vector signed int,
21913 vector unsigned long long vec_mulo (vector unsigned int,
21914 vector unsigned int)
21915 * doc/extend.texi: Fix the documentation for the built-in
21918 2017-06-07 Carl Love <cel@us.ibm.com>
21921 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21924 2017-06-07 Carl Love <cel@us.ibm.com>
21926 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21927 support, Generate doublehv for signed int/float for BE case only.
21929 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21931 * doc/invoke.texi (mcx16): Rewrite.
21933 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21935 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21936 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21937 *mov<mode>_softfloat, and an anonymous splitter): Use
21938 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21940 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21942 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21944 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21945 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21946 (rs6000_debug_reg_global): Adjust.
21947 (rs6000_init_hard_regno_mode_ok): Adjust.
21948 (rs6000_dbx_register_number): Adjust.
21949 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21950 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21951 Remove SPE_ACC and SPEFSCR.
21952 (REG_ALLOC_ORDER): Ditto.
21953 (FRAME_POINTER_REGNUM): Change to 111.
21954 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21955 (REG_CLASS_NAMES): Ditto.
21956 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21957 (REGISTER_NAMES): Ditto.
21958 (ADDITIONAL_REG_NAMES): Ditto.
21959 (rs6000_reg_names): Ditto.
21960 * config/rs6000/rs6000.md: Renumber some register number
21963 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21965 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21967 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21968 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21970 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21971 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21972 Delete the SPE high registers.
21973 (REG_ALLOC_ORDER): Ditto.
21974 (enum reg_class): Remove SPE_HIGH_REGS.
21975 (REG_CLASS_NAMES): Ditto.
21976 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21977 (REGISTER_NAMES): Ditto.
21978 (rs6000_reg_names): Ditto.
21979 * doc/tm.texi.in: Remove SPE as example.
21980 * doc/tm.texi: Regenerate.
21982 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21984 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21985 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21986 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21987 * config/rs6000/rs6000.md (type): Remove "brinc".
21989 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21991 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21992 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21993 * config/rs6000/linuxspe.h: Delete file.
21994 * config/rs6000/rs6000.md: Don't include spe.md.
21995 * config/rs6000/spe.h: Delete file.
21996 * config/rs6000/spe.md: Delete file.
21997 * config/rs6000/t-rs6000: Remove spe.md.
21999 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22001 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22002 (reg_or_none500mem_operand): Delete.
22003 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22004 instead of reg_or_none500mem_operand.
22006 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22008 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22009 handling of SPE flags.
22010 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22012 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22014 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22016 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22017 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22018 paired_divv2sf3): Similar.
22019 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22020 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22021 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22023 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22024 Rename the paired_* instruction patterns.
22025 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22027 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22028 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22029 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22030 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22031 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22032 PAIRED_VECTOR_MODE.
22033 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22034 (spe_func_has_64bit_regs_p): Delete.
22035 (spe_expand_predicate_builtin): Delete.
22036 (spe_expand_evsel_builtin): Delete.
22037 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22038 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22039 (invalid_e500_subreg): Delete.
22040 (rs6000_legitimize_address): Always force_reg op2 as well, for
22041 paired single memory accesses.
22042 (rs6000_member_type_forces_blk): Delete.
22043 (rs6000_spe_function_arg): Delete.
22044 (rs6000_expand_unop_builtin): Delete SPE handling.
22045 (rs6000_expand_binop_builtin): Ditto.
22046 (spe_expand_stv_builtin): Delete.
22047 (bdesc_2arg_spe): Delete.
22048 (spe_expand_builtin): Delete.
22049 (spe_expand_predicate_builtin): Delete.
22050 (spe_expand_evsel_builtin): Delete.
22051 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22052 (spe_init_builtins): Delete.
22053 (spe_func_has_64bit_regs_p): Delete.
22054 (savres_routine_name): Delete "info" parameter. Adjust callers.
22055 (rs6000_emit_stack_reset): Ditto.
22056 (rs6000_dwarf_register_span): Delete.
22057 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22058 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22059 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22061 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22063 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22064 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22065 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22066 mulv2sf3, divv2sf3): Delete expanders.
22068 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22070 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22072 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22074 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22075 * config/rs6000/rs6000.c: Ditto.
22077 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22079 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22080 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22081 comparison_operator.
22083 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22085 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22086 * config/rs6000/rs6000.opt: Ditto.
22087 * config/rs6000/t-rtems: Ditto.
22089 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22091 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22092 TARGET_E500_SINGLE by 0, simplify.
22093 * config/rs6000/rs6000.c: Ditto.
22094 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22095 (spe_build_register_parallel): Delete.
22096 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22097 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22098 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22099 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22100 (E500_CONVERT): Delete.
22101 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22103 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22105 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22106 * config/rs6000/dfp.md: Ditto.
22107 (negdd2, *negdd2_fpr): Merge.
22108 (absdd2, *absdd2_fpr): Merge.
22109 (negtd2, *negtd2_fpr): Merge.
22110 (abstd2, *abstd2_fpr): Merge.
22111 * config/rs6000/e500.h: Delete file.
22112 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22113 TARGET_FPRS by 1 and simplify.
22114 * config/rs6000/rs6000-c.c: Ditto.
22115 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22116 TARGET_DF_SPE by 0.
22117 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22119 * config/rs6000/rs6000.md: Ditto.
22120 (floatdidf2, *floatdidf2_fpr): Merge.
22121 (move_from_CR_gt_bit): Delete.
22122 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22123 (E500_CR_IOR_COMPARE): Delete.
22124 (All patterns that require !TARGET_FPRS): Delete.
22125 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22127 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22129 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22131 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22133 * graphds.c (add_edge): Intitialize edge's attached data.
22134 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22135 pointer parameter. Call pointed function on each edge during
22136 graph traversing. Skip traversing the edge when the function
22138 (graphds_dfs, graphds_scc): Ditto.
22139 (for_each_edge): New parameter. Pass the new parameter to callback
22141 * graphds.h (skip_edge_callback): New function pointer type.
22142 (graphds_dfs, graphds_scc): New function pointer parameter.
22143 (graphds_edge_callback, for_each_edge): New parameter.
22145 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22147 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22148 out code checking if runtime alias check is possible to below ...
22149 Call the new function.
22150 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22151 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22153 2017-06-07 Marek Polacek <polacek@redhat.com>
22156 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22157 TYPE_OVERFLOW_WRAPS check.
22159 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22161 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22162 if versioning is required.
22163 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22164 peeling with the check for versioning.
22166 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22168 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22169 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22170 Set true to new parameter if new ssa variable is defined.
22171 (vect_gen_vector_loop_niters): Refactor. Set range information
22172 for the new vector loop bound variable.
22173 (vect_do_peeling): Ditto.
22175 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22177 * tree-affine.c (ssa.h): Include header file.
22178 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22179 has wrapping overflow behavior.
22181 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22183 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22185 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22187 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22188 (tree_to_aff_combination): ... here.
22190 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22192 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22193 reg_pressure model function.
22194 (ivopts_global_cost_for_size): Delete.
22195 (determine_set_costs, iv_ca_recount_cost): Call new model function
22196 ivopts_estimate_reg_pressure.
22198 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22200 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22201 expensive than udiv. Remove floating point cases from mod.
22203 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22205 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22206 Increase idiv cost.
22208 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22210 * config/aarch64/aarch64.md
22211 (copysignsf3): Fix mask generation.
22213 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22215 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22217 (class dump_manager): Add register_dumps method.
22218 * dumpfile.c: Include langhooks.h.
22219 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22220 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22221 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22222 (dump_manager::dump_register): Start with 512 entries instead of 32.
22223 (dump_manager::register_dumps): New method.
22224 * toplev.c (general_init): Instead of invoking register_dumps
22225 langhook, invoke register_dumps method on the dump manager.
22226 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22229 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22231 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22232 Say that # requires an associated define_split to exist, and that
22233 the define_split must be suitable for use after register allocation.
22235 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22237 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22238 (compute_outgoing_frequencies): Also initialize zero counts.
22239 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22240 around loops; preserve more of profile when nothing changes.
22242 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22244 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22246 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22247 * config/arm/arm-cpu-cdata.h: Regenerate.
22248 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22249 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22250 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22251 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22252 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22254 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22256 * config/arm/t-rmprofile: Likewise.
22257 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22259 2017-06-06 David S. Miller <davem@davemloft.net>
22262 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22263 blockage if function uses alloca.
22265 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22267 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22268 New "uid" fields to hold pretty-print IDs of group and ref.
22269 Memory references are now identified as <group_id>:<ref_id>
22270 instead of using [random] addresses.
22271 (dump_mem_details): Simplify, no functional change.
22272 (dump_mem_ref): Simplify and make output more concise.
22273 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22274 (find_or_create_group): Initialize group uid.
22275 (record_ref): Initialize ref uid. Improve debug output.
22276 (prune_group_by_reuse, should_issue_prefetch_p,)
22277 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22278 (mark_nontemporal_store, determine_loop_nest_reuse):
22279 Improve debug output.
22281 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22283 * dbgcnt.def (prefetch): New debug counter.
22284 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22285 (schedule_prefetches): Stop issueing prefetches if debug counter
22288 2017-06-06 Tom de Vries <tom@codesourcery.com>
22290 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22291 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22293 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22295 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22296 Use aarch64_reg_or_zero predicate for operand 4.
22297 (aarch64_compare_and_swap<mode> define_insn_and_split):
22298 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22299 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22301 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22303 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22304 (arm_compute_save_core_reg_mask): This.
22305 (thumb1_compute_save_reg_mask): Rename into ...
22306 (thumb1_compute_save_core_reg_mask): This.
22307 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22308 (arm_compute_frame_layout): Likewise.
22310 2017-06-06 Richard Biener <rguenther@suse.de>
22312 PR tree-optimization/80974
22313 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22314 keep or clear leaders SSA info.
22316 2017-06-06 Tom de Vries <tom@codesourcery.com>
22318 * config/nvptx/nvptx.c (split_mode_p): New function.
22319 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22321 2017-06-06 Tom de Vries <tom@codesourcery.com>
22323 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22325 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22328 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22331 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22333 * shrink-wrap.c (handle_simple_exit): Update profile.
22334 (try_shrink_wrapping): Upate profile.
22336 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22338 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22339 (tree_guess_outgoing_edge_probabilities): New.
22340 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22341 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22343 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22345 * ipa-split.c (split_function): Initialize return bb profile.
22347 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22349 * profile.c (compute_branch_probabilities): Also initialize
22350 EXIT_BLOCK profile.
22352 2017-06-06 Richard Biener <rguenther@suse.de>
22354 PR tree-optimization/80928
22355 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22356 (vect_analyze_loop_operations): Properly guard analysis for
22358 (vect_transform_loop): Likewise.
22359 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22360 (vect_model_induction_cost): Do not cost for pure SLP.
22361 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22362 of induction in inner loop vectorization.
22363 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22364 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22365 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22367 (vect_analyze_slp_cost_1): Cost induction.
22368 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22369 (vect_get_slp_vect_defs): Likewise.
22370 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22371 (vect_transform_stmt): Handle SLP reductions.
22372 * tree-vectorizer.h (vectorizable_induction): Adjust.
22374 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22376 * config/rs6000/rs6000.c (make_resolver_func): Update
22377 init_lowered_empty_function call.
22379 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22381 * doc/invoke.texi: Document the -fprofile-abs-path option.
22382 * common.opt (fprofile-abs-path): New option.
22383 * gcov-io.h (gcov_write_filename): Declare.
22384 * gcov-io.c (gcov_write_filename): New function.
22385 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22386 * profile.c (output_location): Likewise.
22388 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22390 * shring-wrap.c: Revert accidental commit.
22392 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22394 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22396 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22398 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22400 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22401 profile in callgraph edge.
22402 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22403 (apply_scale): Likewise.
22404 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22405 Also scale profile when inlining function with zero profile.
22406 (initialize_cfun): Update exit block profile even when it is zero.
22407 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22408 when profile is read.
22410 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22412 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22413 (CLONE_*): New constants to define the processors we can generate
22414 code for with the target_clone attribute.
22415 (rs6000_clone_map): New array to identify which clone processors
22416 the current program is running on.
22417 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22418 target_clone attribute.
22419 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22420 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22421 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22422 (cpu_expand_builtin): Add support for target_clone attribute.
22423 (rs6000_valid_attribute_p): Allow "default" attribute.
22424 (get_decl_name): New debug function to simplify printing the
22425 current function name in debugging statements.
22426 (rs6000_clone_priority): New functions to support the target_clone
22427 attribute, and be able to generate code to switch between ISA 2.05
22428 through ISA 3.0 (power6 through power9).
22429 (rs6000_compare_version_priority): Likewise.
22430 (rs6000_get_function_versions_dispatcher): Likewise.
22431 (make_resolver_func): Likewise.
22432 (add_condition_to_bb): Likewise.
22433 (dispatch_function_versions): Likewise.
22434 (rs6000_generate_version_dispatcher_body): Likewise.
22435 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22436 (fusion_gpr_load_p): Fix a spacing issue.
22437 * doc/extend.texi (Common Function Attributes): Document that the
22438 PowerPC supports the target_clone attribute.
22440 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22442 * config/arm/arm.h: explain F symbol found in description of ARM
22443 register allocation in its legend.
22445 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22447 * config/mips/frame-header-opt.c: Include profile-count.h.
22448 * config/riscv/riscv.c: Include profile-count.h
22450 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22452 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22454 (sm_set_flag_if_changed): Add bbs field.
22455 (execute_sm_if_changed_flag_set): Pass BBS.
22456 (execute_sm): Update.
22458 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22460 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22463 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22465 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22466 (peephole2): New peephole2 to emit the above.
22467 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22469 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22471 * config/aarch64/aarch64.c (define_peephole2 above
22472 *sub_<shift>_<mode>): New peephole.
22474 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22476 * config/i386/i386.c (make_resolver_func): Update.
22477 * Makefile.in: Add profile-count.h and profile-count.o
22478 * auto-profile.c (afdo_indirect_call): Update to new API.
22479 (afdo_set_bb_count): Update.
22480 (afdo_propagate_edge): Update.
22481 (afdo_propagate_circuit): Update.
22482 (afdo_calculate_branch_prob): Update.
22483 (afdo_annotate_cfg): Update.
22484 * basic-block.h: Include profile-count.h
22485 (struct edge_def): Turn count to profile_count.
22486 (struct basic_block_def): Likewie.
22487 (REG_BR_PROB_BASE): Move to profile-count.h
22488 (RDIV): Move to profile-count.h
22489 * bb-reorder.c (max_entry_count): Turn to profile_count.
22490 (find_traces): Update.
22491 (rotate_loop):Update.
22492 (connect_traces):Update.
22493 (sanitize_hot_paths):Update.
22494 * bt-load.c (migrate_btr_defs): Update.
22495 * cfg.c (RDIV): Remove.
22496 (init_flow): Use alloc_block.
22497 (alloc_block): Uninitialize count.
22498 (unchecked_make_edge): Uninitialize count.
22499 (check_bb_profile): Update.
22500 (dump_edge_info): Update.
22501 (dump_bb_info): Update.
22502 (update_bb_profile_for_threading): Update.
22503 (scale_bbs_frequencies_int): Update.
22504 (scale_bbs_frequencies_gcov_type): Update.
22505 (scale_bbs_frequencies_profile_count): New.
22506 * cfg.h (update_bb_profile_for_threading): Update.
22507 (scale_bbs_frequencies_profile_count): Declare.
22508 * cfgbuild.c (compute_outgoing_frequencies): Update.
22509 (find_many_sub_basic_blocks): Update.
22510 * cfgcleanup.c (try_forward_edges): Update.
22511 (try_crossjump_to_edge): Update.
22512 * cfgexpand.c (expand_gimple_tailcall): Update.
22513 (construct_exit_block): Update.
22514 * cfghooks.c (verify_flow_info): Update.
22515 (dump_bb_for_graph): Update.
22516 (split_edge): Update.
22517 (make_forwarder_block): Update.
22518 (duplicate_block): Update.
22519 (account_profile_record): Update.
22520 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22521 (get_estimated_loop_iterations): Update.
22522 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22523 (single_likely_exit): Update.
22524 * cfgloopmanip.c (scale_loop_profile): Update.
22526 (set_zero_probability): Update.
22527 (lv_adjust_loop_entry_edge): Update.
22528 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22529 (purge_dead_edges): Update.
22530 (rtl_account_profile_record): Update.
22531 * cgraph.c (cgraph_node::create): Uninitialize count.
22532 (symbol_table::create_edge): Uninitialize count.
22533 (cgraph_update_edges_for_call_stmt_node): Update.
22534 (cgraph_edge::dump_edge_flags): Update.
22535 (cgraph_node::dump): Update.
22536 (cgraph_edge::maybe_hot_p): Update.
22537 * cgraph.h: Include profile-count.h
22538 (create_clone), create_edge, create_indirect_edge): Update.
22539 (cgraph_node): Turn count to profile_count.
22540 (cgraph_edge0: Likewise.
22541 (make_speculative, clone): Update.
22542 (create_edge): Update.
22543 (init_lowered_empty_function): Update.
22544 * cgraphclones.c (cgraph_edge::clone): Update.
22545 (duplicate_thunk_for_node): Update.
22546 (cgraph_node::create_clone): Update.
22547 * cgraphunit.c (cgraph_node::analyze): Update.
22548 (cgraph_node::expand_thunk): Update.
22549 * final.c (dump_basic_block_info): Update.
22550 * gimple-streamer-in.c (input_bb): Update.
22551 * gimple-streamer-out.c (output_bb): Update.
22552 * graphite.c (print_global_statistics): Update.
22553 (print_graphite_scop_statistics): Update.
22554 * hsa-brig.c: Include basic-block.h.
22555 * hsa-dump.c: Include basic-block.h.
22556 * hsa-gen.c (T sum_slice): Update.
22557 (convert_switch_statements):Update.
22558 * hsa-regalloc.c: Include basic-block.h.
22559 * ipa-chkp.c (chkp_produce_thunks): Update.
22560 * ipa-cp.c (struct caller_statistics): Update.
22561 (init_caller_stats): Update.
22562 (gather_caller_stats): Update.
22563 (ipcp_cloning_candidate_p): Update.
22564 (good_cloning_opportunity_p): Update.
22565 (get_info_about_necessary_edges): Update.
22566 (dump_profile_updates): Update.
22567 (update_profiling_info): Update.
22568 (update_specialized_profile): Update.
22569 (perhaps_add_new_callers): Update.
22570 (decide_about_value): Update.
22571 (ipa_cp_c_finalize): Update.
22572 * ipa-devirt.c (struct odr_type_warn_count): Update.
22573 (struct decl_warn_count): Update.
22574 (struct final_warning_record): Update.
22575 (possible_polymorphic_call_targets): Update.
22576 (ipa_devirt): Update.
22577 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22578 * ipa-icf.c (sem_function::merge): Update.
22579 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22580 * ipa-inline.c (compute_uninlined_call_time): Update.
22581 (compute_inlined_call_time): Update.
22582 (want_inline_small_function_p): Update.
22583 (want_inline_self_recursive_call_p): Update.
22584 (edge_badness): Update.
22585 (lookup_recursive_calls): Update.
22586 (recursive_inlining): Update.
22587 (inline_small_functions): Update.
22588 (dump_overall_stats): Update.
22589 (dump_inline_stats): Update.
22590 * ipa-profile.c (ipa_profile_generate_summary): Update.
22591 (ipa_propagate_frequency): Update.
22592 (ipa_profile): Update.
22593 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22594 * ipa-utils.c (ipa_merge_profiles): Update.
22595 * loop-doloop.c (doloop_modify): Update.
22596 * loop-unroll.c (report_unroll): Update.
22597 (unroll_loop_runtime_iterations): Update.
22598 * lto-cgraph.c (lto_output_edge): Update.
22599 (lto_output_node): Update.
22600 (input_node): Update.
22601 (input_edge): Update.
22602 (merge_profile_summaries): Update.
22603 * lto-streamer-in.c (input_cfg): Update.
22604 * lto-streamer-out.c (output_cfg): Update.
22605 * mcf.c (create_fixup_graph): Update.
22606 (adjust_cfg_counts): Update.
22607 (sum_edge_counts): Update.
22608 * modulo-sched.c (sms_schedule): Update.
22609 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22610 * predict.c (maybe_hot_count_p): Update.
22611 (probably_never_executed): Update.
22612 (dump_prediction): Update.
22613 (combine_predictions_for_bb): Update.
22614 (propagate_freq): Update.
22615 (handle_missing_profiles): Update.
22616 (counts_to_freqs): Update.
22617 (rebuild_frequencies): Update.
22618 (force_edge_cold): Update.
22619 * predict.h: Include profile-count.h
22620 (maybe_hot_count_p, counts_to_freqs): UPdate.
22621 * print-rtl-function.c: Do not include cfg.h
22622 * print-rtl.c: Include basic-block.h
22623 * profile-count.c: New file.
22624 * profile-count.h: New file.
22625 * profile.c (is_edge_inconsistent): Update.
22626 (correct_negative_edge_counts): Update.
22627 (is_inconsistent): Update.
22628 (set_bb_counts): Update.
22629 (read_profile_edge_counts): Update.
22630 (compute_frequency_overlap): Update.
22631 (compute_branch_probabilities): Update; Initialize and deinitialize
22633 (branch_prob): Update.
22634 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22635 (edge_gcov_count): New.
22636 (bb_gcov_count): New.
22637 * shrink-wrap.c (try_shrink_wrapping): Update.
22638 * tracer.c (better_p): Update.
22639 * trans-mem.c (expand_transaction): Update.
22640 (ipa_tm_insert_irr_call): Update.
22641 (ipa_tm_insert_gettmclone_call): Update.
22642 * tree-call-cdce.c: Update.
22643 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22644 (gimple_duplicate_sese_tail): Update.
22645 (gimple_account_profile_record): Update.
22646 (execute_fixup_cfg): Update.
22647 * tree-inline.c (copy_bb): Update.
22648 (copy_edges_for_bb): Update.
22649 (initialize_cfun): Update.
22650 (freqs_to_counts): Update.
22651 (copy_cfg_body): Update.
22652 (expand_call_inline): Update.
22653 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22654 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22655 (try_unroll_loop_completely): Update.
22656 (try_peel_loop): Update.
22657 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22658 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22659 * tree-ssa-loop-split.c (connect_loops): Update.
22660 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22661 * tree-ssa-reassoc.c (branch_fixup): Update.
22662 * tree-ssa-tail-merge.c (replace_block_by): Update.
22663 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22664 (compute_path_counts): Update.
22665 (update_profile): Update.
22666 (recompute_probabilities): Update.
22667 (update_joiner_offpath_counts): Update.
22668 (estimated_freqs_path): Update.
22669 (freqs_to_counts_path): Update.
22670 (clear_counts_path): Update.
22671 (ssa_fix_duplicate_block_edges): Update.
22672 (duplicate_thread_path): Update.
22673 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22674 (struct switch_conv_info): Update.
22675 * tree-tailcall.c (decrease_profile): Update.
22676 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22677 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22678 * value-prof.c (check_counter): Update.
22679 (gimple_divmod_fixed_value): Update.
22680 (gimple_mod_pow2): Update.
22681 (gimple_mod_subtract): Update.
22682 (gimple_ic_transform): Update.
22683 (gimple_stringop_fixed_value): Update.
22684 * value-prof.h (gimple_ic): Update.
22686 2017-06-02 Carl Love <cel@us.ibm.com>
22688 * config/rs6000/rs6000-c: Add support for built-in functions
22689 vector double vec_doublee (vector signed int);
22690 vector double vec_doublee (vector unsigned int);
22691 vector double vec_doublee (vector float);
22692 vector double vec_doubleh (vector signed int);
22693 vector double vec_doubleh (vector unsigned int);
22694 vector double vec_doubleh (vector float);
22695 vector double vec_doublel (vector signed int);
22696 vector double vec_doublel (vector unsigned int);
22697 vector double vec_doublel (vector float);
22698 vector double vec_doubleo (vector signed int);
22699 vector double vec_doubleo (vector unsigned int);
22700 vector double vec_doubleo (vector float);.
22701 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22702 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22704 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22705 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22706 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22708 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22709 vec_doublel, vec_doubleh.
22710 * doc/extend.texi: Update the built-in documentation file for the
22711 new built-in functions.
22713 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22716 * ipa-inline-analysis.c (free_growth_caches): Set
22717 edge_removal_hook_holder to NULL after removing it.
22719 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22721 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22722 comparision with zero.
22724 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22725 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22726 for early expansion of vec_min and vec_max builtins.
22727 (builtin_function_type): Add min/max unsigned variants to those
22728 identified as having unsigned arguments.
22730 2017-06-02 Olivier Hainque <hainque@adacore.com>
22732 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22734 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22736 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22737 Use VALL_F16 iterator rather than VALL.
22739 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22741 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22742 Emit CBNZ inside loop when doing a strong exchange and comparing
22743 against zero. Generate the CC flags after the loop.
22745 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22747 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22748 (dl_section_ref): New.
22749 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22750 On AIX, append an expression to subtract the size of the
22751 section length to dl_section_ref.
22753 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22755 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22756 for early expansion of vector absolute builtins.
22758 2017-06-02 Richard Biener <rguenther@suse.de>
22760 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22761 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22763 2017-06-02 Richard Biener <rguenther@suse.de>
22765 PR tree-optimization/80948
22766 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22767 stmt order as well.
22769 2017-06-02 Richard Biener <rguenther@suse.de>
22771 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22773 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22774 for inductions as relevant.
22776 2017-06-02 Richard Biener <rguenther@suse.de>
22778 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22779 (vectorizable_induction): ... this. Remove dead code.
22781 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22783 * builtins. (expand_builtin_alloca): Remove second parameter and
22784 infer its value from the first parameter instead.
22785 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22787 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22789 PR rtl-optimization/80903
22790 * loop-doloop.c (add_test): Unshare sequence.
22792 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22794 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22796 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22798 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22800 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22801 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22802 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22803 (xlogue_layout::get_stub_rtx): Make static.
22804 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22805 (xlogue_layout::compute_stub_managed_regs): Rename to...
22806 (xlogue_layout::count_stub_managed_regs): ...this.
22807 (xlogue_layout::is_stub_managed_reg): New function.
22808 (xlogue_layout::m_stub_names): Rename to...
22809 (xlogue_layout::s_stub_names): ...this, make static.
22810 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22811 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22812 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22813 xlogue_layout::s_stub_names): Instantiate statics.
22814 (stub_managed_regs): Remove.
22815 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22816 (disable_call_ms2sysv_xlogues): Rename to...
22817 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22818 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22820 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22821 change after reload_completed.
22822 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22824 (ix86_expand_prologue): Likewise.
22825 (ix86_expand_epilogue): Likewise.
22826 (ix86_expand_split_stack_prologue): Likewise.
22827 (ix86_compute_frame_layout): Remove frame parameter ...
22828 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22829 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22831 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22832 (ix86_frame): Move from here ...
22833 * config/i386/i386.h (ix86_frame): ... to here.
22834 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22835 complete ix86_frame data structure instead. Remove some_ld_name.
22837 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22839 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22840 symbols that hold a DECL_VALUE_EXPR.
22842 2017-06-01 Martin Jambor <mjambor@suse.cz>
22844 PR tree-optimization/80898
22845 * tree-sra.c (process_subtree_disqualification): Removed.
22846 (disqualify_candidate): Do not acll
22847 process_subtree_disqualification.
22848 (subtree_mark_written_and_enqueue): New function.
22849 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22850 RHS has been disqualified and re-queue LHS if necessary. Apart
22851 from that, ignore disqualified RHS.
22853 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22855 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22856 address fetch for z10 or later.
22858 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22860 * config/arc/arc.md (tst_movb): Add guard when splitting.
22862 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22864 * config/arc/arc.c (arc_can_eliminate): Test against
22865 arc_frame_pointer_needed.
22867 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22869 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22870 to prevent store reordering.
22871 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22872 (type): Add block type.
22873 (stack_tie): Define special instruction to be used in
22876 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22878 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22879 constraint. It is not valid for the pattern.
22880 (noncommutative_binary_comparison): Likewise.
22882 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22884 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22887 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22889 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22890 be used by the reg-alloc.
22892 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22894 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22895 reg-alloc when having mul64 or mul32x16 instructions.
22896 (mulsidi3): Likewise.
22897 (umulsidi3): Likewise.
22898 (mulsi32x16): New pattern.
22899 (mulsi64): Likewise.
22900 (mulsidi64): Likewise.
22901 (umulsidi64): Likewise.
22902 (MUL32x16_REG): Define.
22903 (mul64_600): Use MUL32x16_REG.
22904 (mac64_600): Likewise.
22905 (umul64_600): Likewise.
22906 (umac64_600): Likewise.
22908 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22910 * config/arc/arc.md (mulsi3_700): Make it commutative.
22912 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22914 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22916 (*sign_extendsidi2_insn): Likewise for movstosw.
22918 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22920 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22921 the type of the input discriminant value. Convert the
22922 discriminant value of signedness vary.
22924 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22926 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22927 Add to -Wall section.
22929 2017-06-01 Richard Biener <rguenther@suse.de>
22931 PR middle-end/66313
22932 * fold-const.c (fold_plusminus_mult_expr): If the factored
22933 factor may be zero use a wrapping type for the inner operation.
22934 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22935 and handle moved defs.
22936 (process_assignment): Properly guard the unary op case. Return a
22937 tri-state indicating that moving the stmt before the call may allow
22938 to continue. Pass through to_move.
22939 (find_tail_calls): Handle moving unrelated defs before
22942 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22945 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22946 splitter result in the canonical way.
22948 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22950 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22951 also for 32bit target. Update insn attributes.
22952 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22954 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22956 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22957 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22958 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22959 (_mm_maskz_min_ss): New intrinsics.
22961 2017-05-31 Martin Liska <mliska@suse.cz>
22963 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22964 Change comment style to one we normally use.
22965 (vectorizable_reduction): Likewise.
22966 (vectorizable_induction): Likewise.
22967 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22968 (vectorizable_call): Likewise.
22969 (vectorizable_simd_clone_call): Likewise.
22970 (vectorizable_conversion): Likewise.
22971 (vectorizable_assignment): Likewise.
22972 (vectorizable_shift): Likewise.
22973 (vectorizable_operation): Likewise.
22974 (vectorizable_store): Likewise.
22975 (vectorizable_load): Likewise.
22976 * tree-vectorizer.h: Likewise.
22978 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22980 * passes.c (emergency_dump_function): New.
22981 * tree-pass.h (emergency_dump_function): Declare.
22982 * plugin.c (plugins_internal_error_function): Remove.
22983 * plugin.h (plugins_internal_error_function): Remove declaration.
22984 * toplev.c (internal_error_function): New static function. Use it...
22985 (general_init): ...here.
22987 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22989 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22990 (arc_rtx_costs): Add costs for new patterns.
22991 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22992 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22994 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22996 * tree-ssa-strlen.c (get_next_strinfo): New function.
22997 (get_stridx_plus_constant): Use it.
22998 (zero_length_string): Likewise.
22999 (adjust_related_strinfos): Likewise.
23000 (adjust_last_stmt): Likewise.
23002 2017-05-31 Richard Biener <rguenther@suse.de>
23005 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23006 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23008 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23010 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23011 loop_vinfo argument and use of dependence distance vectors.
23012 Check instead whether the two references differ only in their
23013 initial value and assume that they have the same alignment if the
23014 difference is a multiple of the vector alignment.
23015 (vect_analyze_data_refs_alignment): Update call accordingly.
23017 2017-05-31 Martin Liska <mliska@suse.cz>
23020 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23022 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23024 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23025 (create_intersect_range_checks): Move from ...
23026 * tree-data-ref.c (create_intersect_range_checks_index)
23027 (create_intersect_range_checks): ... to here.
23028 (create_runtime_alias_checks): New function factored from ...
23029 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23030 here. Call above function.
23031 * tree-data-ref.h (create_runtime_alias_checks): New function.
23033 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23035 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23036 segment length for dr_b and compute it in wide_int.
23038 2017-05-31 Richard Biener <rguenther@suse.de>
23040 PR tree-optimization/80906
23041 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23042 and pass through iv_map.
23043 (copy_bb_and_scalar_dependences): Adjust.
23044 (translate_pending_phi_nodes): Likewise.
23045 (copy_loop_close_phi_args): Handle code-generating IVs instead
23048 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23050 * diagnostic-color.c (color_dict): Add "type-diff".
23051 (parse_gcc_colors): Update comment.
23052 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23053 -fdiagnostics-show-template-tree and -fno-elide-type.
23054 (GCC_COLORS): Add type-diff to example.
23056 (-fdiagnostics-show-template-tree): New.
23057 (-fno-elide-type): New.
23058 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23059 the pp_format_decoder callback. Call any m_format_postprocessor's
23061 (pretty_printer::pretty_printer): Initialize
23062 m_format_postprocessor.
23063 (pretty_printer::~pretty_printer): Delete any
23064 m_format_postprocessor.
23065 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23066 (class format_postprocessor): New class.
23067 (struct pretty_printer::format_decoder): Document the new parameters.
23068 (struct pretty_printer::m_format_postprocessor): New field.
23069 * tree-diagnostic.c (default_tree_printer): Update for new
23070 bool and const char ** params.
23071 * tree-diagnostic.h (default_tree_printer): Likewise.
23073 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23075 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23076 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23077 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23078 rs6000_gen_cell_microcode code.
23079 (rs6000_final_prescan_insn): Delete.
23080 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23081 "warn-cell-microcode" entries.
23082 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23083 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23084 throughout. Change cc_reg_not_micro_cr0_operand to
23085 cc_reg_not_cr0_operand throughout.
23086 (*extendhi<mode>2_noload): Delete.
23087 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23088 (mwarn-cell-microcode): Delete.
23089 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23090 -mgen-cell-microcode and -mwarn-cell-microcode.
23092 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23095 * config/i386/constraints.md (Yd): New constraint.
23097 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23098 and (?Yd, r) alternatives. Update insn attributes.
23099 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23100 and (?*Yd, r) alternatives. Update insn attributes.
23101 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23103 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23105 * gimplify.c (gimplify_modify_expr): Don't create a
23106 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23109 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23111 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23113 2017-05-30 Richard Biener <rguenther@suse.de>
23115 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23116 and reduc_def fields.
23117 (STMT_VINFO_REDUC_TYPE): New define.
23118 (STMT_VINFO_REDUC_DEF): Likewise.
23119 (vect_force_simple_reduction): Adjust prototype.
23120 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23121 (vect_is_simple_reduction): Remove check_reduction argument.
23122 (vect_force_simple_reduction): Adjust and set
23123 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23124 (vectorizable_reduction): Do not re-do reduction analysis
23125 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23126 * tree-parloops.c (gather_scalar_reductions): Adjust.
23128 2017-05-30 Richard Biener <rguenther@suse.de>
23130 PR middle-end/80901
23131 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23134 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23136 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23137 Introduce unknown_misalignment parameter and remove vf.
23138 (vect_peeling_hash_get_lowest_cost):
23139 Pass unknown_misalignment parameter.
23140 (vect_enhance_data_refs_alignment):
23141 Fix unsupportable data ref treatment.
23143 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23145 * tree-vect-data-refs.c (vect_get_data_access_cost):
23146 Workaround for SLP handling.
23147 (vect_enhance_data_refs_alignment):
23148 Compute costs for doing no peeling at all, compare to the best
23149 peeling costs so far and avoid peeling if cheaper.
23151 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23153 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23154 Return peeling info and set costs to zero for unlimited cost
23156 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23157 with unknown misalignment. Compute and costs for unknown
23158 misalignment, compare them to the costs for known misalignment
23159 and choose the cheapest for peeling.
23161 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23163 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23164 (vect_get_peeling_costs_all_drs): Create function.
23165 (vect_peeling_hash_get_lowest_cost):
23166 Use vect_get_peeling_costs_all_drs.
23167 (vect_peeling_supportable): Create function.
23168 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23170 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23172 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23173 DR_HAS_NEGATIVE_STEP.
23174 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23175 (vect_enhance_data_refs_alignment): Use.
23176 (vect_duplicate_ssa_name_ptr_info): Use.
23177 * tree-vectorizer.h (dr_misalignment): Use.
23178 (known_alignment_for_access_p): Use.
23180 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23183 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23185 (has_section_name): New function.
23187 2017-05-30 Martin Liska <mliska@suse.cz>
23190 * auto-profile.c (get_function_decl_from_block): Fix
23193 2017-05-30 Richard Biener <rguenther@suse.de>
23195 PR middle-end/80876
23196 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23198 2017-05-30 Martin Liska <mliska@suse.cz>
23200 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23201 * dumpfile.h (struct dump_file_info): Remove ctors.
23203 2017-05-30 Martin Liska <mliska@suse.cz>
23205 * predict.def: Fix GNU coding style.
23207 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23209 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23210 Mark 'to' argument with ATTRIBUTE_UNUSED.
23212 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23214 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23215 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23216 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23219 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23221 * doc/install.texi (Options specification): Restore entry of
23222 --enable-sjlj-exceptions.
23224 2017-05-27 Michael Eager <eager@eagercon.com>
23227 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23229 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23231 * config/microblaze/microblaze.h
23232 (FIXED_REGISTERS): Update in macro.
23233 (CALL_USED_REGISTERS): Update in macro.
23235 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23237 * doc/install.texi: Add links to macOS binary distributions.
23239 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23243 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23245 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23247 2017-05-26 Martin Liska <mliska@suse.cz>
23249 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23251 2017-05-26 Martin Liska <mliska@suse.cz>
23253 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23254 always leading ';; '.
23255 (dump_bb_info): Likewise.
23256 (brief_dump_cfg): Likewise.
23257 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23258 * dumpfile.c: Remove usage of TDF_VERBOSE.
23259 * dumpfile.h (enum dump_kind): Likewise.
23260 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23261 * print-tree.c (debug_verbose): Remove.
23262 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23263 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23264 * tree-diagnostic.c (default_tree_printer): Replace
23265 TDF_DIAGNOSTIC with TDF_SLIM.
23267 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23269 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23270 in parameter loop, rather than loop_vinfo.
23271 (create_intersect_range_checks): Ditto.
23272 (vect_create_cond_for_alias_checks): Update call to above functions.
23274 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23276 PR tree-optimization/80815
23277 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23278 for merging runtime alias checks. Handle negative DR_STEPs.
23280 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23282 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23284 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23285 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23286 out code pruning runtime alias checks.
23287 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23288 factored out from above.
23289 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23291 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23293 (prune_runtime_alias_test_list): New decalaration.
23295 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23297 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23298 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23299 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23300 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23301 (operator==, comp_dr_with_seg_len_pair): Ditto.
23302 (vect_prune_runtime_alias_test_list): Ditto.
23304 2017-05-26 Martin Liska <mliska@suse.cz>
23307 * params.def: Bound partial-inlining-entry-probability param.
23309 2017-05-26 Marek Polacek <polacek@redhat.com>
23312 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23315 2017-05-26 Richard Biener <rguenther@suse.de>
23317 PR tree-optimization/80842
23318 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23321 2017-05-26 Richard Biener <rguenther@suse.de>
23323 PR tree-optimization/80844
23324 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23326 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23328 * doc/md.texi (Machine Constraints): Update x86 family
23329 machine constraints section to match 'config/i386/constraints.md'.
23331 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23333 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23335 2017-05-25 Nathan Sidwell <nathan@acm.org>
23337 * doc/invoke.texi (--enable-languages): Update documentation.
23339 2017-05-25 Martin Liska <mliska@suse.cz>
23341 * dumpfile.c: Add TDF_FOLDING.
23342 * dumpfile.h (enum dump_kind): Likewise.
23343 * genmatch.c (dt_simplify::gen_1): Use it.
23345 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23347 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23349 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23351 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23352 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23354 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23356 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23357 * match.pd (X == C): Rewrite it here.
23358 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23359 with_certain_nonzero_bits2): New predicates.
23360 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23362 2017-05-24 Nathan Sidwell <nathan@acm.org>
23364 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23367 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23370 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23372 * config/powerpcspe: New port. Files are copied from the rs6000
23373 port, with "rs6000" in filenames replaced by "powerpcspe".
23375 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23377 PR rtl-optimization/80754
23378 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23380 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23382 * config/sparc/sparc.md (length): Return the correct value for -mflat
23383 sibcalls to match output_sibcall.
23385 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23389 * config/rs6000/rs6000.c (struct machine_function): Add new field
23391 (rs6000_get_separate_components): Init that field, use it.
23392 (rs6000_components_for_bb): Use the field.
23394 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23396 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23398 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23400 PR middle-end/80823
23401 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23403 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23406 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23407 address against address_operand predicate.
23408 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23410 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23412 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23413 parameters passed indirectly.
23415 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23417 * config/i386/i386.md (*movdi_internal): Remove SSE4
23418 alternative 18 (?r, *v). Update insn attributes.
23419 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23420 Update insn attributes.
23421 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23422 Update insn attributes.
23423 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23424 alternative 1 (r, v). Remove isa attribute.
23425 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23426 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23427 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23429 2017-05-23 Tom de Vries <tom@codesourcery.com>
23431 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23434 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23436 * cgraphunit.c (symbol_table::process_new_functions): Update.
23437 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23438 (inline_generate_summary): Rename to ...
23439 (ipa_fn_summary_generate): ... this one.
23440 (inline_read_summary): Rename to ...
23441 (ipa_fn_summary_read): ... this one.
23442 (inline_write_summary): Rename to ...
23443 (ipa_fn_summary_write): ... this one.
23444 (inline_free_summary): Rename to ...
23445 (ipa_free_fn_summary): ... this one.
23446 (pass_data_local_fn_summary, pass_local_fn_summary,
23447 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23448 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23449 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23450 make_pass_ipa_fn_summary): New.
23451 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23452 inline_write_summary, inline_free_summary): Remove.
23453 (ipa_free_fn_summary) : New.
23454 * ipa-inline.c (ipa_inline): Update.
23455 (pass_ipa_inline): Do not generate summaries.
23456 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23458 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23459 and add pass_ipa_fn_summary.
23460 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23462 (make_pass_inline_parameters): Remove.
23464 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23466 * omp-low.c (struct omp_context): Remove "default_kind" member.
23469 * omp-offload.c (execute_oacc_device_lower): Remove the
23470 parallelism dimensions function attributes for unparallelized
23471 OpenACC kernels constructs.
23473 2017-05-23 Martin Liska <mliska@suse.cz>
23475 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23477 (cgraph_edge::make_speculative): Likewise.
23478 (cgraph_edge::resolve_speculation): Likewise.
23479 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23480 (cgraph_node::dump): Likewise.
23481 * cgraph.h: Likewise.
23482 * cgraphunit.c (analyze_functions): Likewise.
23483 (symbol_table::compile): Likewise.
23484 * ipa-cp.c (print_all_lattices): Likewise.
23485 (determine_versionability): Likewise.
23486 (initialize_node_lattices): Likewise.
23487 (ipcp_verify_propagated_values): Likewise.
23488 (estimate_local_effects): Likewise.
23489 (update_profiling_info): Likewise.
23490 (create_specialized_node): Likewise.
23491 (perhaps_add_new_callers): Likewise.
23492 (decide_about_value): Likewise.
23493 (decide_whether_version_node): Likewise.
23494 (identify_dead_nodes): Likewise.
23495 (ipcp_store_bits_results): Likewise.
23496 * ipa-devirt.c (dump_targets): Likewise.
23497 (ipa_devirt): Likewise.
23498 * ipa-icf.c (sem_item::dump): Likewise.
23499 (sem_function::equals): Likewise.
23500 (sem_variable::equals): Likewise.
23501 (sem_item_optimizer::read_section): Likewise.
23502 (sem_item_optimizer::execute): Likewise.
23503 (congruence_class::dump): Likewise.
23504 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23505 (dump_inline_summary): Likewise.
23506 (estimate_node_size_and_time): Likewise.
23507 (inline_analyze_function): Likewise.
23508 * ipa-inline-transform.c (inline_call): Likewise.
23509 * ipa-inline.c (report_inline_failed_reason): Likewise.
23510 (want_early_inline_function_p): Likewise.
23511 (edge_badness): Likewise.
23512 (update_edge_key): Likewise.
23513 (inline_small_functions): Likewise.
23514 * ipa-profile.c (ipa_profile): Likewise.
23515 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23516 (ipa_make_edge_direct_to_target): Likewise.
23517 (remove_described_reference): Likewise.
23518 (ipa_impossible_devirt_target): Likewise.
23519 (propagate_controlled_uses): Likewise.
23520 (ipa_print_node_params): Likewise.
23521 (ipcp_transform_function): Likewise.
23522 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23523 (propagate_pure_const): Likewise.
23524 * ipa-reference.c (generate_summary): Likewise.
23525 (read_write_all_from_decl): Likewise.
23526 (propagate): Likewise.
23527 (ipa_reference_read_optimization_summary): Likewise.
23528 * ipa-utils.c (ipa_merge_profiles): Likewise.
23529 * ipa.c (walk_polymorphic_call_targets): Likewise.
23530 (symbol_table::remove_unreachable_nodes): Likewise.
23531 (ipa_single_use): Likewise.
23532 * passes.c (execute_todo): Likewise.
23533 * predict.c (drop_profile): Likewise.
23534 * symtab.c (symtab_node::get_dump_name): New function.
23535 (symtab_node::dump_name): Likewise.
23536 (symtab_node::dump_asm_name): Likewise.
23537 (symtab_node::dump_references): Likewise.
23538 (symtab_node::dump_referring): Likewise.
23539 (symtab_node::dump_base): Likewise.
23540 (symtab_node::debug_symtab): Likewise.
23541 * tree-sra.c (convert_callers_for_node): Likewise.
23542 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23543 * value-prof.c (init_node_map): Likewise.
23545 2017-05-23 Martin Liska <mliska@suse.cz>
23547 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23548 and symtab_node::debug_symtab to symbol_table::debug.
23549 * cgraphunit.c (analyze_functions): Use the renamed function.
23550 (symbol_table::compile): Likewise.
23551 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23552 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23553 * passes.c (execute_todo): Likewise.
23554 * symtab.c (symbol_table::dump): New function.
23555 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23557 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23559 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23560 that nonconst implies exec.
23562 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23564 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23565 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23566 (inline_edge_summary_vec): Turn into ...
23567 (ipa_call_summaries): ... this one.
23568 (redirect_to_unreachable, edge_set_predicate,
23569 evaluate_properties_for_edge, inline_summary_alloc,
23570 reset_ipa_call_summary, reset_inline_summary,
23571 inline_summary_t::duplicate): Update.
23572 (inline_edge_duplication_hook): Turn to ...
23573 (ipa_call_summary_t::duplicate): ... this one.
23574 (inline_edge_removal_hook): Turn to ...
23575 (ipa_call_summary_t::remove): ... this one.
23576 (dump_inline_edge_summary): Turn to ...
23577 (dump_ipa_call_summary): ... this one.
23578 (estimate_function_body_sizes): Update.
23579 (inline_update_callee_summaries): Update.
23580 (remap_edge_change_prob): Update.
23581 (remap_edge_summaries): Update.
23582 (inline_merge_summary): Update.
23583 (do_estimate_edge_time): Update.
23584 (inline_generate_summary): Update.
23585 (inline_read_section): Update.
23586 (inline_read_summary): Update.
23587 (inline_free_summary): Update.
23588 * ipa-inline.c (can_inline_edge_p): Update.
23589 (compute_inlined_call_time): Update.
23590 (want_inline_small_function_p): Update.
23591 (edge_badness): Update.
23592 (early_inliner): Update.
23593 * ipa-inline.h (inline_edge_summary): Turn to ...
23594 (ipa_call_summary): ... this one.
23595 (ipa_call_summary_t): New class.
23596 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23597 (ipa_call_summaries): New.
23598 (inline_edge_summary): Remove.
23599 (estimate_edge_growth): Update.
23600 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23601 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23602 * ipa-split.c (execute_split_functions): Update.
23603 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23605 2017-05-23 Tom de Vries <tom@codesourcery.com>
23607 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23608 attributes): Document rdrand effective target.
23610 2017-05-23 Tom de Vries <tom@codesourcery.com>
23612 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23613 attributes): Sort alphabetically.
23615 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23617 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23619 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23622 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23623 V2DF/V2DI splat into two separate patterns, one that handles
23624 registers, and the other that only handles memory. Drop support
23625 for splatting from a GPR on ISA 2.07 and then splitting the
23626 splat into direct move and splat.
23627 (vsx_splat_<mode>_reg): Likewise.
23628 (vsx_splat_<mode>_mem): Likewise.
23630 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23632 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23634 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23636 PR middle-end/80809
23637 * omp-low.c (finish_taskreg_remap): New function.
23638 (finish_taskreg_scan): If unit size of ctx->record_type
23639 is non-constant, unshare the size expression and replace
23640 decls in it with possible outer var refs.
23642 PR middle-end/80809
23643 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23644 GOVD_SHARED rather than GOVD_PRIVATE with it.
23645 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23646 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23648 PR middle-end/80853
23649 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23650 as last argument to build_outer_var_ref for pointer bases of array
23651 section reductions.
23653 2017-05-19 Martin Sebor <msebor@redhat.com>
23655 * print-tree.c (print_node): Print DECL_READ_P flag.
23657 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23659 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23660 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23661 * cgraph.c: Likewise.
23662 * cgraphunit.c: Likewise.
23663 * gengtype.c: Likewise.
23664 * ipa-cp.c: Likewise.
23665 * ipa-devirt.c: Likewise.
23666 * ipa-icf.c: Likewise.
23667 * ipa-predicate.c: Likewise.
23668 * ipa-profile.c: Likewise.
23669 * ipa-prop.c: Likewise.
23670 * ipa-split.c: Likewise.
23672 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23673 edge_predicate_pool, dump_inline_hints,
23674 inline_summary::account_size_time, redirect_to_unreachable,
23675 edge_set_predicate, set_hint_predicate,
23676 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23677 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23678 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23679 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23680 ipa_call_summary_t::remove, initialize_growth_caches,
23681 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23682 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23683 mark_modified, unmodified_parm_1, unmodified_parm,
23684 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23685 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23686 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23687 will_be_nonconstant_predicate, record_modified_bb_info,
23688 get_minimal_bb, record_modified, param_change_prob,
23689 phi_result_unknown_predicate, predicate_for_phi_result,
23690 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23691 estimate_function_body_sizes, compute_inline_parameters,
23692 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23693 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23694 inline_update_callee_summaries, remap_edge_change_prob,
23695 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23696 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23697 inline_analyze_function, inline_summary_t::insert,
23698 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23699 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23700 inline_free_summary): Move to ipa-fnsummary.h
23701 (predicate_t): Remove.
23702 * ipa-fnsummary.c: New file.
23703 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23704 (enum inline_hints_vals, inline_hints, agg_position_info,
23705 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23706 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23707 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23708 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23709 inline_read_summary, inline_write_summary, inline_free_summary,
23710 inline_analyze_function, initialize_inline_failed,
23711 inline_merge_summary, inline_update_overall_summary,
23712 compute_inline_parameters): Move to ipa-fnsummary.h
23713 * ipa-fnsummary.h: New file.
23714 * ipa-inline-transform.h: Include ipa-inline.h.
23715 * ipa-inline.c: LIkewise.
23717 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23719 * ipa-inline.c (edge_badness): Use inlined_time instead of
23720 inline_summaries->get.
23722 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23724 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23726 2017-05-22 Nathan Sidwell <nathan@acm.org>
23728 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23729 (fdump-lang): Document 'raw' option.
23730 * dumpfile.h (TDI_tu): Delete.
23731 * dumpfile.c (dump_files): Remove translation-unit.
23732 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23734 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23736 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23737 command option from $(AWK) call.
23738 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23739 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23740 [FORMAT]: Remove handling of variable.
23741 * config/avr/t-multilib: Regenerate.
23743 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23745 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23747 (dump_inline_summary): Do not print self_time.
23748 (estimate_function_body_sizes): Do not set self_time.
23749 (compute_inline_parameters): Likewise.
23750 (inline_read_section, inline_write_summary): Do not stream self_time.
23751 * ipa-inline.h (inline_summary): Drop self_time.
23753 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23755 * ipa-inline-analysis.c (account_size_time): Rename to ...
23756 (inline_summary::account_size_time): ... this one.
23757 (reset_ipa_call_summary): Turn to ...
23758 (ipa_call_summary::reset): ... this one.
23759 (reset_inline_summary): Turn to ...
23760 (inline_summary::reset): ... this one.
23761 (inline_summary_t::remove): Update.
23762 (inline_summary_t::duplicate): Update.
23763 (ipa_call_summary_t::remove): Update.
23764 (dump_inline_summary): Update.
23765 (estimate_function_body_sizes): Update.
23766 (compute_inline_parameters): Update.
23767 (estimate_node_size_and_time): Update.
23768 (inline_merge_summary): Update.
23769 (inline_update_overall_summary): Update.
23770 (inline_read_section): Update.
23771 (inline_write_summary): Update.
23772 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23773 add account_size_time and reset member functions.
23774 (ipa_call_summary): Add reset function.
23775 * ipa-predicate.h (predicate::operator &): Constify.
23777 2017-05-22 Richard Biener <rguenther@suse.de>
23779 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23781 2017-05-19 Jason Merrill <jason@redhat.com>
23783 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23785 2017-05-19 Marek Polacek <polacek@redhat.com>
23788 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23789 TYPE_OVERFLOW_WRAPS checks.
23791 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23793 * tree-core.h (enum omp_clause_default_kind): Add
23794 "OMP_CLAUSE_DEFAULT_PRESENT".
23795 * tree-pretty-print.c (dump_omp_clause): Handle it.
23796 * gimplify.c (enum gimplify_omp_var_data): Add
23797 "GOVD_MAP_FORCE_PRESENT".
23798 (gimplify_adjust_omp_clauses_1): Map it to
23799 "GOMP_MAP_FORCE_PRESENT".
23800 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23802 * gimplify.c (oacc_default_clause): Clarify.
23804 2017-05-19 Nathan Sidwell <nathan@acm.org>
23806 LANG_HOOK_REGISTER_DUMPS
23807 * toplev.c (general_init): Call register dump lang hook.
23808 * doc/invoke.texi: Document -fdump-lang option family.
23809 * dumpfile.c (dump_files): Remove class dump here.
23810 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23811 * dumpfile.h (tree_dump_index): Remove TDI_class.
23812 * langhooks-def.h (lhd_register_dumps): Declare.
23813 (LANG_HOOKS_REGISTER_DUMPS): Define.
23814 (LANG_HOOKS_INITIALIZER): Add it.
23815 * langhooks.c (lhd_register_dumps): Define.
23816 * langhooks.h (struct lang_hooks): Add register_dumps.
23818 2017-05-19 Nathan Sidwell <nathan@acm.org>
23820 * context.h (context::set_passes): New.
23821 * context.c (context::context): Do not create pass manager.
23822 * toplev.c (general_init): Create pass manager here.
23824 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23826 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23827 use this splitter if two add or or instructions would also work for
23828 the constant we want to generate.
23830 2017-05-19 Richard Biener <rguenther@suse.de>
23833 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23834 predicate evaluation.
23836 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23838 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23840 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23843 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23845 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23846 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23847 (inline_edge_summary_vec): Turn into ...
23848 (ipa_call_summaries): ... this one.
23849 (redirect_to_unreachable, edge_set_predicate,
23850 evaluate_properties_for_edge, inline_summary_alloc,
23851 reset_ipa_call_summary, reset_inline_summary,
23852 inline_summary_t::duplicate): Update.
23853 (inline_edge_duplication_hook): Turn to ...
23854 (ipa_call_summary_t::duplicate): ... this one.
23855 (inline_edge_removal_hook): Turn to ...
23856 (ipa_call_summary_t::remove): ... this one.
23857 (dump_inline_edge_summary): Turn to ...
23858 (dump_ipa_call_summary): ... this one.
23859 (estimate_function_body_sizes): Update.
23860 (inline_update_callee_summaries): Update.
23861 (remap_edge_change_prob): Update.
23862 (remap_edge_summaries): Update.
23863 (inline_merge_summary): Update.
23864 (do_estimate_edge_time): Update.
23865 (inline_generate_summary): Update.
23866 (inline_read_section): Update.
23867 (inline_read_summary): Update.
23868 (inline_free_summary): Update.
23869 * ipa-inline.c (can_inline_edge_p): Update.
23870 (compute_inlined_call_time): Update.
23871 (want_inline_small_function_p): Update.
23872 (edge_badness): Update.
23873 (early_inliner): Update.
23874 * ipa-inline.h (inline_edge_summary): Turn to ...
23875 (ipa_call_summary): ... this one.
23876 (ipa_call_summary_t): New class.
23877 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23878 (ipa_call_summaries): New.
23879 (inline_edge_summary): Remove.
23880 (estimate_edge_growth): Update.
23881 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23882 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23883 * ipa-split.c (execute_split_functions): Update.
23884 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23886 2017-05-19 Richard Biener <rguenther@suse.de>
23888 PR middle-end/80764
23889 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23891 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23893 * config/rs6000/rs6000.c (struct machine_function): Add field
23894 fpr_is_wrapped_separately.
23895 (rs6000_get_separate_components): Use 64 components. Handle the
23896 new FPR components.
23897 (rs6000_components_for_bb): Handle the FPR components.
23898 (rs6000_emit_prologue_components): Handle the FPR components.
23899 (rs6000_emit_epilogue_components): Handle the FPR components.
23900 (rs6000_set_handled_components): Handle the FPR components.
23901 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23902 that are already separately shrink-wrapped.
23903 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23904 that are already separately shrink-wrapped.
23906 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23909 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23912 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23913 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23914 cases where the register allocator uses a move and an offsettable
23915 memory operation to/from a FPR register on ISA 2.06/2.07.
23916 (define_peephole2 for Altivec d-form store): Likewise.
23918 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23921 * config/i386/mmx.md (*mov<mode>_internal): Enable
23922 alternatives 11, 12, 13 and 14 also for 32bit targets.
23923 Remove alternatives 15, 16, 17 and 18.
23924 * config/i386/sse.md (vec_concatv2di): Change
23925 alternative (!x, *y) to (x, ?!*Yn).
23927 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23929 * dumpfile.h (enum dump_kind): Remove stray comma.
23931 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23933 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23934 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23935 predicate::num_conditions
23936 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23937 (CHANGED): turn into predicate::changed.
23938 (agg_position_info): Move to ipa-predicate.h
23939 (add_condition, predicate::add_clause, predicate::operator &=,
23940 predicate::or_with, predicate::evaluate, predicate::probability,
23941 dump_condition, dump_clause, predicate::dump,
23942 predicate::remap_after_duplication, predicate::remap_after_inlining,
23943 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23944 (evaluate_conditions_for_known_args): Update.
23945 (set_cond_stmt_execution_predicate): Update.
23946 * ipa-inline.h: Include ipa-predicate.h
23947 (condition, inline_param_summary, conditions, agg_position_info,
23948 predicate): Move to ipa-predicate.h
23949 * ipa-predicate.c: New file.
23950 * ipa-predicate.h: New file.
23952 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23954 * final.c (leaf_function_p): Check we are not in a sequence.
23956 2017-05-18 Martin Liska <mliska@suse.cz>
23958 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23959 * dumpfile.c (dump_register): Use new enum dump_kind.
23960 (get_dump_file_name): Likewise.
23961 (dump_enable_all): Likewise.
23962 (dump_switch_p_1): Likewise.
23963 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23964 * dumpfile.h (enum dump_kind): New enum type.
23965 (struct dump_file_info): Create constructor and
23966 format fields and comments.
23967 * passes.c (pass_manager::register_one_dump_file):
23969 * statistics.c (statistics_early_init): Likewise.
23970 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23971 TDF_TREE with TDF_SLIM.
23972 (gather_memory_references_ref): Likewise.
23974 2017-05-18 Martin Liska <mliska@suse.cz>
23976 * vec.h (struct vnull): Use it.
23978 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23980 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23981 (true_predicate, false_predicate, true_predicate_p,
23982 false_predicate_p): Remove.
23983 (single_cond_predicate, not_inlined_predicate): Turn to member function
23985 (add_condition): Update.
23986 (add_clause): Turn to...
23987 (predicate::add_clause): ... this one; update; allow passing NULL
23989 (and_predicates): Turn to ...
23990 (predicate::operator &=): ... this one.
23991 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23992 (or_predicates): Turn to ...
23993 (predicate::or_with): ... this one.
23994 (evaluate_predicate): Turn to ...
23995 (predicate::evaluate): ... this one.
23996 (predicate_probability): Turn to ...
23997 (predicate::probability): ... this one.
23998 (dump_condition): Update.
23999 (dump_predicate): Turn to ...
24000 (predicate::dump): ... this one.
24001 (account_size_time): Update.
24002 (edge_set_predicate): Update.
24003 (set_hint_predicate): UPdate.
24004 (evaluate_conditions_for_known_args): Update.
24005 (evaluate_properties_for_edge): Update.
24006 (remap_predicate_after_duplication): Turn to...
24007 (predicate::remap_after_duplication): ... this one.
24008 (remap_hint_predicate_after_duplication): Update.
24009 (inline_summary_t::duplicate): UPdate.
24010 (dump_inline_edge_summary): Update.
24011 (dump_inline_summary): Update.
24012 (set_cond_stmt_execution_predicate): Update.
24013 (set_switch_stmt_execution_predicate): Update.
24014 (compute_bb_predicates): Update.
24015 (will_be_nonconstant_expr_predicate): Update.
24016 (will_be_nonconstant_predicate): Update.
24017 (phi_result_unknown_predicate): Update.
24018 (predicate_for_phi_result): Update.
24019 (array_index_predicate): Update.
24020 (estimate_function_body_sizes): Update.
24021 (estimate_node_size_and_time): Update.
24022 (estimate_ipcp_clone_size_and_time): Update.
24023 (remap_predicate): Rename to ...
24024 (predicate::remap_after_inlining): ... this one.
24025 (remap_hint_predicate): Update.
24026 (inline_merge_summary): Update.
24027 (inline_update_overall_summary): Update.
24028 (estimate_size_after_inlining): Update.
24029 (read_predicate): Rename to ...
24030 (predicate::stream_in): ... this one.
24031 (read_inline_edge_summary): Update.
24032 (write_predicate): Rename to ...
24033 (predicate::stream_out): ... this one.
24034 (write_inline_edge_summary): Update.
24035 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24036 (clause_t): Turn to uint32_t
24037 (predicate): Turn to class; implement constructor and operators
24039 (size_time_entry): Update.
24040 (inline_summary): Update.
24041 (inline_edge_summary): Update.
24043 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24045 * fold-const.c (fold_binary_loc): Move transformation...
24046 * match.pd (C - X CMP X): ... here.
24048 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24050 * config/sparc/sparc.c (sparc_option_override): Set function
24051 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24052 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24054 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24056 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24058 2017-05-18 Marek Polacek <polacek@redhat.com>
24061 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24062 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24063 gimple_assign_load_p.
24065 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24067 PR middle-end/80692
24068 * real.c (do_compare): Give decimal_do_compare preference over
24069 comparing just the signs.
24071 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24073 * doc/md.texi (Canonicalization of Instructions): Describe the
24074 canonical form of instructions that inherently set a condition
24077 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24079 PR middle-end/80775
24080 * tree-cfg.c: Move deletion of unreachable case statements to after
24081 the merging of consecutive case labels.
24083 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24085 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24086 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24087 restoring of callee-saved registers.
24089 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24091 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24092 * config/visium/visium.c (single_set_and_flags): Likewise.
24093 * config/visium/visium.md (Substitutions): Likewise.
24095 2017-05-17 Martin Liska <mliska@suse.cz>
24097 * cfg.c: Introduce dump_flags_t type and
24098 use it instead of int type.
24100 * cfghooks.c: Likewise.
24101 * cfghooks.h (struct cfg_hooks): Likewise.
24102 * cfgrtl.c: Likewise.
24103 * cfgrtl.h: Likewise.
24104 * cgraph.c (cgraph_node::get_body): Likewise.
24105 * coretypes.h: Likewise.
24106 * domwalk.c: Likewise.
24107 * domwalk.h: Likewise.
24108 * dumpfile.c (struct dump_option_value_info): Likewise.
24109 (dump_enable_all): Likewise.
24110 (dump_switch_p_1): Likewise.
24111 (opt_info_switch_p): Likewise.
24112 * dumpfile.h (enum tree_dump_index): Likewise.
24113 (struct dump_file_info): Likewise.
24114 * genemit.c: Likewise.
24115 * generic-match-head.c: Likewise.
24116 * gengtype.c (open_base_files): Likewise.
24117 * gimple-pretty-print.c: Likewise.
24118 * gimple-pretty-print.h: Likewise.
24119 * graph.c (print_graph_cfg): Likewise.
24120 * graphite-scop-detection.c (dot_all_sese): Likewise.
24121 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24122 * loop-unroll.c (report_unroll): Likewise.
24123 * passes.c (pass_manager::register_one_dump_file): Likewise.
24124 * print-tree.c: Likewise.
24125 * statistics.c: Likewise.
24126 * tree-cfg.c: Likewise.
24127 * tree-cfg.h: Likewise.
24128 * tree-dfa.c: Likewise.
24129 * tree-dfa.h: Likewise.
24130 * tree-dump.c (dump_function): Likewise.
24131 * tree-dump.h (struct dump_info): Likewise.
24132 * tree-pretty-print.c: Likewise.
24133 * tree-pretty-print.h: Likewise.
24134 * tree-ssa-live.c: Likewise.
24135 * tree-ssa-live.h: Likewise.
24136 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24137 * tree-vect-loop.c: Likewise.
24138 * tree-vect-slp.c: Likewise.
24140 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24141 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24143 PR tree-optimization/80457
24144 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24145 of all arguments to a statement as scalar_to_vec operations.
24146 (vectorizable_call): Adjust call to vect_model_simple_cost for
24148 (vectorizable_conversion): Likewise.
24149 (vectorizable_assignment): Likewise.
24150 (vectorizable_shift): Likewise.
24151 (vectorizable_operation): Likewise.
24152 (vectorizable_comparison): Likewise.
24153 (vect_is_simple_cond): Record the def types for operands.
24154 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24155 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24156 for statement argument count.
24158 2017-05-16 Carl Love <cel@us.ibm.com>
24160 * config/rs6000/rs6000-c: Add support for built-in functions
24161 vector unsigned long long vec_bperm (vector unsigned long long,
24162 vector unsigned char)
24163 vector signed long long vec_mule (vector signed int,
24165 vector unsigned long long vec_mule (vector unsigned int,
24166 vector unsigned int)
24167 vector signed long long vec_mulo (vector signed int,
24169 vector unsigned long long vec_mulo (vector unsigned int,
24170 vector unsigned int)
24171 vector signed char vec_sldw (vector signed char,
24172 vector signed char,
24174 vector unsigned char vec_sldw (vector unsigned char,
24175 vector unsigned char,
24177 vector signed short vec_sldw (vector signed short,
24178 vector signed short,
24180 vector unsigned short vec_sldw (vector unsigned short,
24181 vector unsigned short,
24183 vector signed int vec_sldw (vector signed int,
24186 vector unsigned int vec_sldw (vector unsigned int,
24187 vector unsigned int,
24189 vector signed long long vec_sldw (vector signed long long,
24190 vector signed long long,
24192 vector unsigned long long vec_sldw (vector unsigned long long,
24193 vector unsigned long long,
24195 * config/rs6000/rs6000-c: Add support for built-in functions
24196 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24197 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24198 * doc/extend.texi: Update the built-in documentation for the
24199 new built-in functions.
24201 2017-05-16 Marek Polacek <polacek@redhat.com>
24205 * tree.c (save_expr): Don't fold the expression.
24207 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24209 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24210 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24211 and (?*y,m). Update insn attributes.
24213 2017-05-16 Martin Liska <mliska@suse.cz>
24215 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24216 flags argument of print_gimple_stmt, print_gimple_expr,
24217 print_generic_stmt and print_generic_expr.
24218 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24219 * coretypes.h: Likewise.
24220 * except.c (dump_eh_tree): Likewise.
24221 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24222 * gimple-pretty-print.h: Likewise.
24223 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24224 (backprop::push_to_worklist): Likewise.
24225 (backprop::pop_from_worklist): Likewise.
24226 (backprop::process_use): Likewise.
24227 (backprop::intersect_uses): Likewise.
24228 (note_replacement): Likewise.
24229 * gimple-ssa-store-merging.c
24230 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24231 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24232 (pass_store_merging::execute): Likewise.
24233 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24234 (ssa_base_cand_dump_callback): Likewise.
24235 (dump_incr_vec): Likewise.
24236 (replace_refs): Likewise.
24237 (replace_mult_candidate): Likewise.
24238 (create_add_on_incoming_edge): Likewise.
24239 (create_phi_basis): Likewise.
24240 (insert_initializers): Likewise.
24241 (all_phi_incrs_profitable): Likewise.
24242 (introduce_cast_before_cand): Likewise.
24243 (replace_one_candidate): Likewise.
24244 * gimplify.c (gimplify_expr): Likewise.
24245 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24246 (set_rename): Likewise.
24247 (rename_uses): Likewise.
24248 (copy_loop_phi_nodes): Likewise.
24249 (add_close_phis_to_merge_points): Likewise.
24250 (copy_loop_close_phi_args): Likewise.
24251 (copy_cond_phi_args): Likewise.
24252 (graphite_copy_stmts_from_block): Likewise.
24253 (translate_pending_phi_nodes): Likewise.
24254 * graphite-poly.c (print_pdr): Likewise.
24255 (dump_gbb_cases): Likewise.
24256 (dump_gbb_conditions): Likewise.
24257 (print_scop_params): Likewise.
24258 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24259 (build_cross_bb_scalars_use): Likewise.
24260 (gather_bbs::before_dom_children): Likewise.
24261 * hsa-dump.c (dump_hsa_immed): Likewise.
24262 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24263 (get_replacement_map): Likewise.
24264 * ipa-inline-analysis.c (dump_condition): Likewise.
24265 (estimate_function_body_sizes): Likewise.
24266 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24267 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24268 * ipa-prop.c (ipa_dump_param): Likewise.
24269 (ipa_print_node_jump_functions_for_edge): Likewise.
24270 (ipa_modify_call_arguments): Likewise.
24271 (ipa_modify_expr): Likewise.
24272 (ipa_dump_param_adjustments): Likewise.
24273 (ipa_dump_agg_replacement_values): Likewise.
24274 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24275 * ipa-pure-const.c (check_stmt): Likewise.
24276 (pass_nothrow::execute): Likewise.
24277 * ipa-split.c (execute_split_functions): Likewise.
24278 * omp-offload.c (dump_oacc_loop_part): Likewise.
24279 (dump_oacc_loop): Likewise.
24280 * trans-mem.c (tm_log_emit): Likewise.
24281 (tm_memopt_accumulate_memops): Likewise.
24282 (dump_tm_memopt_set): Likewise.
24283 (dump_tm_memopt_transform): Likewise.
24284 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24285 (print_loop): Likewise.
24286 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24287 (chkp_gather_checks_info): Likewise.
24288 (chkp_get_check_result): Likewise.
24289 (chkp_remove_check_if_pass): Likewise.
24290 (chkp_use_outer_bounds_if_possible): Likewise.
24291 (chkp_reduce_bounds_lifetime): Likewise.
24292 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24293 (chkp_mark_completed_bounds): Likewise.
24294 (chkp_register_incomplete_bounds): Likewise.
24295 (chkp_mark_invalid_bounds): Likewise.
24296 (chkp_maybe_copy_and_register_bounds): Likewise.
24297 (chkp_build_returned_bound): Likewise.
24298 (chkp_get_bound_for_parm): Likewise.
24299 (chkp_build_bndldx): Likewise.
24300 (chkp_get_bounds_by_definition): Likewise.
24301 (chkp_generate_extern_var_bounds): Likewise.
24302 (chkp_get_bounds_for_decl_addr): Likewise.
24303 * tree-chrec.c (chrec_apply): Likewise.
24304 * tree-data-ref.c (dump_data_reference): Likewise.
24305 (dump_subscript): Likewise.
24306 (dump_data_dependence_relation): Likewise.
24307 (analyze_overlapping_iterations): Likewise.
24308 * tree-inline.c (expand_call_inline): Likewise.
24309 (tree_function_versioning): Likewise.
24310 * tree-into-ssa.c (dump_defs_stack): Likewise.
24311 (dump_currdefs): Likewise.
24312 (dump_names_replaced_by): Likewise.
24313 (dump_update_ssa): Likewise.
24314 (update_ssa): Likewise.
24315 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24316 * tree-parloops.c (build_new_reduction): Likewise.
24317 (try_create_reduction_list): Likewise.
24318 (ref_conflicts_with_region): Likewise.
24319 (oacc_entry_exit_ok_1): Likewise.
24320 (oacc_entry_exit_single_gang): Likewise.
24321 * tree-pretty-print.h: Likewise.
24322 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24323 (get_scalar_evolution): Likewise.
24324 (add_to_evolution): Likewise.
24325 (get_loop_exit_condition): Likewise.
24326 (analyze_evolution_in_loop): Likewise.
24327 (analyze_initial_condition): Likewise.
24328 (analyze_scalar_evolution): Likewise.
24329 (instantiate_scev): Likewise.
24330 (number_of_latch_executions): Likewise.
24331 (gather_chrec_stats): Likewise.
24332 (final_value_replacement_loop): Likewise.
24333 (scev_const_prop): Likewise.
24334 * tree-sra.c (dump_access): Likewise.
24335 (disqualify_candidate): Likewise.
24336 (create_access): Likewise.
24337 (reject): Likewise.
24338 (maybe_add_sra_candidate): Likewise.
24339 (create_access_replacement): Likewise.
24340 (analyze_access_subtree): Likewise.
24341 (analyze_all_variable_accesses): Likewise.
24342 (sra_modify_assign): Likewise.
24343 (initialize_constant_pool_replacements): Likewise.
24344 (find_param_candidates): Likewise.
24345 (decide_one_param_reduction): Likewise.
24346 (replace_removed_params_ssa_names): Likewise.
24347 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24348 * tree-ssa-copy.c (dump_copy_of): Likewise.
24349 (copy_prop_visit_cond_stmt): Likewise.
24350 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24351 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24352 (record_equivalences_from_stmt): Likewise.
24353 * tree-ssa-dse.c (compute_trims): Likewise.
24354 (delete_dead_call): Likewise.
24355 (delete_dead_assignment): Likewise.
24356 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24357 (forward_propagate_into_cond): Likewise.
24358 (pass_forwprop::execute): Likewise.
24359 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24360 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24362 (move_computations_worker): Likewise.
24363 (execute_sm): Likewise.
24364 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24365 (remove_exits_and_undefined_stmts): Likewise.
24366 (remove_redundant_iv_tests): Likewise.
24367 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24368 (adjust_iv_update_pos): Likewise.
24369 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24370 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24371 (value_replacement): Likewise.
24372 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24373 * tree-ssa-pre.c (print_pre_expr): Likewise.
24374 (get_representative_for): Likewise.
24375 (create_expression_by_pieces): Likewise.
24376 (insert_into_preds_of_block): Likewise.
24377 (eliminate_insert): Likewise.
24378 (eliminate_dom_walker::before_dom_children): Likewise.
24379 (eliminate): Likewise.
24380 (remove_dead_inserted_code): Likewise.
24381 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24382 * tree-ssa-reassoc.c (get_rank): Likewise.
24383 (eliminate_duplicate_pair): Likewise.
24384 (eliminate_plus_minus_pair): Likewise.
24385 (eliminate_not_pairs): Likewise.
24386 (undistribute_ops_list): Likewise.
24387 (eliminate_redundant_comparison): Likewise.
24388 (update_range_test): Likewise.
24389 (optimize_range_tests_var_bound): Likewise.
24390 (optimize_vec_cond_expr): Likewise.
24391 (rewrite_expr_tree): Likewise.
24392 (rewrite_expr_tree_parallel): Likewise.
24393 (linearize_expr): Likewise.
24394 (break_up_subtract): Likewise.
24395 (linearize_expr_tree): Likewise.
24396 (attempt_builtin_powi): Likewise.
24397 (attempt_builtin_copysign): Likewise.
24398 (transform_stmt_to_copy): Likewise.
24399 (transform_stmt_to_multiply): Likewise.
24400 (dump_ops_vector): Likewise.
24401 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24402 (print_scc): Likewise.
24403 (set_ssa_val_to): Likewise.
24404 (visit_reference_op_store): Likewise.
24405 (visit_use): Likewise.
24406 (sccvn_dom_walker::before_dom_children): Likewise.
24407 (run_scc_vn): Likewise.
24408 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24410 (expr_hash_elt::print): Likewise.
24411 (const_and_copies::pop_to_marker): Likewise.
24412 (const_and_copies::record_const_or_copy_raw): Likewise.
24413 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24414 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24415 (dump_predicates): Likewise.
24416 (find_uninit_use): Likewise.
24417 (warn_uninitialized_phi): Likewise.
24418 (pass_late_warn_uninitialized::execute): Likewise.
24419 * tree-ssa.c (verify_vssa): Likewise.
24420 (verify_ssa): Likewise.
24421 (maybe_optimize_var): Likewise.
24422 * tree-vrp.c (dump_value_range): Likewise.
24423 (dump_all_value_ranges): Likewise.
24424 (dump_asserts_for): Likewise.
24425 (register_edge_assert_for_2): Likewise.
24426 (vrp_visit_cond_stmt): Likewise.
24427 (vrp_visit_switch_stmt): Likewise.
24428 (vrp_visit_stmt): Likewise.
24429 (vrp_visit_phi_node): Likewise.
24430 (simplify_cond_using_ranges_1): Likewise.
24431 (fold_predicate_in): Likewise.
24432 (evrp_dom_walker::before_dom_children): Likewise.
24433 (evrp_dom_walker::push_value_range): Likewise.
24434 (evrp_dom_walker::pop_value_range): Likewise.
24435 (execute_early_vrp): Likewise.
24437 2017-05-16 Richard Biener <rguenther@suse.de>
24439 * dwarf2out.c (loc_list_from_tree_1): Do not create
24440 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24442 2017-05-16 Richard Biener <rguenther@suse.de>
24444 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24446 (note_variable_value_in_expr): If we resolved the decl ref
24447 do not push to the stack.
24449 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24451 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24452 operations in fast-math mode.
24453 (vaddq_f16): Likewise.
24454 (vmul_f16): Likewise.
24455 (vmulq_f16): Likewise.
24456 (vsub_f16): Likewise.
24457 (vsubq_f16): Likewise.
24458 * config/arm/neon.md (add<mode>3): New.
24460 (fma:<VH:mode>3): New. Also remove outdated comment.
24463 2017-05-16 Martin Liska <mliska@suse.cz>
24467 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24468 (odr_types_equivalent_p): Likewise.
24470 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24472 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24474 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24477 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24478 non-interunit SSE move alternatives with '?'.
24479 (zero-extendsidi peephole2): New peephole to skip intermediate
24480 general register in SSE zero-extend sequence.
24482 2017-05-15 Jeff Law <law@redhat.com>
24484 * reorg.c (relax_delay_slots): Create a new variable to hold
24485 the temporary target rather than clobbering TARGET_LABEL.
24487 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24488 missing argument to extract_bit_field call.
24489 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24491 2017-05-15 Martin Liska <mliska@suse.cz>
24494 * gcc.c (process_command): Do not allow empty argument of -o option.
24496 2017-05-15 Renlin Li <renlin.li@arm.com>
24498 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24499 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24500 * config/aarch64/constraints.md (Usf): Add long call check.
24501 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24502 (call_value): Likewise.
24503 (sibcall): Likewise.
24504 (sibcall_value): Likewise.
24506 (call_value_insn): New.
24507 (sibcall_insn): Update rtx pattern.
24508 (sibcall_value_insn): Likewise.
24509 (call_internal): Remove.
24510 (call_value_internal): Likewise.
24511 (sibcall_internal): Likewise.
24512 (sibcall_value_internal): Likewise.
24513 (call_reg): Likewise.
24514 (call_symbol): Likewise.
24515 (call_value_reg): Likewise.
24516 (call_value_symbol): Likewise.
24518 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24521 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24523 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24525 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24526 compatible with CCGOCmode and with CCZmode.
24528 2017-05-14 Martin Sebor <msebor@redhat.com>
24530 PR middle-end/77671
24531 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24532 (gimple_fold_builtin_snprintf): Same.
24533 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24534 (gimple_fold_builtin_snprintf): Same.
24535 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24536 of character types.
24537 (is_call_safe): New function.
24538 (try_substitute_return_value): Call it.
24539 (try_simplify_call): New function.
24540 (pass_sprintf_length::handle_gimple_call): Call it.
24542 2017-05-14 Martin Sebor <msebor@redhat.com>
24544 PR middle-end/80669
24545 * builtins.c (expand_builtin_stpncpy): Simplify.
24547 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24549 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24550 * config/i386/i386.h
24551 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24552 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24553 (struct machine_function): Add new members call_ms2sysv,
24554 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24555 (struct machine_frame_state): New fields sp_realigned and
24556 sp_realigned_offset.
24557 * config/i386/i386.c
24558 (enum xlogue_stub): New enum.
24559 (enum xlogue_stub_sets): New enum.
24560 (class xlogue_layout): New class.
24561 (struct ix86_frame): New fields stack_realign_allocate_offset,
24562 stack_realign_offset and outlined_save_offset. Modify comments to
24563 detail stack layout when using out-of-line stubs.
24564 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24565 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24566 -mcall-ms2sysv-xlogues.
24567 (stub_managed_regs): New static variable.
24568 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24569 registers managed by out-of-line stub.
24570 (disable_call_ms2sysv_xlogues): New function.
24571 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24572 m->call_ms2sysv when appropriate and compute frame layout for
24574 (sp_valid_at, fp_valid_at): New inline functions.
24575 (choose_basereg): New function.
24576 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24578 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24579 Use align parameter of choose_baseaddr to generated aligned SSE movs
24581 (pro_epilogue_adjust_stack): Modify to track
24582 machine_frame_state::sp_realigned.
24583 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24584 (ix86_nsaved_sseregs): Likewise.
24585 (ix86_emit_save_regs): Likewise.
24586 (ix86_emit_save_regs_using_mov): Likewise.
24587 (ix86_emit_save_sse_regs_using_mov): Likewise.
24588 (get_scratch_register_on_entry): Likewise.
24589 (gen_frame_set): New function.
24590 (gen_frame_load): Likewise.
24591 (gen_frame_store): Likewise.
24592 (emit_outlined_ms2sysv_save): Likewise.
24593 (emit_outlined_ms2sysv_restore): Likewise.
24594 (ix86_expand_prologue): Modify stack re-alignment code and call
24595 emit_outlined_ms2sysv_save when appropriate.
24596 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24597 parameter rtx_insn *insn, which allows the function to be used to only
24598 generate the notes.
24599 (ix86_expand_epilogue): Modify validity checks of frame and stack
24600 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24601 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24602 * config/i386/predicates.md
24603 (save_multiple): New predicate.
24604 (restore_multiple): Likewise.
24605 * config/i386/sse.md
24606 (save_multiple<mode>): New pattern.
24607 (save_multiple_realign<mode>): Likewise.
24608 (restore_multiple<mode>): Likewise.
24609 (restore_multiple_and_return<mode>): Likewise.
24610 (restore_multiple_leave_return<mode>): Likewise.
24611 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24613 2017-05-14 Julia Koval <julia.koval@intel.com>
24615 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24616 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24617 (__builtin_ia32_xsetbv): New builtins.
24618 * config/i386/i386.c (ix86_expand_special_args_builtin):
24620 (ix86_expand_builtin): Special expand for new intrinsics.
24621 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24622 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24623 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24625 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24627 * cfganal.c (inverted_post_order_compute): Change argument type
24629 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24630 * df-core.c (rest_of_handle_df_initialize): Adjust.
24631 (rest_of_handle_df_finish): Likewise.
24632 (df_analyze_1): Likewise.
24633 (df_analyze): Likewise.
24634 (loop_inverted_post_order_compute): Change argument to be a vec *.
24635 (df_analyze_loop): Adjust.
24636 (df_get_n_blocks): Likewise.
24637 (df_get_postorder): Likewise.
24638 * df.h (struct df_d): Change field to be a vec.
24639 * lcm.c (compute_laterin): Adjust.
24640 (compute_available): Likewise.
24641 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24642 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24643 * tree-ssa-pre.c (compute_antic): Likewise.
24645 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24647 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24648 (depth_first_search::depth_first_search): Change structure init
24649 function to this constructor.
24650 (depth_first_search::add_bb): Rename function to this member.
24651 (depth_first_search::execute): Likewise.
24652 (flow_dfs_compute_reverse_finish): Adjust.
24654 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24656 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24657 (longest_simple_path): Likewise.
24658 * shrink-wrap.c (spread_components): Likewise.
24659 (disqualify_problematic_components): Likewise.
24660 (emit_common_heads_for_components): Likewise.
24661 (emit_common_tails_for_components): Likewise.
24662 (insert_prologue_epilogue_for_components): Likewise.
24664 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24666 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24669 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24671 * df-core.c (df_set_blocks): Start using auto_bitmap.
24672 (df_compact_blocks): Likewise.
24673 * df-problems.c (df_rd_confluence_n): Likewise.
24674 * df-scan.c (df_insn_rescan_all): Likewise.
24675 (df_process_deferred_rescans): Likewise.
24676 (df_update_entry_block_defs): Likewise.
24677 (df_update_exit_block_uses): Likewise.
24678 (df_entry_block_bitmap_verify): Likewise.
24679 (df_exit_block_bitmap_verify): Likewise.
24680 (df_scan_verify): Likewise.
24681 * lra-constraints.c (lra_constraints): Likewise.
24682 (undo_optional_reloads): Likewise.
24683 (lra_undo_inheritance): Likewise.
24684 * lra-remat.c (calculate_gen_cands): Likewise.
24685 (do_remat): Likewise.
24686 * lra-spills.c (assign_spill_hard_regs): Likewise.
24687 (spill_pseudos): Likewise.
24688 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24689 (bitmap_set_subtract_values): Likewise.
24691 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24693 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24694 management with auto_bitmap.
24695 (fix_inter_tick): Likewise.
24696 (fix_recovery_deps): Likewise.
24697 * ira.c (add_store_equivs): Likewise.
24698 (find_moveable_pseudos): Likewise.
24699 (split_live_ranges_for_shrink_wrap): Likewise.
24700 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24701 (rtx_reuse_manager::seen_def_p): Likewise.
24702 (rtx_reuse_manager::set_seen_def): Likewise.
24703 * print-rtl.h (class rtx_reuse_manager): Likewise.
24705 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24707 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24709 (migrate_btr_def): Likewise.
24710 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24711 * df-core.c (loop_post_order_compute): Likewise.
24712 (loop_inverted_post_order_compute): Likewise.
24713 * hsa-common.h: Likewise.
24714 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24715 * init-regs.c (initialize_uninitialized_regs): Likewise.
24716 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24717 (inline_small_functions): Likewise.
24718 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24719 * ira.c (combine_and_move_insns): Likewise.
24720 (build_insn_chain): Likewise.
24721 * loop-invariant.c (find_invariants): Likewise.
24722 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24723 * predict.c (tree_predict_by_opcode): Likewise.
24724 (predict_paths_leading_to): Likewise.
24725 (predict_paths_leading_to_edge): Likewise.
24726 (estimate_loops_at_level): Likewise.
24727 (estimate_loops): Likewise.
24728 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24729 (spread_components): Likewise.
24730 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24731 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24732 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24733 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24734 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24735 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24736 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24737 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24738 (mark_threaded_blocks): Likewise.
24739 (thread_through_all_blocks): Likewise.
24740 * tree-ssa.c (verify_ssa): Likewise.
24741 (execute_update_addresses_taken): Likewise.
24742 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24744 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24746 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24748 (post_order_compute): Likewise.
24749 (inverted_post_order_compute): Likewise.
24750 (pre_and_rev_post_order_compute_fn): Likewise.
24752 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24754 * genrecog.c (int_set::int_set): Explicitly construct our
24755 auto_vec base class.
24756 * vec.h (auto_vec::auto_vec): New constructor.
24758 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24760 * bitmap.h (class auto_bitmap): New constructor taking
24761 bitmap_obstack * argument.
24763 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24765 * bitmap.h (class auto_bitmap): Change type of m_bits to
24766 bitmap_head, and adjust ctor / dtor and member operators.
24768 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24770 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24771 when returned register mode doesn't match original mode.
24773 2017-05-12 Jeff Law <law@redhat.com>
24774 Jakub Jelinek <jakub@redhat.com>
24776 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24777 we look for cc setter after the compare-elim changes.
24778 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24779 within the vector to match what compare-elim now expects.
24780 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24781 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24783 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24784 after the compare-elim changes.
24785 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24786 the vector to match what compare-elim now expects.
24787 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24788 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24789 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24790 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24791 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24793 * config/visium/visium.c (single_set_and_flags): Fix where
24794 we look for cc setter after the compare-elim changes.
24795 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24796 with the vector to match what compare-elim now expects.
24797 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24798 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24799 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24800 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24801 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24802 (neg<mode>2_insn_set_overflow): Likewise.
24804 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24806 PR middle-end/79794
24807 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24808 maybe_expand_insn call, set ops[0].target. If still set after call,
24809 set alt_rtl. Add extra arg to recursive calls.
24810 (extract_bit_field): Add alt_rtl argument. Pass to
24812 * expmed.h (extract_bit_field): Fix prototype.
24813 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24814 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24815 to extract_bit_field_calls.
24816 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24817 Pass alt_rtl to extract_bit_field calls.
24818 * calls.c (store_unaligned_arguments_into_psuedos)
24819 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24820 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24822 * optabs.h (struct expand_operand): Add target bitfield.
24824 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24826 * compare-elim.c (try_eliminate_compare): Canonicalize
24827 operation with embedded compare to
24828 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24829 (set (reg) (operation)].
24831 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24833 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24836 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24837 cost of adding a carry flag for ADC instruction.
24838 [case MINUS]: Ignore the cost of subtracting a carry flag
24839 for SBB instruction.
24841 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24843 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24845 * config/rs6000/bmiintrin.h: New file.
24846 * config/rs6000/bmi2intrin.h: New file.
24847 * config/rs6000/x86intrin.h: New file.
24849 2017-05-12 Jeff Law <law@redhat.com>
24851 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24854 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24856 PR middle-end/80707
24857 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24859 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24861 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24862 early expansion of vector divide builtins.
24863 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24864 builtins identified as having unsigned arguments.
24866 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24868 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24869 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24870 expansion of vector logical operations (and, andc, or, xor,
24873 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24875 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24876 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24878 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24880 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24881 early GIMPLE expansion of vector multiplies.
24883 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24885 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24886 TARGET_HAVE_MOVT conditional.
24887 (movt splitter): Likewise.
24889 2017-05-12 Richard Biener <rguenther@suse.de>
24891 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24892 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24893 Fold all stmts not inplace.
24895 2017-05-12 Richard Biener <rguenther@suse.de>
24897 PR tree-optimization/80713
24898 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24899 inserted_exprs bit for not removed stmts.
24901 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24903 PR middle-end/69921
24904 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24905 parallelized" attribute for parallelized OpenACC kernels.
24906 * omp-offload.c (execute_oacc_device_lower): Use it.
24908 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24909 Set "oacc kernels" attribute.
24910 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24911 parameter. Adjust all users.
24912 (oacc_fn_attrib_kernels_p): Remove function.
24913 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24914 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24915 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24916 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24917 assert "oacc kernels" attribute is set.
24919 2017-05-11 Carl Love <cel@us.ibm.com>
24921 * config/rs6000/rs6000-c: Add support for built-in functions
24922 vector unsigned char vec_popcnt (vector signed char)
24923 vector unsigned char vec_popcnt (vector unsigned char)
24924 vector unsigned short vec_popcnt (vector signed short)
24925 vector unsigned short vec_popcnt (vector unsigned short)
24926 vector unsigned int vec_popcnt (vector signed int)
24927 vector unsigned int vec_popcnt (vector unsigned int)
24928 vector unsigned long long vec_popcnt (vector signed long long)
24929 vector unsigned long long vec_popcnt (vector unsigned long long)
24930 vector signed long long vec_slo (vector signed long long,
24931 vector signed char)
24932 vector signed long long vec_slo (vector signed long long,
24933 vector unsigned char)
24934 vector unsigned long long vec_slo (vector unsigned long long,
24935 vector signed char)
24936 vector unsigned long long vec_slo (vector unsigned long long,
24937 vector unsigned char)
24938 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24939 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24940 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24941 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24942 * doc/extend.texi: Update the built-in documentation file for the
24943 new built-in functions.
24945 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24947 * attribs.h (sorted_attr_string): Move machine independent
24948 functions for target clone support from the i386 port to common
24949 code. Rename ix86_function_versions to common_function_versions.
24950 Rename make_name to make_unique_name.
24951 (common_function_versions): Likewise.
24952 (make_unique_name): Likewise.
24953 (make_dispatcher_decl): Likewise.
24954 (is_function_default_version): Likewise.
24955 * attribs.c (attr_strcmp): Likewise.
24956 (sorted_attr_string): Likewise.
24957 (common_function_versions): Likewise.
24958 (make_unique_name): Likewise.
24959 (make_dispatcher_decl): Likewise.
24960 (is_function_default_version): Likewise.
24961 * config/i386/i386.c (attr_strcmp): Likewise.
24962 (sorted_attr_string): Likewise.
24963 (ix86_function_versions): Likewise.
24964 (make_name): Likewise.
24965 (make_dispatcher_decl): Likewise.
24966 (is_function_default_version): Likewise.
24967 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24969 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24972 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24973 Account for direct move costs for vec_construct of integer
24976 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24979 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24980 (UNSPEC_STX_ATOMIC): Ditto.
24981 (loaddi_via_sse): New insn.
24982 (storedi_via_sse): Ditto.
24983 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24984 Update corresponding peephole2 patterns.
24985 (atomic_storedi_fpu): Ditto.
24987 2017-05-11 Julia Koval <julia.koval@intel.com>
24989 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24990 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24992 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24993 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24994 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24996 2017-05-11 Nathan Sidwell <nathan@acm.org>
24998 * graphite-poly.c: Include dumpfile.h.
25000 * dumpfle.h (dump_function): Declare here ...
25001 * tree-dump.h (dump_function): ... not here.
25002 * dumpfile.c: #include tree-cfg.h.
25003 (dump_function): Move here from ...
25004 * tree-dump.c (dump_function): ... here.
25005 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25006 * graphite-poly.c: Don't include tree-dump.h.
25007 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25008 * print-tree.c: Likewise.
25009 * stor-layout.c: Likewise.
25010 * tree-nested.c: Likewise.
25012 * dumpfile.c (dump_start): Use TDF_FLAGS.
25013 (dump_enable_all): Fix TDF_KIND check thinko.
25015 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25017 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25018 array entries to represent two legal parameterizations of the
25019 overloaded __builtin_cmpb function, as represented by the
25020 P6_OV_BUILTIN_CMPB constant.
25021 (altivec_resolve_overloaded_builtin): Add special case handling
25022 for the __builtin_cmpb function, as represented by the
25023 P6_OV_BUILTIN_CMPB constant.
25024 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25025 (BU_P6_64BIT_2): New macro.
25026 (BU_P6_OVERLOAD_2): New macro
25027 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25028 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25029 (CMPB): Add overload support to represent both 32-bit and 64-bit
25030 compare-bytes function.
25031 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25032 support for TARGET_CMPB.
25033 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25034 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25035 documentation of the __builtin_cmpb overloaded built-in function.
25037 2017-05-11 Richard Biener <rguenther@suse.de>
25039 PR tree-optimization/80705
25040 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25041 bases are not vectorizable.
25043 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25045 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25046 when counting register pressure.
25048 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25050 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25051 (iv_ca_more_deps): Renamed to ...
25052 (iv_ca_compare_deps): ... this.
25053 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25055 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25057 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25059 (determine_group_iv_costs): ... here.
25060 (find_inv_vars_cb): Record inv var if it's not recorded before.
25062 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25064 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25065 (get_shiftadd_cost): Ditto.
25067 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25069 * tree-ssa-address.c: Include header file.
25070 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25072 (add_to_parts): Refactor.
25073 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25074 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25077 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25079 PR tree-optimization/53090
25080 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25082 (extract_cond_operands): Detect condition with IV on both sides
25083 and return COMP_IV_EXPR_2.
25084 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25085 (rewrite_use_compare): Simplify by removing call to function
25086 extract_cond_operands.
25088 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25090 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25091 (extract_cond_operands): Detect condition comparing against non-
25092 invariant bound and return appropriate enum value.
25093 (find_interesting_uses_cond): Update use of extract_cond_operands.
25094 Handle its return value accordingly.
25095 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25097 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25099 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25100 nonlinear iv_use computation in loop invariant sensitive way.
25102 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25104 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25105 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25107 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25109 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25110 (dump_cand): Support iv_cand.inv_exprs.
25111 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25113 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25116 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25118 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25120 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25121 as local function. Include necessary header files.
25122 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25124 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25126 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25128 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25130 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25131 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25132 RSHIFT_EXPR and BIT_NOT_EXPR.
25134 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25136 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25137 (adjust_setup_cost): New parameter supporting round up adjustment.
25138 (struct address_cost_data): Delete.
25139 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25140 (split_address_cost, ptr_difference_cost): Delete.
25141 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25142 (struct ainc_cost_data): New struct.
25143 (get_address_cost_ainc): New function.
25144 (get_address_cost, get_computation_cost): Reimplement.
25145 (determine_group_iv_cost_address): Record inv_expr for all uses of
25147 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25148 (iv_ca_has_deps): Reimplemented to ...
25149 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25151 (iv_ca_extend): Call iv_ca_more_deps.
25153 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25155 * tree-ssa-address.c (struct mem_address): Move to header file.
25156 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25157 * tree-ssa-address.h (struct mem_address): Move from C file.
25158 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25160 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25162 * tree-affine.h (aff_combination_type): New interface.
25163 (aff_combination_zero_p): Remove static.
25164 (aff_combination_const_p): New interface.
25165 (aff_combination_singleton_var_p): New interfaces.
25167 2017-05-11 Richard Biener <rguenther@suse.de>
25169 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25170 Skip unreachable blocks and destinations.
25171 (eliminate): Move stmt removal and fixup ...
25172 (fini_eliminate): ... here. Skip inserted exprs.
25173 (pass_pre::execute): Move fini_pre after fini_eliminate.
25174 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25175 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25176 PRE to get rid of dead code that has invalid SSA form and
25177 split critical edges again.
25179 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25181 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25183 2017-05-11 Richard Biener <rguenther@suse.de>
25185 * passes.c (execute_function_todo): Verify loops if they are
25186 said to be up-to-date.
25187 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25188 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25190 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25193 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25194 handle calling assemble_external ourself.
25197 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25198 modes with zero size. Enhance comment.
25200 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25202 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25203 built-ins for vec_xl and vec_xst with short and char pointer
25206 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25208 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25209 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25210 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25211 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25212 (_mm_maskz_min_round_ss): New intrinsics.
25213 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25214 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25215 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25216 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25217 (__builtin_ia32_minss_mask_round): New builtins.
25218 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25219 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25220 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25222 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25223 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25225 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25228 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25230 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25231 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25232 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25233 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25234 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25235 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25236 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25237 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25238 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25239 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25240 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25241 (__builtin_ia32_mulss_mask_round): New builtins.
25242 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25243 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25244 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25246 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25247 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25249 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25252 2017-05-10 Julia Koval <julia.koval@intel.com>
25254 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25255 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25256 (_mm256_setr_m128i): New intrinsics.
25258 2017-05-10 Julia Koval <julia.koval@intel.com>
25260 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25261 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25262 (_mm_maskz_rcp14_ss): New intrinsics.
25263 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25264 (__builtin_ia32_rcp14ss_mask): New builtins.
25265 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25267 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25269 PR tree-optimization/51513
25270 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25271 (assert_unreachable_fallthru_edge_p): Use it.
25272 (group_case_labels_stmt): Likewise.
25273 * tree-cfg.h: Prototype it.
25274 * stmt.c: Include cfghooks.h and tree-cfg.h.
25275 (emit_case_dispatch_table) <gap_label>: New local variable.
25276 Use it to fill dispatch table gaps.
25277 Test for default_label before updating probabilities.
25278 (expand_case) <default_label>: Remove unneeded initialization.
25279 Test for unreachable default case statement and remove its edge.
25280 Set default_label accordingly.
25281 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25283 2017-05-10 Carl Love <cel@us.ibm.com>
25285 * config/rs6000/rs6000-c: Add support for built-in functions
25286 vector signed char vec_neg (vector signed char)
25287 vector signed short int vec_neg (vector short int)
25288 vector signed int vec_neg (vector signed int)
25289 vector signed long long vec_neg (vector signed long long)
25290 vector float vec_neg (vector float)
25291 vector double vec_neg (vector double)
25292 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25294 * config/rs6000/altivec.h: Add define for vec_neg
25295 * doc/extend.texi: Update the built-in documentation for the
25296 new built-in functions.
25298 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25300 PR tree-optimization/77644
25301 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25303 2017-05-10 Nathan Sidwell <nathan@acm.org>
25305 * dumpfile.h (TDI_lang_all): New.
25306 (TDF_KIND): New. Renumber others
25307 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25309 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25311 (get_dump_file_name): Adjust suffix generation.
25312 (dump_enable_all): Use TDF_KIND.
25313 * doc/invoke.texi (-fdump-lang-all): Document.
25315 * dumpfile.h: Tabify.
25317 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25320 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25321 Move member access before delete.
25323 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25325 * tree-inline.c (expand_call_inline): Split block at stmt
25328 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25331 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25332 are now unused after splitting mov{sf,sd}_hardfloat.
25333 (f32_lr2): Likewise.
25334 (f32_lm): Likewise.
25335 (f32_lm2): Likewise.
25336 (f32_li): Likewise.
25337 (f32_li2): Likewise.
25338 (f32_lv): Likewise.
25339 (f32_sr): Likewise.
25340 (f32_sr2): Likewise.
25341 (f32_sm): Likewise.
25342 (f32_sm2): Likewise.
25343 (f32_si): Likewise.
25344 (f32_si2): Likewise.
25345 (f32_sv): Likewise.
25346 (f32_dm): Likewise.
25347 (f32_vsx): Likewise.
25348 (f32_av): Likewise.
25349 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25350 For movsf, order stores so the VSX stores occur before the GPR
25351 store which encourages the register allocator to use a traditional
25352 FPR instead of a GPR. For movsd, order the stores so that the GPR
25353 store comes before the VSX stores to allow the power6 to work.
25354 This is due to the power6 not having a 32-bit integer store
25355 instruction from a FPR.
25356 (movsf_hardfloat): Likewise.
25357 (movsd_hardfloat): Likewise.
25359 2017-05-09 Martin Sebor <msebor@redhat.com>
25361 PR translation/80280
25362 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25365 PR translation/80280
25366 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25367 data member added in r247778.
25368 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25370 2017-05-09 Nathan Sidwell <nathan@acm.org>
25372 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25374 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25377 2017-05-09 Marek Polacek <polacek@redhat.com>
25379 * doc/invoke.texi: Fix typo.
25381 2017-05-09 Richard Biener <rguenther@suse.de>
25383 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25384 (vrp_val_is_min): Likewise.
25385 (set_value_range_to_value): Likewise.
25386 (set_value_range_to_nonnegative): Likewise.
25387 (gimple_assign_nonzero_p): Likewise.
25388 (gimple_stmt_nonzero_p): Likewise.
25389 (vrp_int_const_binop): Likewise. Remove unreachable case.
25390 (adjust_range_with_scev): Adjust comments.
25391 (compare_range_with_value): Likewise.
25392 (extract_range_from_phi_node): Likewise.
25393 (test_for_singularity): Likewise.
25395 2017-05-09 Richard Biener <rguenther@suse.de>
25397 * tree-vrp.c (get_single_symbol): Add assert that we don't
25398 get overflowed constants as invariant part.
25399 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25400 checks. Use wi::cmp instead of recursing for integer constants.
25401 (compare_values): Just ignore whether we assumed undefined
25402 overflow instead of failing the compare.
25403 (extract_range_for_var_from_comparison_expr): Add comment before the
25404 TREE_NO_WARNING sets.
25405 (test_for_singularity): Likewise.
25406 (extract_range_from_comparison): Do not disable optimization
25407 when we assumed undefined overflow.
25408 (extract_range_basic): Remove init of unused var.
25410 2017-05-09 Richard Biener <rguenther@suse.de>
25412 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25413 (extract_range_from_multiplicative_op_1): Adjust.
25414 (extract_range_from_binary_expr_1): Use int_const_binop.
25416 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25419 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25420 rs6000_store_data_bypass_p in seven define_bypass directives and
25421 in several comments.
25422 * config/rs6000/rs6000-protos.h: Add prototype for
25423 rs6000_store_data_bypass_p function.
25424 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25425 function implements slightly different (rs6000-specific) semantics
25426 than store_data_bypass_p, returning false rather than aborting
25427 with assertion error when arguments do not satisfy the
25428 requirements of store data bypass.
25429 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25430 rs6000_store_data_bypass_p.
25432 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25434 * config/xtensa/xtensa-protos.h
25435 (xtensa_initial_elimination_offset): New declaration.
25436 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25437 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25438 macro definition, add case for FRAME_POINTER_REGNUM when
25439 FRAME_GROWS_DOWNWARD.
25440 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25441 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25442 xtensa_initial_elimination_offset.
25444 2017-05-08 Nathan Sidwell <nathan@acm.org>
25446 * doc/invoke.texi: Alphabetize -fdump options.
25448 2017-05-08 Martin Sebor <msebor@redhat.com>
25450 PR translation/80280
25451 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25453 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25455 * target.def (compute_frame_layout): New optional target hook.
25456 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25457 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25458 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25460 * reload1.c (verify_initial_elim_offsets): Likewise.
25461 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25462 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25463 (arm_get_frame_offsets): Split up into this ...
25464 (arm_compute_frame_layout): ... and this function.
25466 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25468 * config/aarch64/constraints.md (Usa): New constraint.
25469 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25471 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25473 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25474 with_multilib_list after it has been checked.
25476 2017-05-08 Richard Biener <rguenther@suse.de>
25478 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25479 (bitmap_set_subtract_values): Likewise.
25481 2017-05-08 Richard Biener <rguenther@suse.de>
25483 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25484 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25486 (gimple_stmt_nonzero_warnv_p): Rename to ...
25487 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25489 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25490 (extract_range_basic): Adjust, do not disable propagation on
25491 strict overflow sensitive simplification.
25492 (vrp_visit_cond_stmt): Likewise.
25494 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25496 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25497 body size unconditionally.
25499 2017-05-07 Jeff Law <law@redhat.com>
25502 2017-05-06 Jeff Law <law@redhat.com>
25503 PR tree-optimization/78496
25504 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25507 PR tree-optimization/78496
25508 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25509 (simplify_stmt_using_ranges): Call it.
25510 (vrp_dom_walker::before_dom_children): Extract equivalences
25511 from an ASSERT_EXPR with an equality comparison against a
25514 2017-05-06 Jeff Law <law@redhat.com>
25516 PR tree-optimization/78496
25517 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25520 PR tree-optimization/78496
25521 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25522 (simplify_stmt_using_ranges): Call it.
25523 (vrp_dom_walker::before_dom_children): Extract equivalences
25524 from an ASSERT_EXPR with an equality comparison against a
25527 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25529 * lra-constraints.c (lra_copy_reg_equiv): New function.
25530 (split_reg): Use it to copy equivalence information from the
25531 original register to the spill register.
25533 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25535 PR rtl-optimization/75964
25536 * simplify-rtx.c (simplify_const_relational_operation): Remove
25537 invalid handling of comparisons of integer ABS.
25539 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25541 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25542 initialize to zero.
25543 (init_regs): Remove declaration.
25544 (function_arg_advance_32): Initialize error_p as boolean variable.
25546 2017-05-05 Nathan Sidwell <nathan@acm.org>
25548 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25549 lines. Use for (;;).
25551 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25553 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25554 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25555 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25556 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25557 VF=2 that require versioning.
25559 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25561 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25564 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25566 * diagnostic.h (diagnostic_override_option_index): Convert from
25567 macro to inline function.
25569 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25571 * diagnostic.c (last_module_changed_p): New function.
25572 (set_last_module): New function.
25573 (diagnostic_report_current_module): Convert macro usage to
25574 the above functions.
25575 * diagnostic.h (diagnostic_context::last_module): Strengthen
25576 from const line_map * to const line_map_ordinary *.
25577 (diagnostic_last_module_changed): Delete macro.
25578 (diagnostic_set_last_module): Delete macro.
25580 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25582 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25583 with diagnostic_report_diagnostic.
25584 (diagnostic_n_impl_richloc): Likewise.
25585 * diagnostic.h (report_diagnostic): Delete macro.
25586 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25587 with diagnostic_report_diagnostic.
25588 * substring-locations.c (format_warning_va): Likewise.
25590 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25592 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25593 save/restor of format_spec. Move option-printing code to...
25594 (print_option_information): ...this new function, and
25595 reimplement by simply printing to the pretty_printer,
25596 rather than appending to the format string.
25598 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25600 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25601 handling logic into...
25602 (update_effective_level_from_pragmas): ...this new function.
25604 2017-05-04 Andrew Waterman <andrew@sifive.com>
25606 * config/riscv/riscv.opt (mstrict-align): New option.
25607 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25608 (SLOW_UNALIGNED_ACCESS): Define.
25609 (riscv_slow_unaligned_access): Declare.
25610 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25612 (riscv_slow_unaligned_access): New variable.
25613 (rocket_tune_info): Set slow_unaligned_access to true.
25614 (optimize_size_tune_info): Set slow_unaligned_access to false.
25615 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25616 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25617 (riscv_option_override): Set riscv_slow_unaligned_access.
25618 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25620 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25622 * config/riscv/riscv.md: Unify indentation.
25624 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25629 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25631 (extendsi<mode>2): Add support for doing sign extension via
25632 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25633 don't have ISA 3.0 instructions.
25634 (extendsi<mode>2 splitter): Likewise.
25635 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25636 generate the normal insns since SImode can now go in vector
25637 registers. Disallow the special UNSPECs needed for previous
25638 machines to hide SImode being used. Add new insns
25639 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25640 (fix_trunc<mode>si2_stfiwx): Likewise.
25641 (fix_trunc<mode>si2_internal): Likewise.
25642 (fixuns_trunc<mode>si2): Likewise.
25643 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25644 (fctiw<u>z_<mode>_smallint): Likewise.
25645 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25646 of floating point to 32-bit integer from doing a direct move to
25647 the GPR registers to do a store.
25648 (fctiwz_<mode>): Break long line.
25650 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25652 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25653 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25654 (addr_list, addr_offset_valid_p): New.
25655 (split_address_groups): Check offset validity with above function.
25656 (gt-tree-ssa-loop-ivopts.h): Include header file.
25658 2017-05-05 Nathan Sidwell <nathan@acm.org>
25660 * config.gcc (arm*-*-*): Add missing 'fi'.
25662 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25664 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25665 included in -fopt-info does not matter.
25666 * doc/optinfo.texi (-fopt-info): Fix description of default
25667 behavour. Explicitly say order of options included in -fopt-info
25670 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25672 * config.gcc: Allow combinations of aprofile and rmprofile values for
25673 --with-multilib-list.
25674 * config/arm/t-multilib: New file.
25675 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25676 variables. Remove setting of ISA and floating-point ABI in
25677 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25678 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25679 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25680 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25682 * config/arm/t-rmprofile: Likewise except for the matches changes.
25683 * doc/install.texi (--with-multilib-list): Document the combination of
25684 aprofile and rmprofile values and warn about pitfalls in doing that.
25686 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25688 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25689 (movdi_aarch64): Likewise.
25691 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25693 PR tree-optimization/80632
25694 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25696 (build_arrays): Initialize it for virtual phis.
25697 (fix_phi_nodes): Use it for virtual phis.
25699 PR tree-optimization/80558
25700 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25701 [x, y] op z into [x op, y op z] for op & or | if conditions
25704 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25705 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25708 * config/arm/arm.md (use_literal_pool): Remove.
25709 (64-bit immediate split): No longer takes cost into consideration
25710 if arm_disable_literal_pool is enabled.
25711 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25712 used when arm_disable_literal_pool is enabled.
25713 (arm_max_const_double_inline_cost): Remove use of
25714 arm_disable_literal_pool.
25715 (push_minipool_fix): Add assert.
25716 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25717 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25718 (no_literal_pool_sf_immediate): New.
25720 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25722 PR tree-optimization/80613
25723 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25724 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25726 2017-05-05 Richard Biener <rguenther@suse.de>
25728 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25730 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25732 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25733 of this flag from insn conditions due to removal from r247495.
25735 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25737 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25739 (arm_early_store_addr_dep_ptr): Likewise.
25740 * config/arm/aarch-common-protos.h
25741 (arm_early_load_addr_dep_ptr): Add prototype.
25742 (arm_early_store_addr_dep_ptr): Likewise.
25743 * config/arm/cortex-a53.md: Add new bypasses.
25745 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25747 * tree.c (next_type_uid): Change type to unsigned.
25748 (type_hash_canon): Decrement back next_type_uid if
25749 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25750 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25753 2017-05-04 Martin Sebor <msebor@redhat.com>
25755 * builtins.c: Fix a trivial typo in a comment.
25757 PR middle-end/79234
25758 * builtins.c (check_sizes): Adjust to handle reading past the end.
25759 Avoid printing excessive upper bound of ranges. Use %E to print
25760 tree nodes instead of converting them to %wu.
25761 (expand_builtin_memchr): New function.
25762 (compute_dest_size): Rename...
25763 (compute_objsize): ...to this.
25764 (expand_builtin_memcpy): Adjust.
25765 (expand_builtin_mempcpy): Adjust.
25766 (expand_builtin_strcat): Adjust.
25767 (expand_builtin_strcpy): Adjust.
25768 (check_strncat_sizes): Adjust.
25769 (expand_builtin_strncat): Adjust.
25770 (expand_builtin_strncpy): Adjust and simplify.
25771 (expand_builtin_memset): Adjust.
25772 (expand_builtin_bzero): Adjust.
25773 (expand_builtin_memcmp): Adjust.
25774 (expand_builtin): Handle memcmp.
25775 (maybe_emit_chk_warning): Check strncat just once.
25777 2017-05-04 Martin Sebor <msebor@redhat.com>
25779 PR preprocessor/79214
25780 PR middle-end/79222
25781 PR middle-end/79223
25782 * builtins.c (check_sizes): Add inlining context and issue
25783 warnings even when -Wno-system-headers is set.
25784 (check_strncat_sizes): Same.
25785 (expand_builtin_strncat): Same.
25786 (expand_builtin_memmove): New function.
25787 (expand_builtin_stpncpy): Same.
25788 (expand_builtin): Handle memmove and stpncpy.
25790 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25792 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25793 which is not used any more.
25795 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25797 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25799 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25801 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25802 (cortexa53_tunings): Likewise.
25803 (cortexa57_tunings): Likewise.
25804 (cortexa72_tunings): Likewise.
25805 (cortexa73_tunings): Likewise.
25807 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25809 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25810 Set loop alignment to 8.
25812 2017-05-04 Martin Sebor <msebor@redhat.com>
25814 PR translation/80280
25815 * builtins.c (expand_builtin_object_size): Add missing quoting to
25816 %D and like directives.
25817 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25818 (hsa_type_for_tree_type): Same.
25819 (verify_function_arguments): Same.
25820 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25821 * varasm.c (get_section): Same.
25824 2017-05-04 Martin Sebor <msebor@redhat.com>
25826 PR translation/80280
25827 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25829 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25831 * config/aarch64/aarch64.c (generic_addrcost_table):
25832 Change HI/TI mode setting.
25834 2017-05-04 Martin Jambor <mjambor@suse.cz>
25836 PR tree-optimization/80622
25837 * tree-sra.c (comes_initialized_p): New function.
25838 (build_accesses_from_assign): Only set write lazily when
25839 comes_initialized_p is false.
25840 (analyze_access_subtree): Use comes_initialized_p.
25841 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25842 instead of testing for PARM_DECL.
25844 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25846 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25847 constraint on operand 0 to allow more general addressing modes.
25848 Adjust output template.
25849 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25851 * config/aarch64/aarch64-protos.h
25852 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25853 * config/aarch64/constraints.md (Dp): New address constraint.
25854 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25857 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25859 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25860 update use of estimate_ipcp_clone_size_and_time.
25861 (estimate_local_effects): Update use of
25862 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25863 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25864 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25865 Return nonspecialized time.
25867 2017-05-04 Richard Biener <rguenther@suse.de>
25869 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25870 for the last VUSE which def dominates the PHI. Directly call
25872 (get_continuation_for_phi_1): Remove.
25874 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25876 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25877 to explain the use of truncating division. Cap the number of
25878 iterations to the maximum given by nb_iterations_upper_bound,
25881 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25883 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25884 * configure: Regenerate.
25885 * config.in: Regenerate.
25886 * config/i386/driver-mingw32.c: new file.
25887 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25888 * config.host: Link driver-mingw32.o on MinGW host.
25889 * doc/install.texi: Document new --enable-mingw-wildcard configure
25892 2017-05-04 Marek Polacek <polacek@redhat.com>
25894 PR tree-optimization/80612
25895 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25897 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25898 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25900 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25901 (movt splitter): Likewise.
25902 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25903 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25904 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25905 block for Thumb-1 with MOVT.
25906 (thumb2_legitimate_address_p): Move code block ...
25907 (can_avoid_literal_pool_for_label_p): ... into this new function.
25908 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25910 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25911 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25912 "M-profile targets with the MOVT instruction".
25914 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25916 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25917 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25918 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25920 2017-05-04 Martin Liska <mliska@suse.cz>
25922 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25923 variable cond_code.
25925 2017-05-04 Richard Biener <rguenther@suse.de>
25927 * tree.c (array_at_struct_end_p): Handle arrays at struct
25928 end with flexarrays more conservatively. Refactor and treat
25929 arrays of arrays or aggregates more strict. Fix
25930 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25931 * tree.c (array_at_struct_end_p): Adjust prototype.
25932 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25933 * gimple-fold.c (get_range_strlen): Likewise.
25934 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25936 2017-05-04 Richard Biener <rguenther@suse.de>
25938 PR tree-optimization/31130
25939 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25941 (supports_overflow_infinity): Likewise.
25942 (is_negative_overflow_infinity): Likewise.
25943 (is_positive_overflow_infinity): Likewise.
25944 (is_overflow_infinity): Likewise.
25945 (stmt_overflow_infinity): Likewise.
25946 (overflow_infinity_range_p): Likewise.
25947 (usable_range_p): Remove as always returning true.
25948 (make_overflow_infinity): Remove.
25949 (negative_overflow_infinity): Likewise.
25950 (positive_overflow_infinity): Likewise.
25951 (avoid_overflow_infinity): Likewise.
25952 (set_value_range): Adjust accordingly.
25953 (set_value_range_to_nonnegative): Likewise, remove now unused
25954 overflow_infinity arg.
25955 (vrp_operand_equal_p): Adjust.
25956 (update_value_range): Likewise.
25957 (range_int_cst_singleton_p): Likewise.
25958 (operand_less_p): Likewise.
25959 (compare_values_warnv): Likewise.
25960 (extract_range_for_var_from_comparison_expr): Likewise.
25961 (vrp_int_const_binop): Likewise.
25962 (zero_nonzero_bits_from_vr): Likewise.
25963 (extract_range_from_multiplicative_op_1): Likewise.
25964 (extract_range_from_binary_expr_1): Likewise.
25965 (extract_range_from_unary_expr): Likewise.
25966 (extract_range_from_comparison): Likewise.
25967 (extract_range_basic): Likewise.
25968 (adjust_range_with_scev): Likewise.
25969 (compare_ranges): Likewise.
25970 (compare_range_with_value): Likewise.
25971 (dump_value_range): Likewise.
25972 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25974 (simplify_cond_using_ranges): Adjust.
25976 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25978 * brig-builtins.def: Added a builtin for class_f64.
25979 * builtin-types.def: Added a builtin type needed by class_f64.
25981 2017-05-03 Jason Merrill <jason@redhat.com>
25983 * timevar.def: Add TV_CONSTEXPR.
25985 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25987 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25989 2017-05-03 Martin Jambor <mjambor@suse.cz>
25991 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25992 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25993 * ipa-cp.c (ipcp_propagate_stage): Do not call
25994 ipa_update_after_lto_read.
25995 * ipa-inline.c (ipa_inline): Likewise.
25997 2017-05-03 Martin Jambor <mjambor@suse.cz>
25999 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26000 tag. Added a default constructor and a destructor.
26001 (ipa_edge_args_sum_t): New class;
26002 (ipa_edge_args_sum): Declare.
26003 (ipa_edge_args_vector): Remove declaration.
26004 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26005 (ipa_free_edge_args_substructures): Remove declaration.
26006 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26007 (ipa_edge_args_info_available_for_edge_p): Likewise.
26008 * ipa-prop.c (ipa_edge_args_vector): Removed.
26009 (edge_removal_hook_holder): Likewise.
26010 (edge_duplication_hook_holder): Likewise.
26011 (ipa_edge_args_sum): New variable.
26012 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26013 ipa_edge_args_vector.
26014 (ipa_free_edge_args_substructures): Likewise.
26015 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26016 ipa_edge_args_vector.
26017 (ipa_edge_removal_hook): Turned into method
26018 ipa_edge_args_sum_t::remove.
26019 (ipa_edge_duplication_hook): Turned into method
26020 ipa_edge_args_sum_t::duplicate.
26021 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26022 registering edge hooks.
26023 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26024 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26025 ipa_edge_args_sum instead of ipa_edge_args_vector.
26026 * ipa-profile.c (ipa_profile): Likewise.
26028 2017-05-03 Martin Jambor <mjambor@suse.cz>
26030 * symbol-summary.h (function_summary): New method exists.
26031 (function_summary::symtab_removal): Deallocate through release.
26032 (call_summary): New class.
26033 (gt_ggc_mx): New overload.
26034 (gt_pch_nx): Likewise.
26035 (gt_pch_nx): Likewise.
26037 2017-05-03 Jeff Law <law@redhat.com>
26039 PR tree-optimization/78496
26040 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26041 from simplify_cond_using_ranges. Split off code to walk
26042 backwards through casts into ...
26043 (simplify_cond_using_ranges_2): New function.
26044 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26045 (execute_vrp): After identifying jump threads, call
26046 simplify_cond_using_ranges_2.
26048 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26051 * ipa-inline.h (inline_summary): Add ctor.
26052 (create_ggc): Do not use ggc_cleared_alloc.
26054 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26055 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26057 * gcc.c (handle_braces): Support escaping in switch matching
26059 * doc/invoke.texi (Spec Files): Document it.
26060 Remove superfluous @code markup in items.
26062 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26064 * diagnostic-show-locus.c (struct column_range): New struct.
26065 (get_affected_columns): New function.
26066 (get_printed_columns): New function.
26067 (struct correction): New struct.
26068 (correction::ensure_capacity): New function.
26069 (correction::ensure_terminated): New function.
26070 (struct line_corrections): New struct.
26071 (line_corrections::~line_corrections): New dtor.
26072 (line_corrections::add_hint): New function.
26073 (layout::print_trailing_fixits): Reimplement in terms of the new
26075 (selftest::test_overlapped_fixit_printing): New function.
26076 (selftest::diagnostic_show_locus_c_tests): Call it.
26078 2017-05-03 Nathan Sidwell <nathan@acm.org>
26080 Canonicalize canonical type hashing
26081 * tree.h (type_hash_canon_hash): Declare.
26082 * tree.c (type_hash_list, attribute_hash_list): Move into
26083 type_hash_canon_hash.
26084 (build_type_attribute_qual_variant): Break out hash code calc into
26085 type_hash_canon_hash.
26086 (type_hash_canon_hash): New. Generic type hash computation.
26087 (build_range_type_1, build_array_type_1, build_function_type,
26088 build_method_type_directly, build_offset_type, build_complex_type,
26089 make_vector_type): Call it.
26091 2017-05-03 Richard Biener <rguenther@suse.de>
26093 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26094 When all DRs have unknown misaligned do not always peel
26095 when there is a store but apply the same costing model as if
26096 there were only loads.
26098 2017-05-03 Richard Biener <rguenther@suse.de>
26101 PR tree-optimization/80492
26102 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26103 compare_base_decls returning dont-know properly.
26105 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26107 * config/arm/iterators.md (CCSI): New mode iterator.
26108 (arch): New mode attribute.
26109 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26110 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26111 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26112 code iterator for success result mode.
26113 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26114 the corresponding new insn generators.
26116 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26119 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26120 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26122 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26124 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26125 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26126 (DDR_A): Wrap DDR argument in brackets.
26127 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26128 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26129 (DDR_REVERSED_P): Likewise.
26131 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26133 PR tree-optimization/79472
26134 * tree-switch-conversion.c (struct switch_conv_info): Add
26135 contiguous_range and default_case_nonstandard fields.
26136 (collect_switch_conv_info): Compute contiguous_range and
26137 default_case_nonstandard fields, don't clear final_bb if
26138 contiguous_range and only the default case doesn't have the required
26140 (check_all_empty_except_final): Set default_case_nonstandard instead
26141 of failing if contiguous_range and the default case doesn't have empty
26143 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26144 and only the default case doesn't have the required constants. Skip
26146 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26147 if default_case_nonstandard.
26148 (build_constructors): Build constant 1 just once. Assert that default
26149 values aren't inserted in between cases if contiguous_range. Skip
26151 (build_arrays): Skip virtual phis.
26152 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26153 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26154 Handle virtual phis.
26155 (gen_inbound_check): Handle default_case_nonstandard case.
26156 (process_switch): Adjust check_final_bb caller. Call
26157 gather_default_values with the first non-default case instead of
26158 default case if default_case_nonstandard.
26160 2017-05-02 Nathan Sidwell <nathan@acm.org>
26162 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26163 check. Fix formatting.
26165 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26167 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26168 errors when comparing specialized and unspecialized times.
26170 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26172 * diagnostic-show-locus.c
26173 (layout::should_print_annotation_line_p): Make private.
26174 (layout::print_annotation_line): Make private.
26175 (layout::annotation_line_showed_range_p): Make private.
26176 (layout::show_ruler): Make private.
26177 (layout::print_source_line): Make private. Pass in line and
26178 line_width, rather than calling location_get_source_line. Drop
26180 (layout::print_leading_fixits): New method.
26181 (layout::print_any_fixits): Rename to...
26182 (layout::print_trailing_fixits): ...this, and make private.
26183 Don't print newline fixits.
26184 (diagnostic_show_locus): Move logic for printing one row into...
26185 (layout::print_line): ...this new function. Move the
26186 location_get_source_line call and error-handling from
26187 print_source_line to here. Call print_leading_fixits, and rename
26188 print_any_fixits to print_trailing_fixits.
26189 (selftest::test_fixit_insert_containing_newline): Update now that
26190 newlines are partially supported.
26191 (selftest::test_fixit_insert_containing_newline_2): New test.
26192 (selftest::test_fixit_replace_containing_newline): Update comments.
26193 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26194 * edit-context.c (class added_line): New class.
26195 (class edited_line): Describe newline handling in comment.
26196 (edited_line::actually_edited_p): New method.
26197 (edited_line::print_content): Delete redundant decl.
26198 (edited_line::m_predecessors): New field.
26199 (edited_file::print_content): Call edited_line::print_content.
26200 (edited_file::print_diff): Update to support newlines.
26201 (edited_file::print_diff_hunk): Likewise.
26202 (edited_file::print_run_of_changed_lines): New function.
26203 (edited_file::print_diff_line): Convert to...
26204 (print_diff_line): ...this.
26205 (edited_file::get_effective_line_count): New function.
26206 (edited_line::edited_line): Initialize new field m_predecessors.
26207 (edited_line::~edited_line): Clean up m_predecessors.
26208 (edited_line::apply_fixit): Handle newlines.
26209 (edited_line::get_effective_line_count): New function.
26210 (edited_line::print_content): New function.
26211 (edited_line::print_diff_lines): New function.
26212 (selftest::test_applying_fixits_insert_containing_newline): New
26214 (selftest::test_applying_fixits_replace_containing_newline): New
26216 (selftest::insert_line): New function.
26217 (selftest::test_applying_fixits_multiple_lines): Add example of
26219 (selftest::edit_context_c_tests): Call the new tests.
26221 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26223 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26224 parameter cand. Update dump information.
26225 (get_computation_cost): Update uses.
26227 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26229 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26230 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26231 (get_computation_at, rewrite_use_address): Update use of
26232 get_computation_aff.
26234 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26236 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26237 (get_computation): Delete.
26238 (get_computation_cost): Implement like get_computation_cost_at.
26239 Use get_computation_at.
26240 (get_computation_cost_at): Delete.
26241 (rewrite_use_nonlinear_expr): Use get_computation_at.
26242 (rewrite_use_compare, remove_unused_ivs): Ditto.
26244 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26246 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26248 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26250 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26251 (ivopts_global_cost_for_size): Rename parameter and update uses.
26252 (iv_ca_recount_cost): Update uses.
26253 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26254 candidates seperately in n_invs and n_cands.
26255 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26257 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26259 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26260 (find_inv_vars_cb): New.
26261 (find_depends): Renamed to ...
26262 (find_inv_vars): ... this.
26263 (add_candidate_1, force_var_cost): Call find_inv_vars.
26264 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26266 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26268 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26269 inv_vars. Add inv_exprs.
26270 (struct iv_cand): Rename depends_on to inv_vars.
26271 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26272 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26273 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26274 (dump_cand): Dump inv_vars.
26275 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26276 (record_invariant, find_depends, add_candidate_1): Ditto.
26277 (set_group_iv_cost, force_var_cost): Ditto.
26278 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26279 (get_computation_cost_at, get_computation_cost): Ditto.
26280 (determine_group_iv_cost_generic): Ditto.
26281 (determine_group_iv_cost_address): Ditto.
26282 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26283 (determine_group_iv_costs): Ditto.
26284 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26285 (iv_ca_set_remove_invariants): Renamed to ...
26286 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26287 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26288 (iv_ca_set_add_invariants): Renamed to ...
26289 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26290 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26291 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26292 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26293 (create_new_ivs): Remove useless dump.
26295 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26297 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26299 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26300 (iv_ca_set_no_cp, create_new_iv): Ditto.
26302 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26304 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26306 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26308 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26309 function tree_check2.
26311 2017-05-02 Martin Liska <mliska@suse.cz>
26313 * doc/gcov.texi: Add missing preposition.
26314 * gcov.c (function_info::function_info): Properly fill up
26315 all member variables.
26317 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26319 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26321 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26323 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26325 2017-05-02 Martin Liska <mliska@suse.cz>
26328 * lto-streamer-in.c (lto_read_tree_1): Remove
26329 LTO_STREAMER_DEBUG.
26330 * lto-streamer.c (struct tree_hash_entry): Likewise.
26331 (struct tree_entry_hasher): Likewise.
26332 (tree_entry_hasher::hash): Likewise.
26333 (tree_entry_hasher::equal): Likewise.
26334 (lto_streamer_init): Likewise.
26335 (lto_orig_address_map): Likewise.
26336 (lto_orig_address_get): Likewise.
26337 (lto_orig_address_remove): Likewise.
26338 * lto-streamer.h: Likewise.
26339 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26340 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26342 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26344 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26345 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26346 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26347 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26348 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26349 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26350 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26351 (mm_maskz_sub_ss): New intrinsics.
26352 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26353 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26354 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26355 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26356 (__builtin_ia32_subss_mask_round): New builtins.
26357 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26358 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26359 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26361 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26362 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26364 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26367 2017-05-02 Martin Jambor <mjambor@suse.cz>
26369 PR tree-optimization/78687
26370 * tree-sra.c (access): New field parent.
26371 (process_subtree_disqualification): New function.
26372 (disqualify_candidate): Call it.
26373 (build_accesses_from_assign): Reset write flag if creating an
26375 (build_access_subtree): Fill in parent field and also prpagate
26376 down grp_write flag.
26377 (create_artificial_child_access): New parameter set_grp_write, set
26378 grp_write to its value.
26379 (propagate_subaccesses_across_link): Also propagate grp_write flag
26381 (propagate_all_subaccesses): Push the closest parent back to work
26382 queue if add_access_to_work_queue returned true.
26384 2017-05-02 Richard Biener <rguenther@suse.de>
26386 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26387 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26388 -fstrict-overflow documentation.
26389 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26390 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26391 flag_strict_overflow.
26392 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26393 * lto-opts.c (lto_write_options): Do not stream it.
26394 * lto-wrapper.c (merge_and_complain): Do not handle it.
26395 * opts.c (default_options_table): Do not set -fstrict-overflow.
26396 (finish_options): Likewise do not clear it when sanitizing.
26397 * simplify-rtx.c (simplify_const_relational_operation): Do not
26398 test flag_strict_overflow.
26400 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26402 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26403 using enabled attribute.
26404 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26405 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26406 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26407 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26408 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26409 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26410 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26411 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26412 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26413 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26415 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26417 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26419 2017-05-02 Richard Biener <rguenther@suse.de>
26421 PR tree-optimization/80591
26423 2017-04-10 Richard Biener <rguenther@suse.de>
26425 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26428 2017-05-02 Richard Biener <rguenther@suse.de>
26430 PR tree-optimization/80549
26431 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26432 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26433 headers if they do not have a preheader.
26435 2017-05-02 Martin Liska <mliska@suse.cz>
26438 * common.opt: Fix typo.
26439 * doc/invoke.texi: Likewise.
26441 2017-05-01 Jan Beulich <jbeulich@suse.com>
26443 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26444 swapping, add (x,x,m,x,n) alternative.
26446 2017-05-01 Nathan Sidwell <nathan@acm.org>
26448 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26449 unnecessary unadjusted_alignment check.
26451 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26454 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26456 * gimplify.c (gimplify_cilk_detach): New function.
26457 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26458 * tree-core.h: Document EXPR_CILK_SPAWN.
26459 * tree.h (EXPR_CILK_SPAWN): Define.
26461 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26463 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26464 to use new fixit_hint representation, using the "replace" logic.
26465 (get_line_span_for_fixit_hint): Likewise.
26466 (layout::print_any_fixits): Likewise.
26467 (selftest::test_one_liner_many_fixits): Rename to...
26468 (selftest::test_one_liner_many_fixits_1): ...this, and update
26469 comment and expected output to reflect that the multiple fix-it
26470 hints are now consolidated into one insertion.
26471 (selftest::test_one_liner_many_fixits_2): New test.
26472 (selftest::test_diagnostic_show_locus_one_liner): Update for
26474 (selftest::test_fixit_consolidation): Update for fix-it API
26476 * diagnostic.c (print_parseable_fixits): Likewise.
26477 * edit-context.c (edited_line::m_line_events): Convert from
26478 auto_vec <line_event *> to auto_vec <line_event>.
26479 (class line_event): Convert from abstract base class to a concrete
26480 class, taking over the role of replace_event.
26481 (class insert_event): Delete.
26482 (class replace_event): Rename to class line_event. Convert to
26484 (edit_context::add_fixits): Reimplement.
26485 (edit_context::apply_insert): Delete.
26486 (edit_context::apply_replace): Rename to...
26487 (edit_context::apply_fixit): ...this. Convert to half-open range.
26488 (edited_file::apply_insert): Delete.
26489 (edited_file::apply_replace): Rename to...
26490 (edited_file::apply_fixit): ...this.
26491 (edited_line::~edited_line): Drop deletion of events.
26492 (edited_line::apply_insert): Delete.
26493 (edited_line::apply_replace): Rename to...
26494 (edited_line::apply_fixit): ...this. Convert to half-open range.
26495 Update for change to type of m_line_events.
26496 * edit-context.h (edit_context::apply_insert): Delete.
26497 (edit_context::apply_replace): Rename to...
26498 (edit_context::apply_fixit): ...this.
26500 2017-05-01 Martin Sebor <msebor@redhat.com>
26502 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26505 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26508 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26509 __get_cpuid_max returns 0.
26510 (__get_cpuid_count): Ditto.
26512 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26514 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26515 replacement expression is another instance of one of its arguments.
26517 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26520 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26521 check for stack push/pop autoinc.
26522 * config/i386/i386.c (ix86_agi_dependent): Return false
26523 if the only reason why modified_in_p returned true is that
26524 addr is SP based and set_insn is a push or pop.
26526 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26528 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26531 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26534 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26535 (account_size_time): Use two predicates - exec_pred and
26537 (evaluate_conditions_for_known_args): Compute both clause and
26539 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26540 (inline_summary_t::duplicate): Update.
26541 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26543 (compute_inline_parameters): Likewise.
26544 (estimate_edge_size_and_time): Update caluclation of time.
26545 (estimate_node_size_and_time): Compute both time and nonspecialized
26547 (estimate_ipcp_clone_size_and_time): Update.
26548 (inline_merge_summary): Update.
26549 (do_estimate_edge_time): Update.
26550 (do_estimate_edge_size): Update.
26551 (do_estimate_edge_hints): Update.
26552 (inline_read_section, inline_write_summary): Stream both new predicates.
26553 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26555 (compute_inlined_call_time): Cleanup.
26556 (big_speedup_p): Update.
26557 (edge_badness): Update.
26558 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26559 (size_time_entry): Replace predicate by exec_predicate and
26560 nonconst_predicate.
26561 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26562 (estimate_edge_time): Return also nonspec_time.
26563 (reset_edge_growth_cache): Update.
26565 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26567 PR rtl-optimization/80491
26568 * ifcvt.c (noce_process_if_block): When looking for x setter
26569 with missing else_bb, don't check only the insn right before
26570 cond_earliest, but look for the last insn that x is modified in
26571 within the same bb.
26573 PR rtl-optimization/80491
26574 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26576 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26578 PR tree-optimization/80487
26579 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26581 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26583 PR tree-optimization/79697
26584 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26585 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26586 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26588 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26589 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26591 2017-04-28 Martin Sebor <msebor@redhat.com>
26593 PR tree-optimization/80523
26594 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26595 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26597 (maybe_warn, format_directive, parse_directive): Use new functions.
26598 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26600 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26602 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26604 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26606 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26607 target_header_dir): Set correctly.
26608 * configure: Regenerated.
26609 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26610 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26611 instead of SYSTEM_HEADER_DIR.
26613 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26615 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26616 (estimate_local_effects): Likewise.
26617 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26618 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26619 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26620 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26621 do_estimate_edge_time, estimate_edge_time): Likewise.
26622 * ipa-inline-analysis.c (estimate_node_size_and_time,
26623 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26624 (estimate_time_after_inlining): Remove.
26626 2017-04-28 Martin Liska <mliska@suse.cz>
26628 * doc/gcov.texi: Enhance documentation of gcov.
26630 2017-04-28 Martin Liska <mliska@suse.cz>
26632 * doc/gcov.texi: Sort options in alphabetic order.
26633 * doc/gcov-dump.texi: Likewise.
26634 * doc/gcov-tool.texi: Likewise.
26635 * gcov.c (print_usage): Likewise.
26636 * gcov-dump.c (print_usage): Likewise.
26637 * gcov-tool.c (print_merge_usage_message): Likewise.
26638 (print_rewrite_usage_message): Likewise.
26639 (print_overlap_usage_message): Likewise.
26641 2017-04-28 Martin Liska <mliska@suse.cz>
26643 PR gcov-profile/53915
26644 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26646 2017-04-28 Martin Liska <mliska@suse.cz>
26648 PR gcov-profile/79891
26649 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26650 is marked by compiler as living on a line.
26651 (get_cycles_count): Remove usage of the union.
26652 (output_intermediate_file): Likewise.
26653 (find_source): Fix GNU coding style.
26654 (accumulate_line_counts): Remove old non-all block mode.
26655 (output_lines): Remove usage of the union.
26656 * profile.c (output_location): Include all BBs, even if
26657 belonging to a same line (and file) as a previous BB.
26659 2017-04-28 Martin Liska <mliska@suse.cz>
26661 * gcov.c (process_args): Handle new argument 'w'.
26662 (read_graph_file): Assign ID to BBs.
26663 (output_branch_count): Display BB # if verbose flag is set.
26664 (output_lines): Likewise for arcs.
26665 (print_usage): Add '--verbose' option help.
26666 * doc/gcov.texi: Document --verbose (-w) option.
26668 2017-04-28 Martin Liska <mliska@suse.cz>
26670 * gcov.c (struct block_location_info): New struct.
26671 (process_file): Fill up the new structure.
26672 (read_graph_file): Replace usage of encoding by the newly added
26674 (add_line_counts): Likewise.
26675 (accumulate_line_counts): Remove usage of the union.
26676 (function_info::function_info): New function.
26677 (function_info::~function_info): Likewise.
26678 (process_file): Call delete instead of release_function.
26679 (release_function): Release the function.
26680 (release_structures): Call delete instead of release_function.
26681 (solve_flow_graph): Replace usage of num_blocks.
26682 (find_exception_blocks): Likewise.
26683 (output_lines): Fix GNU coding style.
26685 2017-04-28 Martin Liska <mliska@suse.cz>
26688 * coverage.c (coverage_remove_note_file): New function.
26689 * coverage.h: Declare the function.
26690 * toplev.c (finalize): Clean if an error has been seen.
26692 2017-04-28 Martin Liska <mliska@suse.cz>
26694 PR gcov-profile/80031
26695 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26696 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26697 * gcov.c (read_graph_file): Read just number of blocks.
26698 * profile.c (branch_prob): Do not stream 0 flags per a basic
26701 2017-04-28 Martin Liska <mliska@suse.cz>
26703 * gcov-dump.c (tag_*): Add new argument to declarations.
26704 (dump_gcov_file): Likewise.
26705 (tag_blocks): Add and use new argument depth.
26706 (tag_arcs): Likewise.
26707 (tag_lines): Likewise.
26708 (tag_counters): Likewise.
26709 (tag_summary): Likewise.
26710 (dump_working_sets): Use depth to do a proper indentation.
26712 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26715 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26716 * symtab.c (symtab_node::debug_symtab): Move definition here.
26718 2017-04-28 Richard Biener <rguenther@suse.de>
26720 * lto-streamer.h (LTO_major_version): Bump to 7.
26722 2017-04-28 Richard Biener <rguenther@suse.de>
26724 * tree-vrp.c (assert_info): New struct.
26725 (add_assert_info): New helper.
26726 (register_edge_assert_for_2): Refactor to add asserts to a vector
26728 (register_edge_assert_for_1): Likewise.
26729 (register_edge_assert_for): Likewise.
26730 (finish_register_edge_assert_for): New helper actually registering
26731 asserts where live on edge.
26732 (find_conditional_asserts): Adjust.
26733 (find_switch_asserts): Likewise.
26734 (evrp_dom_walker::try_find_new_range): Generalize.
26735 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26737 2017-04-27 Marek Polacek <polacek@redhat.com>
26740 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26741 arg10 and arg11 to itype.
26743 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26745 * doc/extend.texi (Object Size Checking): Improve grammar.
26747 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26750 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26751 that the logic for permitting reciprocal estimates matches that
26754 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26757 * tree.c (type_cache_hasher::equal): Only compare
26758 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26759 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26760 non-aggregate element types.
26761 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26762 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26764 2017-04-27 Richard Biener <rguenther@suse.de>
26766 PR middle-end/80533
26767 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26768 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26769 keeping a reference to a trailing array.
26771 2017-04-27 Richard Biener <rguenther@suse.de>
26773 PR middle-end/80539
26774 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26775 being in loop-closed SSA form conservatively.
26776 (chrec_fold_multiply_poly_poly): Likewise.
26778 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26780 PR middle-end/79665
26781 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26782 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26784 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26787 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26788 (aarch64_function_arg_alignment): Return unsigned int again, but still
26789 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26790 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26791 Don't emit -Wpsabi note.
26792 (aarch64_function_arg_boundary): Likewise.
26793 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26796 2017-04-26 Nathan Sidwell <nathan@acm.org>
26798 * tree.h (crc32_unsigned_n): Declare.
26799 (crc32_unsigned, crc32_unsigned): Make inline.
26800 * tree.c (crc32_unsigned_bits): Replace with ...
26801 (crc32_unsigned_n): ... this.
26802 (crc32_unsigned, crc32_byte): Remove.
26803 (crc32_string): Remove unnecessary braces.
26805 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26807 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26808 * ipa-inline-analysis.c (MAX_TIME): Remove.
26809 (account_size_time): Use sreal for time.
26810 (dump_inline_summary): Update.
26811 (estimate_function_body_sizes): Update.
26812 (estimate_edge_size_and_time): Update.
26813 (estimate_calls_size_and_time): Update.
26814 (estimate_node_size_and_time): Update.
26815 (inline_merge_summary): Update.
26816 (inline_update_overall_summary): Update.
26817 (estimate_time_after_inlining): Update.
26818 (inline_read_section): Update.
26819 (inline_write_summary): Update.
26820 * ipa-inline.c (compute_uninlined_call_time): Update.
26821 (compute_inlined_call_time): Update.
26822 (recursive_inlining): Update.
26823 (inline_small_functions): Update.
26824 (dump_overall_stats): Update.
26825 * ipa-inline.h: Include sreal.h.
26826 (size_time_entry): Turn time to sreal.
26827 (inline_summary): Turn self_time nad time to sreal.
26829 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26831 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26833 (sreal::stream_out, sreal::stream_in): New.
26834 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26836 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26838 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26841 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26844 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26846 (dimode_scalar_chain::compute_convert_gain): Ditto.
26847 (dimode_scalar_chain::make_vector_copies): Ditto.
26848 (dimode_scalar_chain::convert_reg): Ditto.
26849 (dimode_scalar_chain::convert_insn): Ditto.
26850 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26851 (VI248_AVX512BW_1): New mode iterator.
26852 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26853 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26856 2017-04-25 Martin Sebor <msebor@redhat.com>
26858 PR tree-optimization/80497
26859 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26860 constants are representable in HOST_WIDE_INT.
26861 (parse_directive): Ditto.
26863 2017-04-25 Martin Sebor <msebor@redhat.com>
26866 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26867 (new_zero_array): Adjust signature.
26868 (dom_info::dom_init): Used unsigned rather that size_t.
26869 (dom_info::dom_info): Same.
26871 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26872 Jakub Jelinek <jakub@redhat.com>
26875 * config/arm/arm.c: Include gimple.h.
26876 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26877 returns negative, increment ncrn only if it returned positive.
26878 (arm_needs_doubleword_align): Return int instead of bool,
26879 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26880 members, but if there is any such non-FIELD_DECL
26881 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26882 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26883 returns negative, increment nregs only if it returned positive.
26884 (arm_setup_incoming_varargs): Likewise.
26885 (arm_function_arg_boundary): Emit -Wpsabi note if
26886 arm_needs_doubleword_align returns negative, return
26887 DOUBLEWORD_ALIGNMENT only if it returned positive.
26889 2017-04-25 Marek Polacek <polacek@redhat.com>
26892 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26893 first argument to type.
26895 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26898 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26899 type checks to test for compatibility instead of equality.
26901 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26902 Jakub Jelinek <jakub@redhat.com>
26905 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26907 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26908 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26909 the alignment computation, but return their maximum in warn_alignment.
26910 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26911 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26913 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26914 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26917 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26919 * config/arc/simdext.md (dmpyh): Fix typo.
26921 2017-04-25 Richard Biener <rguenther@suse.de>
26923 PR tree-optimization/80492
26924 * alias.c (compare_base_decls): Handle registers with asm
26925 specification conservatively.
26926 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26927 compare_base_decls returning dont-know properly.
26929 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26931 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26932 (legitimate_offset_address_p): New function.
26933 (arc_legitimate_address_p): Use above function.
26935 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26937 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26939 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26941 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26942 ACCH registers whenever they are available.
26944 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26946 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26947 double regs fix when not used.
26949 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26951 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26953 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26954 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26956 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26958 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26959 when emitting short ADD instructions.
26961 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26963 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26965 (cmpsi_cc_c_insn): Likewise.
26966 (cbranchsi4_scratch): Compute proper instruction length using
26967 compact_hreg_operand.
26968 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26970 2017-04-25 Richard Biener <rguenther@suse.de>
26972 PR middle-end/80509
26973 * passes.c (pass_manager::pass_manager): Initialize
26974 m_name_to_pass_map.
26976 2017-04-25 Richard Biener <rguenther@suse.de>
26978 PR tree-optimization/79201
26979 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26981 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26984 * config/s390/vector.md: Split MEM->GPR vector moves for
26985 non-s_operand addresses.
26987 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26990 * config/s390/predicates.md (reload_const_wide_int_operand): New
26992 * config/s390/s390.md ("movti"): Remove d/P alternative.
26993 ("movti_bigconst"): New pattern definition.
26995 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26998 * s390-protos.h (s390_expand_cs_hqi): Removed.
26999 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27000 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27001 modes as well as CCZ1mode and CCZmode.
27002 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27003 signature of s390_emit_compare_and_swap.
27004 (s390_expand_cs_hqi): Likewise, make static.
27005 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27006 compare-and-swap, in some cases.
27007 (s390_expand_cs): Wrapper function.
27008 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27010 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27011 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27012 patterns for small and large integers. Forbid symref memory operands.
27013 Move expander to s390.c. Require cc register.
27014 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27015 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27016 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27017 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27018 symref memory operands. Remove CC mode and call s390_match_ccmode
27020 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27022 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27024 * config/s390/s390.md (define_peephole2): New peephole to help
27025 combining the load-and-test pattern with volatile memory.
27027 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27029 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27030 with CCZmode for TARGET_Z196.
27032 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27034 PR rtl-optimization/80501
27035 * combine.c (make_compound_operation_int): Set subreg_code to SET
27036 even for AND with mask of the sign bit of mode.
27038 PR rtl-optimization/80500
27039 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27040 sum's initial value.
27042 2017-04-25 Julian Brown <julian@codesourcery.com>
27043 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27045 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27047 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27049 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27051 2017-04-25 Julian Brown <julian@codesourcery.com>
27052 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27054 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27055 (thunderx2t99_sha): New Reservation.
27057 2017-04-25 Julian Brown <julian@codesourcery.com>
27058 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27060 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27061 type for 1-element load.
27063 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27065 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27067 2017-04-24 Martin Jambor <mjambor@suse.cz>
27069 PR tree-optimization/80293
27070 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27071 char arrays not totally scalarizable if it is false.
27072 (analyze_all_variable_accesses): Pass correct value in the new
27073 parameter. Add a statistics counter.
27075 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27077 PR middle-end/79931
27078 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27080 2017-04-24 Richard Biener <rguenther@suse.de>
27082 PR tree-optimization/80494
27083 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27084 out for complex types.
27086 2017-04-24 Richard Biener <rguenther@suse.de>
27088 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27089 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27090 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27091 (DFS): Adjust and never fail.
27092 (sccvn_dom_walker::fail): Remove.
27093 (sccvn_dom_walker::before_dom_children): Adjust.
27094 (run_scc_vn): Likewise and never fail.
27095 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27096 (pass_fre::execute): Likewise.
27098 2017-04-24 Richard Biener <rguenther@suse.de>
27100 PR tree-optimization/79725
27101 * tree-ssa-sink.c (statement_sink_location): Return whether
27102 failure reason was zero uses. Move that check later.
27103 (sink_code_in_bb): Deal with zero uses by removing the stmt
27106 2017-04-24 Richard Biener <rguenther@suse.de>
27109 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27110 pointer-based references.
27112 2017-04-24 Richard Biener <rguenther@suse.de>
27115 * pass_manager.h (pass_manager::operator new): Remove.
27116 (pass_manager::operator delete): Likewise.
27117 * passes.c (pass_manager::operator new): Remove.
27118 (pass_manager::operator delete): Likewise.
27119 (pass_manager::pass_manager): Zero individual pass members.
27121 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27124 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27125 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27126 Check "XEXP (src, 1)" operand here.
27127 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27128 Check "XEXP (src, 1)" operand here.
27129 (dimode_scalar_chain::make_vector_copies): Detect count register
27130 of a shift instruction. Zero extend count register from QImode
27131 to DImode to satisfy vector shift pattern count operand predicate.
27132 Substitute vector shift count operand with a DImode copy.
27133 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27136 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27138 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27139 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27140 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27141 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27142 (UNSPEC_NOREX_MEM): Remove definition.
27144 2017-04-21 Richard Biener <rguenther@suse.de>
27146 PR tree-optimization/79547
27147 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27148 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27149 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27150 without any constraints.
27152 2017-04-21 Richard Biener <rguenther@suse.de>
27154 PR tree-optimization/78847
27155 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27157 2017-04-21 Richard Biener <rguenther@suse.de>
27159 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27160 (build_distinct_type_copy): Likewise.
27161 (build_variant_type_copy): Likewise.
27162 * tree.c (build_qualified_type): Pass down mem-stat info.
27163 (build_distinct_type_copy): Likewise.
27164 (build_variant_type_copy): Likewise.
27166 2017-04-21 Richard Biener <rguenther@suse.de>
27168 PR tree-optimization/80237
27169 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27171 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27172 for a simplified result.
27174 2016-04-21 Richard Biener <rguenther@suse.de>
27176 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27177 sth as strict as a simple_iv but a chrec without symbols and an
27178 operand defined in the loop we are peeling (and not some subloop).
27179 (propagate_constants_for_unrolling): Propagate all constants.
27181 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27184 * config/i386/i386.c (print_reg): Remove assert for disalowed
27185 regno values, call output_operand_lossage instead.
27187 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27190 * config/i386/constraints.md (Yc): New register constraint.
27191 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27192 Use Yc constraint for alternative 2 of operand 0. Remove
27193 preferred_for_speed attribute.
27195 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27197 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27198 lastprivate clauses in SIMT case.
27200 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27202 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27204 2017-04-20 Richard Biener <rguenther@suse.de>
27206 PR tree-optimization/57796
27207 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27208 as N scalar stores.
27209 (vect_model_load_cost): Cost gathers as N scalar loads.
27211 2017-04-20 Richard Biener <rguenther@suse.de>
27213 * ggc-page.c (ggc_allocated_p): Rename to ...
27214 (safe_lookup_page_table_entry): ... this and return the lookup
27216 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27218 2017-04-20 Richard Biener <rguenther@suse.de>
27220 PR tree-optimization/80453
27221 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27222 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27223 from the conditions.
27224 (vn_phi_eq): Pass them down.
27225 (vn_phi_lookup): Record them.
27226 (vn_phi_insert): Likewise.
27228 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27230 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27231 uninitialized variable warning to avoid buffer overrun.
27233 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27236 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27237 is suppressed for '{ 0 }' in C.
27239 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27241 * BASE-VER: Set to 8.0.0.
27243 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27245 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27246 priority .init_array and .fini_array section with SECTION_NOTYPE
27249 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27251 PR middle-end/80423
27252 * tree.h (build_array_type): Add typeless_storage default argument.
27253 * tree.c (type_cache_hasher::equal): Also compare
27254 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27255 (build_array_type): Add typeless_storage argument, set
27256 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27258 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27259 (build_array_type): Likewise. Add typeless_storage argument.
27261 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27262 Jakub Jelinek <jakub@redhat.com>
27264 PR tree-optimization/80426
27265 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27266 operation on symbolic operands, also compute the overflow for the
27267 invariant part when the operation degenerates into a negation.
27269 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27272 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27273 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27276 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27278 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27281 * config/avr/avr.c (tree.h): Include it.
27282 (cgraph.h): Include it.
27283 (avr_encode_section_info): Don't warn for uninitialized progmem
27284 variable if it's just an alias.
27286 2017-04-19 Richard Biener <rguenther@suse.de>
27289 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27290 when needed by AutoPGO.
27292 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27295 * doc/lto.texi: Remove an extra 'that'.
27297 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27299 PR rtl-optimization/80429
27300 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27301 are only used in debug insns.
27303 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27304 Vladimir Makarov <vmakarov@redhat.com>
27306 * config/sparc/predicates.md (input_operand): Add comment. Return
27307 true for any memory operand when LRA is in progress.
27308 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27310 2017-04-18 Jeff Law <law@redhat.com>
27313 * mips.md ({return,simple_return}_internal): Do not overwrite
27316 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27318 PR tree-optimization/80443
27319 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27320 instead of adding 1, subtract -1 and similarly instead of subtracting
27323 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27325 PR rtl-optimization/80357
27326 * haifa-sched.c (tmp_bitmap): New variable.
27327 (model_recompute): Handle duplicate use records.
27328 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27329 (free_global_sched_pressure_data): Free it.
27331 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27334 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27335 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27336 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27337 instead of SYSTEM_HEADER_DIR.
27339 2017-04-18 Jeff Law <law@redhat.com>
27341 PR middle-end/80422
27342 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27343 predecessors after walking up the insn chain.
27345 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27348 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27349 sizetype type into debug info.
27351 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27354 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27355 unneeded test for TARGET_UPPER_REGS_SF.
27356 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27358 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27361 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27362 instead of gsi_after_labels.
27364 2017-04-18 Jeff Law <law@redhat.com>
27366 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27370 2017-04-13 Jeff Law <law@redhat.com>
27371 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27372 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27374 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27377 * config/avr/avr.c (intl.h): Include it.
27378 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27380 2017-04-18 Martin Liska <mliska@suse.cz>
27382 PR gcov-profile/78783
27383 * gcov-tool.c (gcov_output_files): Validate that destination
27384 file is either removed by the tool or by a user.
27386 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27387 Guy Benyei <guybe@mellanox.com>
27389 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27390 block, and do not negate it, the stored id is already negative.
27392 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27394 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27396 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27399 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27400 masks of options that should be turned off if the VSX vector
27401 options are turned off.
27402 (OTHER_P8_VECTOR_MASKS): Likewise.
27403 (OTHER_VSX_VECTOR_MASKS): Likewise.
27404 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27405 rs6000_disable_incompatible_switches to validate no type switches
27407 (rs6000_incompatible_switch): New function to disallow turning on
27408 other vector options if -mno-vsx, -mno-power8-vector, or
27409 -mno-power9-vector are specified.
27411 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27413 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27415 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27417 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27418 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27419 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27420 (ARG_POINTER_CFA_OFFSET): Likewise.
27422 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27424 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27425 conditions to take advantage of various optimizations.
27427 2017-04-13 Jeff Law <law@redhat.com>
27429 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27430 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27431 (zero_extendsidi2_dext): Likewise.
27433 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27436 * fold-const.c (fold_ternary_loc): Revert
27437 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27440 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27442 PR rtl-optimization/80343
27443 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27444 new scratch pseudo.
27446 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27449 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27450 to ubsan_encode_value.
27452 2017-04-13 Jeff Law <law@redhat.com>
27454 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27455 appearing in DEBUG_INSNs.
27457 2017-04-13 Martin Liska <mliska@suse.cz>
27459 PR gcov-profile/80413
27460 * gcov-io.c (gcov_write_string): Copy to buffer just when
27461 allocated size is greater than zero.
27463 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27466 * dwarf2out.c (decls_for_scope): Ignore declarations of
27467 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27469 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27472 * ipa-visibility.c (non_local_p): Fix typos.
27473 (localize_node): When localizing symbol in same comdat group,
27474 dissolve the group only when we know external symbols are going
27476 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27478 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27480 PR tree-optimization/79390
27481 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27482 order does not result in usable sequence, retry with reversed operand
27488 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27489 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27490 op0 instead of fold_convert_loc (loc, type, arg0).
27492 2017-04-12 Jeff Law <law@redhat.com>
27494 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27495 has a delay slot in the generated code.
27497 * config/cris/cris.md (cris_preferred_reload_class): Return
27498 GENNONACR_REGS rather than GENERAL_REGS.
27500 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27503 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27504 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27505 signedness of the result type.
27507 2017-04-12 Richard Biener <rguenther@suse.de>
27508 Jeff Law <law@redhat.com>
27510 PR tree-optimization/80359
27511 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27512 trim stores to TARGET_MEM_REFs.
27514 2017-04-12 Richard Biener <rguenther@suse.de>
27516 PR tree-optimization/79390
27517 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27518 threading case even more.
27520 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27523 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27524 for quad_address_p for TImode, instead of just not indexed_address.
27526 2017-04-12 Richard Biener <rguenther@suse.de>
27527 Bernd Edlinger <bernd.edlinger@hotmail.de>
27529 PR middle-end/79671
27530 * alias.c (component_uses_parent_alias_set_from): Handle
27531 TYPE_TYPELESS_STORAGE.
27532 (get_alias_set): Likewise.
27533 * tree-core.h (tree_type_common): Add typeless_storage flag.
27534 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27535 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27536 for types containing members with TYPE_TYPELESS_STORAGE.
27537 (place_field): Likewise.
27538 (layout_type): Likewise for ARRAY_TYPE.
27539 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27540 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27541 TYPE_TYPELESS_STORAGE.
27542 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27544 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27547 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27548 first argument to type.
27550 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27554 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27555 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27556 (rs6000_expand_binop_builtin): Likewise.
27557 (rs6000_expand_ternop_builtin): Likewise; also add missing
27558 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27559 vshasigma built-ins.
27560 * doc/extend.texi: Document that vec_xxpermdi's third argument
27561 must be a constant.
27563 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27565 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27566 Use shift_const cost parameter when calculating gain of STV shifts.
27568 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27570 PR rtl-optimization/70478
27571 * lra-constraints.c (process_alt_operands): Check memory for
27572 disfavoring memory insn operand.
27574 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27576 PR middle-end/80100
27577 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27578 left shift in unsigned HOST_WIDE_INT type.
27580 PR rtl-optimization/80385
27581 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27582 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27585 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27586 if they have any depend clauses.
27588 2017-04-11 Martin Liska <mliska@suse.cz>
27591 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27592 * ipa-split.c (split_function): Create a local comdat symbol
27593 if caller is in a comdat group.
27595 2017-04-11 Martin Liska <mliska@suse.cz>
27598 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27601 2017-04-11 Martin Sebor <msebor@redhat.com>
27603 PR middle-end/80364
27604 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27605 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27607 (directive::set_width, directive::set_precision, format_character):
27609 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27612 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27615 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27616 conflict, set target->arch_name instead of target->cpu_name.
27618 2017-04-11 Richard Biener <rguenther@suse.de>
27620 PR tree-optimization/80374
27621 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27622 build_zero_cst, remove fold_convertible_p check again.
27624 2017-04-11 Martin Liska <mliska@suse.cz>
27627 * ubsan.c (instrument_object_size): Do not instrument register
27630 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27633 * config/i386/i386-builtin-types.def
27634 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27635 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27636 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27637 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27638 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27639 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27640 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27641 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27642 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27643 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27644 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27645 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27646 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27647 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27648 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27649 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27650 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27651 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27652 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27653 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27654 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27655 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27656 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27657 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27658 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27659 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27660 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27661 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27662 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27663 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27664 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27665 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27666 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27667 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27668 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27669 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27670 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27671 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27672 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27673 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27674 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27675 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27676 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27677 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27678 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27679 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27681 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27682 flag to second_arg_count, handle 4 argument function type _COUNT
27683 aliases, handle second_arg_count on second argument rather than last.
27685 2017-04-10 Jeff Law <law@redhat.com>
27687 PR tree-optimization/80374
27688 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27689 record anything if we can not convert integer_zero_node to the
27692 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27695 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27696 Enhance special handling given to the TARGET_P9_MINMAX option in
27697 relation to certain other options.
27699 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27701 PR tree-optimization/80153
27702 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27703 remove POINTER_PLUS_EXPR's base part directly, rather than through
27706 2017-04-10 Richard Biener <rguenther@suse.de>
27707 Bin Cheng <bin.cheng@arm.com>
27709 PR tree-optimization/80153
27710 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27711 the first element of pointer type aff_tree. Build result expr in
27713 (add_elt_to_tree): Convert to type unconditionally. Remove other
27714 fold_convert calls.
27715 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27716 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27718 2017-04-10 Richard Biener <rguenther@suse.de>
27720 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27723 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27725 PR rtl-optimization/70478
27726 * lra-constraints.c (curr_small_class_check): New.
27727 (update_and_check_small_class_inputs): New.
27728 (process_alt_operands): Update curr_small_class_check. Disfavor
27729 alternative insn memory operands. Check available regs for small
27732 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27735 * config/mips/mips.opt (-mvirt): Update description.
27736 * doc/invoke.texi (-mvirt): Likewise.
27738 2017-04-10 Richard Biener <rguenther@suse.de>
27740 PR middle-end/80362
27741 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27742 looking for NEGATE_EXPR in -A / -B to A / B folding.
27744 2017-04-10 Martin Liska <mliska@suse.cz>
27746 PR gcov-profile/80224
27747 * gcov.c (print_usage): Fix usage string.
27748 (get_gcov_intermediate_filename): Remove.
27749 (output_gcov_file): Use both for normal and intermediate format.
27750 (generate_results): Do not initialize special file for
27751 intermediate format.
27753 2017-04-10 Richard Biener <rguenther@suse.de>
27755 PR tree-optimization/80304
27756 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27759 2017-04-10 Nathan Sidwell <nathan@acm.org>
27762 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27763 (rs6000_init_builtins): Use it.
27765 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27767 * config/arm/arm.md (<mrc>): Add mode to SET source.
27768 (<mrrc>): Likewise.
27770 2017-04-10 Richard Biener <rguenther@suse.de>
27772 PR middle-end/80344
27773 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27775 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27778 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27779 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27780 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27781 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27782 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27783 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27784 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27785 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27786 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27787 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27788 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27789 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27790 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27791 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27792 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27793 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27794 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27795 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27796 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27797 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27798 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27799 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27800 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27802 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27804 PR rtl-optimization/70478
27805 * lra-constraints.c: Reverse the last patch.
27807 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27809 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27810 Add comment for WCHAR_T.
27812 2017-04-08 Martin Liska <mliska@suse.cz>
27815 2017-04-07 Martin Liska <mliska@suse.cz>
27818 * ipa-split.c (split_function): Add function part to a same comdat
27821 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27824 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27826 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27828 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27829 (vec_pairing): ...to this.
27830 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27831 (rs6000_sched_init): Adjust for name change.
27832 (struct rs6000_sched_context): Likewise.
27833 (rs6000_init_sched_context): Likewise.
27834 (rs6000_set_sched_context): Likewise.
27836 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27842 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27844 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27845 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27846 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27848 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27850 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27852 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27854 PR rtl-optimization/70703
27855 * ira-color.c (update_conflict_hard_regno_costs): Use
27856 int64_t instead of HOST_WIDE_INT.
27858 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27860 PR rtl-optimization/70478
27861 * lra-constraints.c (process_alt_operands): Disfavor alternative
27862 insn memory operands.
27864 2017-04-07 Jeff Law <law@redhat.com>
27866 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27867 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27869 2017-04-07 Martin Liska <mliska@suse.cz>
27872 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27873 Show error message instead of an ICE.
27875 2017-04-07 Martin Liska <mliska@suse.cz>
27878 * ipa-split.c (split_function): Add function part to a same comdat
27881 2017-04-07 Richard Biener <rguenther@suse.de>
27883 PR middle-end/80341
27884 * tree.c (get_unwidened): Also handle ! for_type case for
27886 * convert.c (do_narrow): Split out from ...
27887 (convert_to_integer_1): ... here. Do not pass final truncation
27888 type to get_unwidened for TRUNC_DIV_EXPR.
27890 2017-04-07 Richard Biener <rguenther@suse.de>
27892 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27894 (aff_combination_const): Adjust.
27895 (aff_combination_scale): Likewise.
27896 (aff_combination_add_elt): Likewise.
27897 (aff_combination_add_cst): Likewise.
27898 (aff_combination_convert): Likewise.
27899 (add_elt_to_tree): Likewise. Remove unused argument.
27900 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27902 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27904 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27906 * config/arm/arm.c (arm_default_short_enums): Use
27907 ARM_DEFAULT_SHORT_ENUMS.
27908 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27910 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27913 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27914 members with redundant out-of-class redeclaration.
27916 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27919 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27920 * config/i386/i386.md (*zero_extendsidi2):
27921 Add (?*x,*x) and (?*v,*v) alternatives.
27923 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27926 * config/i386/i386.c (ix86_expand_builtin)
27927 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27928 mode from insn data. Convert operands to insn operand mode.
27929 Copy operands that don't satisfy insn predicate to a register.
27931 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27933 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27936 2017-04-06 Richard Biener <rguenther@suse.de>
27938 PR tree-optimization/80334
27939 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27940 preserve alignment of accesses.
27942 2017-04-06 Richard Biener <rguenther@suse.de>
27944 PR tree-optimization/80262
27945 * tree-sra.c (build_ref_for_offset): Preserve address-space
27947 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27948 Drop useless address-space information on MEM_REF offsets.
27950 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27952 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27954 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27956 PR rtl-optimization/70703
27957 * ira-color.c (update_conflict_hard_regno_costs): Use
27958 HOST_WIDE_INT instead of long.
27960 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27963 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27964 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27966 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27967 for x86_64 target. Handle -m3dnowa option.
27969 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27971 PR rtl-optimization/70703
27972 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27973 (update_conflict_hard_regno_costs): Use long instead of unsigned
27974 arithmetic for cost calculation.
27976 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27977 Bernd Edlinger <bernd.edlinger@hotmail.de>
27980 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27983 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27986 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27987 ptr_mode with Pmode throughout.
27988 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27989 into probe_stack_range and use DImode.
27991 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27994 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27995 call_eh_return is true.
27997 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27999 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28000 Initialize last_match_fntype_index.
28002 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28005 * tree-nvr.c: Include internal-fn.h.
28006 (pass_return_slot::execute): Ignore internal calls without
28009 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28010 Richard Biener <rguenther@suse.de>
28013 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28014 captures used multiple times, except for the last use.
28015 * generic-match-head.c: Include gimplify.h.
28017 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28019 PR tree-optimization/79390
28020 * target.h (struct noce_if_info): Declare.
28021 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28022 * target.def (noce_conversion_profitable_p): New target hook.
28023 * ifcvt.h (struct noce_if_info): New type, moved from ...
28024 * ifcvt.c (struct noce_if_info): ... here.
28025 (noce_conversion_profitable_p): Renamed to ...
28026 (default_noce_conversion_profitable_p): ... this. No longer
28028 (noce_try_store_flag_constants, noce_try_addcc,
28029 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28030 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28031 instead of noce_conversion_profitable_p.
28032 * config/i386/i386.c: Include ifcvt.h.
28033 (ix86_option_override_internal): Don't override
28034 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28035 (ix86_noce_conversion_profitable_p): New function.
28036 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28037 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28038 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28039 * doc/tm.texi: Regenerated.
28041 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28043 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28046 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28049 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28050 instructions for small multiply cores.
28052 2017-04-04 Jeff Law <law@redhat.com>
28054 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28056 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28057 that are not set by the loop over the elements.
28059 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28062 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28063 int mode, convert_modes it to mode as unsigned, otherwise use
28064 lowpart_subreg to mode rather than SImode.
28065 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28066 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28067 Use DImode instead of SImode for the shift count operand.
28068 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28071 2017-04-04 Richard Biener <rguenther@suse.de>
28073 PR middle-end/80281
28074 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28075 arithmetic done for the negate or the plus. Simplify.
28076 (A - (-B) -> A + B): Likewise.
28077 * fold-const.c (split_tree): Make sure to not negate pointers.
28079 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28081 PR rtl-optimization/60818
28082 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28083 a compare of comparisons with the thing compared if this results
28084 in a different machine mode.
28086 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28088 * alias.c (base_alias_check): Fix typo in comment.
28089 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28090 * cgraphunit.c (symbol_table::compile): Likewise.
28091 * collect2.c (maybe_run_lto_and_relink): Likewise.
28092 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28093 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28094 * config/avr/avr.c (avr_map_op_t): Likewise.
28095 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28096 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28097 * config/epiphany/epiphany.md (movcc): Likewise.
28098 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28099 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28101 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28102 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28103 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28104 * config/sh/sh.c (sh_rtx_costs): Likewise.
28105 * fold-const.c (fold_truth_andor): Likewise.
28106 * genautomata.c (collapse_flag): Likewise.
28107 * gengtype.h (struct type::u::s): Likewise.
28108 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28109 * input.c (FORMAT_AMOUNT): Likewise.
28110 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28111 (known_aggs_to_agg_replacement_list): Likewise.
28112 * ipa-inline-analysis.c: Likewise.
28113 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28114 * ipa-polymorphic-call.c
28115 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28116 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28117 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28119 * modulo-sched.c (apply_reg_moves): Likewise.
28120 * omp-expand.c (build_omp_regions_1): Likewise.
28121 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28122 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28123 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28124 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28125 * value-prof.c: Likewise.
28126 * var-tracking.c (val_reset): Likewise.
28128 2017-04-03 Richard Biener <rguenther@suse.de>
28130 PR tree-optimization/80275
28131 * fold-const.c (split_address_to_core_and_offset): Handle
28134 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28136 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28137 descriptors is at least equal to that of functions.
28139 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28141 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28143 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28146 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28147 (mov<IMOD4:mode>): New expander.
28148 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28150 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28152 PR rtl-optimization/79405
28153 * fwprop.c (propagations_left): New variable.
28154 (forward_propagate_into): Decrement it.
28155 (fwprop_init): Initialize it.
28156 (fw_prop): If the variable has reached zero, stop propagating.
28157 (fwprop_addr): Ditto.
28159 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28162 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28163 a FUNCTION_DECL, pass it as decl instead of origin to
28166 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28168 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28171 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28174 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28175 TARGET_VSX_SMALL_INTEGER.
28177 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28179 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28180 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28182 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28184 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28185 extraction from odd-numbered MSA register.
28187 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28189 PR middle-end/80173
28190 * expmed.c (store_bit_field_1): Don't attempt to create
28191 a word subreg out of hard registers wider than word if they
28192 have HARD_REGNO_NREGS of 1 for their mode.
28194 PR middle-end/80163
28195 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28196 conversions to integer types wider than word and pointer.
28199 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28200 (rtx_equal_for_cselib_p): Pass 0 to it.
28201 * cselib.c (cselib_hasher::equal): Likewise.
28202 (rtx_equal_for_cselib_1): Add depth argument. If depth
28203 is 128, don't look up VALUE locs and punt. Increment
28204 depth in recursive calls when walking VALUE locs.
28206 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28208 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28209 (make_gcov_file_name): Use the canonical path name for generating
28211 (read_line): Fix handling of files with ascii null bytes.
28213 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28215 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28216 to initialise a vector register instead
28217 of using a const_int.
28219 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28221 PR translation/80189
28222 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28223 diagnostic messages.
28225 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28228 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28229 (dfp_diex_<mode>): Update mode of operand 1.
28230 * doc/extend.texi (dxex, dxexq): Document change to return type.
28231 (diex, diexq): Document change to argument type.
28233 2017-03-30 Martin Jambor <mjambor@suse.cz>
28236 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28237 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28238 it reflects the signature changes performed at the callee side.
28239 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28240 to cgraph_build_function_type_skip_args.
28241 (build_function_decl_skip_args): Adjust call to the above function.
28243 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28246 * config/i386/sse.md
28247 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28248 register as dest whenever it is a MEM not rtx_equal_p to the
28249 corresponding dup operand, and when forcing into reg move the
28250 reg into the memory afterwards.
28251 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28252 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28253 for the force_reg mode.
28254 (avx512vl_vextractf128<mode>): Use register as dest either
28255 always when a MEM, or when it is a MEM not rtx_equal_p to the
28256 corresponding dup operand, or even not when it is a CONST_VECTOR
28257 depending on the mode and lo vs. hi.
28258 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28260 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28261 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28262 Likewise. Require that operands[2] is even.
28263 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28264 Remove extraneous parens. Require that operands[2] is a multiple
28266 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28267 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28268 disallow memory then.
28270 2017-03-30 Richard Biener <rguenther@suse.de>
28272 PR tree-optimization/77498
28273 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28274 to non-constants over backedges.
28276 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28278 PR rtl-optimization/80233
28279 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28280 as last_combined_insn. Do not test for BARRIER_P separately.
28282 2017-03-29 Andreas Schwab <schwab@suse.de>
28285 * calls.c (prepare_call_address): Convert funexp to Pmode before
28286 copying to temp reg.
28288 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28290 PR tree-optimization/80158
28291 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28292 Handle possible future case of more than one alternate
28294 (replace_rhs_if_not_dup): Likewise.
28295 (replace_one_candidate): Likewise.
28297 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28299 PR rtl-optimization/80193
28300 * ira.c (ira): Do not check allocation for LRA.
28302 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28304 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28305 (nvptx_output_simt_exit): Declare.
28306 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28307 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28308 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28309 (nvptx_declare_function_name): ...here. Emit declaration of local
28310 memory space buffer for omp_simt_enter insn.
28311 (nvptx_output_unisimt_switch): New.
28312 (nvptx_output_softstack_switch): New.
28313 (nvptx_output_simt_enter): New.
28314 (nvptx_output_simt_exit): New.
28315 * config/nvptx/nvptx.h (struct machine_function): New fields
28316 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28317 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28318 (UNSPECV_SIMT_EXIT): Ditto.
28319 (omp_simt_enter_insn): New insn.
28320 (omp_simt_enter): New expansion.
28321 (omp_simt_exit): New insn.
28322 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28324 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28325 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28326 (expand_GOMP_SIMT_EXIT): New.
28327 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28328 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28329 (GOMP_SIMT_EXIT): Ditto.
28330 * target-insns.def (omp_simt_enter): New insn.
28331 (omp_simt_exit): Ditto.
28332 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28334 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28335 (lower_rec_input_clauses): Likewise.
28336 (lower_lastprivate_clauses): Handle SIMT privatization.
28338 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28339 (ompdevlow_adjust_simt_enter): New.
28340 (find_simtpriv_var_op): New.
28341 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28342 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28344 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28345 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28346 (copy_decl_for_dup_finish): Ditto.
28348 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28350 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28353 * config/i386/i386.c (ix86_option_override_internal): Always
28354 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28356 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28358 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28360 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28362 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28363 mark new edge's irreducible flag accordign to it.
28364 (vect_do_peeling): Check loop preheader edge's irreducible flag
28365 and pass it to function slpeel_add_loop_guard.
28367 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28369 PR tree-optimization/80218
28370 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28371 Update block frequencies and counts.
28373 2017-03-28 Richard Biener <rguenther@suse.de>
28375 PR tree-optimization/78644
28376 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28377 of a simplification result we may not use it at all.
28379 2017-03-28 Richard Biener <rguenther@suse.de>
28382 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28383 without arguments, generate default definition of a SSA name.
28385 2017-03-28 Richard Biener <rguenther@suse.de>
28387 PR middle-end/80222
28388 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28389 TYPE_REF_CAN_ALIAS_ALL references.
28390 * fold-const.c (fold_indirect_ref_1): Likewise.
28392 2017-03-28 Martin Liska <mliska@suse.cz>
28395 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28396 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28398 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28399 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28401 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28402 (EXTRA_SPECS): Define.
28403 (SUBTARGET_EXTRA_SPECS): Likewise.
28404 (SUBTARGET_CPP_SPEC): Likewise.
28405 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28406 SUBTARGET_EXTRA_SPECS.
28407 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28409 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28411 * config/arc/simdext.md (vst64_insn): Update pattern.
28412 (vld32wh_insn): Likewise.
28413 (vld32wl_insn): Likewise.
28414 (vld64_insn): Likewise.
28415 (vld32_insn): Likewise.
28417 2017-03-28 Marek Polacek <polacek@redhat.com>
28420 * fold-const.c (fold_comparison): Use protected_set_expr_location
28421 instead of SET_EXPR_LOCATION.
28423 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28425 * tree.c (add_expr): Avoid name lookup warning.
28427 2017-03-27 Jeff Law <law@redhat.com>
28429 PR tree-optimization/80216
28430 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28431 function name. Limit recursion depth.
28432 (record_temporary_equivalences): Corresponding changes.
28434 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28436 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28439 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28442 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28444 * cfgcleanup.c (reg_note_cfa_p): New array.
28445 (insns_have_identical_cfa_notes): New function.
28446 (old_insns_match_p): Don't cross-jump in between /f
28447 and non-/f instructions. If both i1 and i2 are frame related,
28448 verify all CFA notes, their order and content.
28450 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28453 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28454 HImode and SImode with zero extend to DImode to one insn.
28455 (bswap<mode>2_extenddi): Likewise.
28456 (bswapsi2_extenddi): Likewise.
28457 (bswaphi2_extendsi): Likewise.
28458 (bswaphi2): Combine bswap HImode and SImode into one insn.
28459 Separate memory insns from swapping register.
28460 (bswapsi2): Likewise.
28461 (bswap<mode>2): Likewise.
28462 (bswaphi2_internal): Delete, no longer used.
28463 (bswapsi2_internal): Likewise.
28464 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28465 store, and gpr<-gpr swap insns.
28466 (bswap<mode>2_store): Likewise.
28467 (bswaphi2_reg): Register only splitter, combine with the splitter.
28468 (bswaphi2 splitter): Likewise.
28469 (bswapsi2_reg): Likewise.
28470 (bswapsi2 splitter): Likewise.
28471 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28472 the insns into load, store, and register/register insns.
28473 (bswapdi2_ldbrx): Likewise.
28474 (bswapdi2_load): Likewise.
28475 (bswapdi2_store): Likewise.
28476 (bswapdi2_reg): Likewise.
28478 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28480 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28481 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28483 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28486 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28488 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28489 special handling for target option conflicts between dform
28490 options (-mpower9-dform, -mpower9-dform-vector,
28491 -mpower9-dform-scalar) and -mno-direct-move.
28493 2017-03-27 Richard Biener <rguenther@suse.de>
28495 PR tree-optimization/80181
28496 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28498 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28500 * config/arc/predicates.md (move_double_src_operand): Replace the
28501 call to move_double_src_operand with a call to address_operand.
28503 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28505 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28506 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28507 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28509 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28511 * config/arc/predicates.md (long_immediate_loadstore_operand):
28512 Consider scaled addresses cases.
28514 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28516 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28517 restored when in interrupt.
28518 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28519 doesn't have delay slot.
28521 2017-03-27 Richard Biener <rguenther@suse.de>
28524 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28525 inlined thunk clones.
28527 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28530 * asan.c (instrument_derefs): Copy over last operand from
28531 original COMPONENT_REF to the new COMPONENT_REF with
28532 DECL_BIT_FIELD_REPRESENTATIVE.
28533 * ubsan.c (instrument_object_size): Likewise.
28535 2017-03-27 Richard Biener <rguenther@suse.de>
28537 PR tree-optimization/80170
28538 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28539 sure DR/SCEV didnt fold in constants we do not see when looking
28540 at the reference base alignment.
28542 2017-03-27 Richard Biener <rguenther@suse.de>
28544 PR middle-end/80171
28545 * gimple-fold.c (fold_ctor_reference): Properly guard against
28546 NULL return value from canonicalize_constructor_val.
28548 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28551 * config/i386/i386.c (ix86_expand_builtin)
28552 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28553 flags reg setting and flags reg using instructions.
28554 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28555 clobbering instructions to zero extend op2.
28557 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28559 * doc/install.texi (Configuration) <--with-aix-soname>:
28560 Update link to AIX ld.
28562 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28564 PR rtl-optimization/80160
28565 PR rtl-optimization/80159
28566 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28567 reg_alternate_class into account.
28569 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28572 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28573 to consider in curr_insn_transform.
28575 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28577 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28578 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28579 and emit_mode_inner.
28581 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28583 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28584 argument to the overloaded builtin variants. Use the new flag to
28585 deprecate certain builtin variants.
28586 * config/s390/s390-builtin-types.def: Add new builtin types.
28587 * config/s390/s390-builtins.h: Support new flags field for
28588 overloaded builtins.
28589 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28590 (s390_macro_to_expand): Enable vector float data type.
28591 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28592 builtins by incrementing the __VEC__ version number.
28593 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28595 (s390_resolve_overloaded_builtin): Emit error messages depending
28596 on the builtin flags.
28597 * config/s390/s390.c (s390_expand_builtin): Support additional
28598 flags argument. Change error message to match the messages
28599 emitted in s390-c.c.
28600 * config/s390/s390.md: New UNSPEC_* constants.
28601 (op_type): Add new instruction types.
28602 * config/s390/vecintrin.h: Add new builtins and test data class
28604 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28605 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28606 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28607 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28608 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28609 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28611 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28612 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28613 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28614 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28616 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28617 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28618 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28619 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28621 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28622 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28623 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28624 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28625 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28626 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28627 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28629 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28630 ("vec_scatter_element<V_HW_4:mode>_DI")
28631 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28632 ("vec_fpint<mode>", "vflls")
28633 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28634 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28635 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28636 ("*vec_cmphe<mode>_cc"): ... these.
28638 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28639 mode constant instead of magic value.
28641 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28643 * config/s390/s390.c (s390_expand_vec_compare): Support other
28644 vector floating point modes than just V2DF.
28645 (s390_expand_vcond): Likewise.
28646 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28647 (s390_cannot_change_mode_class): Prevent mode changes between TF
28648 and V1TF in vector registers.
28649 * config/s390/s390.md (DF, SF): New mode attributes.
28650 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28651 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28652 SFmode support for VRs.
28653 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28655 (VFT, VF_HW): New mode iterators.
28656 (vw, sdx): New mode attributes.
28657 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28658 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28659 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28660 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28661 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28662 also the new vector floating point modes. Renaming to ...
28664 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28665 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28666 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28667 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28668 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28669 ("vec_unordered<mode>"): ... these.
28671 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28672 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28673 ("*vec_extendv2df"): New insn definitions.
28675 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28677 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28678 ("mulditi3_2", "*muldi3_sign"): New patterns.
28679 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28680 rename the pattern definition.
28682 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28684 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28686 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28688 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28690 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28691 instruction if possible.
28692 * config/s390/vector.md (vec_halfnumelts): New mode
28694 ("*vec_vllezlf<mode>"): New pattern.
28696 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28698 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28699 ("popcountv4si2", "popcountv2di2"): Rename to ...
28700 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28701 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28703 ("popcount<mode>2_vxe"): New pattern.
28705 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28707 * common/config/s390/s390-common.c (processor_flags_table): Add
28709 * config.gcc: Add arch12.
28710 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28711 Default to arch12 for unknown CPU model numbers.
28712 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28713 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28714 PROCESSOR_max sanity check.
28715 * config/s390/s390-opts.h (enum processor_type): Add
28717 * config/s390/s390.c (processor_table): Add arch12.
28718 (s390_expand_builtin): Add check for B_VXE flag.
28719 (s390_issue_rate): Add PROCESSOR_ARCH12.
28720 (s390_get_sched_attrmask): Likewise.
28721 (s390_get_unit_mask): Likewise.
28722 (s390_sched_score): Enable z13 scheduling for arch12.
28723 (s390_sched_reorder): Likewise.
28724 (s390_sched_variable_issue): Likewise.
28725 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28727 (s390_tune_attr): Use z13 scheduling also for arch12.
28728 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28729 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28730 (TARGET_VXE_P): New macros.
28731 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28732 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28733 * config/s390/s390.opt: Add arch12 as processor_type.
28735 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28737 * config/s390/s390.md
28738 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28739 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28740 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28742 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28743 Rename expanders to ...
28745 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28746 ("fixuns_truncdddi2_emu"): ... these.
28748 ("fixuns_trunc<mode>si2_emu"): New expander.
28750 ("*fixuns_truncdfdi2_z13"): Rename to ...
28751 ("*fixuns_truncdfdi2_vx"): ... this.
28753 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28755 * config/s390/2964.md: Remove the single element vector compare
28756 instructions which are no longer used.
28757 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28759 (s390_canonicalize_comparison): Remove handling of DFmode
28761 (s390_expand_vec_compare_scalar): Remove function.
28762 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28763 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28765 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28767 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28769 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28771 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28772 will anyway by matched by mov<mode>_64dfp.
28774 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28776 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28777 vlef/vstef. Add missing operand to vleif.
28779 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28781 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28782 pair for all vector types with 64 bit elements.
28783 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28784 * config/s390/vector.md (V_HW_64): ... here.
28785 (V_128_NOSINGLE): New mode iterator.
28786 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28787 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28788 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28789 ("*vec_load_pairv2di"): Change to ...
28790 ("*vec_load_pair<mode>"): ... this one.
28792 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28794 * config/s390/constraints.md: Add comments.
28795 (jKK): Reject element sizes > 8 bytes.
28796 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28798 * config/s390/s390.md: Add the s_operand checks formerly in
28799 s390_split_ok_p to various splitters where they are still
28801 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28802 for 128 bit vectors. Plus two splitters.
28804 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28806 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28809 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28812 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28813 error if the boundary argument is not constant.
28815 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28817 PR rtl-optimization/80112
28818 * loop-doloop.c (doloop_condition_get): Don't check condition
28819 if cmp isn't SET with IF_THEN_ELSE src.
28821 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28823 PR tree-optimization/80158
28824 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28825 replacing a candidate statement, also replace it for the
28826 candidate's alternate interpretation.
28827 (replace_rhs_if_not_dup): Likewise.
28828 (replace_one_candidate): Likewise.
28830 2017-03-24 Richard Biener <rguenther@suse.de>
28832 PR tree-optimization/80167
28833 * graphite-isl-ast-to-gimple.c
28834 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28836 (translate_isl_ast_to_gimple::get_rename): Likewise.
28838 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28841 handling of certain combinations of target options, including the
28842 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28843 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28845 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28848 * config/arm/arm.md (*load_multiple): Add reload_completed to
28849 matching condition.
28851 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28852 Richard Biener <rguenth@suse.de>
28854 PR tree-optimization/79908
28855 PR tree-optimization/80136
28856 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28857 been cast away, gimplify_and_add suffices.
28859 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28861 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28863 2017-03-23 Richard Biener <rguenther@suse.de>
28865 PR tree-optimization/80032
28866 * gimplify.c (gimple_push_cleanup): Forced unconditional
28867 cleanups still have to go to the conditional_cleanups
28870 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28872 PR tree-optimization/80072
28873 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28875 (next_operand_entry_id): Change type to unsigned int.
28876 (sort_by_operand_rank): Make sure to return the right return value
28877 even if unsigned fields are bigger than INT_MAX.
28878 (struct oecount): Change cnt and id type to unsigned int.
28879 (oecount_hasher::equal): Formatting fix.
28880 (oecount_cmp): Make sure to return the right return value
28881 even if unsigned fields are bigger than INT_MAX.
28882 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28885 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28886 TREE_READONLY on result if writing it more than once.
28889 * doc/invoke.texi (-fsanitize=thread): Document that with
28890 -fnon-call-exceptions atomics are not able to throw
28894 * tsan.c: Include tree-eh.h.
28895 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28896 maybe_clean_or_replace_eh_stmt where needed.
28897 (instrument_memory_accesses): Add cfg_changed argument.
28898 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28899 if it returned true.
28900 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28902 PR rtl-optimization/63191
28903 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28904 wrapper function, moved the whole old content into ...
28905 (ix86_delegitimize_address_1): ... this. New inline function.
28906 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28907 true as last argument instead of ix86_delegitimize_address.
28909 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28911 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28912 cortexa57_branch_cost.
28914 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28916 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28918 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28921 * doc/md.texi (Constraints): Document wA constraint.
28922 * config/rs6000/constraints.md (wA): New.
28923 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28924 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28925 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28926 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28928 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28931 * gimplify.c (is_oacc_declared): New function.
28932 (oacc_default_clause): Use it to set default flags for acc declared
28933 variables inside parallel regions.
28934 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28935 declared variables.
28936 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28937 declare attribute to any decl as necessary.
28939 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28942 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28943 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28944 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28945 (arm_arch_lpae): This.
28946 * config/arm/arm.c (arm_arch7ve): Rename into ...
28947 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28948 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28951 2017-03-22 Martin Liska <mliska@suse.cz>
28954 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28955 error message instead of an ICE.
28957 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28959 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28961 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28963 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28965 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28968 2017-03-21 Martin Sebor <msebor@redhat.com>
28970 * doc/extend.texi: Use "cannot" instead of "can't."
28971 * doc/hostconfig.texi: Same.
28972 * doc/install.texi: Same.
28973 * doc/invoke.texi: Same.
28974 * doc/loop.texi: Same.
28975 * doc/md.texi: Same.
28976 * doc/objc.texi: Same.
28977 * doc/rtl.texi: Same.
28978 * doc/tm.texi: Same.
28979 * doc/tm.texi.in: Same.
28980 * doc/trouble.texi: Same.
28982 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28985 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28986 (collect_checksum_attributes): Set it.
28987 (die_checksum_ordered): Use it.
28989 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28991 PR tree-optimization/79908
28992 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28993 change: For a VA_ARG whose LHS has been cast away, use
28994 force_gimple_operand to construct the side effects.
28996 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28998 PR translation/80001
28999 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29000 more amenable to translation.
29001 (oacc_loop_auto_partitions): Likewise.
29003 2017-03-21 Marek Polacek <polacek@redhat.com>
29004 Martin Sebor <msebor@redhat.com>
29006 PR tree-optimization/80109
29007 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29008 on INTEGRAL_TYPE_P.
29010 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29011 Segher Boessenkool <segher@kernel.crashing.org>
29014 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29015 check reg_used_between_p between insn and one of succ or succ2
29016 depending on if succ is artificial insn not inserted into insn
29019 2017-03-21 Martin Liska <mliska@suse.cz>
29021 PR gcov-profile/80081
29022 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29023 * doc/gcc.texi: Include gcov-dump stuff.
29024 * doc/gcov-dump.texi: New file.
29026 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29028 PR rtl-optimization/79150
29029 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29030 conditional jump, if the jump is the last insn of the loop.
29032 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29033 Richard Biener <rguenth@suse.de>
29035 PR tree-optimization/79908
29036 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29037 been cast away, use force_gimple_operand to construct the side
29040 2017-03-21 Martin Liska <mliska@suse.cz>
29042 PR libfortran/79956
29043 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29046 2017-03-21 Brad Spengler <spender@grsecurity.net>
29049 * plugin.c (htab_hash_plugin): New function.
29050 (add_new_plugin): Use it and adjust.
29051 (parse_plugin_arg_opt): Adjust.
29052 (init_one_plugin): Likewise.
29054 2017-03-21 Richard Biener <rguenther@suse.de>
29056 PR tree-optimization/80032
29057 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29058 if set force the cleanup to happen unconditionally.
29059 (gimplify_target_expr): Push inserted clobbers with force_uncond
29060 to avoid them being removed by control-dependent DCE.
29062 2017-03-21 Richard Biener <rguenther@suse.de>
29064 PR tree-optimization/80122
29065 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29066 va_arg_pack_len when the inlined call stmt requires pack
29068 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29070 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29073 * tsan.c (instrument_builtin_call): If the memory model argument
29074 is not a constant, assume it is valid.
29077 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29080 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29082 PR rtl-optimization/79910
29083 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29084 if its dest is used by an insn before I2 (other than the combined
29085 insns themselves, which are properly handled already).
29087 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29090 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29092 * combine.c (record_used_regs): New static function.
29093 (try_combine): Handle situations where there is an additional
29094 instruction between I2 and I3 which needs to have a LOG_LINK
29098 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29100 * combine.c (try_combine): Delete redundant i1 test. Call
29101 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29103 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29106 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29107 alternatives 13/14.
29109 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29111 PR tree-optimization/80054
29112 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29113 the optimization if a PHI or any of its arguments is not dominated
29114 by the candidate's basis. Use gphi* rather than gimple* as
29116 (replace_profitable_candidates): Clean up a gimple* variable that
29117 should be a gphi* variable.
29119 2017-03-20 Martin Sebor <msebor@redhat.com>
29122 * doc/extend.texi (attribute constructor): Document present limitation.
29124 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29127 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29128 __POWER9_VECTOR__ #ifdef control, change template definition to
29129 use Power9-specific built-in function.
29130 (vec_any_eq): Likewise.
29131 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29132 to control outcomes from this test.
29133 (vector_ae_<mode>p): For VEC_F modes, likewise.
29135 2017-03-20 Ian Lance Taylor <iant@google.com>
29137 * config/i386/i386.c (ix86_function_regparm): Save an extra
29138 register for -fsplit-stack with DECL_STATIC_CHAIN.
29140 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29143 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29144 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29146 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29148 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29150 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29153 2017-03-20 Marek Polacek <polacek@redhat.com>
29156 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29158 2017-03-20 Richard Biener <rguenther@suse.de>
29160 PR tree-optimization/80113
29161 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29162 allocate extra SSA name for PHI def.
29163 (add_close_phis_to_outer_loops): Likewise.
29164 (add_close_phis_to_merge_points): Likewise.
29165 (copy_loop_close_phi_args): Likewise.
29166 (copy_cond_phi_nodes): Likewise.
29168 2017-03-20 Martin Liska <mliska@suse.cz>
29170 PR middle-end/79753
29171 * tree-chkp.c (chkp_build_returned_bound): Do not build
29172 returned bounds for a LHS that's not a BOUNDED_P type.
29174 2017-03-20 Martin Liska <mliska@suse.cz>
29178 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29179 COMPLEX_CST and VECTOR_CST.
29181 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29184 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29185 target operand. A new splitter adds the clobber statement in case
29186 the target operand is dead anyway.
29188 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29190 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29191 to age-old versions of binutils and glibc.
29193 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29195 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29197 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29199 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29201 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29203 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29204 requirement for binutils 2.13.
29206 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29208 * combine.c (try_combine): Delete redundant i1 test. Call
29209 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29211 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29213 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29214 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29216 <riscv64-*-elf>: Re-arrange section
29217 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29218 <riscv32-*-linux>: Likewise.
29219 <riscv64-*-elf>: Likewise
29220 <riscv64-*-linux>: Likewise.
29222 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29225 * aarch64.opt(verbose-cost-dump): Fix typo.
29227 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29230 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29231 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29233 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29235 * reload.c (find_reloads): When reloading a nonoffsettable address,
29236 use RELOAD_OTHER for it and its address reloads.
29238 PR rtl-optimization/79910
29239 * combine.c (record_used_regs): New static function.
29240 (try_combine): Handle situations where there is an additional
29241 instruction between I2 and I3 which needs to have a LOG_LINK
29244 2017-03-17 Jeff Law <law@redhat.com>
29246 PR tree-optimization/71437
29247 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29248 conditional in the hash table first.
29249 (vrp_dom_walker::before_dom_children): Extract condition from
29250 ASSERT_EXPR. Record condition, its inverion and any implied
29251 conditions as well.
29253 2017-03-17 Marek Polacek <polacek@redhat.com>
29254 Markus Trippelsdorf <markus@trippelsdorf.de>
29256 PR tree-optimization/80079
29257 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29260 2017-03-17 Richard Biener <rguenther@suse.de>
29262 PR middle-end/80075
29263 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29264 Properly verify the LHS before the RHS possibly claims to be
29266 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29269 2017-03-17 Martin Jambor <mjambor@suse.cz>
29271 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29272 (List of -O2 options): Likewise.
29273 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29276 2017-03-17 Tom de Vries <tom@codesourcery.com>
29278 * gcov-dump.c (print_usage): Print bug_report_url.
29280 2017-03-17 Richard Biener <rguenther@suse.de>
29282 PR middle-end/80050
29283 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29284 (parser::peek): Likewise.
29286 2017-03-17 Richard Biener <rguenther@suse.de>
29288 PR tree-optimization/80048
29289 * sese.c (free_sese_info): Properly release rename_map and
29290 copied_bb_map elements.
29292 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29294 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29295 Add linked-list forward and backlinks. Insert on
29296 construction, remove on destruction.
29297 (class pass_store_merging): Add m_stores_head field.
29298 (pass_store_merging::terminate_and_process_all_chains):
29299 Iterate over m_stores_head list.
29300 (pass_store_merging::terminate_all_aliasing_chains):
29302 (pass_store_merging::execute): Check for debug stmts first.
29303 Push new chains onto the m_stores_head stack.
29305 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29308 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29309 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29310 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29312 2017-03-16 Jeff Law <law@redhat.com>
29314 PR tree-optimization/71437
29315 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29316 member function. Implementation moved into after_dom_children
29317 member function and into the threader's thread_outgoing_edges
29319 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29320 some code into new thread_outgoing_edges.
29321 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29322 definition. Simplify marker handling (do it here). Assume we always
29323 have the available expression and the const/copies tables.
29324 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29326 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29327 * tree-vrp.c (equiv_stack): No longer file scoped.
29328 (vrp_dom_walker): New class.
29329 (vrp_dom_walker::before_dom_children): New member function.
29330 (vrp_dom_walker::after_dom_children): Likewise.
29331 (identify_jump_threads): Setup domwalker. Use it rather than
29332 walking edges in a random order by hand. Simplify setup/finalization.
29333 (finalize_jump_threads): Remove.
29334 (vrp_finalize): Do not call identify_jump_threads here.
29335 (execute_vrp): Do it here instead and call thread_through_all_blocks
29338 PR tree-optimization/71437
29339 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29341 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29343 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29344 (dom_opt_dom_walker::thread_across_edge): Remove
29345 handle_dominating_asserts argument. All callers changed.
29346 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29347 changes. Remove calls to lhs_of_dominating_assert. Other
29348 uses of handle_dominating_asserts turn into unconditional code
29349 (simplify_control_stmt_condition_1): Likewise.
29350 (simplify_control_stmt_condition): Likewise.
29351 (thread_through_normal_block, thread_across_edge): Likewise.
29352 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29353 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29354 object if it is not an SSA_NAME.
29355 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29356 before calling into the VRP specific simplifiers.
29357 (identify_jump_threads): Remove handle_dominating_asserts
29360 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29363 * tree-diagnostic.c (default_tree_printer): No longer static.
29364 * tree-diagnostic.h (default_tree_printer): New prototype.
29366 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29368 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29369 Change ins into fmov.
29371 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29373 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29374 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29375 Use h_con constraint for operand 1.
29376 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29377 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29379 2017-03-15 Jeff Law <law@redhat.com>
29381 PR tree-optimization/71437
29382 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29383 (record_temporary_equivalences): Use it.
29385 PR tree-optimization/71437
29386 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29387 tree-ssa-scopedtables.
29388 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29389 (record_conditions, record_cond, vuse_eq): Likewise.
29390 (record_edge_info): Adjust to API tweak of record_conditions.
29391 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29392 (record_temporary_equivalences, optimize_stmt): Likewise.
29393 (eliminate_redundant_computations): Likewise.
29394 (record_equivalences_from_stmt): Likewise.
29395 * tree-ssa-scopedtables.c: Include options.h and params.h.
29396 (vuse_eq): New function, moved from tree-ssa-dom.c
29397 (build_and_record_new_cond): Likewise.
29398 (record_conditions): Likewise. Accept vector of conditions rather
29399 than edge_equivalence structure for first argument.
29400 for the first argument.
29401 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29402 from tree-ssa-dom.c.
29403 (avail_exprs_stack::record_cond): Likewise.
29404 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29405 from tree-ssa-dom.c.
29406 (avail_exprs_stack): Add new member functions lookup_avail_expr
29408 (record_conditions): Declare.
29410 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29413 * lra-constraints.c (process_alt_operands): Increase reject for
29414 reloading an input/output operand.
29416 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29419 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29420 insns to convert from signed/unsigned char/short to IEEE 128-bit
29422 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29424 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29427 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29428 subreg of inner mode for values already in registers.
29430 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29432 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29433 iteration reg is used after the loop.
29435 2017-03-14 Martin Sebor <msebor@redhat.com>
29437 PR tree-optimization/79800
29438 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29439 precision in negative-positive range.
29440 (format_floating): Call non-const overload with adjusted precision.
29442 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29445 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29448 2017-03-14 Martin Sebor <msebor@redhat.com>
29450 PR middle-end/80020
29451 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29452 * builtins.def (aligned_alloc): Use it.
29455 * Makefile.in (GTFILES): Add calls.c.
29456 * calls.c: Include "gt-calls.h".
29458 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29460 PR rtl-optimization/79728
29461 * regs.h (struct target_regs): New field
29462 x_contains_allocatable_regs_of_mode.
29463 (contains_allocatable_regs_of_mode): New macro.
29464 * reginfo.c (init_reg_sets_1): Initialize it, and change
29465 contains_reg_of_mode so it includes global regs as well.
29466 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29467 rather than contains_regs_of_mode.
29469 2017-03-14 Martin Liska <mliska@suse.cz>
29471 * doc/invoke.texi: Document options that can't be combined with
29472 -fcheck-pointer-bounds.
29474 2017-03-14 Martin Liska <mliska@suse.cz>
29476 PR middle-end/79831
29477 * doc/invoke.texi (-Wchkp): Document the option.
29479 2017-03-14 Martin Liska <mliska@suse.cz>
29481 * Makefile.in: Install gcov-dump.
29483 2017-03-14 Martin Liska <mliska@suse.cz>
29485 * multiple_target.c (expand_target_clones): Bail out for
29486 an invalid attribute.
29488 2017-03-14 Richard Biener <rguenther@suse.de>
29490 * alias.c (struct alias_set_entry): Pack properly.
29491 * cfgloop.h (struct loop): Likewise.
29492 * cse.c (struct set): Likewise.
29493 * ipa-utils.c (struct searchc_env): Likewise.
29494 * loop-invariant.c (struct invariant): Likewise.
29495 * lra-remat.c (struct cand): Likewise.
29496 * recog.c (struct change_t): Likewise.
29497 * rtl.h (struct address_info): Likewise.
29498 * symbol-summary.h (function_summary): Likewise.
29499 * tree-loop-distribution.c (struct partition): Likewise.
29500 * tree-object-size.c (struct object_size_info): Likewise.
29501 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29502 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29503 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29504 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29505 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29506 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29507 (struct _stmt_vec_info): Likewise.
29509 2017-03-14 Martin Liska <mliska@suse.cz>
29512 * multiple_target.c (create_dispatcher_calls): Check that
29513 a target can create a function dispatcher.
29515 2017-03-14 Martin Liska <mliska@suse.cz>
29518 * multiple_target.c (expand_target_clones): Drop local.local
29519 flag for default implementation.
29521 2017-03-14 Richard Biener <rguenther@suse.de>
29523 PR tree-optimization/80030
29524 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29526 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29528 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29529 gcc_fallthrough() instead of __attribute__((fallthrough));
29531 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29533 * doc/gcc.texi: Remove "up" link to (DIR).
29534 * doc/gccint.texi: Ditto.
29536 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29538 * doc/install.texi (Specific) <avr>: Remove reference to
29541 2017-03-13 Jeff Law <law@redhat.com>
29543 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29544 attribute rather than comments.
29546 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29547 match_scratch operand is highest.
29549 2017-03-13 Martin Liska <mliska@suse.cz>
29551 PR middle-end/78339
29552 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29553 is a CHKP clone, use original declaration.
29555 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29557 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29558 (arc_conditional_register_usage): Use a different allocation order
29559 when optimizing for size.
29560 * common/config/arc/arc-common.c (arc_option_optimization_table):
29561 Section anchors default on when optimizing for size.
29563 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29565 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29567 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29569 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29570 * config/arc/arc.md (cpu_facility): Add cd variant.
29571 (*movqi_insn): Add code density variant.
29572 (*movhi_insn): Likewise.
29573 (*movqi_insn): Likewise.
29574 (*addsi3_mixed): Likewise.
29575 (subsi3_insn): Likewise.
29577 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29579 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29581 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29583 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29584 expressions with MINUS and UNARY ops.
29586 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29589 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29591 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29592 between vec_select and vector argument.
29593 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29594 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29595 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29596 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29597 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29598 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29600 2017-03-13 Richard Biener <rguenther@suse.de>
29603 * params.def (vect-max-peeling-for-alignment): Fix typo.
29605 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29607 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29608 issue that only occurred with binutils below 2.18.
29610 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29612 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29613 refer to binutils 2.11/2.12 minimum.
29615 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29617 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29618 ftp.kernel.org and simplify binutils requirement.
29620 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29622 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29624 (Optimize Options): Ditto. Also remove redundancy.
29626 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29628 PR translation/79848
29629 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29631 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29632 to G_ to avoid double translation.
29634 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29636 PR translation/79923
29637 * auto-profile.c (get_combined_location): Convert leading
29638 character of diagnostics to lower case and remove trailing period.
29639 (read_profile): Likewise for various diagnostics.
29640 * config/arm/arm.c (arm_option_override): Remove trailing period
29641 from various diagnostics.
29642 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29643 (msp430_expand_delay_cycles): Likewise.
29645 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29648 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29649 full command-line argument, rather than just "str".
29650 (aarch64_validate_march): Likewise.
29651 (aarch64_validate_mtune): Likewise.
29653 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29655 PR rtl-optimization/78911
29656 * lra-assigns.c (must_not_spill_p): New function.
29657 (spill_for): Use it.
29659 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29661 PR tree-optimization/79981
29662 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29663 ATOMIC_COMPARE_EXCHANGE ifn result.
29664 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29665 IFN_ATOMIC_COMPARE_EXCHANGE.
29667 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29670 * opts.c (parse_sanitizer_options): Add missing question mark to
29671 "did you mean" message.
29673 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29675 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29677 (VMULEUH_UNS): Likewise.
29678 (VMULOUB_UNS): Likewise.
29679 (VMULOUH_UNS): Likewise.
29680 * config/rs6000/rs6000.c (builtin_function_type): Remove
29681 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29683 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29686 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29687 x with result of extra_parsing_for_operand_code_0.
29688 (function_reader::extra_parsing_for_operand_code_0): Convert
29689 return type from void to rtx, returning x. When reading
29690 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29691 larger size containing struct block_symbol.
29693 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29695 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29696 -mfloat128-hardware without -m64.
29698 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29701 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29702 entries to the case statement that marks unsigned arguments to
29703 overloaded functions.
29705 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29707 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29708 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29710 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29713 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29714 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29716 2017-03-10 Martin Liska <mliska@suse.cz>
29720 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29721 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29724 2017-03-10 Olivier Hainque <hainque@adacore.com>
29726 * tree-switch-conversion (array_value_type): Start by resetting
29727 candidate type to it's main variant.
29729 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29731 PR rtl-optimization/79909
29732 * combine.c (try_combine): Use simplify_replace_rtx on individual
29733 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29734 of the whole CALL_INSN_FUNCTION_USAGE.
29736 PR tree-optimization/79972
29737 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29738 get_range_info on SSA_NAMEs. Formatting fixes.
29740 2017-03-10 Richard Biener <rguenther@suse.de>
29741 Jakub Jelinek <jakub@redhat.com>
29743 PR tree-optimization/77975
29744 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29745 edge to be constant.
29746 (get_val_for): For constant x return it. Formatting fix.
29747 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29748 would use the same bases as the current one.
29750 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29752 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29753 instead of vec_select for V1TImode.
29754 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29756 (VSX_LE_128): Add V1TI to this mode iterator.
29757 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29758 (*vsx_le_perm_store_<mode>): Likewise.
29759 (pre-reload splitter for VSX stores): Likewise.
29760 (post-reload splitter for VSX stores): Likewise.
29761 (*vsx_xxpermdi2_le_<mode>): Likewise.
29762 (*vsx_lxvd2x2_le_<mode>): Likewise.
29763 (*vsx_stxvd2x2_le_<mode>): Likewise.
29765 2017-03-09 Michael Eager <eager@eagercon.com>
29767 Correct failures with --enable-checking=yes,rtl.
29769 * config/microblaze/microblaze.c (microblaze_expand_shift):
29770 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29771 test for const0_rtx.
29772 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29773 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29775 2017-03-09 Richard Biener <rguenther@suse.de>
29777 PR tree-optimization/79977
29778 * graphite-scop-detection.c (scop_detection::merge_sese):
29779 Handle the case of extra exits to blocks dominating the entry.
29781 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29783 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29786 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29788 PR rtl-optimization/79949
29789 * lra-constraints.c (process_alt_operands): Check memory when
29790 trying to predict a cycle. Print about the overall increase.
29792 2017-03-09 Richard Biener <rguenther@suse.de>
29794 PR middle-end/79971
29795 * gimple-expr.c (useless_type_conversion_p): Preserve
29796 TYPE_SATURATING for fixed-point types.
29798 2017-03-09 Richard Biener <rguenther@suse.de>
29801 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29802 alignment of BLKmode params.
29804 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29807 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29808 (VALL_NO_V2Q): Likewise.
29810 * config/aarch64/aarch64-simd.md
29811 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29813 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29814 VALL_NO_V2Q mode iterator.
29815 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29817 2017-03-09 Martin Liska <mliska@suse.cz>
29819 PR tree-optimization/79631
29820 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29821 tree_int_cst_sign_bit just for INTEGER constants.
29823 2017-03-09 Martin Liska <mliska@suse.cz>
29827 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29830 2017-03-09 Marek Polacek <polacek@redhat.com>
29833 * tree.c (inchash::add_expr): Handle TREE_VEC.
29835 2017-03-09 Martin Liska <mliska@suse.cz>
29838 (chkp_narrow_size_and_offset): New function.
29839 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29840 (void chkp_parse_bit_field_ref): New function.
29841 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29842 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29844 2017-03-09 Martin Liska <mliska@suse.cz>
29847 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29848 (chkp_find_bounds_1): Remove gcc_unreachable.
29850 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29853 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29854 BUILT_IN_SYNC*, determine the access type from the size suffix and
29855 always build a MEM_REF with that type. Handle forgotten
29856 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29859 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29860 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29861 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29862 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29863 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29864 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29865 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29866 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29867 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29868 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29869 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29870 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29871 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29872 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29873 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29874 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29875 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29876 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29877 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29878 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29879 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29880 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29881 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29882 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29883 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29884 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29885 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29886 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29887 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29888 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29889 definitions outside of __OPTIMIZE__ guarded section.
29892 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29893 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29894 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29895 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29898 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29900 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29901 ("vfenez<mode>"): Add missing constraints.
29903 2017-03-08 Martin Sebor <msebor@redhat.com>
29906 * config/nds32/nds32.c (nds32_option_override):
29907 Fix misspelled diagnostic.
29909 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29912 * gimplify.c (gimplify_omp_for): Replace index var in outer
29913 taskloop statement with an artificial variable and add
29914 OMP_CLAUSE_PRIVATE clause for it.
29916 2017-03-08 Richard Biener <rguenther@suse.de>
29918 PR tree-optimization/79955
29919 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29920 for accesses that are completely outside of the variable.
29922 2017-03-08 Andrew Haley <aph@redhat.com>
29924 PR tree-optimization/79943
29925 * tree-ssa-loop-split.c (compute_new_first_bound): When
29926 calculating the new upper bound, (END-BEG) should be added, not
29929 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29931 * config/avr/avr.md (setmemhi): Make sure match_dup
29932 operand number comes before match_scratch.
29934 2017-03-08 Richard Biener <rguenther@suse.de>
29936 PR tree-optimization/79920
29937 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29938 with ncopies == 1 to ...
29939 (vect_transform_slp_perm_load): ... here. Properly compute
29940 all element loads by iterating VF times over the group. Do
29941 not handle ncopies (computed in a broken way) in
29942 vect_create_mask_and_perm.
29944 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29947 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29948 is a uniform vector, use uniform_vector_p return value instead of
29949 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29951 2017-03-07 Marek Polacek <polacek@redhat.com>
29953 PR middle-end/79809
29954 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29955 (alloca_call_type): Likewise.
29957 2017-03-07 Martin Liska <mliska@suse.cz>
29959 * gcov.c (process_args): Put comment to correct location.
29961 2017-03-07 Martin Liska <mliska@suse.cz>
29963 PR middle-end/68270
29964 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29965 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29966 (chkp_narrow_bounds_for_field): Likewise.
29967 (chkp_parse_array_and_component_ref): Pass one more argument to
29970 2017-03-07 Richard Biener <rguenther@suse.de>
29972 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29975 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29977 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29978 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29980 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29983 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29984 to end of description.
29985 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29987 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29989 PR rtl-optimization/79901
29990 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29992 (*avx512f_<code><mode>3<mask_name>): ... this.
29993 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29994 iterator instead of VI8_AVX2_AVX512BW.
29996 PR rtl-optimization/79901
29997 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29998 min/max expander, expand it using expand_vec_cond_expr.
30001 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30004 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30007 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30008 to void * for PCH reasons.
30009 * dwarf2out.c (output_loc_operands, output_die): Cast
30010 v.val_vec.array to unsigned char *.
30012 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30015 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30018 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30020 PR rtl-optimization/79571
30021 * lra-constraints.c (process_alt_operands): Calculate static
30022 reject and subtract it from overall when only addresses will be
30025 2017-03-06 Julia Koval <julia.koval@intel.com>
30028 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30029 incoming stack boundary to 128 for 64-bit targets.
30031 2017-03-06 Richard Biener <rguenther@suse.de>
30033 PR tree-optimization/79894
30034 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30035 to NULL after folding it.
30037 2017-03-06 Richard Biener <rguenther@suse.de>
30039 PR tree-optimization/79824
30040 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30041 check disabling peeling for gaps.
30043 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30045 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30046 attributes): Document gettimeofday.
30048 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30050 * config/s390/s390.c (s390_option_override_internal): Set
30051 PARAM_MIN_VECT_LOOP_BOUND
30053 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30055 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30056 * config/s390/s390.md: Likewise.
30058 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30061 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30062 (<avx2_avx512>_perm<mode>): Rename to ...
30063 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30065 (<avx512>_perm<mode>_mask): Rename to ...
30066 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30068 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30069 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30070 instead of VI8F_256_512.
30071 (avx512f_perm<mode>): New define_expand.
30072 (avx512f_perm<mode>_mask): Likewise.
30073 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30074 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30076 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30078 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30079 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30081 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30083 2017-03-06 Martin Liska <mliska@suse.cz>
30086 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30087 when having a SSA NAME w/o VAR_DECL assigned to it.
30089 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30091 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30092 msa_dpsub_<su>_d): Fix MODE for vec_select.
30094 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30096 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30098 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30100 2017-03-06 Richard Biener <rguenther@suse.de>
30102 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30103 * plugin.c (register_plugin_info): Likewise.
30104 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30106 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30108 * config/i386/sse.md (sse_storehps, sse_storelps,
30109 avx_<castmode><avxsizesuffix>_<castmode>,
30110 avx512f_<castmode><avxsizesuffix>_<castmode>,
30111 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30112 in condition that at least one operand is not a MEM.
30114 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30116 PR middle-end/79805
30117 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30118 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30120 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30121 gimple_call_nothrow_p flag based on whether original builtin can throw.
30122 If it can, emit following stmts on the fallthrough edge.
30123 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30124 don't create new bb if inserting just debug stmts on the edge, try to
30125 insert them on the fallthru bb or just reset debug stmts.
30127 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30130 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30131 restore recog_data (including the operand rtxes inside it) around
30132 the call to get_insn_template.
30134 2017-03-03 Martin Sebor <msebor@redhat.com>
30136 PR tree-optimization/79699
30137 * context.c (context::~context): Free MPFR caches to avoid
30138 a memory leak on program exit.
30140 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30142 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30143 Use wide_int::ulow () instead of .elt (0).
30145 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30147 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30148 (*pushxf): Limit oF constraint to 32bit targets and add oC
30149 constraint for 64bit targets.
30150 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30151 (*pushdf): Change rmF constraint to rmC.
30153 2017-03-03 Martin Liska <mliska@suse.cz>
30155 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30156 Remove unused variable.
30158 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30161 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30162 is a memory operand, increase num_memory.
30163 (ix86_expand_args_builtin): Likewise.
30165 2017-03-03 Jan Hubicka <jh@suse.cz>
30168 * ipa-devirt.c (maybe_record_node): Properly handle
30169 __cxa_pure_virtual visibility.
30171 2017-03-03 Martin Liska <mliska@suse.cz>
30173 PR tree-optimization/79803
30174 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30176 (pass_loop_prefetch::execute): Disabled optimization if an
30177 assumption about L1 cache size is not met.
30179 2017-03-03 Martin Liska <mliska@suse.cz>
30181 PR rtl-optimization/79574
30182 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30183 (hash_scan_set): Likewise.
30184 (dump_hash_table): Likewise.
30185 (hoist_code): Likewise.
30187 2017-03-03 Richard Biener <rguenther@suse.de>
30189 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30190 in place of uhigh ().
30191 (fixed_convert_from_real): Likewise.
30193 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30196 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30198 2017-03-03 Richard Biener <rguenther@suse.de>
30200 PR middle-end/79818
30201 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30202 TYPE_OVERFLOW_UNDEFINED check.
30204 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30206 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30208 (vector_ae_<mode>_p): Likewise.
30209 (vector_nez_<mode>_p): Likewise.
30210 (vector_ne_v2di_p): Likewise.
30211 (vector_ae_v2di_p): Likewise.
30212 (vector_ne_<mode>_p): Likewise.
30213 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30215 (vsx_tsqrt<mode>2_fe): Likewise.
30217 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30220 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30222 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30224 PR rtl-optimization/79780
30225 * cprop.c (one_cprop_pass): When second and further conditional trap
30226 in a single basic block is turned into an unconditional trap, turn it
30227 into a deleted note to avoid RTL verification failures.
30229 2017-03-02 Richard Biener <rguenther@suse.de>
30231 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30233 2017-03-02 Richard Biener <rguenther@suse.de>
30235 PR tree-optimization/79345
30237 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30238 param and abort the walk, returning -1 if it is hit.
30239 (walk_aliased_vdefs): Take a limit param and pass it on.
30240 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30241 defaulting to 0 and return a signed int.
30242 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30243 (check_defs): New helper.
30244 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30245 about uninitialized memory.
30246 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30247 bogus uninitialized warning.
30248 (fixed_convert_from_real): Likewise.
30250 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30252 PR tree-optimization/66768
30253 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30254 iv_use if base object can't be determined.
30256 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30258 PR tree-optimization/79345
30259 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30260 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30261 (get_pattern_stats): Initialize it.
30262 * genemit.c (gen_expand): Verify match_scratch numbers come after
30263 match_operand/match_dup numbers.
30264 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30265 match_scratch numbers.
30266 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30268 * config/s390/s390.md (trunctdsd2): Likewise.
30270 2017-03-02 Richard Biener <rguenther@suse.de>
30272 * wide-int.h (wide_int_storage::operator=): Implement in terms
30275 2017-03-02 Richard Biener <rguenther@suse.de>
30277 PR tree-optimization/79777
30278 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30279 the to insert expression to sth existing.
30281 2017-03-01 Martin Sebor <msebor@redhat.com>
30283 PR middle-end/79692
30284 * gimple-ssa-sprintf.c
30285 (directive::known_width_and_precision): New function.
30286 (format_integer): Use it.
30287 (get_mpfr_format_length): Consider the full range of precision
30288 when computing %g output with the # flag. Set the likely byte
30289 count to 3 rather than 1 when precision is indeterminate.
30290 (format_floating): Correct the lower bound of precision.
30292 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30294 * doc/invoke.texi: Document default code model for 64-bit Linux.
30296 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30299 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30300 udiv rather than div since input pattern is unsigned.
30302 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30304 * config/i386/i386.c (print_reg): Warn for values of
30305 unsupported size in integer register.
30307 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30310 * config/rs6000/predicates.md (current_file_function_operand): Do
30311 not allow self calls to be local if the function is replaceable.
30313 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30316 * config/rs6000/altivec.h (vec_ctz and others): Change the
30317 preprocessor macro that controls conditional compilation from
30318 _ARCH_PWR9 to __POWER9_VECTOR__.
30319 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30320 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30321 control (instead of _ARCH_PWR9 control) so that template
30322 definition uses power9-specific function.
30323 (vec_any_eq): Likewise.
30324 (vec_all_ne): Change macro definition to use a power9-specific
30325 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30326 _ARCH_PWR9 control).
30327 (vec_any_eq) Likewise.
30328 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30329 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30330 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30331 support for xvcmpnedp instruction.
30332 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30333 macro expansion so that Power9 implementation of vec_all_ne does
30334 not use the AltiVec predicate framework.
30335 (VCMPNEH_P): Likewise.
30336 (VCMPNEW_P): Likewise.
30337 (VCMPNED_P): Likewise.
30338 (VCMPNEFP_P): Likewise.
30339 (VCMPNEDP_P): Likewise.
30340 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30341 implementation of vec_any_eq to not use AltiVec predicate
30343 (VCMPAEH_P): Likewise.
30344 (VCMPAEW_P): Likewise.
30345 (VCMPAED_P): Likewise.
30346 (VCMPAEFP_P): Likewise.
30347 (VCMPAEDP_P): Likewise.
30348 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30349 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30350 not use the AltiVec predicate framework.
30351 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30352 of vec_any_eq to not use AltiVec predicate framework.
30353 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30354 support for predefined __POWER9_VECTOR__ macro to indicate that
30355 Power9 instruction selection is enabled.
30356 (altivec_overloaded_builtins): Remove extraneous
30357 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30358 function argument types RS6000_BTI_bool_V16QI and
30359 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30360 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30361 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30362 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30363 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30364 Power9 for implementations of vec_cmpne. Change the signature for
30365 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30366 (representing vec_all_ne) to remove the previously described first
30367 argument of type RS6000_BTI_INTSI, as this was an artifact of
30368 reliance on the AltiVec predicate framework, which is no longer
30369 used in the implementation of these functions. Add
30370 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30371 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30372 since, unlike the AltiVec predicate framework implementation, we
30373 do not share function descriptors between vec_alle and vec_anyeq.
30374 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30375 set of modes that receive special treatment even when
30376 TARGET_P9_VECTOR is true. The special treatment emits code that
30377 does not depend on Power9 instructions.
30378 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30379 define_expand to not rely on AltiVec predicate framework.
30380 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30382 (vector_ne_v2di_p): Change this define_expand to not rely on
30383 AltiVec predicate framework.
30384 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30386 (vector_ne_<mode>_p): Change this define_expand to not rely on
30387 AltiVec predicate framework.
30388 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30390 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30391 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30392 define_insn pattern.
30393 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30394 define_insn pattern because the xvcmpne<VSs>. instruction is not
30396 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30397 instruction is not supported.
30399 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30401 * config/nvptx/nvptx.c: Include intl.h.
30403 2017-03-01 Martin Jambor <mjambor@suse.cz>
30406 * ipa-prop.h (ipa_bits): Removed field known.
30407 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30408 to pointers. Adjusted their comments to warn about their sharing.
30409 (ipcp_transformation_summary): Change bits to a vector of pointers.
30410 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30411 (ipa_get_ipa_bits_for_value): Declare.
30412 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30413 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30414 (ipa_bits_hash_table): Likewise.
30415 (ipa_vr_ggc_hash_traits): Likewise.
30416 (ipa_vr_hash_table): Likewise.
30417 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30418 being pointers and vr_known being removed.
30419 (ipa_set_jf_unknown): Likewise.
30420 (ipa_get_ipa_bits_for_value): New function.
30421 (ipa_set_jfunc_bits): Likewise.
30422 (ipa_get_value_range): New overloaded functions.
30423 (ipa_set_jfunc_vr): Likewise.
30424 (ipa_compute_jump_functions_for_edge): Use the above functions to
30425 construct bits and vr parts of jump functions.
30426 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30427 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30429 (ipcp_grow_transformations_if_necessary): Also allocate
30430 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30432 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30433 them. Fix too long lines.
30434 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30435 vr_known being removed.
30436 (ipa_read_jump_function): Use new setter functions to construct bits
30437 and vr parts of jump functions or set them to NULL.
30438 (write_ipcp_transformation_info): Adjust for bits being pointers.
30439 (read_ipcp_transformation_info): Likewise.
30440 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30442 Include gt-ipa-prop.h.
30443 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30445 (ipcp_store_bits_results): Likewise.
30446 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30447 Do not write to existing jump functions but use a temporary instead.
30449 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30452 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30453 attempt to use its first operand as BIT_FIELD_REF base.
30455 2017-03-01 Richard Biener <rguenther@suse.de>
30457 PR middle-end/79721
30458 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30459 interpolating formula in wrapping arithmetic.
30460 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30462 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30464 PR tree-optimization/79734
30465 * tree-vect-generic.c (expand_vector_condition): Optimize
30466 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30467 Handle VEC_COND_EXPR where comparison has different inner width from
30468 type's inner width.
30470 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30472 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30473 markup, and similar issues. Remove @opindex entries for things
30474 that aren't options. Add missing -mmpy-option entries.
30476 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30478 PR tree-optimization/79737
30479 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30480 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30481 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30482 instead of byte_size. Formatting fix.
30483 (shift_bytes_in_array_right): Formatting fix.
30485 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30488 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30489 condition on optimize for the leaf function test.
30491 2017-02-28 Martin Liska <mliska@suse.cz>
30494 * read-rtl-function.c (function_reader::handle_unknown_directive):
30495 Bail out when one uses -flto.
30497 2017-02-28 Martin Liska <mliska@suse.cz>
30499 * common.opt: Replace space with tabular for options of <number>
30501 * config/i386/i386.opt: Show <number> value for
30502 -mlarge-data-threshold.
30503 * opts.c (print_filtered_help): Do not display number in hexadecimal
30506 2017-02-28 Martin Liska <mliska@suse.cz>
30508 * common.opt: Fix --help=option -Q for options which are of
30511 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30513 * config/i386/i386.c (print_reg): Error out for values
30514 of 8-bit size in invalid integer register.
30516 2017-02-28 Martin Sebor <msebor@redhat.com>
30518 PR tree-optimization/79691
30519 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30521 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30524 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30525 gcc_unreachable with output_operand_lossage.
30527 2017-02-28 Richard Biener <rguenther@suse.de>
30529 PR tree-optimization/79740
30530 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30532 (visit_nary_op): Insert the nary into the hashtable if we
30533 pattern-matched sth.
30534 * tree-ssa-pre.c (eliminate_insert): Robustify.
30536 2017-02-28 Richard Biener <rguenther@suse.de>
30538 PR middle-end/79731
30539 * fold-const.c (decode_field_reference): Reject out-of-bound
30542 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30544 * config/i386/i386.c: Include intl.h.
30545 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30546 instead of just cond ? "..." : "...".
30547 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30548 * coverage.c (read_counts_file): Likewise.
30549 * omp-offload.c: Include intl.h.
30550 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30551 of just cond ? "..." : "...".
30552 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30553 of just cond ? "..." : "...".
30555 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30558 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30560 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30561 'tune for' CPU name.
30562 * config/arm/arm-cpu-data.h: Regenerated.
30564 2017-02-28 Richard Biener <rguenther@suse.de>
30566 PR tree-optimization/79732
30567 * tree-inline.c (expand_call_inline): Do not shadow var.
30569 2017-02-28 Richard Biener <rguenther@suse.de>
30571 PR tree-optimization/79723
30572 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30573 address-space properly.
30575 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30577 * doc/optinfo.texi (Optimization groups): Fix option used for
30579 * doc/invoke.texi (-fopt-info): Document "omp".
30580 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30581 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30582 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30583 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30584 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30586 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30588 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30589 OPTGROUP_OMP with "omp".
30591 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30594 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30595 for arithmetic shift of unsigned V2DI.
30597 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30599 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30600 arc/linux.h headers.
30601 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30602 (LINK_SPEC): Likewise.
30603 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30604 (EXTRA_SPECS): Likewise.
30605 (STARTFILE_SPEC): Likewise.
30606 (ENDFILE_SPEC): Likewise.
30607 (LIB_SPEC): Likewise.
30608 (TARGET_SDATA_DEFAULT): Likewise.
30609 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30610 (MULTILIB_DEFAULTS): Likewise.
30611 (DWARF2_UNWIND_INFO): Likewise.
30612 * config/arc/big.h: New file.
30613 * config/arc/elf.h: Likewise.
30614 * config/arc/linux.h: Likewise.
30615 * config/arc/t-uClibc: Remove.
30617 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30619 PR tree-optimization/77536
30620 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30621 (tree_transform_and_unroll_loop): Use above function to compute the
30622 estimated niter of unrolled loop and use it when scaling profile.
30623 Also use count info rather than frequency if it's non-zero.
30624 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30625 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30626 (vect_transform_loop): Call above function.
30628 2017-02-27 Richard Biener <rguenther@suse.de>
30630 PR tree-optimization/45397
30631 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30632 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30633 (visit_nary_op): Add pattern matching for CSEing sign-changed
30634 or truncated operations with wider ones.
30636 2017-02-27 Richard Biener <rguenther@suse.de>
30638 PR tree-optimization/79690
30639 * tree-vect-stmts.c (vectorizable_store): Use vector type
30640 built from the DR with address-space.
30642 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30644 * doc/invoke.texi (Optimize Options): Refine the description
30645 of asan-use-after-return.
30647 2017-02-25 Alan Modra <amodra@gmail.com>
30649 PR rtl-optimization/79584
30650 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30651 base, not ad->base_term, the reg within base. Remove assertion
30652 that ad->base == ad->base_term. Replace gen_int_mode using
30653 bogus mode with const0_rtx.
30655 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30657 PR middle-end/79396
30658 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30659 FMA_EXPR like tcc_binary or tcc_unary.
30661 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30664 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30666 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30667 (output_loc_operands): Handle DW_OP_call_ref and
30668 DW_OP_GNU_variable_value.
30669 (struct variable_value_struct): New type.
30670 (struct variable_value_hasher): Likewise.
30671 (variable_value_hash): New variable.
30672 (string_types): Remove.
30673 (copy_loc_descr): New function.
30674 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30675 (prepend_loc_descr_to_each): New function.
30676 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30677 instead of add_loc_descr_to_each if the first argument is single
30678 location list and the second has multiple.
30679 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30680 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30681 when looking for variable value which doesn't have other location info.
30682 (loc_list_from_tree): Formatting fix.
30683 (gen_array_type_die): Simplify DW_AT_string_length handling.
30684 (adjust_string_types): Remove.
30685 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30686 string_types. Call resolve_variable_values.
30687 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30688 (resolve_addr_in_expr): Likewise. Add A argument.
30689 (copy_deref_exprloc): Remove deref argument. Adjust for the
30690 original expression being DW_OP_GNU_variable_value with optionally
30691 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30692 optionally after it.
30693 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30694 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30695 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30696 (variable_value_hasher::hash, variable_value_hasher::equal): New
30698 (resolve_variable_value_in_expr, resolve_variable_value,
30699 resolve_variable_values, note_variable_value_in_expr,
30700 note_variable_value): New functions.
30701 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30704 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30707 * opts.h (handle_generated_option): Add GENERATED_P argument.
30708 * opts-common.c (handle_option): Adjust function comment.
30709 (handle_generated_option): Add GENERATED_P argument, pass it to
30711 (control_warning_option): Pass false to handle_generated_option
30713 * opts.c (maybe_default_option): Pass true to handle_generated_option
30715 * optc-gen.awk: Likewise.
30717 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30719 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30720 a REG, look at the REG it is a SUBREG of.
30721 (splitter for cmpeqsi_t): Ditto.
30723 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30725 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30726 the special USEs with the pattern of the insn, not the insn itself.
30728 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30731 * doc/invoke.texi: Document -mload-store-pairs.
30733 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30734 Sandra Loosemore <sandra@codesourcery.com>
30736 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30737 argument isn't a CONST_INT.
30738 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30739 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30740 (nios2_validate_compare): Bypass alternate compare logic if *op2
30741 is not a CONST_INT.
30742 (ldstwm_operation_p): Return false if first_base is not a REG or
30743 if first_offset is not a CONST_INT.
30745 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30747 * config/cris/cris.md: Use correct operand in a define_peephole2.
30749 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30751 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30753 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30755 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30756 this_insn if it is an INSN or JUMP_INSN.
30757 (force_offsettable): Look at base, not at addr.
30758 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30759 on things that aren't necessarily CONST_INTs.
30761 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30763 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30764 -mfpmath=sse is the default also for x86-32 targets with SSE2
30765 instruction set when @option{-ffast-math} is enabled
30767 2017-02-24 Jeff Law <law@redhat.com>
30769 PR rtl-optimizatoin/79286
30770 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30773 2017-02-24 Richard Biener <rguenther@suse.de>
30775 PR tree-optimization/79389
30776 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30779 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30781 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30782 function comment to reflect reality.
30783 (loop_exits_before_overflow): Fix typo in function description.
30785 2017-02-24 Richard Biener <rguenther@suse.de>
30787 PR tree-optimization/79389
30788 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30789 properly that a threading opportunity exists. Detect conditional
30790 copy/constant propagation opportunities.
30792 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30794 * config/visium/visium.md (type): Add trap.
30795 (b): New mode attribute.
30796 (*btst): Rename into...
30797 (*btst<mode>): ...this and adjust.
30798 (*cbranchsi4_btst_insn): Rename into...
30799 (*cbranch<mode>4_btst_insn): ...this and adjust.
30800 (trap): New define_insn.
30802 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30804 PR tree-optimization/79389
30805 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30806 (noce_reversed_cond_code): New function.
30807 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30808 reversed_comparison_code. Formatting fix.
30809 (noce_try_store_flag): Test rev_cond != NULL in addition to
30810 reversed_comparison_code.
30811 (noce_try_store_flag_constants): Likewise.
30812 (noce_try_store_flag_mask): Likewise.
30813 (noce_try_addcc): Use rev_cond if non-NULL instead of
30814 reversed_comparison_code.
30815 (noce_try_cmove_arith): Likewise. Formatting fixes.
30816 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30817 (noce_find_if_block): Initialize rev_cond.
30818 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30819 instead of false as last argument never attempt to reverse it
30822 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30824 PR tree-optimization/79663
30825 * tree-predcom.c (combine_chains): Process refs in reverse order
30826 only for ZERO length chains, and add explaining comment.
30828 2017-02-23 Jeff Law <law@redhat.com>
30830 PR tree-optimization/79578
30831 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30832 in call to operand_equal_p.
30834 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30837 * config/i386/cpuid.h: Fix another undefined behavior.
30839 2017-02-23 Richard Biener <rguenther@suse.de>
30841 PR tree-optimization/79683
30842 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30843 vector types for data-refs.
30845 2017-02-23 Martin Liska <mliska@suse.cz>
30847 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30849 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30851 PR middle-end/79665
30852 * internal-fn.c (get_range_pos_neg): Moved to ...
30853 * tree.c (get_range_pos_neg): ... here. No longer static.
30854 * tree.h (get_range_pos_neg): New prototype.
30855 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30856 are known to be in between 0 and signed maximum inclusive, try to
30857 expand both unsigned and signed divmod and use the cheaper one from
30860 2017-02-22 Jeff Law <law@redhat.com>
30862 PR tree-optimization/79578
30863 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30864 to compare base operands.
30866 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30869 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30870 gpc_reg_operand instead of fpr_reg_operand.
30872 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30874 * config/mips/mips.c (mips_return_in_memory): Force FP
30875 vector types to be returned in memory for o32 ABI.
30877 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30879 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30880 instead of DW_TAG_member for static data member declarations and don't
30881 set no_linkage_name for static inline data members.
30882 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30885 2017-02-22 Martin Liska <mliska@suse.cz>
30887 * doc/invoke.texi: Replace inequality signs with square brackets
30890 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30893 * lra-constraints.c (simplify_operand_subreg): Handle
30894 WORD_REGISTER_OPERATIONS targets.
30896 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30899 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30900 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30901 elimination by swapping fld*.
30903 2017-02-22 Richard Biener <rguenther@suse.de>
30905 PR tree-optimization/79673
30906 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30907 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30908 irrelevant address-space qualifiers and avoiding a
30909 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30911 2017-02-22 Richard Biener <rguenther@suse.de>
30913 PR tree-optimization/79666
30914 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30915 to not symbolically negate if that may introduce undefined
30918 2017-02-22 Martin Liska <mliska@suse.cz>
30921 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30922 * data-streamer-out.c (streamer_write_gcov_count_stream):
30924 * value-prof.c (stream_out_histogram_value): Make assert more
30925 precise based on type of counter.
30927 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30930 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30931 Use nonimmediate_operand instead of memory_operand for operand 1.
30932 (float-extend standard_x87sse_constant_load splitter): Ditto.
30934 2017-02-21 Jeff Law <law@redhat.com>
30936 PR tree-optimization/79621
30937 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30938 blocks with edges to themselves.
30940 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30943 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30944 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30945 Use gimple_call_builtin_p.
30948 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30949 on temporarily removed DEBUG_INSNs.
30951 PR tree-optimization/79649
30952 * tree-loop-distribution.c (classify_partition): Give up on
30953 non-generic address space loads/stores.
30955 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30957 * doc/loop.texi (Loop manipulation): Remove nonexistent
30958 tree_ssa_loop_version from the documentation.
30959 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30961 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30964 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30965 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30966 * config/rs6000/rs6000.c: Include except.h.
30967 (rs6000_expand_split_stack_prologue): Call
30968 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30970 2017-02-21 Martin Jambor <mjambor@suse.cz>
30973 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30974 have been analyzed.
30976 2017-02-21 Martin Jambor <mjambor@suse.cz>
30978 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30979 for backward compatibility only.
30980 * doc/invoke.texi (Option Summary): Remove all references to
30981 -fipa-cp-alignment.
30983 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30987 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30989 * lra-constraints.c (curr_insn_transform): Handle
30990 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30992 2017-02-21 Martin Liska <mliska@suse.cz>
30994 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30996 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30998 PR translation/79638
30999 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31001 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31004 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31005 (arm_function_ok_for_sibcall): Return false for an indirect call by
31006 descriptor if all the argument registers are used.
31007 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31008 alignment of the function.
31010 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31012 PR tree-optimization/61441
31013 * simplify-rtx.c (simplify_const_unary_operation): For
31014 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31015 the sNaN unmodified.
31017 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31019 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31020 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31021 instead of SYSTEM_HEADER_DIR.
31023 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31024 Martin Liška <mliska@suse.cz>
31026 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31027 Fix typos and grammar, use active voice, and clarify.
31029 2017-02-20 Marek Polacek <polacek@redhat.com>
31031 PR middle-end/79537
31032 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31035 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31037 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31040 * config/i386/i386.c (ix86_expand_builtin): Handle
31041 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31042 ix86_builtins_isa[fcode].isa as a requirement of those
31043 flags and any other flag in the bitmask.
31044 (ix86_init_mmx_sse_builtins): Use 0 instead of
31045 ~OPTION_MASK_ISA_64BIT as mask.
31046 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31047 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31048 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31049 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31051 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31054 * lra-constraints.c (split_reg): Check requested split mode
31055 is supported by the register.
31057 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31059 * lra-constraints.c (simplify_operand_subreg): Remove early
31062 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31065 * lra-constraints.c (curr_insn_transform): Tighten condition
31066 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31068 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31071 * lra-constraints.c (curr_insn_transform): Handle
31072 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31074 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31077 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31079 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31081 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31084 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31087 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31089 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31092 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31094 * final.c (last_columnnum, override_columnnum): New variables.
31095 (final_start_function): Set last_columnnum, pass it to begin_prologue
31096 hook and pass 0 to dwarf2out_begin_prologue.
31097 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31098 to source_line debug hook.
31099 (notice_source_line): Compute last_columnnum and for debug_column_info
31100 return true on column changes.
31101 * debug.h (struct gcc_debug_hooks): Add column argument to
31102 source_line and begin_prologue hooks.
31103 (debug_nothing_int_charstar_int_bool): Remove prototype.
31104 (debug_nothing_int_int_charstar,
31105 debug_nothing_int_int_charstar_int_bool): New prototypes.
31106 (dwarf2out_begin_prologue): Add column argument.
31107 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31108 begin_prologue hooks.
31109 (debug_nothing_int_charstar_int_bool): Remove.
31110 (debug_nothing_int_int_charstar,
31111 debug_nothing_int_int_charstar_int_bool): New functions.
31112 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31113 through to dwarf2out_source_line.
31114 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31115 (dwarf2out_source_line): Add column argument, emit it if requested.
31116 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31118 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31119 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31120 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31121 through to dwarf2out_begin_prologue.
31122 (vmsdbgout_source_line): Add column argument, pass it through to
31123 dwarf2out_source_line.
31124 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31125 dbxout_source_line caller.
31126 (dbxout_source_line): Add column argument.
31128 * common.opt (gno-column-info, gcolumn-info): New options.
31129 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31130 (check_die): Also test for multiple DW_AT_decl_column attributes.
31131 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31132 DW_AT_decl_column if requested.
31133 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31135 (gen_variable_die): Likewise.
31136 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31137 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31140 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31141 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31142 (ix86_handle_option): Handle OPT_m3dnowa.
31143 * doc/invoke.texi (-m3dnowa): Document.
31144 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31145 -m3dnowa instead of -m3dnow -march=athlon.
31148 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31149 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31151 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31154 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31155 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31156 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31157 generator for vsx_xxpermdi_<mode>_be.
31158 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31159 force big-endian semantics.
31160 (vsx_xxpermdi_<mode>_be): New define_expand with same
31161 implementation as previous version of vsx_xxpermdi_<mode>.
31163 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31165 PR tree-optimization/79327
31166 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31167 variable, its initialization and use.
31169 2017-02-17 Julia Koval <julia.koval@intel.com>
31171 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31172 (OPTION_MASK_ISA_PKU_UNSET): New.
31173 (ix86_handle_option): Handle -mrdpid.
31174 * config/i386/cpuid.h (bit_RDPID): New.
31175 * config/i386/driver-i386.c (host_detect_local_cpu):
31176 Detect RDPID feature.
31177 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31178 * config/i386/i386-c.c (ix86_target_macros_internal):
31180 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31181 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31182 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31183 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31184 * config/i386/i386.md (define_insn "rdpid"): New.
31185 * config/i386/i386.opt Add -mrdpid.
31186 * config/i386/immintrin.h (_rdpid_u32): New.
31188 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31190 PR rtl-optimization/79541
31191 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31192 instead of transforming it into USE.
31194 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31196 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31197 If HONOR_SNANS (SFmode) force the input to a register.
31198 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31199 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31200 an frsp or similar insn.
31202 2017-02-17 Martin Liska <mliska@suse.cz>
31204 PR rtl-optimization/79577
31205 * params.def (selsched-max-sched-times): Increase minimum to 1.
31207 2017-02-17 Martin Liska <mliska@suse.cz>
31209 PR rtl-optimization/79574
31210 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31212 2017-02-17 Martin Liska <mliska@suse.cz>
31214 PR tree-optimization/79529
31215 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31216 ssa_defined_default_def_p to handle cases which are implicitly
31218 * tree-ssa.c (ssa_defined_default_def_p): New function.
31219 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31220 which are implicitly defined.
31221 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31223 2017-02-17 Richard Biener <rguenther@suse.de>
31225 PR middle-end/79576
31226 * params.def (max-ssa-name-query-depth): Limit to 10.
31228 2017-02-17 Richard Biener <rguenther@suse.de>
31230 PR tree-optimization/79552
31231 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31234 2017-02-17 Richard Biener <rguenther@suse.de>
31237 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31239 2017-02-17 Marek Polacek <polacek@redhat.com>
31241 PR middle-end/79536
31242 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31243 (fold_negate_expr): New wrapper.
31245 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31247 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31248 Correct terminology and de-emphasize pre-standard behavior.
31250 2017-02-16 Alan Modra <amodra@gmail.com>
31252 PR rtl-optimization/79286
31253 * ira.c (def_dominates_uses): New function.
31254 (update_equiv_regs): Don't create an equivalence for insns that
31255 may trap where the register def does not dominate the use.
31257 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31259 PR rtl-optimization/78127
31260 * lra.c (lra): Call lra_eliminate before finish the loop after
31263 2017-02-16 Richard Biener <rguenther@suse.de>
31265 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31267 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31268 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31269 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31270 (isl_val_int_from_wi): New function.
31271 (extract_affine_gmp): Rename to ...
31272 (extract_affine_wi): ... this, take a widest_int.
31273 (extract_affine_int): Just wrap extract_affine_wi.
31274 (add_param_constraints): Use isl_val_int_from_wi.
31275 (add_loop_constraints): Likewise, and extract_affine_wi.
31277 2017-02-15 Jeff Law <law@redhat.com>
31279 PR middle-end/79521
31280 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31281 ira_init_register_move_cost_if_necessary.
31283 2017-02-15 Martin Sebor <msebor@redhat.com>
31285 PR middle-end/32003
31286 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31287 removed in a prior commit.
31289 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31291 PR tree-optimization/79347
31292 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31293 counters during peeling.
31295 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31297 * Makefile.in (site.exp): Remove "set ISLVER".
31299 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31302 * real.c (real_from_integer): Call real_convert even for decimal.
31304 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31307 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31309 2017-02-14 Andrew Pinski <apinski@cavium.com>
31311 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31312 cores and change the partno/implementer to be correct.
31313 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31314 the 'B" as the implementer.
31315 * config/aarch64/aarch64-tune.md: Regenerate.
31317 2017-02-14 Carl Love <cel@us.ibm.com>
31319 * config/rs6000/rs6000.c: Add case statement entry to make the
31320 xvcvuxdsp built-in argument unsigned.
31321 * config/rs6000/vsx.md: Fix the source and return operand types so they
31322 match the instruction definitions from the ISA document. Fix typo
31323 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31326 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31329 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31330 member early_clobber_alts.
31331 * lra-lives.c (reg_early_clobber_p): New.
31332 (process_bb_lives): Use it.
31333 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31334 (debug_operand_data): Initialize early_clobber_alts.
31335 (setup_operand_alternative): Set up early_clobber_alts.
31336 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31337 alternatives to new_insn_reg.
31338 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31340 (lra_update_insn_regno_info): Pass the new arg.
31342 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31344 PR middle-end/79505
31345 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31346 (new_oacc_loop_raw): Don't clear already cleared fields.
31349 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31350 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31351 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31353 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31356 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31358 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31361 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31362 the extra instruction to the right place to store 128-bit constant
31365 2017-02-14 Martin Sebor <msebor@redhat.com>
31367 PR middle-end/79448
31368 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31369 warning for strings of unknown length.
31371 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31373 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31375 2017-02-14 Jeff Law <law@redhat.com>
31378 * ira-costs.c (scan_one_insn): Initialize register move costs
31379 for pseudos seen in USE/CLOBBER insns.
31381 PR tree-optimization/79095
31382 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31383 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31384 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31385 if the operands are known to be not equal, then the resulting range
31387 (intersect_ranges): If the new range is ~[0,0] and the old range is
31388 wide, then prefer ~[0,0].
31389 * tree-vrp.c (overflow_comparison_p_1): New function.
31390 (overflow_comparison_p): New function.
31391 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31392 if NAME is used in an overflow test.
31393 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31394 overflow check that can be expressed as an equality test, then adjust
31395 ops to be that equality test.
31397 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31399 * config/s390/s390-builtin-types.def: Remove flags argument.
31400 * config/s390/s390.c (s390_init_builtins): Likewise.
31402 2017-02-14 Martin Liska <mliska@suse.cz>
31404 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31405 vector. Fix trailing white spaces.
31407 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31409 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31412 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31414 PR rtl-optimization/68664
31415 * config/arm/arm.c (arm_sched_can_speculate_insn):
31416 New function. Declare prototype.
31417 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31419 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31421 PR rtl-optimization/68664
31422 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31424 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31426 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31428 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31429 max skip bytes for function, loop and jump.
31431 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31433 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31434 ABS_EXPR for gimple dump.
31436 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31439 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31441 PR tree-optimization/79408
31442 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31443 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31444 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31445 also if rhs1 is INTEGER_CST.
31447 2017-02-14 Richard Biener <rguenther@suse.de>
31449 PR middle-end/79432
31450 * tree-into-ssa.c (insert_phi_nodes): When the function can
31451 have abnormal edges rewrite SSA names with broken use-def
31452 dominance out of SSA and register them for PHI insertion.
31454 2017-02-13 Martin Sebor <msebor@redhat.com>
31456 PR middle-end/79496
31457 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31458 clearing info.nowrite flag when snprintf size argument is a range.
31460 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31462 * cprop.c (cprop_jump): Add missing space in string literal.
31463 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31464 (get_constraint_for_component_ref): Likewise.
31465 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31466 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31467 * lra-constraints.c (process_alt_operands): Likewise.
31468 * ipa-inline.c (inline_small_functions): Likewise.
31469 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31470 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31471 * trans-mem.c (diagnose_tm_1_op): Likewise.
31472 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31473 (grid_parallel_clauses_gridifiable): Likewise.
31475 * config/nvptx/mkoffload.c (process): Add space in between
31478 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31479 "MOD4_SSE_REGS" and "ALL_REGS".
31481 * spellcheck.c (test_data): Add , in between "foo" and "food".
31483 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31486 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31487 boundary crossing check and subsequent code generation agree.
31489 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31491 * config/aarch64/aarch64.c (has_memory_op): Delete.
31492 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31495 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31497 PR rtl-optimization/79388
31498 PR rtl-optimization/79450
31499 * combine.c (distribute_notes): When removing TEM_INSN for which
31500 corresponding dest has last value recorded, invalidate that last
31503 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31505 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31506 of explicit '@'. Add missing assembly comment marker on branch costs
31509 2017-02-13 Nathan Sidwell <nathan@acm.org>
31511 * gengtype-lex.l (<in_struct>): Add '/'.
31513 2017-02-13 Martin Liska <mliska@suse.cz>
31516 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31518 2017-02-13 Richard Biener <rguenther@suse.de>
31520 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31522 * configure: Re-generate.
31523 * config.in: Likewise.
31524 * graphite-dependences.c: Simplify as if
31525 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31526 * graphite-isl-ast-to-gimple.c: Likewise.
31527 * graphite-optimize-isl.c: Likewise.
31528 * graphite-poly.c: Likewise.
31529 * graphite-sese-to-poly.c: Likewise.
31530 * graphite.h: Likewise.
31531 * toplev.c: Include isl/version.h and use isl_version () for
31532 printing the ISL version.
31533 * doc/install.texi: Update ISL requirement.
31535 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31537 * doc/standards.texi (Standards): Update reference to
31540 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31542 * doc/extend.texi (Named Address Spaces): sourceware.org now
31544 * doc/install.texi (Binaries): Ditto.
31547 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31549 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31550 standard terminology "stringize"/"stringizing" throughout.
31551 * doc/cppinternals.texi: Likewise.
31553 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31555 * doc/extend.texi: Fix some spelling mistakes and typos.
31556 * doc/invoke.texi: Likewise.
31558 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31561 * params.def (inline-min-speedup) Change from 10 to 8.
31563 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31565 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31568 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31571 * ipa-inline-analysis.c (get_minimal_bb): New function.
31572 (record_modified): Use it.
31573 (remap_edge_change_prob): Handle also ancestor functions.
31575 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31577 * doc/contrib.texi (Contributors): Remove broken link into
31578 the Mauve CVS repository.
31580 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31582 PR middle-end/79454
31583 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31584 result computation whenever lhs doesn't have vector mode, not
31585 just when it has BLKmode.
31587 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31589 * doc/makefile.texi (profiledbootstrap): Refer to the
31590 installation instructions only in textual form.
31592 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31595 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31597 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31599 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31600 (Specific): Update mingw-w64 reference.
31602 (Specific): Remove broken link to Renesas RX processor.
31604 2017-02-10 Richard Biener <rguenther@suse.de>
31606 * toplev.c (process_options): Do not mention obsolete graphite
31607 options when printing sorry message about missing graphite support.
31608 Mention -floop-nest-optimize.
31610 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31612 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31613 (vtst_p16): Likewise.
31614 (vtstq_p8): Likewise.
31615 (vtstq_p16): Likewise.
31617 (vtstq_p64): Likewise.
31618 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31619 (vset_lane_p64): New.
31620 (vsetq_lane_p64): New.
31622 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31624 PR tree-optimization/79411
31625 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31626 stmt operands are SSA_NAMEs used in abnormal phis.
31627 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31630 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31633 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31636 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31638 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31640 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31642 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31646 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31647 not arbitrary TREE_CONSTANT.
31650 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31651 "omp declare target link" attribute unless is_global_var.
31652 * omp-offload.c (find_link_var_op): Likewise.
31654 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31655 Chung-Lin Tang <cltang@codesourcery.com>
31657 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31659 (gimplify_adjust_omp_clauses): Don't delete TILE.
31660 (gimplify_omp_for): Deal with TILE.
31661 * internal-fn.c (expand_GOACC_TILE): New function.
31662 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31664 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31665 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31667 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31668 avoid DIV for outermost collapse var.
31669 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31670 Remove out of date comments, fix whitespace.
31671 * omp-general.c (omp_extract_for_data): Deal with tiling.
31672 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31673 adjust OLF_DIM_BASE value.
31674 (struct omp_for_data): Add tiling field.
31675 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31676 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31677 for auto loops. Remove default auto determining, moved to
31678 oacc_loop_fixed_partitions.
31679 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31680 stmts, add e_mask field.
31681 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31682 (oacc_thread_numbers): Use oacc_dim_call.
31683 (oacc_xform_tile): New.
31684 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31685 (finish_oacc_loop): Adjust for ifns vector.
31686 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31687 add case for GOACC_TILE fns.
31688 (oacc_loop_xform_loop): Delete.
31689 (oacc_loop_process): Iterate over call list directly, and add
31690 handling for GOACC_TILE fns.
31691 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31693 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31694 vector partitioning to outer loops. Assign 2 partitions to loops
31695 when available. Add TILE handling.
31696 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31697 (execite_oacc_device_lower): Process GOACC_TILE fns,
31698 ignore unknown specs.
31699 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31700 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31701 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31703 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31705 * configure.ac (ACX_BUGURL): Update.
31706 * configure: Regenerate.
31708 2017-02-09 Richard Biener <rguenther@suse.de>
31710 PR tree-optimization/69823
31711 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31712 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31714 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31716 * config/arc/arc-c.def: Add __NPS400__ definition.
31717 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31718 (TARGET_NPS400): Define.
31720 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31722 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31724 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31725 pointer, arch_info.
31726 (arc_cpu_types): Fill the arch_info field with a pointer into the
31727 arc_arch_types table.
31728 (arc_selected_cpu): Declare.
31729 * config/arc/arc.c (arc_selected_cpu): Make global.
31730 (arc_selected_arch): Delete.
31731 (arc_base_cpu): Delete.
31732 (arc_override_options): Remove references to deleted variables,
31733 update access to arch information.
31734 (ARC_OPT): Update access to arch information.
31735 (ARC_OPTX): Likewise.
31736 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31737 (TARGET_ARC600): Update access to arch information.
31738 (TARGET_ARC601): Likewise.
31739 (TARGET_ARC700): Likewise.
31740 (TARGET_EM): Likewise.
31741 (TARGET_HS): Likewise.
31742 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31745 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31748 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31749 condition/operands for integer GE/LE/GEU/LEU operations.
31751 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31753 PR translation/79397
31754 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31757 2017-02-08 Martin Jambor <mjambor@suse.cz>
31760 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31761 whether allocation happened.
31762 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31763 nothing was allocated.
31765 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31767 PR tree-optimization/79408
31768 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31769 constant, but SSA_NAME with a known integer range, use the minimum
31770 of that range instead of op1 to determine if modulo can be replaced
31771 with its first operand.
31773 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31775 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31777 2017-02-08 Richard Biener <rguenther@suse.de>
31779 PR tree-optimization/71824
31780 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31781 Check all loops contained in the merged region.
31783 2017-02-07 Andrew Pinski <apinski@cavium.com>
31785 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31787 2017-02-07 Andrew Pinski <apinski@cavium.com>
31789 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31790 (thunderxt88): Likewise.
31791 (thunderxt81): Disable LSE and change v8.1 to v8.
31792 (thunderxt83): Likewise.
31794 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31795 Richard Biener <rguenther@suse.de>
31797 PR middle-end/79399
31798 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31799 type from int to size_t.
31800 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31802 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31804 PR rtl-optimization/79386
31805 * cprop.c (bypass_conditional_jumps): Initialize
31806 bypass_last_basic_block already before splitting bbs after
31807 unconditional traps...
31808 (bypass_conditional_jumps): ... rather than here.
31811 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31812 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31813 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31814 fix -masm=intel patterns.
31816 2017-02-07 Richard Biener <rguenther@suse.de>
31818 PR tree-optimization/79256
31819 PR middle-end/79278
31820 * builtins.c (get_object_alignment_2): Use min_align_of_type
31821 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31822 and ADJUST_FIELD_ALIGN.
31824 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31826 * doc/tm.texi: Regenerate.
31827 * stor-layout.c (layout_decl): Adjust.
31828 (update_alignment_for_field): Likewise.
31829 (place_field): Likewise.
31830 (min_align_of_type): Likewise.
31831 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31832 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31833 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31834 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31835 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31836 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31837 * config/i386/i386.c (x86_field_alignment): Likewise.
31838 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31839 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31840 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31841 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31842 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31843 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31847 2017-01-30 Richard Biener <rguenther@suse.de>
31849 PR tree-optimization/79256
31850 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31851 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31854 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31856 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31857 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31858 builtins to SImode and emit a zero-extend, if necessary.
31860 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31862 * docs/invoke.texi (RISC-V Options): Alphabetize.
31864 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31866 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31869 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31871 * config/riscv/riscv.c: New file.
31872 * common/config/riscv/riscv-common.c: Likewise.
31873 * config.gcc: Likewise.
31874 * config/riscv/constraints.md: Likewise.
31875 * config/riscv/elf.h: Likewise.
31876 * config/riscv/generic.md: Likewise.
31877 * config/riscv/linux.h: Likewise.
31878 * config/riscv/multilib-generator: Likewise.
31879 * config/riscv/peephole.md: Likewise.
31880 * config/riscv/pic.md: Likewise.
31881 * config/riscv/predicates.md: Likewise.
31882 * config/riscv/riscv-builtins.c: Likewise.
31883 * config/riscv/riscv-c.c: Likewise.
31884 * config/riscv/riscv-ftypes.def: Likewise.
31885 * config/riscv/riscv-modes.def: Likewise.
31886 * config/riscv/riscv-opts.h: Likewise.
31887 * config/riscv/riscv-protos.h: Likewise.
31888 * config/riscv/riscv.h: Likewise.
31889 * config/riscv/riscv.md: Likewise.
31890 * config/riscv/riscv.opt: Likewise.
31891 * config/riscv/sync.md: Likewise.
31892 * config/riscv/t-elf-multilib: Likewise.
31893 * config/riscv/t-linux: Likewise.
31894 * config/riscv/t-linux-multilib: Likewise.
31895 * config/riscv/t-riscv: Likewise.
31896 * configure.ac: Likewise.
31897 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31898 Waterman as RISC-V maintainers.
31899 * doc/install.texi: Add RISC-V entries.
31900 * doc/invoke.texi: Add RISC-V options section.
31901 * doc/md.texi: Add RISC-V constraints section.
31902 * configure: Regenerated.
31904 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31907 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31908 false values to be constant vectors with all 0 or all 1 bits set.
31909 (vcondu<mode><mode>): Likewise.
31910 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31912 (fpmask_comparison_operator): Update comment.
31913 (vecint_comparison_operator): New predicate.
31914 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31915 vector conditionals when the true and false values are constant
31916 vectors with all 0 bits or all 1 bits set.
31918 2017-02-06 Martin Sebor <msebor@redhat.com>
31920 PR tree-optimization/79376
31921 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31923 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31925 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31926 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31927 to simplify split condition.
31929 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31931 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31932 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31935 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31937 PR rtl-optimization/68664
31938 * target.def (can_speculate_insn): New hook.
31939 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31940 * doc/tm.texi: Regenerate.
31941 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31942 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31943 (rs6000_sched_can_speculate_insn): New function.
31945 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31947 PR tree-optimization/79284
31948 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31949 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31950 vectorizable_mask_load_store, vectorizable_operation,
31951 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31952 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31953 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31954 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31955 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31956 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31957 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31958 is_gimple_assign (stmt). Replace another such test with
31959 is_gimple_assign (stmt).
31961 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31964 * config/avr/avr.c (rtl-iter.h): Include it.
31965 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31966 (avr_legitimate_combined_insn): ...and implementation.
31968 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31970 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31971 * config/s390/s390.c (s390_const_operand_ok)
31972 (s390_canonicalize_comparison, s390_extract_part)
31973 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31974 (s390_contiguous_bitmask_p, s390_rtx_costs)
31975 (legitimize_pic_address): Likewise.
31976 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31977 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31978 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31979 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31980 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31982 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31984 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31985 REGNO($0) == REGNO($1).
31987 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31989 * config/s390/linux.h(SIZE_TYPE): Add comment.
31991 2017-02-06 Julian Brown <julian@codesourcery.com>
31992 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31993 Virendra Pathak <virendra.pathak@broadcom.com>
31995 * config/aarch64/aarch64-cores.def: Change the scheduler
31997 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31998 * config/aarch64/thunderx2t99.md: New file.
32000 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32002 * doc/standards.texi (Go Language): Update link to language
32005 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32007 * tree-eh.c (lower_resx): Sanitize profile.
32008 (cleanup_empty_eh_move_lp): Likewise.
32010 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32013 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32015 * cfgloopmanip.h (loop_version): Update prototype.
32016 * modulo-sched.c (sms_schedule): Update call of loop_version.
32017 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32018 * tree-parloops.c (gen_parallel_loop): Likewise.
32019 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32020 * tree-ssa-loop-split.c (split_loop): Likewise.
32021 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32022 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32024 2017-02-05 Martin Liska <mliska@suse.cz>
32027 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32029 (print_operand_address): Initialize a struct to zero.
32031 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32033 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32034 garbage collector only in textual form.
32036 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32038 * doc/extend.texi (x86 specific memory model extensions for
32039 transactional memory): Simplify a phrase.
32041 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32044 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32045 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32046 (atomic_storedi_1): Likewise.
32048 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32050 PR tree-optimization/79338
32051 * tree-parloops.c (gather_scalar_reductions): Don't call
32052 vect_analyze_loop_form for loop->inner before destroying loop's
32055 2017-02-03 Martin Sebor <msebor@redhat.com>
32057 PR tree-optimization/79327
32058 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32059 when precision has resulted in leading zeros.
32060 (format_integer): Adjust the likely counter to assume an unknown
32061 argument that may be zero is non-zero.
32063 2017-02-03 Jason Merrill <jason@redhat.com>
32066 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32067 avoid copying non-taken branch.
32069 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32071 PR tree-optimization/79340
32072 * tree-vect-loop.c (vectorizable_reduction): Release
32073 vec_defs elements after safe_splicing them into other vectors.
32076 PR tree-optimization/79327
32077 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32078 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32080 (format_integer): Use wide_int_to_tree instead of build_int_cst
32081 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32082 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32083 of shortest and longest sequence.
32085 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32087 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32088 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32090 2017-02-03 Walter Lee <walt@tilera.com>
32093 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32094 after initial stackframe link reg save.
32095 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32097 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32100 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32101 wu for stxssp alternative.
32103 2017-02-03 Martin Sebor <msebor@redhat.com>
32105 PR tree-optimization/79352
32106 * gimple-fold.c (get_range_strlen): Add argument.
32107 (get_range_strlen): Change return type to bool.
32108 (get_maxval_strlen): Pass in a dummy argument.
32109 * gimple-fold.h (get_range_strlen): Change return type to bool.
32110 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32111 * tree.h (array_at_struct_end_p): Add argument.
32112 * tree.c (array_at_struct_end_p): Handle it.
32114 2017-02-03 Martin Liska <mliska@suse.cz>
32117 * multiple_target.c (create_dispatcher_calls): Redirect edge
32118 from a caller of a dispatcher.
32119 (expand_target_clones): Make the clones local.
32120 (ipa_target_clone): Do both target clones and resolvers.
32121 (ipa_dispatcher_calls): Remove the pass.
32122 (pass_dispatcher_calls::gate): Likewise.
32123 (make_pass_dispatcher_calls): Likewise.
32124 * passes.def (pass_target_clone): Put as very first IPA early
32127 2017-02-03 Martin Liska <mliska@suse.cz>
32129 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32130 in case of a function with ifunc attribute.
32132 2017-02-03 Martin Liska <mliska@suse.cz>
32134 * cgraph.c (cgraph_node::dump): Dump function version info.
32135 * symtab.c (symtab_node::dump_base): Add missing new line.
32137 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32139 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32140 (ifcombine_ifandif): Use it.
32142 2017-02-03 Martin Liska <mliska@suse.cz>
32144 * doc/invoke.texi: Document default value for
32145 use-after-scope-direct-emission-threshold.
32147 2017-02-03 Martin Liska <mliska@suse.cz>
32149 PR tree-optimization/79339
32150 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32151 (format_floating): Likewise.
32153 2017-02-03 Martin Liska <mliska@suse.cz>
32156 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32158 (ipa_node_params_t::remove): Likewise.
32159 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32160 initialization from removed ipa_node_params_t::insert.
32161 (ipa_node_params::~ipa_node_params): Move from removed
32162 ipa_node_params_t::release.
32163 * symbol-summary.h (symbol_summary::m_released): New member.
32164 Do not release a summary twice. Do not allow to call finalizer
32165 for types of a summary that live in GGC memory.
32167 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32169 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32172 2017-02-02 Martin Sebor <msebor@redhat.com>
32174 PR middle-end/79275
32175 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32176 (format_string): Tighten up the range of output for non-constant
32177 strings and correct the expected range for wide non-constant strings.
32179 2017-02-02 Martin Sebor <msebor@redhat.com>
32181 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32183 PR middle-end/32003
32184 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32186 (-fdump-tree-@var): Add to index and document how to come up
32187 with pass-specific option and dump file names.
32188 (-fdump-passes): Clarify where to look for output.
32190 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32192 PR middle-end/77445
32193 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32194 statistics of the analyzed path; allow threading for speed when
32195 any of BBs along the path are optimized for speed.
32197 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32199 PR middle-end/78468
32200 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32201 settings of the virtual registers.
32204 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32206 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32207 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32210 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32212 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32213 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32215 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32217 * config/s390/s390.md: Add missing comments with the expanded
32219 * config/s390/vector.md: Likewise.
32220 * config/s390/vx-builtins.md: Likewise.
32222 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32225 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32226 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32227 conditions on a single line.
32229 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32231 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32232 __S390_VX__ to __VX__.
32234 2017-02-01 Andrew Pinski <apinski@cavium.com>
32236 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32237 stmt_info to record_stmt_cost.
32238 (vect_get_known_peeling_cost): Pass stmt_info if known to
32240 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32241 cpu_vector_cost field into
32242 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32243 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32244 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32245 splitting of scalar_stmt_cost and vec_stmt_cost.
32246 (thunderx_vector_cost): Likewise.
32247 (cortexa57_vector_cost): LIkewise.
32248 (exynosm1_vector_cost): Likewise.
32249 (xgene1_vector_cost): Likewise.
32250 (thunderx2t99_vector_cost): Improve after the splitting of the two
32252 (aarch64_builtin_vectorization_cost): Update for the splitting of
32253 scalar_stmt_cost and vec_stmt_cost.
32255 2017-02-01 Torvald Riegel <triegel@redhat.com>
32256 Richard Henderson <rth@redhat.com>
32258 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32259 conditional on existance of a fast atomic load.
32260 * optabs-query.c (can_atomic_load_p): New function.
32261 * optabs-query.h (can_atomic_load_p): Declare it.
32262 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32263 no fast atomic load is available for the particular size of access.
32264 (expand_atomic_compare_and_swap): Likewise.
32265 (expand_atomic_load): Likewise.
32266 (expand_atomic_store): Likewise.
32267 (expand_atomic_fetch_op): Likewise.
32268 * testsuite/lib/target-supports.exp
32269 (check_effective_target_sync_int_128): Remove x86 because it provides
32270 no fast atomic load.
32271 (check_effective_target_sync_int_128_runtime): Likewise.
32273 2017-02-01 Richard Biener <rguenther@suse.de>
32275 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32276 (graphite_transform_loops): Provide opt-info for optimized nests.
32277 * tree-parloop.c (parallelize_loops): Provide opt-info for
32278 parallelized loops.
32280 2017-02-01 Richard Biener <rguenther@suse.de>
32282 PR middle-end/79315
32283 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32284 was not set before.
32286 2017-02-01 Richard Biener <rguenther@suse.de>
32288 PR tree-optimization/71824
32289 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32290 Verify the loops are valid in the merged SESE region.
32291 (scop_detection::can_represent_loop_1): Check analyzing the
32292 evolution of the number of iterations in the region succeeds.
32294 2017-01-31 Ian Lance Taylor <iant@golang.org>
32296 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32297 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32299 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32301 PR preprocessor/79210
32302 * input.c (get_substring_ranges_for_loc): Replace line_width
32303 assertion with error-handling.
32305 2017-01-31 Richard Biener <rguenther@suse.de>
32307 PR tree-optimization/77318
32308 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32309 (create_pw_aff_from_tree): Take loop parameter.
32310 (add_condition_to_pbb): Pass loop of the condition to
32311 create_pw_aff_from_tree.
32313 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32315 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32316 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32318 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32322 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32324 (convert_int_to_float128): Likewise.
32325 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32326 (convert_int_to_float128): Likewise.
32327 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32328 (UNSPEC_IEEE128_CONVERT): Likewise.
32329 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32330 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32331 Use local variables for IBM extended format.
32332 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32333 (fix_trunc<mode>si2_fprs): Likewise.
32334 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32335 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32336 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32337 to know that we can now have integers of all sizes in vector
32339 (fix<uns>_<mode>di2_hw): Likewise.
32340 (float<uns>_<mode>si2_hw): Likewise.
32341 (fix_<mode>si2_hw): Likewise.
32342 (fixuns_<mode>si2_hw): Likewise.
32343 (float<uns>_<mode>di2_hw): Likewise.
32344 (float_<mode>di2_hw): Likewise.
32345 (float_<mode>si2_hw): Likewise.
32346 (floatuns_<mode>di2_hw): Likewise.
32347 (floatuns_<mode>si2_hw): Likewise.
32348 (xscvqp<su>wz_<mode>): Delete, no longer used.
32349 (xscvqp<su>dz_<mode>): Likewise.
32350 (xscv<su>dqp_<mode>): Likewise.
32351 (ieee128_mfvsrd_64bit): Likewise.
32352 (ieee128_mfvsrd_32bit): Likewise.
32353 (ieee128_mfvsrwz): Likewise.
32354 (ieee128_mtvsrw): Likewise.
32355 (ieee128_mtvsrd_64bit): Likewise.
32356 (ieee128_mtvsrd_32bit): Likewise.
32358 2017-01-31 Martin Liska <mliska@suse.cz>
32361 * ipa-prop.c (ipa_free_all_node_params): Call release method
32362 instead of ~sumbol_summary to not to trigger double times
32365 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32367 PR tree-optimization/71691
32368 * bitmap.h (class auto_bitmap): New.
32369 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32370 is_maybe_undefined instead of ssa_undefined_value_p.
32372 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32374 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32375 __S390_ARCH_LEVEL__ to __ARCH__.
32377 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32379 PR tree-optimization/79267
32380 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32381 if should_remove_lhs_p is true.
32383 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32386 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32387 (add_alignment_attribute): New.
32388 (base_type_die): Add alignment attribute.
32389 (subrange_type_die): Likewise.
32390 (modified_type_die): Likewise.
32391 (gen_array_type_die): Likewise.
32392 (gen_descr_array_type_die: Likewise.
32393 (gen_enumeration_type_die): Likewise.
32394 (gen_subprogram_die): Likewise.
32395 (gen_variable_die): Likewise.
32396 (gen_field_die): Likewise.
32397 (gen_ptr_to_mbr_type_die): Likewise.
32398 (gen_struct_or_union_type_die): Likewise.
32399 (gen_subroutine_type_die): Likewise.
32400 (gen_typedef_die): Likewise.
32401 (base_type_cmp): Compare alignment attribute.
32403 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32406 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32407 (setb_unsigned) New pattern for setb with CCUNS.
32408 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32409 subfc./subfe sequence to avoid overflow problems. Generate a
32410 shorter sequence with cmpld/setb for power9.
32411 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32412 for generating subfc. instruction.
32413 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32414 now uses this instruction.
32416 2017-01-30 Ian Lance Taylor <iant@google.com>
32419 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32420 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32422 2017-01-30 Martin Sebor <msebor@redhat.com>
32424 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32425 Move constant to the right of a relational operator.
32426 (get_mpfr_format_length, format_character, format_string): Ditto.
32427 (should_warn_p, maybe_warn): Same.
32429 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32431 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32434 * asan.c (get_translation_unit_decl): Remove function.
32435 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32437 2017-01-30 Martin Liska <mliska@suse.cz>
32439 PR gcov-profile/79259
32440 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32441 -fprofile-generate.
32443 2017-01-30 Martin Liska <mliska@suse.cz>
32446 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32447 Initialize variables with NULL value.
32449 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32452 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32454 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32456 2017-01-30 Richard Biener <rguenther@suse.de>
32458 PR tree-optimization/79276
32459 * tree-vrp.c (process_assert_insertions): Properly adjust common
32460 when removing a duplicate.
32462 2017-01-30 Richard Biener <rguenther@suse.de>
32464 PR tree-optimization/79256
32465 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32466 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32468 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32470 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32473 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32474 ("*r<noxa>sbg_<mode>_sll_bitmask")
32475 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32476 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32477 Use contiguous_bitmask_nowrap_operand.
32479 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32482 * config/rs6000/altivec.h (vec_xl): Revise #define.
32483 (vec_xst): Likewise.
32485 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32487 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32489 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32491 PR rtl-optimization/79194
32492 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32493 traps before call to bypass_conditional_jumps.
32495 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32497 PR tree-optimization/71374
32498 * lra-constraints.c (check_conflict_input_operands): New.
32499 (match_reload): Use it.
32501 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32504 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32505 account to calculate conflict_set.
32507 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32509 PR rtl-optimization/78559
32510 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32511 other_insn in combine.
32513 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32515 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32516 uint16_type_node for BT_UINT16.
32518 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32520 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32521 "RTL Tests" to menu.
32522 (GIMPLE Tests): New node.
32523 (RTL Tests): New node.
32525 2017-01-27 Richard Biener <rguenther@suse.de>
32527 PR tree-optimization/79245
32528 * tree-loop-distribution.c (distribute_loop): Apply cost
32529 modeling also to detected patterns.
32531 2017-01-27 Richard Biener <rguenther@suse.de>
32533 PR tree-optimization/71433
32534 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32535 (compare_assert_loc): New function.
32536 (process_assert_insertions): Sort and optimize assert locations
32537 to remove duplicates and push down identical assertions on
32538 edges to their destination block.
32540 2017-01-27 Richard Biener <rguenther@suse.de>
32542 PR tree-optimization/79244
32543 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32544 out SSA names even if abnormal.
32546 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32548 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32549 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32550 instead of MPFR_RNDN.
32552 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32555 * arm.c (arm_option_override): Don't call build_target_option_node
32556 until after doing all option overrides.
32557 (arm_valid_target_attribute_tree): Likewise.
32559 2017-01-27 Martin Liska <mliska@suse.cz>
32561 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32562 for {cd}tors and C++ {cd}tors.
32564 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32566 * config/s390/s390.md ("*setmem_long_and")
32567 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32569 2017-01-26 Martin Sebor <msebor@redhat.com>
32571 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32574 2017-01-26 Martin Sebor <msebor@redhat.com>
32576 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32577 HAVE_DFmode before using XFmode or DFmode.
32578 (parse_directive): Avoid using the z length modifier to avoid
32579 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32581 PR middle-end/78703
32582 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32583 to accept adjustment as an array.
32584 (get_int_range): New function.
32585 (struct directive): Make width and prec arrays.
32586 (directive::set_width, directive::set_precision): Call get_int_range.
32587 (format_integer, format_floating): Handle width and precision ranges.
32588 (format_string, parse_directive): Same.
32590 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32593 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32594 comdat_type_p set, just clone them, but keep the children in the
32598 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32599 which have direct callers with -fvar-tracking-assignments enabled
32601 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32602 inside of type units.
32604 2017-01-26 Martin Sebor <msebor@redhat.com>
32606 PR middle-end/78703
32607 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32609 (struct format_result): Replace number_chars, number_chars_min,
32610 and number_chars_max with a single member of struct result_range.
32612 (format_result::operator+=): Adjust.
32613 (struct fmtresult): Remove bounded. Handle likely and unlikely
32615 (fmtresult::adjust_for_width_or_precision): New function.
32616 (fmtresult:type_max_digits): New function.
32617 (bytes_remaining): Handle likely and unlikely counters.
32618 (min_bytes_remaining): Remove.
32619 (format_percent): Simplify.
32620 (format_integer, format_floating): Set likely and unlikely counters.
32621 (get_string_length, format_character, format_string): Same.
32622 (format_plain, should_warn_p): New function.
32623 (maybe_warn): Call should_warn_p. Update diagnostic messages
32624 and handle those for all directives, including plain strings.
32625 (format_directive): Handle likely and unlikely counters.
32626 Remove unnecessary quoting from diagnostics. Add an informational
32628 (add_bytes): Remove.
32629 (pass_sprintf_length::compute_format_length): Simplify.
32630 (try_substitute_return_value): Handle likely and unlikely counters.
32632 2017-01-26 Carl Love <cel@us.ibm.com>
32634 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32635 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32637 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32640 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32641 endianess for subregs into account.
32642 * lra-constraints.c (lra_constraints): Do risky transformations
32643 always on the first iteration.
32644 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32645 last_call_used_reg_set.
32646 (process_bb_lives): Define and use last_call_used_reg_set.
32647 * lra.c (lra): Always continue after lra_constraints on the first
32650 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32652 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32654 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32656 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32658 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32659 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32660 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32661 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32662 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32663 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32664 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32665 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32666 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32668 2017-01-26 Marek Polacek <polacek@redhat.com>
32671 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32672 for the third operand.
32674 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32676 PR middle-end/79236
32677 * omp-low.c (struct omp_context): Add simt_stmt field.
32678 (scan_omp_for): Return omp_context *.
32679 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32680 context to the _simt_ SIMD stmt.
32681 (lower_omp_for): For combined SIMD with sibling _simt_
32682 SIMD, make sure to use the same decls in _looptemp_
32683 clauses as in the sibling.
32685 2017-01-26 David Sherwood <david.sherwood@arm.com>
32687 PR middle-end/79212
32688 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32691 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32694 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32695 emit fld b; fld a; if possible.
32697 * brig-builtins.def: Update copyright years.
32698 * config/arm/arm_acle_builtins.def: Update copyright years.
32700 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32703 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32704 constraint instead of o for the stxsd instruction.
32706 2017-01-25 Carl Love <cel@us.ibm.com>
32708 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32709 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32711 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32713 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32715 2017-01-25 Richard Biener <rguenther@suse.de>
32717 PR tree-optimization/69264
32718 * target.def (vector_alignment_reachable): Improve documentation.
32719 * doc/tm.texi: Regenerate.
32720 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32722 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32723 earlier changes with respect to TYPE_USER_ALIGN.
32724 (vector_alignment_reachable_p): Likewise. Improve dumping.
32726 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32729 * config/arm/arm.md (xordi3): Force constant operand into a register
32732 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32734 * doc/invoke.texi (-fstore-merging): Correct default optimization
32735 levels at which it is enabled.
32736 (-O): Move -fstore-merging from list to...
32739 2017-01-25 Richard Biener <rguenther@suse.de>
32742 * omp-expand.c: Include debug.h.
32743 (expand_omp_taskreg): Make sure to generate early debug before
32744 outlining anything from a function.
32745 (expand_omp_target): Likewise.
32746 (grid_expand_target_grid_body): Likewise.
32748 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32751 * asan.c (get_translation_unit_decl): New function.
32752 (asan_add_global): Extract modules file name from globals
32753 TRANSLATION_UNIT_DECL name.
32755 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32758 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32759 for long calls with APCS frame and VFP.
32761 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32763 * cfg.c (original_copy_tables_initialized_p): New function.
32764 * cfg.h (original_copy_tables_initialized_p): New decl.
32765 * cfgrtl.c (relink_block_chain): Guard the call to
32766 free_original_copy_tables with a call to
32767 original_copy_tables_initialized_p.
32768 * cgraph.h (symtab_node::native_rtl_p): New decl.
32769 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32770 (symtab_node::needed_p): Don't assert for early assembly output
32771 for __RTL functions.
32772 (cgraph_node::finalize_function): Set "force_output" for __RTL
32774 (cgraph_node::analyze): Bail out early for __RTL functions.
32775 (analyze_functions): Update assertion to support __RTL functions.
32776 (cgraph_node::expand): Bail out early for __RTL functions.
32777 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32779 * function.h (struct function): Update comment for field
32781 * gimple-expr.c: Include "tree-pass.h".
32782 (gimple_has_body_p): Return false for __RTL functions.
32783 * Makefile.in (OBJS): Add run-rtl-passes.o.
32784 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32786 (gcc::pass_manager::get_clean_slate): New accessor.
32787 * passes.c: Include "insn-addr.h".
32788 (should_skip_pass_p): Add logging. Update logic for running
32789 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32790 property-provider override so it is only done for gimple passes.
32792 (skip_pass): New function.
32793 (execute_one_pass): Call skip_pass when skipping passes.
32794 * read-md.c (md_reader::read_char): Support filtering
32795 the input to a subset of line numbers.
32796 (md_reader::md_reader): Initialize fields
32797 m_first_line and m_last_line.
32798 (md_reader::read_file_fragment): New function.
32799 * read-md.h (md_reader::read_file_fragment): New decl.
32800 (md_reader::m_first_line): New field.
32801 (md_reader::m_last_line): New field.
32802 * read-rtl-function.c (function_reader::create_function): Only
32803 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32804 curr_properties. Set DECL_INITIAL to a dummy block.
32805 (read_rtl_function_body_from_file_range): New function.
32806 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32808 * run-rtl-passes.c: New file.
32809 * run-rtl-passes.h: New file.
32811 2017-01-24 Jeff Law <law@redhat.com>
32813 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32816 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32818 PR tree-optimization/79159
32819 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32820 (record_nonwrapping_iv): Improve boundary using above function if no
32821 value range information.
32823 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32824 Martin Jambor <mjambor@suse.cz>
32826 * brig-builtins.def: New file.
32827 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32828 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32829 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32830 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32831 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32832 * builtin-types.def (BT_INT8): New.
32833 (BT_INT16): Likewise.
32834 (BT_UINT8): Likewise.
32835 (BT_UINT16): Likewise.
32836 (BT_FN_ULONG): Likewise.
32837 (BT_FN_UINT_INT): Likewise.
32838 (BT_FN_UINT_ULONG): Likewise.
32839 (BT_FN_UINT_LONG): Likewise.
32840 (BT_FN_UINT_PTR): Likewise.
32841 (BT_FN_ULONG_PTR): Likewise.
32842 (BT_FN_INT8_FLOAT): Likewise.
32843 (BT_FN_INT16_FLOAT): Likewise.
32844 (BT_FN_UINT32_FLOAT): Likewise.
32845 (BT_FN_UINT16_FLOAT): Likewise.
32846 (BT_FN_UINT8_FLOAT): Likewise.
32847 (BT_FN_UINT64_FLOAT): Likewise.
32848 (BT_FN_UINT16_UINT32): Likewise.
32849 (BT_FN_UINT32_UINT16): Likewise.
32850 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32851 (BT_FN_INT_PTR_INT): Likewise.
32852 (BT_FN_UINT_PTR_UINT): Likewise.
32853 (BT_FN_LONG_PTR_LONG): Likewise.
32854 (BT_FN_ULONG_PTR_ULONG): Likewise.
32855 (BT_FN_VOID_UINT64_UINT64): Likewise.
32856 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32857 (BT_FN_INT8_INT8_INT8): Likewise.
32858 (BT_FN_INT16_INT16_INT16): Likewise.
32859 (BT_FN_INT_INT_INT): Likewise.
32860 (BT_FN_UINT_FLOAT_UINT): Likewise.
32861 (BT_FN_FLOAT_UINT_UINT): Likewise.
32862 (BT_FN_ULONG_UINT_UINT): Likewise.
32863 (BT_FN_ULONG_UINT_PTR): Likewise.
32864 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32865 (BT_FN_UINT_UINT_UINT): Likewise.
32866 (BT_FN_VOID_UINT_PTR): Likewise.
32867 (BT_FN_UINT_UINT_PTR: Likewise.
32868 (BT_FN_UINT32_UINT64_PTR): Likewise.
32869 (BT_FN_INT_INT_UINT_UINT): Likewise.
32870 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32871 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32872 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32873 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32874 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32875 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32876 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32877 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32878 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32879 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32880 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32881 * doc/frontends.texi: List BRIG FE.
32882 * doc/install.texi (Testing): Add BRIG tesring requirements.
32883 * doc/invoke.texi (Overall Options): Mention BRIG.
32884 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32886 2017-01-24 Richard Biener <rguenther@suse.de>
32888 PR translation/79208
32889 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32891 2017-01-24 Martin Jambor <mjambor@suse.cz>
32894 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32895 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32896 and known_contexts.
32898 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32900 PR middle-end/79123
32901 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32902 casts from signed to unsigned really don't have a range.
32904 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32906 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32907 GMP_RNDx for compatiblity.
32909 2017-01-24 Martin Liska <mliska@suse.cz>
32912 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32913 that would prevent us to call alloca with -1 as argument.
32915 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32917 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32918 Avoid -Wformat-security warning.
32920 2017-01-23 Andrew Pinski <apinski@cavium.com>
32922 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32925 2017-01-23 Martin Sebor <msebor@redhat.com>
32927 PR middle-end/78703
32928 * gimple-ssa-sprintf.c (warn_level): New global.
32929 (format_integer): Use it here and throughout the rest of the file.
32930 Use the same switch to compute sign as base.
32931 (maybe_warn): New function.
32932 (format_directive): Factor out warnings into maybe_warn.
32933 Add debugging output. Use warn_level.
32934 (add_bytes): Use warn_level.
32935 (pass_sprintf_length::compute_format_length): Add debugging output.
32936 (try_substitute_return_value): Same.
32937 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32939 PR middle-end/78703
32940 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32941 (struct fmtresult, format_integer, format_floating): Adjust.
32942 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32943 (get_string_length, format_string,format_directive): Same.
32944 (pass_sprintf_length::compute_format_length): Same.
32945 (try_substitute_return_value): Simplify slightly.
32947 PR middle-end/78703
32948 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32949 (fmtresult::operator+=): Outlined.
32950 (struct fmtresult): Add ctors.
32951 (struct conversion_spec): Rename...
32952 (struct directive): ...to this. Add and remove data members.
32953 (directive::set_width, directive::set_precision): New functions.
32954 (format_percent): Use fmtresult ctor.
32955 (get_width_and_precision): Remove.
32956 (format_integer): Make naming changes. Avoid computing width and
32958 (format_floating): Same. Adjust indentation.
32959 (format_character, format_none): New functions.
32960 (format_string): Moved character handling to format_character.
32961 (format_directive): Remove arguments, change return type.
32962 (parse_directive): New function.
32963 (pass_sprintf_length::compute_format_length): Move directive
32964 parsing to parse_directive.
32966 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32968 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32969 (assign_assembler_name_if_needed): ... this.
32970 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32971 (assign_assembler_name_if_needed): ... this.
32972 (free_lang_data_in_cgraph): Adjust callers.
32973 * cgraphunit.c (cgraph_node::analyze): Likewise.
32974 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32977 2017-01-23 Richard Biener <rguenther@suse.de>
32979 PR tree-optimization/79088
32980 PR tree-optimization/79188
32981 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32982 resetting loop bounds after last path deletion. Reset loop
32983 bounds of the target loop, make code match the comments.
32984 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32985 Make sure loops need no fixups.
32987 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32989 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32990 exponent support with double type for first argument.
32991 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32992 type returned by __builtin_vec_extract_sig,
32993 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32994 functions from "vector int" to "vector unsigned int" or from
32995 "vector long long int" to "vector unsigned long long int".
32996 Changed type returned by __builtin_vec_extract_exp,
32997 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32998 functions from "vector int" to "vector unsigned int" or from
32999 "vector long long int" to "vector unsigned long long int".
33000 Changed return type of __builtin_vec_test_data_class,
33001 __builtin_vec_test_data_class_sp, and
33002 __builtin_vec_test_data_class_dp from "vector int" to
33003 "vector bool int" or from "vector long long int" to "vector bool
33004 long long int" and changed second argument type from "unsigned
33005 int" to "int". Added new overloaded function forms "vector float
33006 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33007 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33008 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33009 double, vector unsigned long long int)" and "vector double
33010 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33011 long int)". Changed return type of
33012 __builtin_scalar_test_data_class and
33013 __builtin_scalar_test_data_class_sp and
33014 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33015 int" and changed second argument from "unsigned int" to "int".
33016 Changed type returned by __builtin_scalar_test_neg,
33017 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33018 from "int" to "bool int". Added new overloaded function form
33019 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33020 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33021 exponent double-precision with floating point first argument.
33022 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33023 documentation of scalar_test_data_class, scalar_test_neg,
33024 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33025 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33026 vec_test_data_class built-in functions to reflect refinements in
33027 their type signatures.
33029 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33031 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33033 (aarch64_elf_asm_destructor): Likewise.
33035 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33037 PR rtl-optimization/78634
33038 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33039 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33040 * ifcvt.c (noce_try_cmove): Add missing cost check.
33042 PR rtl-optimization/71724
33043 * combine.c (if_then_else_cond): Look for situations where it is
33044 beneficial to undo the work of one of the recursive calls.
33046 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33048 PR tree-optimization/70754
33049 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33050 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33051 combined stmt before it if not NULL.
33052 (combine_chains): Process refs reversely and compute dominance point
33055 2017-01-23 Martin Liska <mliska@suse.cz>
33057 PR tree-optimization/79196
33058 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33059 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33061 (strlen_optimize_stmt): Call the renamed function.
33063 2017-01-23 Michael Matz <matz@suse.de>
33065 PR tree-optimization/78384
33066 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33068 2017-01-23 Richard Biener <rguenther@suse.de>
33070 PR tree-optimization/79186
33071 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33072 both incoming edges before moving an assert.
33074 2017-01-23 Martin Jambor <mjambor@suse.cz>
33076 * ipa-prop.c (load_from_param_1): Removed.
33077 (load_from_unmodified_param): Bits from load_from_param_1 put back
33079 (load_from_param): Removed.
33080 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33081 with stmt. Reverted back to use of load_from_unmodified_param.
33083 2017-01-23 Martin Jambor <mjambor@suse.cz>
33086 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33087 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33088 field a pointer to garbage collected vector, mark lattices and
33089 ipcp_orig_node with GTY((skip)).
33090 (ipa_get_param_count): Adjust to descriptors being a pointer.
33091 (ipa_get_param): Likewise.
33092 (ipa_get_type): Likewise.
33093 (ipa_get_param_move_cost): Likewise.
33094 (ipa_set_param_used): Likewise.
33095 (ipa_get_controlled_uses): Likewise.
33096 (ipa_set_controlled_uses): Likewise.
33097 (ipa_is_param_used): Likewise.
33098 (ipa_node_params_t): Move into garbage collector. New methods insert
33100 (ipa_node_params_sum): Annotate wth GTY(()).
33101 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33103 (ipa_load_from_parm_agg): Adjust declaration.
33104 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33105 * ipa-profile.c (ipa_profile): Likewise.
33106 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33107 (ipa_populate_param_decls): Make descriptors parameter garbage
33109 (ipa_dump_param): Adjust to descriptors being a pointer.
33110 (ipa_alloc_node_params): Likewise.
33111 (ipa_initialize_node_params): Likewise.
33112 (load_from_param_1): Make descriptors parameter garbage collected.
33113 (load_from_unmodified_param): Likewise.
33114 (load_from_param): Likewise.
33115 (ipa_load_from_parm_agg): Likewise.
33116 (ipa_node_params::~ipa_node_params): Removed.
33117 (ipa_free_all_node_params): Remove call to delete operator.
33118 (ipa_node_params_t::insert): New.
33119 (ipa_node_params_t::remove): Likewise.
33120 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33121 copy known_csts and known_contexts vectors.
33122 (ipa_read_node_info): Adjust to descriptors being a pointer.
33123 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33125 (ipcp_transform_function): Make descriptors variable garbage
33128 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33130 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33131 * config/i386/avx512dqintrin.h: Ditto.
33132 * config/i386/avx512fintrin.h: Ditto.
33133 * config/i386/i386.c: Handle new builtins.
33134 * config/i386/i386-builtin.def: Add new builtins.
33135 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33136 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33138 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33139 Martin Liska <mliska@suse.cz>
33141 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33142 * asan.c (asan_expand_poison_ifn): Support stores and use
33143 appropriate ASAN report function.
33144 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33145 * internal-fn.def (ASAN_POISON_USE): Declare.
33146 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33147 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33148 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33149 ASAN_POISON calls w/o LHS.
33150 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33151 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33152 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33153 * gimplify.c (asan_poison_variables): Add attribute
33154 use_after_scope_memory to variables that really needs to live
33156 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33157 having the attribute.
33159 2017-01-23 Martin Liska <mliska@suse.cz>
33161 * asan.c (create_asan_shadow_var): New function.
33162 (asan_expand_poison_ifn): Likewise.
33163 * asan.h (asan_expand_poison_ifn): New declaration.
33164 * internal-fn.c (expand_ASAN_POISON): Likewise.
33165 * internal-fn.def (ASAN_POISON): New builtin.
33166 * sanopt.c (pass_sanopt::execute): Expand
33167 asan_expand_poison_ifn.
33168 * tree-inline.c (copy_decl_for_dup_finish): Make function
33170 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33171 * tree-ssa.c (is_asan_mark_p): New function.
33172 (execute_update_addresses_taken): Rewrite local variables
33173 (identified just by use-after-scope as addressable) into SSA.
33175 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33177 * doc/install.texi (Specific): opensource.apple.com uses https
33178 now. Remove trailing slash.
33180 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33182 * README.Portability: Remove note on an Irix compatibility issue.
33184 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33186 * gcov.c (INCLUDE_ALGORITHM): Define.
33187 (INCLUDE_VECTOR): Define.
33188 No longer include <vector> and <algorithm> directly.
33190 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33192 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33194 * doc/invoke.texi (Code Gen Options): Ditto.
33196 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33199 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33201 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33203 rtl-optimization/79125
33204 * cprop.c (local_cprop_pass): Handle cases where we make an
33205 unconditional trap.
33207 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33211 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33212 read from, for big endian.
33214 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33216 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33217 register pauth builtins for LP64 only.
33219 2017-01-20 Marek Polacek <polacek@redhat.com>
33222 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33225 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33227 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33229 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33230 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33231 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33233 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33236 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33237 in big-endian mode when they are not a single duplicated value.
33239 2017-01-20 Richard Biener <rguenther@suse.de>
33241 * BASE-VER: Bump to 7.0.1.
33243 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33245 * omp-low.c (omplow_simd_context): New struct. Use it...
33246 (lower_rec_simd_input_clauses): ...here and...
33247 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33248 references to idx, lane, max_vf, is_simt.
33250 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33252 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33255 2017-01-20 Martin Jambor <mjambor@suse.cz>
33257 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33258 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33260 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33261 (GTFILES): Rename hsa.c to hsa-common.c.
33262 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33263 * hsa-dump.c: Likewise.
33264 * hsa-gen.c: Likewise.
33265 * hsa-regalloc.c: Likewise.
33266 * ipa-hsa.c: Likewise.
33267 * omp-expand.c: Likewise.
33268 * omp-low.c: Likewise.
33269 * toplev.c: Likewise.
33271 2017-01-20 Marek Polacek <polacek@redhat.com>
33274 * doc/invoke.texi: Document -Wduplicated-branches.
33275 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33276 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33277 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33278 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33279 return 0 only when not OEP_LEXICOGRAPHIC.
33280 (fold_build_cleanup_point_expr): Use the expression
33281 location when building CLEANUP_POINT_EXPR.
33282 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33283 * tree.c (add_expr): Handle error_mark_node.
33285 2017-01-20 Martin Liska <mliska@suse.cz>
33288 * tree-profile.c (init_ic_make_global_vars): Do not call
33290 (gimple_init_gcov_profiler): Likewise.
33292 2017-01-20 Martin Liska <mliska@suse.cz>
33295 * cgraph.h (maybe_create_reference): Remove argument and
33297 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33299 * ipa-cp.c (create_specialized_node): Likewise.
33300 * symtab.c (symtab_node::maybe_create_reference): Handle
33301 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33303 2017-01-20 Martin Liska <mliska@suse.cz>
33305 * read-rtl-function.c (function_reader::create_function): Use
33306 build_decl instread of build_decl_stat.
33308 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33310 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33311 * config/i386/avx512dqintrin.h: Ditto.
33312 * config/i386/avx512fintrin.h: Ditto.
33313 * config/i386/i386-builtin-types.def: Add new types.
33314 * config/i386/i386.c: Handle new types.
33315 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33316 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33317 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33318 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33319 (__builtin_ia32_kshiftridi): New.
33320 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33322 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33326 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33327 define to rs6000_init_stack_protect_guard.
33328 (rs6000_init_stack_protect_guard): New function.
33330 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33331 Yunqiang Su <yunqiang.su@imgtec.com>
33333 * config.gcc (supported_defaults): Add madd4.
33334 (with_madd4): Add validation.
33335 (all_defaults): Add madd4.
33336 * config/mips/mips.opt (mmadd4): New option.
33337 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33339 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33341 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33342 (ISA_HAS_FUSED_MADD4): Likewise.
33343 * doc/invoke.texi (-mmadd4): Document the new option.
33344 * doc/install.texi (--with-madd4): Document the new option.
33346 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33348 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33349 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33350 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33351 (aarch64_init_pauth_hint_builtins): New.
33352 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33353 (aarch64_expand_builtin): Expand new builtins.
33355 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33357 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33358 * combine-stack-adj.c (no_unhandled_cfa): Handle
33359 REG_CFA_TOGGLE_RA_MANGLE.
33360 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33361 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33362 info for return address signing.
33363 (aarch64_expand_epilogue): Likewise.
33365 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33367 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33368 * config/aarch64/aarch64-protos.h
33369 (aarch64_return_address_signing_enabled): New declaration.
33370 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33372 (aarch64_expand_prologue): Sign return address before it's pushed onto
33374 (aarch64_expand_epilogue): Authenticate return address fetched from
33376 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33377 (aarch64_attributes): New function attributes for "sign-return-address".
33378 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33379 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33380 ("*do_return"): Generate combined instructions according to key index.
33381 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33382 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33384 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33385 * config/aarch64/aarch64.opt (msign-return-address=): New.
33386 * doc/extend.texi (AArch64 Function Attributes): Documents
33387 "sign-return-address=".
33388 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33390 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33392 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33393 overall option summary.
33395 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33397 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33398 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33399 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33400 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33402 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33404 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33405 -mpower9-minmax by default for -mcpu=power9.
33406 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33407 128-bit floating point.
33409 2017-01-20 Alan Modra <amodra@gmail.com>
33411 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33412 optimizing for size.
33414 2017-01-20 Alan Modra <amodra@gmail.com>
33417 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33418 for strcmp and strncmp from corresponding builtin decl.
33420 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33422 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33423 instead of i386/rtems-64.h.
33424 * config/i386/rtems-64.h: Remove.
33426 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33430 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33432 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33434 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33436 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33437 Change int to HOST_WIDE_INT.
33438 * config/aarch64/aarch64-protos.h
33439 (aarch64_simd_gen_const_vector_dup): Likewise.
33440 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33442 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33444 * langhooks-def.h (lhd_type_for_size): New decl.
33445 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33446 * langhooks.c (lhd_type_for_size): New function, taken from
33449 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33451 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33452 define_bypass for CR latency.
33453 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33454 (power9-alu2): Add define_bypass for CR latency.
33456 (power9-mul): Update insn latency.
33457 (power9-mul-compare): Update insn latency, bypass latency and remove
33460 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33462 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33464 * config/aarch64/aarch64.md
33465 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33466 aarch64_nopcrelative_literal_loads.
33467 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33469 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33471 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33472 TARGET_LOONGSON_3A.
33473 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33475 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33478 * config.gcc (supported_defaults): Add lxc1-sxc1.
33479 (with_lxc1_sxc1): Add validation.
33480 (all_defaults): Add lxc1-sxc1.
33481 * config/mips/mips.opt (mlxc1-sxc1): New option.
33482 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33484 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33485 __mips_no_lxc1_sxc1.
33486 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33487 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33488 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33490 2017-01-19 Richard Biener <rguenther@suse.de>
33492 PR tree-optimization/72488
33493 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33494 sure to restore SSA info.
33495 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33497 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33499 PR rtl-optimization/79121
33500 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33501 of the inner type when shifting an extended value.
33503 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33506 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33507 interposable aliases.
33509 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33512 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33513 Use the evmergelohi instruction.
33514 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33515 (mov_sitf_e500_subreg8_2_be): Likewise.
33516 (mov_sitf_e500_subreg12_2_le): Likewise.
33517 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33518 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33519 (mov_sitf_e500_subreg8_2_le): Likewise.
33520 (mov_sitf_e500_subreg12_2_be): Likewise.
33522 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33524 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33525 attribute from vecsimple to vecperm.
33526 (altivec_vbpermq2): Likewise.
33528 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33531 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33533 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33534 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33535 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33536 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33537 case where N arg is SIZE_MAX.
33538 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33539 (cmpstrsi): Add pattern.
33541 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33543 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33544 __builtin_vec_revb builtins.
33545 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33546 built-in functions to support generation of the ISA 3.0 XXBR<x>
33547 vector byte reverse instructions.
33548 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33549 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33550 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33551 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33552 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33553 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33554 (P9V_BUILTIN_VEC_REVB): Likewise.
33555 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33556 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33557 (p9_xxbrq_v16qi): Likewise.
33558 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33559 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33560 (p9_xxbrh_v8hi): Likewise.
33561 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33562 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33563 vec_revb built-in functions.
33565 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33567 PR rtl-optimization/78952
33568 * config/i386/i386.md (any_extract): New code iterator.
33569 (*insvqi_2): Use any_extract for source operand.
33570 (*insvqi_3): Use any_shiftrt for source operand.
33572 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33574 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33576 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33578 2017-01-18 Matthias Klose <doko@ubuntu.com>
33580 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33582 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33584 * config/rs6000/altivec.h (vec_bperm): Change #define.
33585 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33586 (altivec_vbpermq2): New define_insn.
33587 (altivec_vbpermd): Likewise.
33588 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33589 function interface.
33590 (VBPERMD): Likewise.
33591 (VBPERM): New polymorphic function interface.
33592 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33593 Add entries for P9V_BUILTIN_VEC_VBPERM.
33594 * doc/extend.texi: Add interfaces for vec_bperm.
33596 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33598 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33599 first letter of error messages.
33600 (s390_resolve_overloaded_builtin): Likewise.
33601 * config/s390/s390.c (s390_expand_builtin): Likewise.
33602 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33603 (s390_valid_target_attribute_inner_p): Likewise.
33604 * config/s390/s390.md ("tabort"): Likewise.
33606 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33608 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33609 (ISA_AVOID_DIV_HILO): New macro.
33610 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33611 (ISA_HAS_DDIV): Likewise.
33613 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33615 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33617 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33619 * doc/invoke.texi (fabi-version): Spelling fix.
33621 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33624 * doc/invoke.texi (fabi-version): Mention mangling fix for
33627 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33630 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33632 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33635 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33636 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33638 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33640 (stack_protect_test): Ditto.
33641 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33642 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33644 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33645 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33646 -mstack-protector-guard-offset=.
33647 (RS/6000 and PowerPC Options): Ditto.
33649 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33651 * config/i386/i386.h (MASK_CLASS_P): New define.
33652 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33653 there are no registers from different register sets also when
33654 mask registers are used. Update function comment.
33655 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33656 to (*k/*r) and (*k/*km) alternatives.
33658 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33660 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33661 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33662 (EH_RETURN_HANDLER_RTX): New define.
33663 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33664 Force frame pointer in EH return functions.
33665 (aarch64_expand_epilogue): Add barrier for eh_return.
33666 (aarch64_final_eh_return_addr): Remove.
33667 (aarch64_eh_return_handler_rtx): New function.
33668 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33670 (aarch64_eh_return_handler_rtx): New prototype.
33672 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33674 * config/rs6000/altivec.h (vec_rlmi): New #define.
33675 (vec_vrlnm): Likewise.
33676 (vec_rlnm): Likewise.
33677 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33678 (UNSPEC_VRLNM): Likewise.
33679 (VIlong): New mode iterator.
33680 (altivec_vrl<VI_char>mi): New define_insn.
33681 (altivec_vrl<VI_char>nm): Likewise.
33682 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33684 (VRLDNM): Likewise.
33685 (RLNM): New polymorphic function entry.
33686 (VRLWMI): New monomorphic function entry.
33687 (VRLDMI): Likewise.
33688 (RLMI): New polymorphic function entry.
33689 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33690 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33691 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33694 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33697 * dwarf2out.c (field_byte_offset): Restore the
33698 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33699 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33700 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33703 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33706 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33708 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33711 * dwarf2out.c (add_data_member_location_attribute): For constant
33712 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33713 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33714 DW_AT_byte_size attributes.
33716 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33718 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33719 after forcing to constant memory when the code model is medium.
33721 2017-01-17 Julia Koval <julia.koval@intel.com>
33724 * config/i386/avx512fintrin.h
33725 (_mm512_i32gather_ps): Change __addr type to void const*.
33726 (_mm512_mask_i32gather_ps): Ditto.
33727 (_mm512_i32gather_pd): Ditto.
33728 (_mm512_mask_i32gather_pd): Ditto.
33729 (_mm512_i64gather_ps): Ditto.
33730 (_mm512_mask_i64gather_ps): Ditto.
33731 (_mm512_i64gather_pd): Ditto.
33732 (_mm512_mask_i64gather_pd): Ditto.
33733 (_mm512_i32gather_epi32): Ditto.
33734 (_mm512_mask_i32gather_epi32): Ditto.
33735 (_mm512_i32gather_epi64): Ditto.
33736 (_mm512_mask_i32gather_epi64): Ditto.
33737 (_mm512_i64gather_epi32): Ditto.
33738 (_mm512_mask_i64gather_epi32): Ditto.
33739 (_mm512_i64gather_epi64): Ditto.
33740 (_mm512_mask_i64gather_epi64): Ditto.
33741 (_mm512_i32scatter_ps): Change __addr type to void*.
33742 (_mm512_mask_i32scatter_ps): Ditto.
33743 (_mm512_i32scatter_pd): Ditto.
33744 (_mm512_mask_i32scatter_pd): Ditto.
33745 (_mm512_i64scatter_ps): Ditto.
33746 (_mm512_mask_i64scatter_ps): Ditto.
33747 (_mm512_i64scatter_pd): Ditto.
33748 (_mm512_mask_i64scatter_pd): Ditto.
33749 (_mm512_i32scatter_epi32): Ditto.
33750 (_mm512_mask_i32scatter_epi32): Ditto.
33751 (_mm512_i32scatter_epi64): Ditto.
33752 (_mm512_mask_i32scatter_epi64): Ditto.
33753 (_mm512_i64scatter_epi32): Ditto.
33754 (_mm512_mask_i64scatter_epi32): Ditto.
33755 (_mm512_i64scatter_epi64): Ditto.
33756 (_mm512_mask_i64scatter_epi64): Ditto.
33757 * config/i386/avx512pfintrin.h
33758 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33759 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33760 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33761 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33762 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33763 (_mm512_prefetch_i32scatter_ps): Ditto.
33764 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33765 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33766 (_mm512_prefetch_i64scatter_pd): Ditto.
33767 (_mm512_prefetch_i64scatter_ps): Ditto.
33768 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33769 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33770 * config/i386/avx512vlintrin.h
33771 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33772 (_mm_mmask_i32gather_ps): Ditto.
33773 (_mm256_mmask_i32gather_pd): Ditto.
33774 (_mm_mmask_i32gather_pd): Ditto.
33775 (_mm256_mmask_i64gather_ps): Ditto.
33776 (_mm_mmask_i64gather_ps): Ditto.
33777 (_mm256_mmask_i64gather_pd): Ditto.
33778 (_mm_mmask_i64gather_pd): Ditto.
33779 (_mm256_mmask_i32gather_epi32): Ditto.
33780 (_mm_mmask_i32gather_epi32): Ditto.
33781 (_mm256_mmask_i32gather_epi64): Ditto.
33782 (_mm_mmask_i32gather_epi64): Ditto.
33783 (_mm256_mmask_i64gather_epi32): Ditto.
33784 (_mm_mmask_i64gather_epi32): Ditto.
33785 (_mm256_mmask_i64gather_epi64): Ditto.
33786 (_mm_mmask_i64gather_epi64): Ditto.
33787 (_mm256_i32scatter_ps): Change __addr type to void*.
33788 (_mm256_mask_i32scatter_ps): Ditto.
33789 (_mm_i32scatter_ps): Ditto.
33790 (_mm_mask_i32scatter_ps): Ditto.
33791 (_mm256_i32scatter_pd): Ditto.
33792 (_mm256_mask_i32scatter_pd): Ditto.
33793 (_mm_i32scatter_pd): Ditto.
33794 (_mm_mask_i32scatter_pd): Ditto.
33795 (_mm256_i64scatter_ps): Ditto.
33796 (_mm256_mask_i64scatter_ps): Ditto.
33797 (_mm_i64scatter_ps): Ditto.
33798 (_mm_mask_i64scatter_ps): Ditto.
33799 (_mm256_i64scatter_pd): Ditto.
33800 (_mm256_mask_i64scatter_pd): Ditto.
33801 (_mm_i64scatter_pd): Ditto.
33802 (_mm_mask_i64scatter_pd): Ditto.
33803 (_mm256_i32scatter_epi32): Ditto.
33804 (_mm256_mask_i32scatter_epi32): Ditto.
33805 (_mm_i32scatter_epi32): Ditto.
33806 (_mm_mask_i32scatter_epi32): Ditto.
33807 (_mm256_i32scatter_epi64): Ditto.
33808 (_mm256_mask_i32scatter_epi64): Ditto.
33809 (_mm_i32scatter_epi64): Ditto.
33810 (_mm_mask_i32scatter_epi64): Ditto.
33811 (_mm256_i64scatter_epi32): Ditto.
33812 (_mm256_mask_i64scatter_epi32): Ditto.
33813 (_mm_i64scatter_epi32): Ditto.
33814 (_mm_mask_i64scatter_epi32): Ditto.
33815 (_mm256_i64scatter_epi64): Ditto.
33816 (_mm256_mask_i64scatter_epi64): Ditto.
33817 (_mm_i64scatter_epi64): Ditto.
33818 (_mm_mask_i64scatter_epi64): Ditto.
33819 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33820 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33821 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33822 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33823 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33824 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33825 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33826 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33827 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33828 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33829 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33830 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33831 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33832 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33833 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33834 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33835 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33836 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33837 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33838 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33839 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33840 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33841 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33842 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33843 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33844 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33845 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33846 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33847 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33848 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33849 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33850 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33851 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33852 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33853 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33854 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33855 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33856 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33857 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33858 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33859 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33860 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33861 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33862 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33863 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33864 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33865 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33866 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33867 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33868 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33869 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33870 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33871 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33872 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33873 definitions accordingly.
33875 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33876 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33879 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33882 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33885 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33886 hard regs for inner regno.
33888 2017-01-17 Martin Liska <mliska@suse.cz>
33891 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33892 assumption and add comment.
33894 2017-01-17 Nathan Sidwell <nathan@acm.org>
33896 * ipa-visibility.c (localize_node): New function, broken out of ...
33897 (function_and_variable_visibility): ... here. Call it.
33899 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33901 PR middle-end/77445
33902 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33903 correctly set frequency of oudgoing edge.
33904 (duplicate_thread_path): Fix profile updating.
33906 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33909 * configure.ac: Add GCC_BASE_VER.
33910 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33911 version from BASE-VER file.
33912 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33913 (gcc.o): Depend on $(BASEVER).
33914 * common.opt (dumpfullversion): New option.
33915 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33916 * doc/invoke.texi: Document -dumpfullversion.
33917 * doc/install.texi: Document --with-gcc-major-version-only.
33918 * configure: Regenerated.
33920 2017-01-17 Richard Biener <rguenther@suse.de>
33922 PR tree-optimization/71433
33923 * tree-vrp.c (register_new_assert_for): Merge same asserts
33924 on all incoming edges.
33925 (process_assert_insertions_for): Handle insertions at the
33928 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33930 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33931 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33933 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33936 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33939 2017-01-17 Alan Modra <amodra@gmail.com>
33942 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33943 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33944 symbolic stack limit when pic.
33946 2017-01-16 Martin Sebor <msebor@redhat.com>
33948 PR tree-optimization/78608
33949 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33951 2017-01-16 Jeff Law <law@redhat.com>
33954 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33955 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33956 for several include directories that may be relative to sysroot.
33957 * config/i386/x-mingw32 (gplus_includedir): Define.
33958 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33959 (native_system_includedir): Likewise.
33960 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33961 override if TARGET_SYSTEM_ROOT is defined.
33962 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33964 PR tree-optimization/79090
33965 PR tree-optimization/33562
33966 PR tree-optimization/61912
33967 PR tree-optimization/77485
33968 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33969 and computed trims into the dump file.
33971 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33973 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33975 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33978 * gimplify.c (gimplify_init_constructor): If want_value and
33979 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33983 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33984 sequence. Formatting fixes.
33985 (doloop_optimize): Formatting fixes.
33988 * gcc.c (debug_level_greater_than_spec_func): New function.
33989 (static_spec_functions): Add debug-level-gt spec function.
33990 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33992 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33993 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33994 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33995 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33996 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33997 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33999 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34001 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34002 QImode fixups to general and mask registers only.
34004 2017-01-16 Carl Love <cel@us.ibm.com>
34006 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34007 for built-in functions
34008 vector signed char vec_nabs (vector signed char)
34009 vector signed short vec_nabs (vector signed short)
34010 vector signed int vec_nabs (vector signed int)
34011 vector signed long long vec_nabs (vector signed long long)
34012 vector float vec_nabs (vector float)
34013 vector double vec_nabs (vector double)
34014 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34016 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34017 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34018 * doc/extend.texi: Update the documentation file for the new built-in
34021 2017-01-16 Martin Sebor <msebor@redhat.com>
34023 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34026 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34028 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34029 UNSPEC_VSX__XXSPLTD to require special splat handling.
34031 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34034 * system.h: Poison strndup.
34036 2017-01-16 Alan Modra <amodra@gmail.com>
34039 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34042 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34044 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34046 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34048 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34049 call recog here. Assert that INSN_CODE (insn) is non-negative.
34051 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34054 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34056 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34057 in the currently scheduled RTL fragment.
34059 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34061 PR rtl-optimization/78751
34062 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34065 2017-01-14 Jeff Law <law@redhat.com>
34067 PR tree-optimization/79090
34068 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34069 variable length stores.
34070 (compute_trims): Delete dead assignment to *trim_tail.
34071 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34074 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34076 PR rtl-optimization/78626
34077 PR rtl-optimization/78727
34078 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34079 of a block, and split such blocks after everything else is finished.
34081 2017-01-14 Alan Modra <amodra@gmail.com>
34084 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34085 target legitimate_combined_insn.
34086 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34087 (rs6000_legitimate_combined_insn): New function.
34088 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34090 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34091 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34092 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34094 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34096 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34098 2017-01-13 Jeff Law <law@redhat.com>
34100 PR tree-optimization/33562
34101 PR tree-optimization/61912
34102 PR tree-optimization/77485
34103 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34105 (delete_dead_assignment): Likewise.
34106 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34107 statement to delete_dead_call and delete_dead_assignment.
34109 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34112 * substring-locations.c (format_warning_va): Strengthen case 1 so
34113 that both endpoints of the substring must be within the format
34114 range for just the substring to be printed.
34116 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34118 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34119 * config/i386/i386.c (ix86_target_string): Add missing options
34120 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34121 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34122 flags_other and ix86_target_other to flags2_other. Display unknown
34124 (ix86_valid_target_attribute_inner_p): Add missing options and
34125 reorder options by implied ISAs, as in ix86_target_string.
34127 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34129 * hash-table.h (hash_table::too_empty_p): New function.
34130 (hash_table::expand): Use it.
34131 (hash_table::traverse): Likewise.
34132 (hash_table::empty_slot): Use sizeof (value_type) instead of
34133 sizeof (PTR) to convert bytes to elements. Shrink the table
34134 if the current size is excessive for the current number of
34137 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34139 * ira-costs.c (record_reg_classes): Break from the inner loop
34140 early once alt_fail is known to be true. Update outer loop
34141 handling accordingly.
34143 2017-01-13 Jeff Law <law@redhat.com>
34145 * tree-ssa-dse.c (decrement_count): New function.
34146 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34147 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34148 when we know the partially dead statement is a mem* function.
34150 PR tree-optimization/61912
34151 PR tree-optimization/77485
34152 * tree-ssa-dse.c: Include expr.h.
34153 (maybe_trim_constructor_store): New function.
34154 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34156 PR tree-optimization/33562
34157 PR tree-optimization/61912
34158 PR tree-optimization/77485
34159 * doc/invoke.texi: Document new dse-max-object-size param.
34160 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34161 * tree-ssa-dse.c: Include params.h.
34162 (dse_store_status): New enum.
34163 (initialize_ao_ref_for_dse): New, partially extracted from
34165 (valid_ao_ref_for_dse, normalize_ref): New.
34166 (setup_live_bytes_from_ref, compute_trims): Likewise.
34167 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34168 (maybe_trim_partially_dead_store): Likewise.
34169 (maybe_trim_complex_store): Likewise.
34170 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34171 Track what bytes live from the original store. Return tri-state
34172 for dead, partially dead or live.
34173 (dse_dom_walker): Add constructor, destructor and new private members.
34174 (delete_dead_call, delete_dead_assignment): New extracted from
34176 (dse_optimize_stmt): Make a member of dse_dom_walker.
34177 Use initialize_ao_ref_for_dse.
34179 PR tree-optimization/33562
34180 PR tree-optimization/61912
34181 PR tree-optimization/77485
34182 * sbitmap.h (bitmap_count_bits): Prototype.
34183 (bitmap_clear_range, bitmap_set_range): Likewise.
34184 * sbitmap.c (bitmap_clear_range): New function.
34185 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34187 2017-01-13 Martin Liska <mliska@suse.cz>
34190 * function.c (set_cfun): Add new argument force.
34191 * function.h (set_cfun): Likewise.
34192 * ipa-inline-transform.c (inline_call): Use the function when
34193 strict alising from is dropped for function we inline to.
34195 2017-01-13 Richard Biener <rguenther@suse.de>
34197 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34198 for dumping GIMPLE INTEGER_CSTs.
34200 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34202 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34203 to 201112L since C++17.
34205 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34208 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34209 if -fsanitize=kernel-address is present.
34211 2017-01-13 Richard Biener <rguenther@suse.de>
34213 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34214 as _Literal ( type ) number in case usual suffixes do not
34215 preserve all information.
34217 2017-01-13 Richard Biener <rguenther@suse.de>
34219 PR tree-optimization/77283
34220 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34221 and ssa-iterators.h.
34222 (is_feasible_trace): Implement a cost model based on joiner
34225 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34228 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34229 char or short to __float128/_Float128 directly.
34231 2017-01-12 Martin Sebor <msebor@redhat.com>
34233 to -Wformat-overflow.
34234 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34235 (min_bytes_remaining): Same.
34236 (get_string_length): Same.
34237 (format_string): Same.
34238 (format_directive): Same.
34240 (pass_sprintf_length::handle_gimple_call): Same.
34242 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34244 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34245 info.nowrite calls with no lhs that can't throw. Return bool
34246 whether gsi_remove has been called or not.
34247 (pass_sprintf_length::handle_gimple_call): Return bool whether
34248 try_substitute_return_value called gsi_remove. Formatting fix.
34249 (pass_sprintf_length::execute): Don't use gsi_remove if
34250 handle_gimple_call returned true.
34253 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34254 be removed due to side-effects, don't remove following barrier nor
34255 turn the successor edge into fallthru edge.
34257 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34260 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34261 element-reversing loads and stores as not swappable.
34263 2017-01-12 Nathan Sidwell <nathan@acm.org>
34264 Nicolai Stange <nicstange@gmail.com>
34266 * combine.c (try_combine): Don't ignore result of overlap checking
34267 loop. Combine overlap & asm check into single loop.
34269 2017-01-12 Richard Biener <rguenther@suse.de>
34271 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34272 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34274 2017-01-12 Richard Biener <rguenther@suse.de>
34276 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34277 and TS_TARGET_OPTION directly derive from TS_BASE.
34278 * tree-core.h (tree_optimization_option): Derive from tree_base.
34279 (tree_target_option): Likewise.
34281 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34283 * config/i386/i386.c (memory_address_length): Increase len
34284 only when rip_relative_addr_p returns false.
34286 2017-01-11 Julia Koval <julia.koval@intel.com>
34288 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34289 (OPTION_MASK_ISA_SGX_SET): New.
34290 (ix86_handle_option): Handle OPT_msgx.
34291 * config.gcc: Added sgxintrin.h.
34292 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34293 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34294 * config/i386/i386.c (ix86_target_string): Add -msgx.
34296 (ix86_option_override_internal): Handle new options.
34297 (ix86_valid_target_attribute_inner_p): Add sgx.
34298 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34299 * config/i386/i386.opt: Add msgx.
34300 * config/i386/sgxintrin.h: New file.
34301 * config/i386/x86intrin.h: Add sgxintrin.h.
34303 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34306 * fold-const.c (maybe_nonzero_address): Return 1 for function
34308 (tree_single_nonzero_warnv_p): Don't handle function local objects
34312 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34315 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34318 * opts.c: Include "spellcheck.h"
34319 (struct string_fragment): New struct.
34320 (struct edit_distance_traits<const string_fragment &>): New
34322 (get_closest_sanitizer_option): New function.
34323 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34325 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34327 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34329 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34330 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34331 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34332 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34333 for initial die_offset if dwarf_split_debug_info.
34334 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34335 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34337 (output_skeleton_debug_sections): Formatting fix. Use
34338 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34339 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34341 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34343 * config/arm/cortex-a53.md: Add bypasses for
34344 cortex_a53_r2f_cvt.
34345 (cortex_a53_r2f): Only use for transfers.
34346 (cortex_a53_f2r): Likewise.
34347 (cortex_a53_r2f_cvt): Add reservation for conversions.
34348 (cortex_a53_f2r_cvt): Likewise.
34350 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34352 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34353 to all inlined functions, change static to extern.
34355 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34358 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34360 (arm_assemble_integer): Likewise.
34362 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34364 * config.gcc: Use new awk script to check CPU, FPU and architecture
34365 parameters for --with-... options.
34366 * config/arm/parsecpu.awk: New file
34367 * config/arm/arm-cpus.in: New file.
34368 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34370 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34372 * config/arm/t-arm: Update dependency rules.
34373 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34374 of processing .def files.
34375 * config/arm/genopt.sh: Deleted.
34376 * config/arm/gentune.sh: Deleted.
34377 * config/arm/arm-cores.def: Deleted.
34378 * config/arm/arm-arches.def: Deleted.
34379 * config/arm/arm-fpus.def: Deleted.
34380 * config/arm/arm-tune.md: Regenerated.
34381 * config/arm/arm-tables.opt: Regenerated.
34382 * config/arm/arm-cpu.h: New generated file.
34383 * config/arm/arm-cpu-data.h: New generated file.
34384 * config/arm/arm-cpu-cdata.h: New generated file.
34386 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34389 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34391 (input_varpool_node): Unpack dynamically_initialized bit.
34393 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34395 PR rtl-optimization/79032
34396 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34397 the alignment of the adjusted memory reference against that of MODE,
34398 instead of the alignment of the original memory reference.
34400 2017-01-11 Martin Jambor <mjambor@suse.cz>
34402 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34404 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34405 decorated functions.
34407 2017-01-11 Richard Biener <rguenther@suse.de>
34409 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34410 set range/nonnull info for PHI results. Do not set it on
34411 stmts marked for removal.
34413 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34415 * expr.c (store_field): In the bitfield case, fetch the return value
34416 from the registers before applying a single big-endian adjustment.
34417 Always do a final load for a BLKmode value not larger than a word.
34419 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34422 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34423 that we correctly handle column numbers greater than
34424 LINE_MAP_MAX_COLUMN_NUMBER.
34426 2017-01-10 Martin Sebor <msebor@redhat.com>
34428 PR middle-end/78245
34429 * gimple-ssa-sprintf.c (get_destination_size): Call
34430 {init,fini}object_sizes.
34431 * tree-object-size.c (addr_object_size): Adjust.
34432 (pass_through_call): Adjust.
34433 (pass_object_sizes::execute): Adjust.
34434 * tree-object-size.h (fini_object_sizes): Declare.
34436 2017-01-10 Martin Sebor <msebor@redhat.com>
34438 PR tree-optimization/78775
34439 * builtins.c (get_size_range): Move...
34440 * calls.c: ...to here.
34441 (alloc_max_size): Accept zero argument.
34442 (operand_signed_p): Remove.
34443 (maybe_warn_alloc_args_overflow): Call get_size_range.
34444 * calls.h (get_size_range): Declare.
34446 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34448 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34449 from TI's devices.csv file as of September 2016.
34450 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34452 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34454 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34455 * doc/invoke.texi: Likewise.
34456 * doc/md.texi: Likewise.
34457 * doc/objc.texi: Likewise.
34459 2017-01-10 Joshua Conner <joshconner@google.com>
34461 * config/arm/fuchsia-elf.h: New file.
34462 * config/fuchsia.h: New file.
34463 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34464 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34466 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34468 2016-01-10 Richard Biener <rguenther@suse.de>
34470 PR tree-optimization/79034
34471 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34472 Propagate out degenerate PHIs in the joiner.
34474 2017-01-10 Martin Liska <mliska@suse.cz>
34476 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34477 (sort_congruence_classes_by_decl_uid): Likewise.
34478 (sort_congruence_class_groups_by_decl_uid): Likewise.
34479 (sem_item_optimizer::merge_classes): Sort class, groups in these
34480 classes and members in the groups by DECL_UID of declarations.
34481 This would make merge operations stable.
34483 2017-01-10 Martin Liska <mliska@suse.cz>
34485 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34486 usage of m_classes_vec.
34487 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34488 (sem_item_optimizer::get_group_by_hash): Likewise.
34489 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34490 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34491 (sem_item_optimizer::verify_classes): Likewise.
34492 (sem_item_optimizer::process_cong_reduction): Likewise.
34493 (sem_item_optimizer::dump_cong_classes): Likewise.
34494 (sem_item_optimizer::merge_classes): Likewise.
34495 * ipa-icf.h (congruence_class_hash): Rename from
34496 congruence_class_group_hash. Remove declaration of m_classes_vec.
34498 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34500 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34501 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34502 * config.gcc: Add avx512vpopcntdqintrin.h.
34503 * config/i386/avx512vpopcntdqintrin.h: New.
34504 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34505 * config/i386/i386-builtin-types.def: Add new types.
34506 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34507 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34508 __builtin_ia32_vpopcountq_v8di_mask): New.
34509 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34510 __AVX512VPOPCNTDQ__.
34511 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34512 (PTA_AVX512VPOPCNTDQ): Define.
34513 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34514 TARGET_AVX512VPOPCNTDQ_P): Define.
34515 * config/i386/i386.opt: Add mavx512vpopcntdq.
34516 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34517 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34519 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34521 PR middle-end/77484
34522 * predict.def (PRED_CALL): Set to 67.
34524 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34526 * expr.c (store_field): In the bitfield case, if the value comes from
34527 a function call and is of an aggregate type returned in registers, do
34528 not modify the field mode; extract the value in all cases if the mode
34529 is BLKmode and the size is not larger than a word.
34531 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34534 * config/i386/cpuid.h: Fix undefined behavior.
34536 2017-01-04 Jeff Law <law@redhat.com>
34538 PR tree-optimization/79007
34539 PR tree-optimization/67955
34540 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34541 conservative for pt.null when flag_non_call_exceptions is on.
34543 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34545 PR translation/79019
34546 PR translation/79020
34547 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34548 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34549 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34551 * config/avr/avr.opt (maccumulate-args): Likewise.
34552 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34553 * common.opt (freport-bug): Likewise.
34554 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34555 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34556 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34557 translatable string.
34558 * config/i386/i386.c (function_value_32): Likewise.
34559 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34560 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34562 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34563 * common/config/msp430/msp430-common.c (msp430_handle_option):
34565 * symtab.c (symtab_node::verify_base): Likewise.
34566 * opts.c (set_debug_level): Likewise.
34567 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34568 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34569 missing whitespace to translatable strings.
34570 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34571 * config/sh/superh.h: Likewise.
34572 * config/i386/xopintrin.h: Likewise.
34573 * config/i386/znver1.md: Likewise.
34574 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34575 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34576 * double-int.h (struct double_int): Likewise.
34577 * double-int.c (div_and_round_double): Likewise.
34578 * wide-int.cc: Likewise.
34579 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34580 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34581 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34582 (crossjumps_occurred): ... this.
34583 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34586 PR tree-optimization/78899
34587 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34588 returning bool return struct loop *, NULL for failure and the new
34590 (versionable_outer_loop_p): Don't version outer loop if it has
34591 dont_vectorized bit set.
34592 (tree_if_conversion): When versioning outer loop, ensure
34593 tree_if_conversion is performed also on the inner loop of the
34594 non-vectorizable outer loop copy.
34595 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34596 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34597 prevent vectorization of it.
34598 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34599 the outer loop vectorization of the non-scalar version is attempted
34600 before vectorization of the inner loop in scalar version. If
34601 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34602 vectorization of its inner loop.
34603 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34604 has 2 inner loops, rename also on edges from bb whose single pred
34605 is outer_loop->header. Fix typo in function comment.
34607 2017-01-09 Martin Sebor <msebor@redhat.com>
34610 * asan.c (asan_emit_stack_protection): Increase local buffer size
34611 to avoid snprintf truncation warning.
34613 2017-01-09 Andrew Pinski <apinski@cavium.com>
34615 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34616 to reference thunderx2t99 for the tuning structure
34617 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34619 (thunderx2t99_extra_costs): This.
34620 * config/aarch64/aarch64-tune.md: Regenerate.
34621 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34622 (vulcan_addrcost_table): This.
34623 (vulcan_regmove_cost): Rename to ...
34624 (thunderx2t99_regmove_cost): This.
34625 (vulcan_vector_cost): Rename to ...
34626 (thunderx2t99_vector_cost): this.
34627 (vulcan_branch_cost): Rename to ...
34628 (thunderx2t99_branch_cost): This.
34629 (vulcan_tunings): Rename to ...
34630 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34631 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34633 2017-01-09 Martin Jambor <mjambor@suse.cz>
34637 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34638 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34639 (propagate_vr_accross_jump_function): Use the above function for all
34640 value range computations for pass-through jump functions and type
34641 converasion from explicit value range values.
34642 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34643 parameters from TYPE_ARG_TYPES.
34644 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34645 (ipa_write_node_info): Stream type of the actual argument.
34646 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34648 2017-01-09 Martin Liska <mliska@suse.cz>
34651 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34652 (lookup_compiler): Do not show error message with have_E.
34654 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34656 PR tree-optimization/78938
34657 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34658 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34659 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34660 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34663 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34665 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34668 2017-01-09 Richard Biener <rguenther@suse.de>
34670 PR tree-optimization/78997
34671 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34672 name condition properly.
34674 2017-01-09 Richard Biener <rguenther@suse.de>
34677 * dwarf2out.c (is_cxx): New overload with context.
34678 (is_naming_typedef_decl): Use it.
34680 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34682 * invoke.texi (Option Summary): Correct spacing in option lists
34683 and add line breaks to fix over-long lines.
34685 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34687 PR middle-end/17660
34689 * extend.texi (Common Variable Attributes): Add xref to GCC
34690 Internals manual to explain mode attribute keywords.
34692 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34695 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34696 and Preprocessor Options.
34697 (Options for Linking): Document -pthread here....
34698 (RS/6000 and PowerPC Options): ...not here.
34699 (Solaris 2 Options): ...or here.
34700 * doc/cppopts.texi: Document -pthread.
34702 2017-01-08 Martin Sebor <msebor@redhat.com>
34704 PR middle-end/77708
34705 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34706 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34707 New member functions.
34708 (format_directive): Used them.
34710 (pass_sprintf_length::handle_gimple_call): Same.
34711 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34712 to avoid truncation for any argument.
34713 (extract_affine_mul): Same.
34714 * tree.c (get_file_function_name): Same.
34716 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34718 PR middle-end/77484
34719 * predict.def (PRED_INDIR_CALL): Set to 86.
34721 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34723 PR preprocessor/54124
34724 * doc/cppopts.texi: Reformat -d subtable to list the full name
34725 of the options. Add cross-reference to the docs for the general
34726 compiler -d options.
34727 * doc/invoke.texi (Developer Options): Add cross-reference to the
34728 preprocessor-specific -d option documentation.
34730 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34732 PR preprocessor/13498
34733 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34734 redudant material, and reflect new command-line options.
34735 (System Headers): Likewise.
34737 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34739 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34740 -isystem, and -idirafter. Copy-edit.
34741 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34742 default for -ftrack-macro-expansion. Delete obsolete and
34743 badly-formatted implementation details about -fdebug-cpp output.
34744 * doc/cppwarnopts.texi: Copy-edit.
34746 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34749 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34750 that the transition from a max line width >= 1<<10 to narrower
34751 lines works correctly.
34753 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34755 * doc/options.texi (PerFunction): New.
34756 * opt-functions.awk (switch_flags): Map both Optimization and
34757 PerFunction to CL_OPTIMIZATION.
34758 * opth-gen.awk: Test for PerFunction flag along with
34760 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34761 it only when the latter is present. Skip those that don't in
34762 the hash function generator.
34763 * common.opt (fvar-tracking): Mark as PerFunction instead of
34765 (fvar-tracking-assignments): Likewise.
34766 (fvar-tracking-assignments-toggle): Likewise.
34767 (fvar-tracking-uninit): Likewise.
34769 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34771 PR translation/79018
34772 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34775 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34778 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34779 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34780 TARGET_LONG_JUMP_TABLE_OFFSETS.
34781 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34782 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34783 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34784 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34785 * config/m68k/m68k.md (tablejump expander): Likewise.
34786 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34787 TARGET_LONG_JUMP_TABLE_OFFSETS.
34788 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34789 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34791 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34792 David Holsgrove <david.holsgrove@xilinx.com>
34794 * common/config/microblaze/microblaze-common.c
34795 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34796 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34798 * config/microblaze/microblaze.c (microblaze_must_save_register)
34799 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34801 (microblaze_eh_return): New function.
34802 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34803 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34804 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34805 * config/microblaze/microblaze.md (eh_return): New pattern.
34807 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34809 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34810 GCC_DIAGNOSTIC_STRINGIFY): Define.
34812 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34814 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34816 * config/arm/arm.md (<mcrr>): New.
34818 * config/arm/arm.c (arm_arch5te): New.
34819 (arm_option_override): Set arm_arch5te.
34820 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34822 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34823 (arm_mcrr_qualifiers): ... this. New.
34824 (MRRC_QUALIFIERS): Define to...
34825 (arm_mrrc_qualifiers): ... this. New.
34826 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34828 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34829 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34830 (MRRCI, mrrc, MRRC): New.
34831 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34832 VUNSPEC_MRRC2): New.
34834 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34836 * config/arm/arm.md (<mcr>): New.
34838 * config/arm/arm.c (arm_coproc_builtin_available): Add
34839 support for mcr, mrc, mcr2 and mrc2.
34840 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34841 (arm_mcr_qualifiers): ... this. New.
34842 (MRC_QUALIFIERS): Define to ...
34843 (arm_mrc_qualifiers): ... this. New.
34844 (MCR_QUALIFIERS): Define to ...
34845 (arm_mcr_qualifiers): ... this. New.
34846 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34848 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34849 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34850 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34851 VUNSPEC_MRC2): New.
34853 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34855 * config/arm/arm.md (*ldc): New.
34859 * config/arm/arm.c (arm_coproc_builtin_available): Add
34860 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34861 (arm_coproc_ldc_stc_legitimate_address): New.
34862 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34863 'qualifier_const_pointer'.
34864 (LDC_QUALIFIERS): Define to...
34865 (arm_ldc_qualifiers): ... this. New.
34866 (STC_QUALIFIERS): Define to...
34867 (arm_stc_qualifiers): ... this. New.
34868 * config/arm/arm-protos.h
34869 (arm_coproc_ldc_stc_legitimate_address): New.
34870 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34871 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34872 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34873 stc2, stcl, stc2l): New.
34874 * config/arm/constraints.md (Uz): New.
34875 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34876 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34877 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34878 VUNSPEC_STC2L): New.
34880 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34882 * config/arm/arm.md (<cdp>): New.
34883 * config/arm/arm.c (neon_const_bounds): Rename this ...
34884 (arm_const_bounds): ... this.
34885 (arm_coproc_builtin_available): New.
34886 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34887 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34888 (CDP_QUALIFIERS): Define to...
34889 (arm_cdp_qualifiers): ... this. New.
34891 (arm_expand_builtin_args): Add case for 6 arguments.
34892 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34893 (arm_const_bounds): ... this.
34894 (arm_coproc_builtin_available): New.
34895 * config/arm/arm_acle.h (__arm_cdp): New.
34897 * config/arm/arm_acle_builtins.def (cdp): New.
34899 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34900 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34901 'arm_const_bounds'.
34902 * config/arm/types.md (coproc): New.
34903 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34904 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34905 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34906 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34908 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34910 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34911 (UBINOP_QUALIFIERS): New.
34913 (acle_builtin_data): New. Change comment.
34914 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34915 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34916 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34917 arm_acle_builtins.def.
34918 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34919 (arm_init_acle_builtins): New.
34920 (CRC32_BUILTIN): Remove.
34921 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34922 crc32cb, crc32ch and crc32cw.
34923 (arm_init_crc32_builtins): Remove.
34924 (arm_init_builtins): Use arm_init_acle_builtins rather
34925 than arm_init_crc32_builtins.
34926 (arm_expand_acle_builtin): New.
34927 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34928 * config/arm/arm_acle_builtins.def: New.
34930 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34932 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34933 (arm_builtin_datum): ... this.
34934 (arm_init_neon_builtin): Rename to ...
34935 (arm_init_builtin): ... this. Add a new parameters PREFIX
34936 and USE_SIG_IN_NAME.
34937 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34938 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34939 'arm_builtin_datum'.
34940 (arm_init_vfp_builtins): Likewise.
34941 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34942 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34943 (arm_expand_neon_args): Rename to ...
34944 (arm_expand_builtin_args): ... this. Rename builtin_arg
34945 enum values and differentiate between ARG_BUILTIN_MEMORY
34946 and ARG_BUILTIN_NEON_MEMORY.
34947 (arm_expand_neon_builtin_1): Rename to ...
34948 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34949 values, arm_expand_builtin_args and add bool parameter NEON.
34950 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34951 (arm_expand_vfp_builtin): Likewise.
34952 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34954 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34956 PR middle-end/77484
34957 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34958 * predict.c (tree_estimate_probability_bb): Reverse direction of
34959 polymorphic call predictor.
34961 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34963 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34964 (determine_pass_name_match): ...this new function and...
34965 (should_skip_pass_p): ...this new function.
34967 2017-01-06 Nathan Sidwell <nathan@acm.org>
34969 * ipa-visibility.c (function_and_variable_visibility): Reformat
34970 comments and long lines. Remove extrneous if.
34971 * symtab.c (symtab_node::make_decl_local): Fix code format.
34972 (symtab_node::set_section_for_node): Fix comment typo.
34974 2017-01-06 Martin Liska <mliska@suse.cz>
34977 * lra-constraints.c: Rename invariant to lra_invariant.
34978 * predict.c (set_even_probabilities): Initialize e to NULL.
34980 2017-01-05 Martin Sebor <msebor@redhat.com>
34982 PR tree-optimization/78910
34983 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34984 (format_integer): Correct off-by-one error in the handling
34985 of precision with negative numbers in signed conversions..
34987 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34989 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34991 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34993 PR tree-optimization/71016
34994 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34995 factor_out_conditional_conversion. Formatting fix.
34996 (factor_out_conditional_conversion): Add cond_stmt argument.
34997 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34998 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35001 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35003 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35004 read-rtl-function.o, and selftest-rtl.o.
35005 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35006 (selftest::aarch64_test_loading_full_dump): New function.
35007 (selftest::aarch64_run_selftests): New function.
35008 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35009 selftest::aarch64_run_selftests.
35010 * config/i386/i386.c
35011 (selftest::ix86_test_loading_dump_fragment_1): New function.
35012 (selftest::ix86_test_loading_call_insn): New function.
35013 (selftest::ix86_test_loading_full_dump): New function.
35014 (selftest::ix86_test_loading_unspec): New function.
35015 (selftest::ix86_run_selftests): Call the new functions.
35016 * emit-rtl.c (maybe_set_max_label_num): New function.
35017 * emit-rtl.h (maybe_set_max_label_num): New decl.
35018 * function.c (instantiate_decls): Guard call to
35019 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35020 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35022 * gensupport.c (gen_reader::gen_reader): Pass "false"
35023 for new "compact" param of rtx_reader.
35024 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35025 rather than an empty string for NULL strings.
35026 * read-md.c: Potentially include config.h rather than bconfig.h.
35027 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35028 (have_error): New global, copied from errors.c.
35029 (md_reader::read_name): Rename to...
35030 (md_reader::read_name_1): ...this, adding "out_loc" param,
35031 and converting "missing name or number" to returning false, rather
35033 (md_reader::read_name): Reimplement in terms of read_name_1.
35034 (md_reader::read_name_or_nil): New function.
35035 (md_reader::read_string): Handle "(nil)" by returning NULL.
35036 (md_reader::md_reader): Add new param "compact".
35037 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35038 (md_reader::read_file): New method.
35039 * read-md.h (md_reader::md_reader): Add new param "compact".
35040 (md_reader::read_file): New method.
35041 (md_reader::is_compact): New accessor.
35042 (md_reader::read_name): Convert return type from void to file_location.
35043 (md_reader::read_name_or_nil): New decl.
35044 (md_reader::read_name_1): New decl.
35045 (md_reader::m_compact): New field.
35046 (noop_reader::noop_reader): Pass "false" for new "compact" param
35048 (rtx_reader::rtx_reader): Add new "compact" param.
35049 (rtx_reader::read_rtx_operand): Make virtual and convert return
35050 type from void to rtx.
35051 (rtx_reader::read_until): New decl.
35052 (rtx_reader::handle_any_trailing_information): New virtual function.
35053 (rtx_reader::postprocess): New virtual function.
35054 (rtx_reader::finalize_string): New virtual function.
35055 (rtx_reader::m_in_call_function_usage): New field.
35056 (rtx_reader::m_reuse_rtx_by_id): New field.
35057 * read-rtl-function.c: New file.
35058 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35059 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35060 (selftest::verify_three_block_rtl_cfg): New decl.
35061 * read-rtl-function.h: New file.
35062 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35063 For host, include function.h, memmodel.h, and emit-rtl.h.
35064 (one_time_initialization): New function.
35065 (struct compact_insn_name): New struct.
35066 (compact_insn_names): New array.
35067 (find_code): Handle insn codes in compact dumps.
35068 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35069 (bind_subst_iter_and_attr): Likewise.
35070 (add_condition_to_string): Likewise.
35071 (add_condition_to_rtx): Likewise.
35072 (apply_attribute_uses): Likewise.
35073 (add_current_iterators): Likewise.
35074 (apply_iterators): Likewise.
35075 (initialize_iterators): Guard usage of apply_subst_iterator with
35076 #ifdef GENERATOR_FILE.
35077 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35078 (md_reader::read_mapping): Likewise.
35079 (add_define_attr_for_define_subst): Likewise.
35080 (add_define_subst_attr): Likewise.
35081 (read_subst_mapping): Likewise.
35082 (check_code_iterator): Likewise.
35083 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35085 (one_time_initialization): New function.
35086 (rtx_reader::read_until): New method.
35087 (read_flags): New function.
35088 (parse_reg_note_name): New function.
35089 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35090 Handle reuse_rtx ids.
35091 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35092 Add parsing support for RTL dumps, mirroring the special-cases in
35093 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35094 values, and calling handle_any_trailing_information.
35095 (rtx_reader::read_rtx_operand): Convert return type from void
35096 to rtx, returning return_rtx. Handle case 'e'. Call
35097 finalize_string on XSTR and XTMPL fields.
35098 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35099 "(nil)" values were omitted. Call the postprocess vfunc on the
35101 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35102 class ctor. Initialize m_in_call_function_usage. Call
35103 one_time_initialization.
35104 * rtl-tests.c (selftest::test_uncond_jump): Call
35105 set_new_first_and_last_insn.
35106 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35107 * selftest-rtl.c: New file.
35108 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35109 (selftest::get_insn_by_uid): New decl.
35110 * selftest-run-tests.c (selftest::run_tests): Call
35111 read_rtl_function_c_tests.
35112 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35113 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35116 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35118 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35119 operands in a special way. Assert that pos+len <= mode precision.
35121 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35123 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35124 3 argument Alias with unlimited for the negative form.
35125 (fno-vect-cost-model): Removed.
35127 2017-01-05 Martin Liska <mliska@suse.cz>
35129 * hsa-gen.c (gen_hsa_divmod): New function.
35130 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35132 2017-01-05 Martin Liska <mliska@suse.cz>
35135 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35138 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35140 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35141 small constant length operands.
35143 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35145 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35146 between loop iterations.
35148 2017-01-05 Martin Liska <mliska@suse.cz>
35151 * gimplify.c (gimplify_decl_expr): Compare to
35152 asan_poisoned_variables instread of checking flags.
35153 (gimplify_target_expr): Likewise.
35154 (gimplify_expr): Likewise.
35155 (gimplify_function_tree): Conditionally initialize
35156 asan_poisoned_variables.
35158 2017-01-04 Jeff Law <law@redhat.com>
35160 PR tree-optimizatin/78812
35161 * rtl.h (contains_mem_rtx_p): Prototype.
35162 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35163 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35164 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35165 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35166 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35168 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35170 * input.c (assert_char_at_range): Default-initialize actual_range.
35172 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35174 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35175 to match the caller.
35177 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35179 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35180 insns after final jump in test to emit dummy move.
35182 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35184 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35185 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35187 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35189 * multiple_target.c (create_dispatcher_calls): Init e_next.
35190 * tree-ssa-loop-split.c (split_loop): Init border.
35191 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35194 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35199 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35200 (altivec_register_operand): Do not return true if the operand
35201 contains a SUBREG mixing SImode and SFmode.
35202 (vsx_register_operand): Likewise.
35203 (vsx_reg_sfsubreg_ok): New predicate.
35204 (vfloat_operand): Do not return true if the operand contains a
35205 SUBREG mixing SImode and SFmode.
35206 (vint_operand): Likewise.
35207 (vlogical_operand): Likewise.
35208 (gpc_reg_operand): Likewise.
35209 (int_reg_operand): Likewise.
35210 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35211 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35212 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35214 (rs6000_emit_move_si_sf_subreg): New helper function.
35215 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35216 fixup SUBREGs involving SImode and SFmode.
35217 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35218 numbers for the new peephole2 optimization.
35219 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35220 the GLIBC math library that do AND/IOR/XOR operations on single
35221 precision floating point.
35222 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35223 target macros to say whether we need to avoid SUBREGs mixing
35225 (TARGET_ALLOW_SF_SUBREG): Likewise.
35226 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35227 (UNSPEC_SI_FROM_SF): Likewise.
35228 (iorxor): Change spacing.
35229 (and_ior_xor): New iterator for AND, IOR, and XOR.
35230 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35231 (movdi_from_sf_zero_ext): Likewise.
35232 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35233 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35234 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35235 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35236 (fms<mode>4): Likewise.
35237 (fnma<mode>4): Likewise.
35238 (fnms<mode>4): Likewise.
35239 (nfma<mode>4): Likewise.
35240 (nfms<mode>4): Likewise.
35242 2017-01-04 Marek Polacek <polacek@redhat.com>
35245 * doc/invoke.texi: Document -Wpointer-compare.
35247 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35249 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35252 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35253 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35256 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35259 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35260 documentation of the powerpc_popcntb_ok attribute.
35261 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35262 code to issue warning messages if a requested CPU configuration is
35263 not supported by the binary (assembler and loader) toolchain.
35264 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35265 made to define a built-in function that has been disabled.
35266 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35267 made to define a built-in function that has been disabled.
35268 (altivec_init_builtins): Add comment explaining why definition
35269 of the DST built-in functions is not preceded by an assertion
35270 check. Add assertions to prevent ICE if attempts are made to
35271 define an altivec predicate or an abs* built-in function that has
35273 (htm_init_builtins): Add comment explaining why definition of the
35274 htm built-in functions is not preceded by an assertion check.
35276 2017-01-04 Jeff Law <law@redhat.com>
35278 PR tree-optimizatin/67955
35279 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35280 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35281 the points-to solution does not include pt_null. Use DECL_PT_UID
35284 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35286 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35287 Use gen_int_mode instead of gen_lopwart for const_int operands.
35289 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35291 PR tree-optimization/71563
35292 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35293 out of range value - has low bits known to be zero.
35295 2017-01-04 Alan Modra <amodra@gmail.com>
35297 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35298 * configure: Regenerate.
35299 * config.in: Regenerate.
35301 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35304 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35305 a substring of the message, but strcmp with the whole message. Ifdef
35306 ENABLE_NLS, translate the message first using dgettext.
35308 2017-01-03 Jeff Law <law@redhat.com>
35310 PR tree-optimizatin/78856
35311 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35312 (mark_threaded_blocks): Remove code to truncate thread paths that
35313 cross multiple loop headers. Instead invalidate the cached loop
35314 iteration information and handle case of a thread path walking
35315 into an irreducible region.
35317 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35320 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35321 assertions. Add support for doing the signbit if the IEEE 128-bit
35322 floating point value is in a GPR.
35323 * config/rs6000/rs6000.md (Fsignbit): Delete.
35324 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35325 Update the length attribute if the value is in a GPR.
35326 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35327 the sign or zero extension instruction, since the value is always 0/1.
35328 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35331 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35332 extracting SImode to a GPR register so that we can generate a
35333 store, limit the vector to be in a traditional Altivec register
35334 for the vextuwrx instruction.
35336 2017-01-03 Ian Lance Taylor <iant@google.com>
35338 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35340 2017-01-03 Martin Sebor <msebor@redhat.com>
35342 PR tree-optimization/78696
35343 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35344 precision. Use MPFR for %f for greater fidelity. Correct handling
35346 (pass_sprintf_length::compute_format_length): Set width and precision
35347 specified by asrerisk to void_node for vararg functions.
35348 (try_substitute_return_value): Adjust dump output.
35350 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35352 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35354 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35356 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35357 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35358 -mlra/-mno-lra was passed to the compiler.
35360 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35362 PR rtl-optimization/65618
35363 * emit-rtl.c (try_split): Move initialization of "before" and
35364 "after" to just before the call to emit_insn_after_setloc.
35366 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35368 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35370 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35372 * dwarf2out.c (gen_enumeration_type_die): When
35373 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35375 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35377 PR tree-optimization/78965
35378 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35379 Change first argument from const call_info & to call_info &. For %n
35380 set info.nowrite to false.
35382 PR middle-end/78901
35383 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35384 possibly throwing calls.
35386 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35387 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35388 and fns handling, rather than in a separate case SSA_NAME.
35390 2017-01-02 Jeff Law <law@redhat.com>
35392 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35393 fixes for first_period and second_period variables.
35395 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35398 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35399 (*insvqi_1): New insn pattern.
35400 (*insvqi_1_mem_rex64): Ditto.
35401 (*insvqi_2): Ditto.
35402 (*insvqi_3): Rename from *insvqi.
35404 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35406 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35408 * doc/cfg.texi (Edges): Remove reference to Java.
35409 (Maintaining the CFG): Ditto.
35411 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35413 PR middle-end/77674
35414 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35415 transparent aliases.
35417 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35419 PR middle-end/77484
35420 * predict.def (PRED_CALL): Update hitrate.
35421 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35422 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35423 into direct/indirect/polymorphic variants.
35425 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35427 Update copyright years.
35429 * gcc.c (process_command): Update copyright notice dates.
35430 * gcov-dump.c (print_version): Ditto.
35431 * gcov.c (print_version): Ditto.
35432 * gcov-tool.c (print_version): Ditto.
35433 * gengtype.c (create_file): Ditto.
35434 * doc/cpp.texi: Bump @copying's copyright year.
35435 * doc/cppinternals.texi: Ditto.
35436 * doc/gcc.texi: Ditto.
35437 * doc/gccint.texi: Ditto.
35438 * doc/gcov.texi: Ditto.
35439 * doc/install.texi: Ditto.
35440 * doc/invoke.texi: Ditto.
35442 Copyright (C) 2017 Free Software Foundation, Inc.
35444 Copying and distribution of this file, with or without modification,
35445 are permitted in any medium without royalty provided the copyright
35446 notice and this notice are preserved.