1 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
8 * predict.c (combine_predictions_for_bb): Preserve zero predicted
10 (expensive_function_p): Remove useless assert.
11 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
13 2017-11-16 Martin Sebor <msebor@redhat.com>
15 PR tree-optimization/82588
16 PR tree-optimization/82583
17 * tree-vrp.c (check_array_ref): Handle flexible array members,
18 string literals, and inner indices.
19 (search_for_addr_array): Add detail to diagnostics.
21 2017-11-16 Nathan Sidwell <nathan@acm.org>
25 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
26 (SET_DECL_ASSEMBLER_NAME): Forward to
27 overwrite_decl_assembler_name.
28 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
29 (overwrite_decl_assembler_name): Declare.
30 * tree.c (overwrite_decl_assembler_name): New.
31 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
32 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
33 (LANG_HOOKS_INITIALIZER): Add it.
34 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
35 * langhooks.c (lhd_set_decl_assembler_name): Use
36 SET_DECL_ASSEMBLER_NAME.
37 (lhd_overwrite_decl_assembler_name): Default implementation.
39 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
40 Jackson Woodruff <jackson.woodruff@arm.com>
42 PR tree-optimization/71026
43 * match.pd: Canonicalize constant multiplies in division.
45 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
47 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
48 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
49 Remove OPT_fomit_frame_pointer entry.
50 * common/config/arc/arc-common.c: Likewise.
51 * common/config/arm/arm-common.c: Likewise.
52 * common/config/avr/avr-common.c: Likewise.
53 * common/config/c6x/c6x-common.c: Likewise.
54 * common/config/cr16/cr16-common.c: Likewise.
55 * common/config/cris/cris-common.c: Likewise.
56 * common/config/epiphany/epiphany-common.c: Likewise.
57 * common/config/fr30/fr30-common.c: Likewise.
58 * common/config/frv/frv-common.c: Likewise.
59 * common/config/ia64/ia64-common.c: Likewise.
60 * common/config/iq2000/iq2000-common.c: Likewise.
61 * common/config/lm32/lm32-common.c: Likewise.
62 * common/config/m32r/m32r-common.c: Likewise.
63 * common/config/mcore/mcore-common.c: Likewise.
64 * common/config/microblaze/microblaze-common.c: Likewise.
65 * common/config/mips/mips-common.c: Likewise.
66 * common/config/mmix/mmix-common.c: Likewise.
67 * common/config/mn10300/mn10300-common.c: Likewise.
68 * common/config/nios2/nios2-common.c: Likewise.
69 * common/config/pa/pa-common.c: Likewise.
70 * common/config/pdp11/pdp11-common.c: Likewise.
71 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
72 * common/config/riscv/riscv-common.c: Likewise.
73 * common/config/rs6000/rs6000-common.c: Likewise.
74 * common/config/rx/rx-common.c: Likewise.
75 * common/config/s390/s390-common.c: Likewise.
76 * common/config/sh/sh-common.c: Likewise.
77 * common/config/sparc/sparc-common.c: Likewise.
78 * common/config/tilegx/tilegx-common.c: Likewise.
79 * common/config/tilepro/tilepro-common.c: Likewise.
80 * common/config/v850/v850-common.c: Likewise.
81 * common/config/visium/visium-common.c: Likewise.
82 * common/config/xstormy16/xstormy16-common.c: Likewise.
83 * common/config/xtensa/xtensa-common.c: Likewise.
84 * invoke.texi (-fomit-frame-pointer): Update documentation.
86 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
88 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
90 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
92 * cfg.c (scale_bbs_frequencies_int,
93 cale_bbs_frequencies_gcov_type): Remove.
94 * cfg.h (scale_bbs_frequencies_int,
95 cale_bbs_frequencies_gcov_type): Remove.
97 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
99 * tree-ssa-loop-manip.c
100 (scale_dominated_blocks_in_loop): Update to profile counts.
101 (tree_transform_and_unroll_loop): Likewise.
103 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
105 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
106 scale_bbs_frequencies_int.
108 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
110 * final.c (compute_alignments): Use counts rather than frequencies.
112 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
114 * cfgloopanal.c: Include sreal.h
115 (average_num_loop_insns): Use counts and sreal for accounting.
117 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
119 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
122 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
123 Kito Cheng <kito.cheng@gmail.com>
125 * config/nds32/constraints.md: Provide more constraints.
126 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
127 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
128 support constraints usage.
130 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
132 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
134 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
135 Kito Cheng <kito.cheng@gmail.com>
137 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
138 * config/nds32/nds32.opt: Refine the layout.
139 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
140 TARGET_EXT_STRING): Support new options.
141 * config/nds32/nds32.h: Likewise.
142 * config/nds32/nds32.md: Likewise.
143 * config/nds32/nds32-predicates.c: Likewise.
144 * config/nds32/constraints.md: Likewise.
145 * common/config/nds32/nds32-common.c: Likewise.
147 2017-11-16 Julia Koval <julia.koval@intel.com>
150 * config/i386/gfniintrin.h: Add sse check.
151 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
153 2017-11-16 Julia Koval <julia.koval@intel.com>
155 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
156 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
157 (ix86_handle_option): Handle -mavx512vbmi2.
158 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
159 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
160 * config/i386/i386-c.c (__AVX512VBMI2__): New.
161 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
162 (ix86_valid_target_attribute_inner_p): Ditto.
163 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
164 * config/i386/i386.opt (mavx512vbmi2): New option.
165 * doc/invoke.texi: Add new option.
167 [2017-11-16 Julia Koval <julia.koval@intel.com>
169 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
170 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
171 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
172 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
173 _mm512_gf2p8mul_epi8): New intrinsics.
174 * config/i386/i386-builtin-types.def
175 (V64QI_FTYPE_V64QI_V64QI): New type.
176 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
177 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
178 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
179 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
180 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
181 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
183 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
185 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
186 explicitly as a stream of bytes.
188 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
190 * config/rs6000/altivec.h (vec_xst_be): New #define.
191 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
192 and externalize from *altivec_vperm_<mode>_internal.
193 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
195 (XL_BE_V8HI): Likewise.
196 (XL_BE_V4SI): Likewise.
197 (XL_BE_V4SI): Likewise.
198 (XL_BE_V2DI): Likewise.
199 (XL_BE_V4SF): Likewise.
200 (XL_BE_V2DF): Likewise.
201 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
202 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
203 all array entries with these keys: VSX_BUILTIN_VEC_XL,
204 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
205 VSX_BUILTIN_VEC_XST_BE.
206 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
207 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
209 (altivec_init_builtins): Replace conditional calls to def_builtin
210 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
211 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
212 with unconditional calls. Remove calls to def_builtin for
213 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
214 __builtin_vec_xst_be.
215 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
216 to define_expand, and add alternate RTL generation for P8.
217 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
219 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
220 add alternate RTL generation for P8.
221 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
222 vsx_ld_elemrev_v16qi.
223 (vsx_st_elemrev_v8hi): Convert define_insn
224 to define_expand, and add alternate RTL generation for P8.
225 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
227 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
228 add alternate RTL generation for P8.
229 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
230 vsx_st_elemrev_v16qi.
232 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
235 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
236 TARGET_AVX512ER check.
237 (ix86_option_override_internal): Set MASK_VZEROUPPER if
238 neither -mzeroupper nor -mno-zeroupper is used and
239 TARGET_EMIT_VZEROUPPER is set.
240 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
241 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
243 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
245 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
246 folding of vector compares.
247 (fold_build_vec_cmp): New helper function.
248 (fold_compare_helper): New helper function.
249 (builtin_function_type): Add compare builtins to the list of functions
250 having unsigned arguments. Cosmetic updates to comment indentation.
251 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
252 the not+eq combination.
254 2017-11-15 Bin Cheng <bin.cheng@arm.com>
256 PR tree-optimization/82726
257 PR tree-optimization/70754
258 * tree-predcom.c (order_drefs_by_pos): New function.
259 (combine_chains): Move code setting has_max_use_after to...
260 (try_combine_chains): ...here. New parameter. Sort combined chains
261 according to position information.
262 (tree_predictive_commoning_loop): Update call to above function.
263 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
265 2017-11-15 Bin Cheng <bin.cheng@arm.com>
267 PR tree-optimization/82726
269 2017-01-23 Bin Cheng <bin.cheng@arm.com>
271 PR tree-optimization/70754
272 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
273 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
274 combined stmt before it if not NULL.
275 (combine_chains): Process refs reversely and compute dominance point
279 2017-02-23 Bin Cheng <bin.cheng@arm.com>
281 PR tree-optimization/79663
282 * tree-predcom.c (combine_chains): Process refs in reverse order
283 only for ZERO length chains, and add explaining comment.
285 2017-11-15 Tamar Christina <tamar.christina@arm.com>
287 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
288 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
289 * doc/invoke.texi (ARM Options): Add armv8.3-a.
291 2017-11-15 Tamar Christina <tamar.christina@arm.com>
293 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
295 2017-11-15 Martin Liska <mliska@suse.cz>
297 * tree-cfg.c (pass_warn_function_return::execute):
298 Compare warn_return_type for greater than zero.
300 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
304 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
305 to return true on Xeon and not on Xeon Phi.
306 (ix86_check_avx256_register): Changed to ...
307 (ix86_check_avx_upper_register): ... this. Add extra check for
308 VALID_AVX512F_REG_OR_XI_MODE.
309 (ix86_avx_u128_mode_needed): Changed
310 ix86_check_avx256_register to ix86_check_avx_upper_register.
311 (ix86_check_avx256_stores): Changed to ...
312 (ix86_check_avx_upper_stores): ... this. Changed
313 ix86_check_avx256_register to ix86_check_avx_upper_register.
314 (ix86_avx_u128_mode_after): Changed
315 avx_reg256_found to avx_upper_reg_found. Changed
316 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
317 (ix86_avx_u128_mode_entry): Changed
318 ix86_check_avx256_register to ix86_check_avx_upper_register.
319 (ix86_avx_u128_mode_exit): Ditto.
320 * config/i386/i386.h: (host_detect_local_cpu): New define.
322 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
324 * config/arm/xgene1.md (xgene1): Split into automatons
325 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
326 (xgene1_f_load): Adjust reservations and/or types.
327 (xgene1_f_store): Likewise.
328 (xgene1_load_pair): Likewise.
329 (xgene1_store_pair): Likewise.
330 (xgene1_fp_load1): Likewise.
331 (xgene1_load1): Likewise.
332 (xgene1_store1): Likewise.
333 (xgene1_move): Likewise.
334 (xgene1_alu): Likewise.
335 (xgene1_simd): Likewise.
336 (xgene1_bfm): Likewise.
337 (xgene1_neon_load1): Likewise.
338 (xgene1_neon_store1): Likewise.
339 (xgene1_neon_logic): Likewise.
340 (xgene1_neon_st1): Likewise.
341 (xgene1_neon_ld1r): Likewise.
342 (xgene1_alu_cond): Added.
343 (xgene1_shift_reg): Likwise.
344 (xgene1_bfx): Likewise.
345 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
347 2017-11-15 Jakub Jelinek <jakub@redhat.com>
350 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
352 (can_widen_mult_without_libcall): New function.
353 (expand_mul_overflow): If only checking unsigned mul overflow,
354 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
355 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
356 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
357 (expand_DIVMOD): Formatting fix.
358 * expmed.h (expand_mult): Add NO_LIBCALL argument.
359 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
360 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
362 PR tree-optimization/82977
363 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
364 constructed temporary to strlen_to_stridx.put.
366 2017-11-15 Martin Liska <mliska@suse.cz>
368 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
369 * configure: Regenerate.
371 2017-11-15 Martin Liska <mliska@suse.cz>
374 * config/sh/sh-mem.cc: Use proper probability for
377 2017-11-14 Jeff Law <law@redhat.com>
379 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
380 the red zone for stack_clash_protection_final_dynamic_probe targets
381 when the total dynamic stack size is zero bytes.
383 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
384 blocks is post order.
386 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
388 * dumpfile.h (TDF_COMPARE_DEBUG): New.
389 * final.c (rest_of_clean_state): Set it for the
390 -fcompare-debug dump.
391 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
392 class when TDF_COMPARE_DEBUG is set.
394 * dwarf2out.c (gen_producer_string): Discard
397 2017-11-15 Joseph Myers <joseph@codesourcery.com>
400 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
401 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
402 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
403 (__TGMATH_REAL_2_3): Remove macros.
404 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
405 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
407 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
409 (remquo): Define using __TGMATH_REAL_3.
411 2017-11-14 Jeff Law <law@redhat.com>
413 * vr-values.c: New file with contents extracted from tree-vrp.c.
414 * Makefile.in (OBJS): Add vr-values.o
415 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
416 (set_value_range, set_and_canonicalize_value_range): Likewise.
417 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
418 (value_range_constant_singleton, symbolic_range_p): Likewise.
419 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
420 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
421 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
422 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
423 (find_case_label_range, find_case_label_index): Likewise.
424 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
425 (range_int_cst_singleton_p, value_inside_range): Likewise.
426 (get_single_symbol): Likewise.
427 (switch_update): Move structure definition here.
428 (to_remove_edges, to_update_switch_stmts): Provide externs.
429 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
430 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
431 (vrp_val_is_min, set_value_range): Likewise.
432 (set_and_canonicalize_value_range, copy_value_range): Likewise.
433 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
434 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
435 (range_is_nonnull, range_int_cst_p): Likewwise.
436 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
437 (get_single_symbol, operand_less_p): Likewise
438 (compare_values_warnv, compare_values): Likewise.
439 (value_inside_range, value_range_constant_singleton): Likewise.
440 (zero_nonzero_bitgs_from_vr): Likewise.
441 (extract_range_from_binary_expr_1): Likewise.
442 (overflow_comparison_p): Likewise.
443 (to_remove_edges, to_update_switch_stmts): Likewise.
444 (find_case_label-index, find_case_label_range): Likewise.
445 (switch_update, set_value_range_to_nonnegative): Remove.
446 (set_value_range_to_truthvalue): Likewise.
447 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
448 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
449 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
450 (find_case_label_ranges, test_for_singularity): Likewise.
451 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
452 (x_vr_values): Move to its remaining use site.
454 2017-11-10 Jeff Law <law@redhat.com>
456 * vr-values.h (VR_INITIALIZER): Move #define here.
457 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
458 * Makefile.in (OBJS): Add tree-evrp.o
459 * tree-vrp.h (assert_info): Move structure definition here.
460 (set_value_range_to_varying): Prototype.
461 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
462 (infer_value_range, register_edge_assert_for): Likewise.
463 (stmt_interesting_for_vrp): Likewise.
464 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
465 (set_value_range_to_varying): No longer static.
466 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
467 (infer_value_range, register_edge_assert_for): Likewise.
469 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
471 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
472 generate the XXBRD instruction.
474 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
475 (rs6000_builtin_type_compatible): Treat _Float128 and long double
476 as being compatible if -mabi=ieeelongdouble.
477 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
478 to setup float128 built-ins with hardware support.
479 (BU_FLOAT128_HW_2): Likewise.
480 (BU_FLOAT128_HW_3): Likewise.
481 (BU_FLOAT128_HW_VSX_1): Likewise.
482 (BU_FLOAT128_HW_VSX_2): Likewise.
483 (scalar_extract_expq): Change float128 built-in functions to
484 accommodate having both KFmode and TFmode functions. Use the
485 KFmode variant as the default.
486 (scalar_extract_sigq): Likewise.
487 (scalar_test_neg_qp): Likewise.
488 (scalar_insert_exp_q): Likewise.
489 (scalar_insert_exp_qp): Likewise.
490 (scalar_test_data_class_qp): Likewise.
491 (sqrtf128_round_to_odd): Delete processing the round to odd
492 built-in functions as special built-in functions, and define them
493 as float128 built-ins. Use the KFmode variant as the default.
494 (truncf128_round_to_odd): Likewise.
495 (addf128_round_to_odd): Likewise.
496 (subf128_round_to_odd): Likewise.
497 (mulf128_round_to_odd): Likewise.
498 (divf128_round_to_odd): Likewise.
499 (fmaf128_round_to_odd): Likewise.
500 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
501 support for KFmode and TFmode xststdcqp calls.
502 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
503 point, switch the built-in handlers for the get/set float128
504 exponent, get float128 mantissa, float128 test built-ins, and the
505 float128 round to odd built-in functions. Eliminate creating the
506 float128 round to odd built-in functions as special built-ins.
507 (rs6000_init_builtins): Eliminate special creation of the float128
508 round to odd built-in functions.
509 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
510 function insns to support both TFmode and KFmode variants.
511 (xsxsigqp_<mode>): Likewise.
512 (xsiexpqpf_<mode>): Likewise.
513 (xsiexpqp_<mode>): Likewise.
514 (xststdcqp_<mode>): Likewise.
515 (xststdcnegqp_<mode>): Likewise.
516 (xststdcqp_<mode>): Likewise.
518 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
520 * tree-ssa-threadupdate.c (compute_path_counts): Remove
521 unused path_in_freq_ptr parameter.
522 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
524 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
526 * ipa-inline.c (edge_badness): Dump sreal frequency.
527 (compute_inlined_call_time): Match natural implementaiton ...
528 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
529 forgotten division by CGRAPH_FREQ_BASE.
531 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
533 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
534 Solaris 11. Update comment.
535 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
537 * config/sol2.h (STARTFILE_SPEC): Likewise.
538 * configure: Regenerate.
540 2017-11-14 Carl Love <cel@us.ibm.com>
542 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
543 le_ and be_ prefixes to swap* variables. Remove
544 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
547 2017-11-14 Jason Merrill <jason@redhat.com>
549 Support GTY((cache)) on hash_map.
550 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
551 (ggc_cache_remove): Override it instead of ggc_mx.
552 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
553 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
554 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
555 (simple_cache_map_traits): Override maybe_mx.
556 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
557 (hash_map): Friend gt_cleare_cache.
558 (gt_cleare_cache): New.
559 * tree.h (tree_cache_traits): New hash_map traits class.
560 (tree_cache_map): New typedef.
562 2017-11-14 Richard Biener <rguenther@suse.de>
564 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
565 paramter and handling.
566 (cleanup_control_flow_bb): Likewise.
567 (cleanup_control_flow_pre): New helper performing a DFS walk
568 to call cleanup_control_flow_bb in PRE order.
569 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
570 via cleanup_control_flow_pre.
572 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
574 * config/aarch64/aarch64-simd.md
575 (aarch64_simd_bsl<mode>_internal): Remove DImode.
576 (*aarch64_simd_bsl<mode>_alt): Likewise.
577 (aarch64_simd_bsldi_internal): New.
578 (aarch64_simd_bsldi_alt): Likewise.
580 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
582 * tracer.c (better_p): Do not compare frequencies.
583 * reg-stack.c (better_edge): Likewise.
584 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
587 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
589 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
590 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
591 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
592 * lto-streamer-in.c (input_function): Use update_max_bb_count.
593 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
594 * predict.c (maybe_hot_frequency_p): Inline to ...
595 (maybe_hot_count_p): ... here; rewrite to counts.
596 (counts_to_freqs): Rename to ...
597 (update_max_bb_count): ... this one.
598 (expensive_function_p): Use counts.
599 (estimate_bb_frequencies): Update.
600 (rebuild_frequencies): Update.
601 * predict.h (counts_to_freqs): Rename to ...
602 (update_max_bb_count): ... this one.
603 * profile.c (compute_branch_probabilities): Add debug info
604 * tree-inline.c (expand_call_inline): Update debug info.
605 (optimize_inline_calls): Use update_max_bb_count..
606 (tree_function_versioning): Use update_max_bb_count..
607 * value-prof.c (gimple_value_profile_transformations):
608 Do not use update_max_bb_count.
610 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
612 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
613 always use frequencies.
615 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
617 * bb-reorder.c: Remove frequencies from comments.
618 (better_edge_p): Use profile counts.
619 (find_traces): Dump profile counts.
620 (rotate_loop): Use profile counts.
621 (find_traces_1_round): Likewise.
622 (connect_better_edge_p): Use counts instead of probabilities for
624 (copy_bb_p): Drop early check for non-0 frequency.
625 (sanitize_hot_paths): Update comments.
627 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
629 * ipa-split.c (struct split_point): Add count.
630 (consider_split): Do not compute incoming frequency; compute incoming
631 count and store it to split_point.
632 (split_function): Set count of the call to split part correctly.
634 2017-11-13 Carl Love <cel@us.ibm.com>
636 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
638 2017-11-13 Tom Tromey <tom@tromey.com>
640 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
642 2017-11-13 Carl Love <cel@us.ibm.com>
644 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
645 Add support for builtins:
646 unsigned int vec_first_{,miss}_match_{,or_eos}index,
647 vector {un,}signed {char,int,short},
648 vector {un,}signed {char,int,short}) arguments.
649 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
650 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
651 Add BU_P9V_AV_2 expansions for the builtins.
652 * config/rs6000/altivec.h (vec_first_match_index,
653 vec_first_mismatch_index, vec_first_match_or_eos_index,
654 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
655 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
656 new extern declaration.
657 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
658 * config/rs6000/vsx.md (first_match_index_<mode>,
659 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
660 first_mismatch_or_eos_index_<mode>): Add define expand.
661 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
662 * doc/extend.texi: Update the built-in documenation file for the new
665 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
667 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
668 into the min/max operations for _Float<N> and _Float<N>X types.
670 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
673 * dwarf2out.c (do_eh_frame): New static variable.
674 (dwarf2out_begin_prologue): Set it.
675 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
677 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
679 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
681 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
683 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
686 * debug.h (dwarf2out_do_eh_frame): Declare.
687 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
688 (dwarf2out_do_frame): Use it.
689 (dwarf2out_do_cfi_asm): Likewise.
690 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
691 (dwarf2out_assembly_start): Likewise.
692 (dwarf2out_begin_prologue): Fix comment.
693 * toplev.c (compile_file): Always call dwarf2out_frame_finish
694 if the target needs either debug or unwind DWARF2 info.
695 * lto-opts.c (lto_write_options): Do not save -fexceptions,
696 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
697 -ftrapping-math, -ftrapv and -fwrapv.
699 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
701 * cgraph.c (cgraph_edge::sreal_frequency): New function.
702 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
703 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
704 (estimate_edge_size_and_time): Likewise.
705 (ipa_merge_fn_summary_after_inlining): Likewise.
706 * ipa-inline.c (cgraph_freq_base_rec): Remove.
707 (compute_uninlined_call_time): Use sreal_frequency.
708 (compute_inlined_call_time): Likewise.
709 (ipa_inline): Do not initialize cgraph_freq_base_rec.
710 * profile-count.c: Include sreal.h.
711 (profile_count::to_sreal_scale): New.
712 * profile-count.h: Forward declare sreal.
713 (profile_count::to_sreal_scale): Declare.
715 2017-11-13 Nathan Sidwell <nathan@acm.org>
717 * diagnostic.c (maybe_line_and_column): New.
718 (diagnostic_get_location_text): Use it.
719 (diagnostic_report_current_module): Likewise.
720 (test_diagnostic_get_location_text): Add tests.
722 2017-11-13 Luis Machado <luis.machado@linaro.org>
724 * doc/md.texi (Specifying processor pipeline description): Fix
725 incorrect latency for the div instruction example.
727 2017-11-13 Jakub Jelinek <jakub@redhat.com>
729 PR tree-optimization/78821
730 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
731 that bit_not_p is the same.
732 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
733 (split_group): Count precisely bit_not_p bits in each statement.
734 (invert_op): New function.
735 (imm_store_chain_info::output_merged_store): Use invert_op to
736 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
737 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
739 2017-11-13 Martin Liska <mliska@suse.cz>
741 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
742 (struct source_info): Likewise.
743 (add_branch_counts): Likewise.
744 (add_line_counts): Likewise.
745 (function_summary): Likewise.
746 (output_intermediate_line): Likewise.
747 (generate_results): Likewise.
749 2017-11-13 Martin Liska <mliska@suse.cz>
751 * gcov.c (struct block_info): Remove typedef for block_t.
752 (struct line_info): Likewise.
753 (line_info::has_block): Likewise.
754 (EXIT_BLOCK): Likewise.
757 (get_cycles_count): Likewise.
758 (process_file): Likewise.
759 (read_graph_file): Likewise.
760 (solve_flow_graph): Likewise.
761 (find_exception_blocks): Likewise.
762 (add_line_counts): Likewise.
763 (accumulate_line_info): Likewise.
764 (output_line_details): Likewise.
766 2017-11-13 Martin Liska <mliska@suse.cz>
768 * gcov.c (struct arc_info): Remove typedef for arc_t.
769 (struct line_info): Likewise.
770 (add_branch_counts): Likewise.
771 (output_branch_count): Likewise.
772 (function_info::~function_info): Likewise.
774 (output_intermediate_line): Likewise.
775 (read_graph_file): Likewise.
776 (solve_flow_graph): Likewise.
777 (find_exception_blocks): Likewise.
778 (add_line_counts): Likewise.
779 (accumulate_line_info): Likewise.
780 (output_line_details): Likewise.
781 (output_function_details): Likewise.
783 2017-11-13 Martin Liska <mliska@suse.cz>
785 * gcov.c (struct function_info): Remove typedef for function_t.
786 (struct source_info): Likewise.
787 (source_info::get_functions_at_location): Likewise.
788 (solve_flow_graph): Likewise.
789 (find_exception_blocks): Likewise.
790 (add_line_counts): Likewise.
791 (output_intermediate_file): Likewise.
792 (process_file): Likewise.
793 (generate_results): Likewise.
794 (release_structures): Likewise.
795 (read_graph_file): Likewise.
796 (read_count_file): Likewise.
797 (accumulate_line_counts): Likewise.
798 (output_lines): Likewise.
800 2017-11-13 Martin Liska <mliska@suse.cz>
802 * gcov.c (function_info::function_info): Remove num_counts
803 and add vector<gcov_type>.
804 (function_info::~function_info): Use the vector.
805 (process_file): Likewise.
806 (read_graph_file): Likewise.
807 (read_count_file): Likewise.
808 (solve_flow_graph): Likewise.
810 2017-11-13 Martin Liska <mliska@suse.cz>
812 * gcov.c (function_info::is_artificial): New function.
813 (process_file): Erase all artificial early.
814 (generate_results): Skip as all artificial are already
817 2017-11-13 Martin Liska <mliska@suse.cz>
819 * gcov.c (read_graph_file): Store to global vector of functions.
820 (read_count_file): Iterate the vector.
821 (process_file): Likewise.
822 (generate_results): Likewise.
823 (release_structures): Likewise.
825 2017-11-13 Jakub Jelinek <jakub@redhat.com>
827 PR tree-optimization/82954
828 * gimple-ssa-store-merging.c
829 (imm_store_chain_info::coalesce_immediate_stores): If
830 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
832 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
834 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
835 Upddate call to ENDIAN_LANE_N.
836 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
837 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
838 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
839 and use aarch64_endian_lane_rtx.
840 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
842 2017-11-12 Tom de Vries <tom@codesourcery.com>
844 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
846 2017-11-12 Tom de Vries <tom@codesourcery.com>
848 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
850 2017-11-12 Tom de Vries <tom@codesourcery.com>
852 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
854 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
855 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
856 * defaults.h (ASM_OUTPUT_LABELREF): Same.
858 2017-11-11 Martin Sebor <msebor@redhat.com>
861 * doc/extend.texi (attribute nonstring): Remove spurious argument.
864 * prefic.c (translate_name): Replace strncpy with memcpy to
865 avoid -Wstringop-truncation.
867 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
869 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
871 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
873 * predict.c (maybe_hot_frequency_p): Do not use cfun.
875 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
877 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
880 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
882 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
883 enable generating XXBRH if the value is in a vector register.
884 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
885 value is in a vector register.
886 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
887 register to register bswap64's instead of doing the GPR sequence
888 used on previous machines.
889 (bswapdi2_xxbrd): New insn.
890 (bswapdi2_reg): Disallow on ISA 3.0.
891 (register to register bswap64 splitter): Do not split the insn on
892 ISA 3.0 systems that use XXBRD.
894 2017-11-10 Martin Sebor <msebor@redhat.com>
897 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
899 (find_subframework_file): Same.
901 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
903 * auto-profile.c (afdo_indirect_call): Drop frequency.
904 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
905 (cgraph_node::create_edge): Drop frequency argument.
906 (cgraph_node::create_indirect_edge): Drop frequency argument.
907 (cgraph_edge::make_speculative): Drop frequency arguments.
908 (cgraph_edge::resolve_speculation): Do not update frequencies
909 (cgraph_edge::dump_edge_flags): Do not dump frequency.
910 (cgraph_node::dump): Check consistency in IPA mode.
911 (cgraph_edge::maybe_hot_p): Use IPA counter.
912 (cgraph_edge::verify_count_and_frequency): Rename to ...
913 (cgraph_edge::verify_count): ... this one; drop frequency checking.
914 (cgraph_node::verify_node): Update.
915 * cgraph.h (struct cgraph_edge): Drop frequency.
916 (cgraph_edge::frequency): New function.
917 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
919 (cgraph_edge::rebuild_edges): Likewise.
920 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
921 (duplicate_thunk_for_node): Do not pass frequency.
922 (cgraph_node::create_clone): Scale only counts.
923 (cgraph_node::create_virtual_clone): Do not pass frequency.
924 (cgraph_node::create_edge_including_clones): Do not pass frequency.
925 (cgraph_node::create_version_clone): Do not pass frequency.
926 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
927 (cgraph_node::expand_thunk): Do not pass frequency.
928 (cgraph_node::create_wrapper): Do not pass frequency.
929 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
931 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
932 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
933 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
934 (gather_caller_stats): Use frequency function.
935 (ipcp_cloning_candidate_p): Use frequency function.
936 (ipcp_propagate_stage): Use frequency function.
937 (get_info_about_necessary_edges): Use frequency function.
938 (update_profiling_info): Update only IPA profile.
939 (update_specialized_profile): Use frequency functoin.
940 (perhaps_add_new_callers): Update only IPA profile.
941 * ipa-devirt.c (ipa_devirt): Use IPA profile.
942 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
943 (dump_ipa_call_summary): Use frequency function.
944 (estimate_edge_size_and_time): Use frequency function.
945 (ipa_merge_fn_summary_after_inlining): Use frequency function.
946 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
947 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
948 (update_noncloned_counts): ... ths one; scale counts only.
949 (clone_inlined_nodes): Do not scale frequency.
950 (inline_call): Do not pass frequency.
951 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
952 (compute_inlined_call_time): Use IPA profile.
953 (want_inline_small_function_p): Use IPA profile.
954 (want_inline_self_recursive_call_p): Use IPA profile.
955 (edge_badness): Use IPA profile.
956 (lookup_recursive_calls): Use IPA profile.
957 (recursive_inlining): Do not pass frequency.
958 (resolve_noninline_speculation): Do not update frequency.
959 (inline_small_functions): Collect max of IPA profile.
960 (dump_overall_stats): Dump IPA porfile.
961 (dump_inline_stats): Dump IPA porfile.
962 (ipa_inline): Collect IPA stats.
963 * ipa-inline.h (clone_inlined_nodes): Update prototype.
964 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
965 (ipa_propagate_frequency): Use frequency function.
966 (ipa_profile): Cleanup.
967 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
968 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
969 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
970 (input_node): Do not stream frequency.
971 (input_edge): Do not stream frequency.
972 (merge_profile_summaries): Scale only IPA profiles.
973 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
974 * predict.c (drop_profile): Do not recompute frequency.
975 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
976 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
977 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
978 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
979 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
980 * tree-inline.c (copy_bb): Do not scale frequency.
981 (expand_call_inline): Do not scale frequency.
982 (tree_function_versioning): Do not scale frequency.
983 * ubsan.c (ubsan_create_edge): Do not pass frequency.
985 2017-11-10 Julia Koval <julia.koval@intel.com>
987 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
988 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
989 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
990 (_mm256_maskz_gf2p8affine_epi64_epi8)
991 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
992 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
993 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
994 (__builtin_ia32_vgf2p8affineqb_v32qi)
995 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
996 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
998 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1002 (arm_option_override): Refactor.
1003 (arm_option_reconfigure_globals): New.
1004 (arm_options_perform_arch_sanity_checks): New.
1005 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1007 (arm_options_perform_arch_sanity_checks): Likewise
1009 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1011 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1012 (power9-qpmul): New.
1013 * rs6000/rs6000.md ("type" attr): Add qmul.
1014 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1015 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1016 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1018 2017-11-10 Martin Sebor <msebor@redhat.com>
1021 * builtins.c (compute_objsize): Handle arrays that
1022 compute_builtin_object_size likes to fail for. Make extern.
1023 * builtins.h (compute_objsize): Declare.
1024 (check_strncpy_sizes): New function.
1025 (expand_builtin_strncpy): Call check_strncpy_sizes.
1026 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1027 -Wstringop-truncation.
1028 (gimple_fold_builtin_strncat): Same.
1029 * gimple.c (gimple_build_call_from_tree): Set call location.
1030 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1031 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1032 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1033 (handle_builtin_strlen): Use strlen_to_stridx.
1034 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1036 Use strlen_to_stridx.
1037 (pass_strlen::execute): Release strlen_to_stridx.
1038 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1039 (-Wstringop-truncation): Document new option.
1041 2017-11-10 Martin Liska <mliska@suse.cz>
1043 PR gcov-profile/82702
1044 * gcov.c (main): Handle intermediate files in a different
1046 (get_gcov_intermediate_filename): New function.
1047 (output_gcov_file): Remove support of intermediate files.
1048 (generate_results): Allocate intermediate file.
1049 (release_structures): Clean-up properly fn_end.
1050 (output_intermediate_file): Start iterating with line 1.
1052 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1054 PR tree-optimization/82929
1055 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1056 ops_swapped_p non-static data member.
1057 (store_immediate_info::store_immediate_info): Clear it.
1058 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1059 ops set ops_swapped_p.
1060 (count_multiple_uses): Handle ops_swapped_p.
1062 2017-11-10 Martin Liska <mliska@suse.cz>
1064 * coverage.c (coverage_init): Stream information about
1065 support of has_unexecuted_blocks.
1066 * doc/gcov.texi: Document that.
1067 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1068 * gcov.c (read_graph_file): Likewise.
1069 (output_line_beginning): Fix a small issue with
1072 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1074 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1075 reference of trivial component.
1077 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1080 * gimple-ssa-store-merging.c
1081 (pass_store_merging::terminate_all_aliasing_chains): For
1082 gimple_store_p stmts also call refs_output_dependent_p.
1084 PR rtl-optimization/82913
1085 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1088 2017-11-09 Jeff Law <law@redhat.com>
1090 * vr-values.h: New file with vr_values class.
1091 * tree-vrp.c: Include vr-values.h
1092 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1093 data objects into the vr_values class.
1094 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1095 (get_value_range): Make it a member function within vr_values class.
1096 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1097 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1098 (op_with_constant_singleton_value_range): Likewise.
1099 (extract_range_for_var_from_comparison_expr): Likewise.
1100 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1101 (extract_range_from_binary_expr): Likewise.
1102 (extract_range_from_unary_expr): Likewise.
1103 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1104 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1105 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1106 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1107 (compare_name_with_value, compare_names): Likewise.
1108 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1109 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1110 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1111 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1112 (extract_range_from_phi_node): Likewise.
1113 (simplify_truth_ops_using_ranges): Likewise.
1114 (simplify_div_or_mod_using_ranges): Likewise.
1115 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1116 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1117 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1118 (simplify_float_conversion_using_ranges): Likewise.
1119 (simplify_internal_call_using_ranges): Likewise.
1120 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1121 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1122 poitner via x_vr_values for calls into gimple folder.
1123 (vrp_initialize_lattice): Make this the vr_values ctor.
1124 (vrp_free_lattice): Make this the vr_values dtor.
1125 (set_vr_value): New function.
1126 (class vrp_prop): Add vr_values data member. Add various member
1127 functions as well as member functions that delegate to vr_values.
1128 (check_array_ref): Make a member function within vrp_prop class.
1129 (search_for_addr_array, vrp_initialize): Likewise.
1130 (vrp_finalize): Likewise. Revamp to avoid direct access to
1131 vr_value, values_propagated, etc.
1132 (check_array_bounds): Extract vrp_prop class instance pointer from
1133 walk info structure. Use it to call member functions.
1134 (check_all_array_refs): Make a member function within vrp_prop class.
1135 Smuggle class instance pointer via walk info structure.
1136 (x_vr_values): New local static.
1137 (vrp_valueize): Use x_vr_values to get class instance.
1138 (vr_valueize_1): Likewise.
1139 (class vrp_folder): Add vr_values data member. Add various member
1140 functions as well as member functions that delegate to vr_values.
1141 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1142 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1143 class instance from vr_values. Use it to call member functions.
1144 (vrp_dom_walker): Add vr_values data member.
1145 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1146 instance via x_vr_values.
1147 (identify_jump_threads): Accept vr_values as argument. Store
1148 it into the walker structure.
1149 (evrp_dom_walker): Add vr_values class data member. Add various
1151 (evrp_dom_walker::try_find_new_range): Use vr_values data
1152 member to access the memory allocator.
1153 (evrp_dom_walker::before_dom_children): Store vr_values class
1154 instance into the vrp_folder class.
1155 (evrp_dom_walker::push_value_range): Rework to avoid direct
1156 access to num_vr_values and vr_value.
1157 (evrp_dom_walker::pop_value_range): Likewise.
1158 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1159 Use vr_values to get to dump_all_value_ranges member function.
1160 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1161 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1162 Pass vr_values class instance down to identify_jump_threads.
1163 Remove call to vrp_free_lattice.
1164 (debug_all_value_ranges): Remove.
1166 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1167 (vrp_folder): Likewise.
1169 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1170 Get it from the existing bitmap instead.
1171 (vrp_intersect_ranges_1): Likewise.
1173 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1175 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1177 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1178 set bit_not_p to it.
1179 (imm_store_chain_info::coalesce_immediate_stores): Break group
1180 if bit_not_p is different.
1181 (count_multiple_uses, split_group,
1182 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1183 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1184 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1185 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1186 to store_immediate_info ctor.
1188 2017-11-09 Jim Wilson <jimw@sifive.com>
1190 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1191 (scan_prog_file): Likewise.
1193 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1195 * bb-reorder.c (max_entry_frequency): Remove.
1196 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1197 connect_traces, push_to_next_round_p): Remove prototypes.
1198 (find_traces_1_round): Use counts only.
1199 (push_to_next_round_p): Likewise.
1200 (find_traces): Likewise.
1201 (rotate_loop): Likewise.
1202 (find_traces_1_round): Likewise.
1203 (connect_traces): Likewise.
1204 (edge_order): Likewise.
1206 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1208 * config/arm/arm.c (output_return_instruction): Add comments to
1209 indicate requirement for cmse_nonsecure_entry return to account
1210 for the size of clearing instruction output here.
1211 (thumb_exit): Likewise.
1212 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1213 return in hardfloat mode.
1215 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1217 * config/rs6000/rs6000.c (machine_function): Add a bool,
1218 "toc_is_wrapped_separately".
1219 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1220 if it wasn't explicitly set or unset, we are optimizing for speed, and
1221 doing separate shrink-wrapping.
1222 (rs6000_get_separate_components): Enable the TOC component if
1223 saving the TOC register in the prologue.
1224 (rs6000_components_for_bb): Handle the TOC component.
1225 (rs6000_emit_prologue_components): Store the TOC register where needed.
1226 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1227 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1229 2017-11-09 Martin Jambor <mjambor@suse.cz>
1231 * ipa-param-manipulation.c: New file.
1232 * ipa-param-manipulation.h: Likewise.
1233 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1234 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1235 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1236 (ipa_parm_adjustment): Likewise.
1237 (ipa_parm_adjustment_vec): Likewise.
1238 (ipa_get_vector_of_formal_parms): Moved declaration to
1239 ipa-param-manipulation.h.
1240 (ipa_get_vector_of_formal_parm_types): Likewise.
1241 (ipa_modify_formal_parameters): Likewise.
1242 (ipa_modify_call_arguments): Likewise.
1243 (ipa_combine_adjustments): Likewise.
1244 (ipa_dump_param_adjustments): Likewise.
1245 (ipa_modify_expr): Likewise.
1246 (ipa_get_adjustment_candidate): Likewise.
1247 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1248 ipa-param-manipulation.c.
1249 (ipa_get_vector_of_formal_parm_types): Likewise.
1250 (ipa_modify_formal_parameters): Likewise.
1251 (ipa_modify_call_arguments): Likewise.
1252 (ipa_modify_expr): Likewise.
1253 (get_ssa_base_param): Likewise.
1254 (ipa_get_adjustment_candidate): Likewise.
1255 (index_in_adjustments_multiple_times_p): Likewise.
1256 (ipa_combine_adjustments): Likewise.
1257 (ipa_dump_param_adjustments): Likewise.
1258 * tree-sra.c: Also include ipa-param-manipulation.h
1259 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1262 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1263 Alan Hayward <alan.hayward@arm.com>
1264 David Sherwood <david.sherwood@arm.com>
1266 * doc/sourcebuild.texi (vect_masked_store): Document.
1268 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1269 Alan Hayward <alan.hayward@arm.com>
1270 David Sherwood <david.sherwood@arm.com>
1272 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1274 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1275 Alan Hayward <alan.hayward@arm.com>
1276 David Sherwood <david.sherwood@arm.com>
1278 * doc/sourcebuild.texi (vect_variable_length): Document.
1280 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1281 Alan Hayward <alan.hayward@arm.com>
1282 David Sherwood <david.sherwood@arm.com>
1284 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1286 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1287 Alan Hayward <alan.hayward@arm.com>
1288 David Sherwood <david.sherwood@arm.com>
1290 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1292 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1293 Alan Hayward <alan.hayward@arm.com>
1294 David Sherwood <david.sherwood@arm.com>
1296 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1297 previously undocumented selectors.
1298 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1300 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1302 * doc/rtl.texi (const_vector): Say that elements can be
1303 const_wide_ints too.
1304 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1305 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1306 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1307 * optabs.c (expand_vector_broadcast): Likewise.
1309 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1310 Alan Hayward <alan.hayward@arm.com>
1311 David Sherwood <david.sherwood@arm.com>
1313 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1314 scaled index even if the unscaled address was invalid.
1315 Don't increase the complexity of using a scale in that case.
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/rtl.texi: Rewrite the subreg rules so that they partition
1322 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1323 UNITS_PER_WORD bytes.
1324 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1325 based on REGMODE_NATURAL_SIZE of the inner mode.
1326 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1327 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1328 * expmed.c (lowpart_bit_field_p): Divide the value up into
1329 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1330 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1331 whether something is likely to occupy more than one register.
1333 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1336 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1337 profile_count::adjust_for_ipa_scaling.
1338 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1339 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1341 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1343 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1345 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1346 (split_group): Add total_orig and total_new arguments, estimate the
1347 number of statements related to the store group without store merging
1348 and with store merging.
1349 (imm_store_chain_info::output_merged_store): Adjust split_group
1350 callers, punt if estimated number of statements with store merging
1351 is not smaller than estimated number of statements without it.
1353 (handled_load): Remove has_single_use checks.
1354 (pass_store_merging::process_store): Likewise.
1356 2017-11-09 Richard Biener <rguenther@suse.de>
1358 PR tree-optimization/82902
1359 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1361 2017-11-09 Martin Liska <mliska@suse.cz>
1364 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1367 2017-11-09 Martin Liska <mliska@suse.cz>
1369 PR tree-optimization/82669
1370 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1372 2017-11-09 Martin Liska <mliska@suse.cz>
1374 PR gcov-profile/48463
1375 * coverage.c (coverage_begin_function): Output also end locus
1376 of a function and information whether the function is
1378 * gcov-dump.c (tag_function): Parse and print the information.
1379 * gcov.c (INCLUDE_MAP): Add include.
1380 (INCLUDE_SET): Likewise.
1381 (struct line_info): Move earlier in the source file because
1382 of vector<line_info> in function_info structure.
1383 (line_info::line_info): Likewise.
1384 (line_info::has_block): Likewise.
1385 (struct source_info): Add new member index.
1386 (source_info::get_functions_at_location): New function.
1387 (function_info::group_line_p): New function.
1388 (output_intermediate_line): New function.
1389 (output_intermediate_file): Use the mentioned function.
1390 (struct function_start): New.
1391 (struct function_start_pair_hash): Likewise.
1392 (process_file): Add code that identifies group functions.
1393 Assign lines either to global or function scope.
1394 (generate_results): Skip artificial functions.
1395 (find_source): Assign index for each source file.
1396 (read_graph_file): Read new flag artificial and end_line.
1397 (add_line_counts): Assign it either to global of function scope.
1398 (accumulate_line_counts): Isolate core of the function to
1399 accumulate_line_info and call it for both function and global
1401 (accumulate_line_info): New function.
1402 (output_line_beginning): Fix GNU coding style.
1403 (print_source_line): New function.
1404 (output_line_details): Likewise.
1405 (output_function_details): Likewise.
1406 (output_lines): Iterate both source (global) scope and function
1408 (struct function_line_start_cmp): New class.
1409 * doc/gcov.texi: Reflect changes in documentation.
1411 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1414 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1415 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1416 and similarly for not instead of neg.
1418 2017-11-08 Andi Kleen <ak@linux.intel.com>
1420 * config/i386/i386.opt: Add -mforce-indirect-call.
1421 * config/i386/predicates.md: Check for flag_force_indirect_call.
1422 * doc/invoke.texi: Document -mforce-indirect-call
1424 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1426 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1428 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1429 config/riscv/riscv.c (predict.h): New include.
1430 (riscv_slow_unaligned_access_p): No longer static.
1431 (riscv_block_move_straight): Add require.
1432 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1434 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1437 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1438 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1439 nonimmediate_operand predicate for operand 1 instead of
1442 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1444 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1446 * config/aarch64/constraints.md (Uml): New constraint.
1447 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1450 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1452 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1453 of two vec_duplicates into a vec_concat.
1455 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1457 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1458 Simplify vec_merge of vec_duplicate and vec_concat.
1459 * config/aarch64/constraints.md (Utq): New constraint.
1460 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1463 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1465 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1466 Simplify vec_merge of vec_duplicate and const_vector.
1467 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1469 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1470 mode iterator. Update predicate on operand 1 to
1471 handle non-const_vec constants. Delete constraints.
1472 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1474 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1476 PR tree-optimization/78821
1477 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1479 (store_operand_info::store_operand_info): Initialize it to false.
1480 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1481 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1482 store in the group, and if chain_info is non-NULL, to ignore altogether
1484 (compatible_load_p): Fail if bit_not_p does not match.
1485 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1486 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1487 (pass_store_merging::process_store): Adjust
1488 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1489 call terminate_all_aliasing_chains newly when adding a store into
1490 a chain with non-NULL chain_info.
1492 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1494 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1496 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1498 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1500 (aarch64_layout_frame): Initialise emit_frame_chain.
1501 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1502 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1504 2017-11-08 Martin Liska <mliska@suse.cz>
1506 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1507 of gimple_call_internal_p.
1509 2017-11-07 Tom de Vries <tom@codesourcery.com>
1511 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1513 2017-11-07 Tom de Vries <tom@codesourcery.com>
1515 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1518 2017-11-08 Martin Liska <mliska@suse.cz>
1521 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1523 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1525 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1527 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1530 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1531 (*cmp<mode>_ccz_1): New insn with $k alternative.
1534 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1535 RTX_COMM_COMPARE as commutative as well.
1536 (ix86_binary_operator_ok): Formatting fix.
1537 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1538 *<code><mode>3<mask_name><round_saeonly_name>,
1539 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1540 *<s>mul<mode>3_highpart<mask_name>,
1541 *vec_widen_umult_even_v16si<mask_name>,
1542 *vec_widen_umult_even_v8si<mask_name>,
1543 *vec_widen_umult_even_v4si<mask_name>,
1544 *vec_widen_smult_even_v16si<mask_name>,
1545 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1546 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1547 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1548 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1549 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1550 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1551 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1552 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1553 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1554 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1555 ix86_binary_operator_ok. Formatting fixes.
1556 (*<plusminus_insn><mode>3<mask_name><round_name>,
1557 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1560 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1562 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1563 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1564 eq and ne if TARGET_ISEL.
1565 (cmp): New code_iterator.
1566 (UNS, UNSU_, UNSIK): New code_attrs.
1567 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1568 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1570 ("eq<mode>3"): ... this.
1571 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1573 ("ne<mode>3"): ... this.
1575 2017-11-07 Julia Koval <julia.koval@intel.com>
1578 * common/config/i386/i386-common.c
1579 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1580 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1581 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1582 * config/i386/i386.opt: Ditto.
1583 * config/i386/i386.c (ix86_target_string): Ditto.
1584 (ix86_option_override_internal): Ditto.
1585 (ix86_init_mpx_builtins): Move MPX to args2.
1586 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1587 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1588 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1589 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1590 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1591 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1592 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1594 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1597 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1598 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1599 (zero-extendsidi peephole2): Remove peephole.
1601 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1604 * stor-layout.c: Include attribs.h.
1605 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1606 explicit lookup of "aligned" attribute.
1608 2017-11-07 Andrew Waterman <andrew@sifive.com>
1610 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1611 (riscv_expand_block_move): Likewise.
1612 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1614 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1615 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1616 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1617 (riscv_adjust_block_mem): Likewise.
1618 (riscv_block_move_loop): Likewise.
1619 (riscv_expand_block_move): Likewise.
1620 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1622 2017-11-07 Michael Clark <michaeljclark@mac.com>
1624 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1625 (MUSL_DYNAMIC_LINKER): Likewise.
1627 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1629 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1630 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1633 2017-11-07 Richard Biener <rguenther@suse.de>
1635 * match.pd: Fix build.
1637 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1638 Jackson Woodruff <jackson.woodruff@arm.com>
1640 PR tree-optimization/71026
1641 * match.pd: Canonicalize negate in division.
1643 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1646 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1648 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1650 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1651 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1653 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1655 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1656 non-scalar integral types.
1657 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1658 (-(A-B), -(~A)): New transformations.
1660 2017-11-07 Tom de Vries <tom@codesourcery.com>
1662 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1663 semicolon after "do {} while (0)".
1664 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1665 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1666 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1667 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1668 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1670 2017-11-07 Tom de Vries <tom@codesourcery.com>
1672 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1673 after "do {} while (0)".
1674 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1675 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1676 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1677 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1678 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1680 2017-11-07 Tom de Vries <tom@codesourcery.com>
1683 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1685 (arm_rtx_costs_internal): Add missing semicolon after
1686 HANDLE_NARROW_SHIFT_ARITH call.
1688 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1690 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1691 disable isel if it was not set explicitly.
1693 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1695 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1697 (add_type_attribute) likewise.
1699 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1701 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1702 of struct ix86_frame.
1703 (ix86_initial_elimination_offset): Likewise.
1704 (ix86_expand_split_stack_prologue): Likewise.
1706 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1708 * tree-vrp.h (enum value_range_type): Update stale comment.
1710 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1711 Alan Hayward <alan.hayward@arm.com>
1712 David Sherwood <david.sherwood@arm.com>
1714 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1715 (aarch64_expand_vec_perm_const): Take the number of units too.
1716 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1717 (aarch64_expand_vec_perm_const): Likewise.
1718 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1719 (vec_perm<mode>): Update accordingly.
1721 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1722 Alan Hayward <alan.hayward@arm.com>
1723 David Sherwood <david.sherwood@arm.com>
1725 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1726 Take the number of units too.
1727 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1728 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1729 but check for a vector mode before rather than after the call.
1730 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1731 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1732 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1733 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1734 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1735 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1736 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1737 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1738 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1739 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1740 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1741 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1742 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1743 (aarch64_sqdmull2_n<mode>): Update accordingly.
1745 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1746 Alan Hayward <alan.hayward@arm.com>
1747 David Sherwood <david.sherwood@arm.com>
1749 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1750 the number of units too.
1751 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1752 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1753 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1754 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1755 (vec_store_lanesxi<mode>): Update accordingly.
1757 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1758 Alan Hayward <alan.hayward@arm.com>
1759 David Sherwood <david.sherwood@arm.com>
1761 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1762 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1763 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1764 of units rather than the mode.
1765 * config/aarch64/iterators.md (nunits): New mode attribute.
1766 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1767 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1768 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1769 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1770 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1771 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1772 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1773 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1774 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1775 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1776 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1777 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1778 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1779 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1780 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1781 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1782 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1783 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1784 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1785 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1786 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1787 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1788 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1789 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1790 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1791 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1792 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1793 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1794 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1795 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1796 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1797 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1798 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1799 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1800 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1801 (aarch64_simd_vec_setv2di): Likewise.
1803 2017-11-06 Carl Love <cel@us.ibm.com>
1805 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1807 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1808 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1810 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1812 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1813 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1814 Add power 8 macro expansions.
1815 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1816 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1817 power 8 instructions. (VSX_XXBR): Add iterator.
1819 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1821 * config/arm/arm.md (predicable_short_it): Change default to "no",
1822 improve documentation, remove uses that are identical to the default.
1823 (enabled_for_depr_it): Rename to enabled_for_short_it.
1824 * gcc/config/arm/arm-fixed.md (predicable_short_it):
1825 Remove default uses.
1826 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1827 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1828 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1829 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1831 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1834 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1835 float128 helper macros, which are no longer used after deleting
1836 the old 'q' built-in functions, and moving the round to odd
1837 built-in functions to being special built-in functions.
1838 (BU_FLOAT128_2): Likewise.
1839 (BU_FLOAT128_1_HW): Likewise.
1840 (BU_FLOAT128_2_HW): Likewise.
1841 (BU_FLOAT128_3_HW): Likewise.
1842 (FABSQ): Delete old 'q' built-in functions.
1843 (COPYSIGNQ): Likewise.
1844 (SQRTF128_ODD): Move round to odd built-in functions to be
1845 special built-in functions, so that we can handle
1846 -mabi=ieeelongdouble.
1847 (TRUNCF128_ODD): Likewise.
1848 (ADDF128_ODD): Likewise.
1849 (SUBF128_ODD): Likewise.
1850 (MULF128_ODD): Likewise.
1851 (DIVF128_ODD): Likewise.
1852 (FMAF128_ODD): Likewise.
1853 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1854 built-in names to 'f128'.
1855 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1856 old 'q' built-in functions, as the machine independent code for
1857 'f128' built-in functions handles this.
1858 (rs6000_expand_builtin): Add expansion for float128 round to odd
1859 functions, keying off on -mabi=ieeelongdouble of whether to use
1860 the KFmode or TFmode variant.
1861 (rs6000_init_builtins): Initialize the _Float128 round to odd
1863 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1864 _Float128 'q' built-in functions are now mapped into the new
1865 'f128' built-in functions.
1867 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1869 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1870 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1872 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1875 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1876 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1877 * toplev.c: Include "ipa-fnsummary.h".
1878 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1880 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1882 PR tree-optimization/82838
1883 * gimple-ssa-store-merging.c
1884 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1885 on a separate gimple_seq which is then appended to seq.
1887 2017-11-06 Jeff Law <law@redhat.com>
1890 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1891 (get_probe_interval): New functions.
1892 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1893 (ix86_adjust_stack_and_probe): Likewise.
1894 (output_adjust_stack_and_probe): Likewise.
1895 (ix86_emit_probe_stack_range): Likewise.
1896 (ix86_expand_prologue): Likewise.
1898 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1900 PR tree-optimization/82816
1901 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1902 if the modes of the two types are the same.
1903 (convert_plusminus_to_widen): Likewise.
1905 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1907 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1909 (usadv16qi): New define_expand.
1910 (usadv8hi): New define_expand.
1912 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1915 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1917 (inline_transform): Likewise.
1918 * predict.c (counts_to_freqs): Remove useless conditional.
1919 * profile-count.h (profile_count::apply_scale): Move sanity check.
1920 * tree-inline.c (copy_bb): Always scale.
1921 (copy_cfg_body): Likewise.
1923 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1926 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1928 (*cmp_ite0): Add enabled_for_depr_it attribute.
1929 (*cmp_ite1): Likewise.
1931 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1933 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1936 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1938 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1939 return the value by pointer.
1940 (extract_range_from_multiplicative_op_1): Update accordingly.
1941 Return as soon as an operation fails.
1943 2017-11-05 Tom de Vries <tom@codesourcery.com>
1946 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1947 (DEF_SANITIZER_BUILTIN): ... here.
1948 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1949 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1951 2017-11-05 Tom de Vries <tom@codesourcery.com>
1954 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1956 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1957 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1958 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1960 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1961 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1962 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1964 2017-11-05 Tom de Vries <tom@codesourcery.com>
1967 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1970 2017-11-04 Michael Clark <michaeljclark@mac.com>
1972 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1973 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1975 (*addsi3_extended): Likewise.
1976 (*addsi3_extended2): Likewise.
1977 (<optab>si3): Likewise.
1978 (<optab>di3): Likewise.
1979 (<optab><mode>3): Likewise.
1980 (<*optabe>si3_internal): Likewise.
1981 (zero_extendqi<SUPERQI:mode>2): Likewise.
1982 (*add<mode>hi3): Likewise.
1983 (*xor<mode>hi3): Likewise.
1984 (<optab>di3): Likewise.
1985 (*<optab>si3_extend): Likewise.
1986 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1987 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1988 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1990 2017-11-04 Andrew Waterman <andrew@sifive.com>
1992 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1993 TARGET_STRICT_ALIGN based upon -mtune argument.
1995 2017-11-04 Andrew Waterman <andrew@sifive.com>
1997 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1999 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2001 config/i386/i386.c (choose_basereg): Use optional scratch
2002 register and add assertion.
2003 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2004 needed, and don't allocate stack.
2005 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2006 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2007 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2009 2017-11-03 Jeff Law <law@redhat.com>
2011 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2012 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2013 to probe at the start of a noreturn function.
2015 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2017 PR tree-optimization/78821
2018 * gimple-ssa-store-merging.c: Update the file comment.
2019 (MAX_STORE_ALIAS_CHECKS): Define.
2020 (struct store_operand_info): New type.
2021 (store_operand_info::store_operand_info): New constructor.
2022 (struct store_immediate_info): Add rhs_code and ops data members.
2023 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2024 and op1r arguments to the ctor, initialize corresponding data members.
2025 (struct merged_store_group): Add load_align_base and load_align
2027 (merged_store_group::merged_store_group): Initialize them.
2028 (merged_store_group::do_merge): Update them.
2029 (merged_store_group::apply_stores): Pick the constant for
2030 encode_tree_to_bitpos from one of the two operands, or skip
2031 encode_tree_to_bitpos if neither operand is a constant.
2032 (class pass_store_merging): Add process_store method decl. Remove
2033 bool argument from terminate_all_aliasing_chains method decl.
2034 (pass_store_merging::terminate_all_aliasing_chains): Remove
2035 var_offset_p argument and corresponding handling.
2036 (stmts_may_clobber_ref_p): New function.
2037 (compatible_load_p): New function.
2038 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2039 if there is overlap and rhs_code is not INTEGER_CST. For
2040 non-overlapping stores terminate group if rhs is not mergeable.
2041 (get_alias_type_for_stmts): Change first argument from
2042 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2043 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2044 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2046 (get_location_for_stmts): Change first argument from
2047 auto_vec<gimple *> & to vec<gimple *> &.
2048 (struct split_store): Remove orig_stmts data member, add orig_stores.
2049 (split_store::split_store): Create orig_stores rather than orig_stmts.
2050 (find_constituent_stmts): Renamed to ...
2051 (find_constituent_stores): ... this. Change second argument from
2052 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2053 to info structures rather than the statements.
2054 (split_group): Rename ALLOW_UNALIGNED argument to
2055 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2056 it. Adjust find_constituent_stores caller.
2057 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2058 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2059 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2060 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2061 (mem_valid_for_store_merging): New function.
2062 (handled_load): New function.
2063 (pass_store_merging::process_store): New method.
2064 (pass_store_merging::execute): Use process_store method. Adjust
2065 terminate_all_aliasing_chains caller.
2067 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2069 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2070 Return true for more constants, symbols and label references.
2071 (aarch64_valid_floating_const): Remove unused function.
2073 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2076 * config/aarch64/aarch64.c (aarch64_layout_frame):
2077 Undo forcing of LR at bottom of frame.
2079 2017-11-03 Jeff Law <law@redhat.com>
2082 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2083 for int_registers_saved.
2085 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2086 extracted from tree-ssa-dom.c.
2087 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2088 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2089 (record_equivalences_from_incoming_edge): Add additional argument
2090 to single_pred_edge_ignoring_loop_edges call.
2091 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2092 (uncprop_dom_walker::before_dom_children): Add additional argument
2093 to single_pred_edge_ignoring_loop_edges call.
2094 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2095 single_pred_edge_ignoring_loop_edges rather than open coding.
2096 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2098 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2100 * match.pd (-(-A)): Rewrite.
2102 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2104 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2105 (rs6000_emit_int_cmove): New declaration.
2106 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2107 (rs6000_emit_sISEL): Delete.
2108 (rs6000_emit_int_cmove): Make non-static.
2109 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2110 instead of rs6000_emit_sISEL.
2112 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2114 * asan.c (create_cond_insert_point): Maintain profile.
2115 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2117 * basic-block.h (struct basic_block_def): Remove frequency.
2118 (EDGE_FREQUENCY): Use to_frequency
2119 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2121 (find_traces): Update to use to_frequency.
2122 (find_traces_1_round): Likewise; use only IPA counts.
2123 (bb_to_key): Likewise.
2124 (connect_traces): Use IPA counts only.
2125 (copy_bb_p): Update to use to_frequency.
2126 (fix_up_crossing_landing_pad): Likewise.
2127 (sanitize_hot_paths): Likewise.
2128 * bt-load.c (basic_block_freq): Likewise.
2129 * cfg.c (init_flow): Set count_max to uninitialized.
2130 (check_bb_profile): Remove frequencies; check counts.
2131 (dump_bb_info): Do not dump frequencies.
2132 (update_bb_profile_for_threading): Update counts only.
2133 (scale_bbs_frequencies_int): Likewise.
2134 (MAX_SAFE_MULTIPLIER): Remove.
2135 (scale_bbs_frequencies_gcov_type): Update counts only.
2136 (scale_bbs_frequencies_profile_count): Update counts only.
2137 (scale_bbs_frequencies): Update counts only.
2138 * cfg.h (struct control_flow_graph): Add count-max.
2139 (update_bb_profile_for_threading): Update prototype.
2140 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2141 (find_many_sub_basic_blocks): Likewise.
2142 * cfgcleanup.c (try_forward_edges): Likewise.
2143 (try_crossjump_to_edge): Likewise.
2144 * cfgexpand.c (expand_gimple_cond): Likewise.
2145 (expand_gimple_tailcall): Likewise.
2146 (construct_init_block): Likewise.
2147 (construct_exit_block): Likewise.
2148 * cfghooks.c (verify_flow_info): Check consistency of counts.
2149 (dump_bb_for_graph): Do not dump frequencies.
2150 (split_block_1): Do not update frequencies.
2151 (split_edge): Do not update frequencies.
2152 (make_forwarder_block): Do not update frequencies.
2153 (duplicate_block): Do not update frequencies.
2154 (account_profile_record): Do not update frequencies.
2155 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2156 for global heuristics.
2157 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2158 (expected_loop_iterations_unbounded): Use counts only.
2159 * cfgloopmanip.c (scale_loop_profile): Simplify.
2160 (create_empty_loop_on_edge): Simplify
2162 (duplicate_loop_to_header_edge): Simplify
2163 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2164 (update_br_prob_note): Take care of removing note when profile
2166 (relink_block_chain): Do not dump frequency.
2167 (rtl_account_profile_record): Use to_frequency.
2168 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2169 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2170 (cgraph_update_edges_for_call_stmt_node): Likewise.
2171 (cgraph_edge::verify_count_and_frequency): Update.
2172 (cgraph_node::verify_node): Temporarily disable frequency verification.
2173 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2174 to_cgraph_frequency.
2175 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2176 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2178 (cgraph_node::expand_thunk): Update profile.
2179 * except.c (dw2_build_landing_pads): Do not update frequency.
2180 * final.c (compute_alignments): Use to_frequency.
2181 (dump_basic_block_info): Do not dump frequency.
2182 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2183 (dump_gimple_bb_header): Do not dump frequency.
2184 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2186 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2187 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2188 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2189 (init_before_recovery): Do not update frequency.
2190 (sched_create_recovery_edges): Do not update frequency.
2191 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2192 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2193 (ipa_cp_c_finalize): Set max_count to uninitialized.
2194 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2195 (param_change_prob): Use counts.
2196 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2198 * ipa-split.c (consider_split): Use to_frequency.
2199 (split_function): Use to_frequency.
2200 * ira-build.c (loop_compare_func): Likewise.
2201 (mark_loops_for_removal): Likewise.
2202 (mark_all_loops_for_removal): Likewise.
2203 * loop-doloop.c (doloop_modify): Do not update frequency.
2204 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2206 * lto-streamer-in.c (input_function): Update count_max.
2207 * omp-expand.c (expand_omp_taskreg): Update count_max.
2208 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2209 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2210 (maybe_hot_count_p): Use ipa counts only.
2211 (maybe_hot_bb_p): Simplify.
2212 (maybe_hot_edge_p): Simplify.
2213 (probably_never_executed): Do not take frequency argument.
2214 (probably_never_executed_bb_p): Do not pass frequency.
2215 (probably_never_executed_edge_p): Likewise.
2216 (combine_predictions_for_bb): Check that profile is nonzero.
2217 (propagate_freq): Do not set frequency.
2218 (drop_profile): Simplify.
2219 (counts_to_freqs): Simplify.
2220 (expensive_function_p): Use to_frequency.
2221 (propagate_unlikely_bbs_forward): Simplify.
2222 (determine_unlikely_bbs): Simplify.
2223 (estimate_bb_frequencies): Add hack to silence graphite issues.
2224 (compute_function_frequency): Use ipa counts.
2225 (pass_profile::execute): Update.
2226 (rebuild_frequencies): Use counts only.
2227 (force_edge_cold): Use counts only.
2228 * profile-count.c (profile_count::dump): Dump new count types.
2229 (profile_count::differs_from_p): Check compatiblity.
2230 (profile_count::to_frequency): New function.
2231 (profile_count::to_cgraph_frequency): New function.
2232 * profile-count.h (struct function): Declare.
2233 (enum profile_quality): Add profile_guessed_local and
2234 profile_guessed_global0.
2235 (class profile_proability): Decrease number of bits to 29;
2236 update from_reg_br_prob_note and to_reg_br_prob_note.
2237 (class profile_count: Update comment; decrease number of bits
2238 to 61. Check compatibility.
2239 (profile_count::compatible_p): New private member function.
2240 (profile_count::ipa_p): New member function.
2241 (profile_count::operator<): Handle global zero correctly.
2242 (profile_count::operator>): Handle global zero correctly.
2243 (profile_count::operator<=): Handle global zero correctly.
2244 (profile_count::operator>=): Handle global zero correctly.
2245 (profile_count::nonzero_p): New member function.
2246 (profile_count::force_nonzero): New member function.
2247 (profile_count::max): New member function.
2248 (profile_count::apply_scale): Handle IPA scalling.
2249 (profile_count::guessed_local): New member function.
2250 (profile_count::global0): New member function.
2251 (profile_count::ipa): New member function.
2252 (profile_count::to_frequency): Declare.
2253 (profile_count::to_cgraph_frequency): Declare.
2254 * profile.c (OVERLAP_BASE): Delete.
2255 (compute_frequency_overlap): Delete.
2256 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2257 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2258 * sched-ebb.c (rank): Use counts only.
2259 * shrink-wrap.c (handle_simple_exit): Use counts only.
2260 (try_shrink_wrapping): Use counts only.
2261 (place_prologue_for_one_component): Use counts only.
2262 * tracer.c (find_best_predecessor): Use to_frequency.
2263 (find_trace): Use to_frequency.
2264 (tail_duplicate): Use to_frequency.
2265 * trans-mem.c (expand_transaction): Do not update frequency.
2266 * tree-call-cdce.c: Do not update frequency.
2267 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2268 (gimple_merge_blocks): Likewise.
2269 (gimple_split_edge): Likewise.
2270 (gimple_duplicate_sese_region): Likewise.
2271 (gimple_duplicate_sese_tail): Likewise.
2272 (move_sese_region_to_fn): Likewise.
2273 (gimple_account_profile_record): Likewise.
2274 (insert_cond_bb): Likewise.
2275 * tree-complex.c (expand_complex_div_wide): Likewise.
2276 * tree-eh.c (lower_resx): Update profile.
2277 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2279 (initialize_cfun): Do not initialize frequencies
2280 (freqs_to_counts): Delete.
2281 (copy_cfg_body): Ignore count parameter.
2282 (copy_body): Update.
2283 (expand_call_inline): Update count_max.
2284 (optimize_inline_calls): Update count_max.
2285 (tree_function_versioning): Update count_max.
2286 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2287 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2289 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2290 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2291 (try_peel_loop): Likewise.
2292 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2294 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2295 (tree_transform_and_unroll_loop): Do not use frequencies
2296 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2297 Use reliable prediction only.
2298 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2299 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2300 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2301 probability scaling.
2302 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2303 not update frequency
2304 (any_remaining_duplicated_blocks): Likewise.
2305 (update_profile): Likewise.
2306 (estimated_freqs_path): Delete.
2307 (freqs_to_counts_path): Delete.
2308 (clear_counts_path): Delete.
2309 (ssa_fix_duplicate_block_edges): Likewise.
2310 (duplicate_thread_path): Likewise.
2311 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2312 * tree-tailcall.c (decrease_profile): Do not update frequency.
2313 (eliminate_tail_call): Likewise.
2314 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2315 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2316 (optimize_mask_stores): Likewise.
2317 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2318 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2319 (ubsan_expand_ptr_ifn): Update profile.
2320 * value-prof.c (gimple_ic): Simplify.
2321 * value-prof.h (gimple_ic): Update prototype.
2322 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2323 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2325 (want_inline_self_recursive_call_p): Likewise.
2326 (resolve_noninline_speculation): Only cummulate defined counts.
2327 (inline_small_functions): Use nonzero_p.
2328 (ipa_inline): Do not access freed node.
2330 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2332 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2333 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2335 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2337 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2338 non-legitimate address.
2340 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2342 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2343 (*lt0_<mode>di, *lt0_<mode>si): New.
2345 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2347 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2348 TARGET_PAIRED_FLOAT.
2350 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2351 Jim Wilson <jim.wilson@linaro.org>
2353 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2354 * config/aarch64/aarch64-tune.md: Regenerated.
2355 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2356 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2358 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2360 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2361 (arc_expand_prologue): Restore blink for millicode.
2362 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2364 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2367 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2368 gen_vec_duplicate after forcing the scalar into a register.
2370 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2372 * combine (try_combine): Print the insns input to try_combine to the
2375 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2378 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2379 Remove second argument from aarch64_process_target_attr call.
2380 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2382 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2384 (aarch64_handle_attr_arch): Remove second argument.
2385 (aarch64_handle_attr_cpu): Ditto.
2386 (aarch64_handle_attr_tune): Ditto.
2387 (aarch64_handle_attr_isa_flags): Ditto.
2388 (aarch64_process_one_target_attr): Ditto.
2389 (aarch64_process_target_attr): Ditto.
2390 (aarch64_option_valid_attribute_p): Remove second argument.
2391 on aarch64_process_target_attr call.
2393 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2395 * diagnostic.c: Include "selftest-diagnostic.h".
2396 (selftest::assert_location_text): New function.
2397 (selftest::test_diagnostic_get_location_text): New function.
2398 (selftest::diagnostic_c_tests): Call it.
2400 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2402 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2403 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2404 (class selftest::test_diagnostic_context): Move to...
2405 * selftest-diagnostic.c: New file.
2406 * selftest-diagnostic.h: New file.
2408 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2410 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2411 offset range for FT32B.
2412 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2413 * config/ft32/ft32.md: Add TARGET_NOPM.
2414 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2415 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2417 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2419 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2421 2017-11-02 Jeff Law <law@redhat.com>
2423 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2424 virtual keyword on FINAL OVERRIDE members.
2426 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2427 virtuals together. Add virtual destructor.
2428 (substitute_and_fold_engine): Similarly.
2430 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2432 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2434 2017-11-02 Richard Biener <rguenther@suse.de>
2436 PR tree-optimization/82795
2437 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2439 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2441 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2443 (gcc_GAS_CHECK_FEATURE): Remove.
2444 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2445 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2446 * configure: Regenerate.
2448 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2450 * config/arc/arc.c (hwloop_optimize): Account for empty
2453 2017-11-02 Richard Biener <rguenther@suse.de>
2456 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2457 Truncate ARRAY_REF index and element size.
2459 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2461 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2463 2017-11-01 Jeff Law <law@redhat.com>
2465 * tree-ssa-ccp.c (ccp_folder): New class derived from
2466 substitute_and_fold_engine.
2467 (ccp_folder::get_value): New member function.
2468 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2469 (ccp_fold_stmt): Remove prototype.
2470 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2471 * tree-ssa-copy.c (copy_folder): New class derived from
2472 substitute_and_fold_engine.
2473 (copy_folder::get_value): Renamed from get_value.
2474 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2475 * tree-vrp.c (vrp_folder): New class derived from
2476 substitute_and_fold_engine.
2477 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2478 (vrp_folder::get_value): New member function.
2479 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2480 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2481 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2482 provide a class interface to folder/substitute routines.
2483 (ssa_prop_fold_stmt_fn): Remove typedef.
2484 (ssa_prop_get_value_fn): Likewise.
2485 (subsitute_and_fold): Remove prototype.
2486 (replace_uses_in): Likewise.
2487 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2488 Renamed from replace_uses_in. Call the virtual member function
2489 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2490 (substitute_and_fold_dom_walker): Remove initialization of
2491 data member entries for calbacks. Add substitute_and_fold_engine
2492 member and initialize it.
2493 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2494 member functions for get_value, replace_phi_args_in c
2495 replace_uses_in, and fold_stmt calls.
2496 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2497 substitute_and_fold. Remove assert. Update ctor call.
2499 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2500 (ssa_prop_visit_phi_fn): Likewise.
2501 (class ssa_propagation_engine): New class to provide an interface
2503 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2505 (ssa_prop_visit_phi): Likewise.
2506 (ssa_propagation_engine::simulate_stmt): Moved into class.
2507 Call visit_phi/visit_stmt from the class rather than via
2508 file scoped static variables.
2509 (ssa_propagation_engine::simulate_block): Moved into class.
2510 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2511 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2512 set file scoped statics for the visit_stmt/visit_phi callbacks.
2513 * tree-complex.c (complex_propagate): New class derived from
2514 ssa_propagation_engine.
2515 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2516 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2517 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2519 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2520 ssa_propagation_engine.
2521 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2522 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2523 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2524 * tree-ssa-copy.c (copy_prop): New class derived from
2525 ssa_propagation_engine.
2526 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2527 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2528 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2529 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2530 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2531 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2532 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2534 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2536 PR rtl-optimization/82778
2537 PR rtl-optimization/82597
2538 * compare-elim.c (struct comparison): Add in_a_setter field.
2539 (find_comparison_dom_walker::before_dom_children): Remove killed
2540 bitmap and df_simulate_find_defs call, instead walk the defs.
2541 Compute last_setter and initialize in_a_setter. Merge definitions
2542 with first initialization for a few variables.
2543 (try_validate_parallel): Use insn_invalid_p instead of
2544 recog_memoized. Return insn rather than just the pattern.
2545 (try_merge_compare): Fix up comment. Don't uselessly test if
2546 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2548 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2551 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2552 Alan Hayward <alan.hayward@arm.com>
2553 David Sherwood <david.sherwood@arm.com>
2555 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2556 aarch64_hard_regno_nregs to get the number of registers
2559 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2560 Alan Hayward <alan.hayward@arm.com>
2561 David Sherwood <david.sherwood@arm.com>
2563 * config/aarch64/constraints.md (Upl): Rename to...
2565 * config/aarch64/aarch64.md
2566 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2569 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2570 Alan Hayward <alan.hayward@arm.com>
2571 David Sherwood <david.sherwood@arm.com>
2573 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2574 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2577 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2578 Alan Hayward <alan.hayward@arm.com>
2579 David Sherwood <david.sherwood@arm.com>
2581 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2582 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2583 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2585 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2586 of a single element.
2587 * config/aarch64/iterators.md: Add a comment above the permute
2588 unspecs to say that they are generated directly by
2589 aarch64_expand_vec_perm_const.
2590 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2592 2017-11-01 Nathan Sidwell <nathan@acm.org>
2594 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2596 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2598 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2599 and medany code models, and describe what they do.
2601 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2603 Revert accidental duplicate:
2605 * combine.c (can_change_dest_mode): Reject changes in
2606 REGMODE_NATURAL_SIZE.
2608 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2610 PR rtl-optimization/64682
2611 PR rtl-optimization/69567
2612 PR rtl-optimization/69737
2613 PR rtl-optimization/82683
2614 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2615 register mentioned in the note, drop the note, unless it came from I3,
2616 in which case it should go to I3 again.
2618 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2620 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2621 and return false if not.
2622 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2624 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2626 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2627 range is known to be empty.
2629 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2630 Alan Hayward <alan.hayward@arm.com>
2631 David Sherwood <david.sherwood@arm.com>
2633 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2634 and CONST_VECTOR_NUNITS instead of computing the number of units from
2635 the byte sizes of the vector and element.
2636 (simplify_binary_operation_1): Likewise.
2637 (simplify_const_binary_operation): Likewise.
2638 (simplify_ternary_operation): Likewise.
2640 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2641 Alan Hayward <alan.hayward@arm.com>
2642 David Sherwood <david.sherwood@arm.com>
2644 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2645 (int_mem_offset): ...this new function.
2646 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2647 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2648 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2651 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2652 Alan Hayward <alan.hayward@arm.com>
2653 David Sherwood <david.sherwood@arm.com>
2655 * lower-subreg.c (interesting_mode_p): New function.
2656 (compute_costs, find_decomposable_subregs, decompose_register)
2657 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2658 (resolve_clobber, dump_choices): Use it.
2660 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2661 Alan Hayward <alan.hayward@arm.com>
2662 David Sherwood <david.sherwood@arm.com>
2664 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2666 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2667 Alan Hayward <alan.hayward@arm.com>
2668 David Sherwood <david.sherwood@arm.com>
2670 * alias.c (find_base_value, find_base_term): Only process integer
2671 truncations. Check the precision rather than the size.
2673 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2674 Alan Hayward <alan.hayward@arm.com>
2675 David Sherwood <david.sherwood@arm.com>
2677 * machmode.h (is_narrower_int_mode): New function
2678 * optabs.c (expand_float, expand_fix): Use it.
2679 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2681 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2682 Alan Hayward <alan.hayward@arm.com>
2683 David Sherwood <david.sherwood@arm.com>
2685 * rtl.h (narrower_subreg_mode): New function.
2686 * ira-color.c (update_costs_from_allocno): Use it.
2688 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2689 Alan Hayward <alan.hayward@arm.com>
2690 David Sherwood <david.sherwood@arm.com>
2692 * optabs-query.h (convert_optab_p): New function, split out from...
2693 (convert_optab_handler): ...here.
2694 (widening_optab_handler): Delete.
2695 (find_widening_optab_handler): Remove permit_non_widening parameter.
2696 (find_widening_optab_handler_and_mode): Likewise. Provide an
2697 override that operates on mode class wrappers.
2698 * optabs-query.c (widening_optab_handler): Delete.
2699 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2700 parameter. Assert that the two modes are the same class and that
2701 the "from" mode is narrower than the "to" mode. Use
2702 convert_optab_handler instead of widening_optab_handler.
2703 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2704 instead of widening_optab_handler.
2705 * expr.c (expand_expr_real_2): Update calls to
2706 find_widening_optab_handler.
2707 * optabs.c (expand_widen_pattern_expr): Likewise.
2708 (expand_binop_directly): Take the insn_code as a parameter.
2709 (expand_binop): Only call find_widening_optab_handler for
2710 conversion optabs; use optab_handler otherwise. Update calls
2711 to find_widening_optab_handler and expand_binop_directly.
2712 Use convert_optab_handler instead of widening_optab_handler.
2713 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2714 find_widening_optab_handler and use scalar_mode rather than
2716 (convert_plusminus_to_widen): Likewise.
2718 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2719 Alan Hayward <alan.hayward@arm.com>
2720 David Sherwood <david.sherwood@arm.com>
2722 * machmode.h (fixed_size_mode): New class.
2723 * rtl.h (get_pool_mode): Return fixed_size_mode.
2724 * gengtype.c (main): Add fixed_size_mode.
2725 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2726 (get_raw_arg_mode): Likewise.
2727 * doc/tm.texi: Regenerate.
2728 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2729 * targhooks.c (default_get_reg_raw_mode): Likewise.
2730 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2731 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2732 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2733 (msp430_get_raw_result_mode): Likewise.
2734 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2735 * dbxout.c (dbxout_parms): Require fixed-size modes.
2736 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2737 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2738 * omp-low.c (lower_oacc_reductions): Likewise.
2739 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2740 (simplify_subreg): Update accordingly.
2741 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2742 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2743 that aren't fixed-size.
2744 (get_pool_mode): Return a fixed_size_mode.
2745 (output_constant_pool_2): Take a fixed_size_mode.
2747 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2748 Alan Hayward <alan.hayward@arm.com>
2749 David Sherwood <david.sherwood@arm.com>
2751 * doc/rtl.texi (vec_series): Document.
2752 (const): Say that the operand can be a vec_series.
2753 * rtl.def (VEC_SERIES): New rtx code.
2754 * rtl.h (const_vec_series_p_1): Declare.
2755 (const_vec_series_p): New function.
2756 * emit-rtl.h (gen_const_vec_series): Declare.
2757 (gen_vec_series): Likewise.
2758 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2759 (gen_vec_series): Likewise.
2760 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2761 * simplify-rtx.c (simplify_unary_operation): Handle negations
2763 (simplify_binary_operation_series): New function.
2764 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2765 (test_vector_ops_series): New function.
2766 (test_vector_ops): Call it.
2767 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2768 gen_const_vec_series.
2769 (altivec_lvsr): Likewise.
2770 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2772 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2773 Alan Hayward <alan.hayward@arm.com>
2774 David Sherwood <david.sherwood@arm.com>
2776 * doc/rtl.texi (const): Update description of address constants.
2777 Say that vector constants are allowed too.
2778 * common.md (E, F): Use CONSTANT_P instead of checking for
2780 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2781 checking for CONST_VECTOR.
2782 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2784 * expr.c (expand_expr_real_2): Check for vector modes instead
2785 of checking for CONST_VECTOR.
2786 * rtl.h (const_vec_p): New function.
2787 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2788 (unwrap_const_vec_duplicate): Handle them here too.
2790 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2791 David Malcolm <dmalcolm@redhat.com>
2792 Alan Hayward <alan.hayward@arm.com>
2793 David Sherwood <david.sherwood@arm.com>
2795 * rtl.h (vec_duplicate_p): New function.
2796 * selftest-rtl.c (assert_rtx_eq_at): New function.
2797 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2798 (assert_rtx_eq_at): Declare.
2799 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2800 * selftest-run-tests.c (selftest::run_tests): Call it.
2801 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2802 (simplify_unary_operation_1): Recursively handle vector duplicates.
2803 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2805 (simplify_subreg): Handle subregs of vector duplicates.
2806 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2807 (selftest::simplify_rtx_c_tests): New functions.
2809 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2810 Alan Hayward <alan.hayward@arm.com>
2811 David Sherwood <david.sherwood@arm.com>
2813 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2814 (gen_vec_duplicate): Likewise.
2815 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2817 (gen_const_vector): ...here.
2818 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2819 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2820 whose elements are all equal.
2821 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2822 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2823 (simplify_relational_operation): Likewise.
2824 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2826 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2827 (aarch64_expand_vector_init): Likewise.
2828 * config/arm/arm.c (neon_vdup_constant): Likewise.
2829 (neon_expand_vector_init): Likewise.
2830 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2831 (arm_block_set_unaligned_vect): Likewise.
2832 (arm_block_set_aligned_vect): Likewise.
2833 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2834 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2835 (expand_vec_perm_even_odd_pack): Likewise.
2836 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2837 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2838 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2839 gen_const_vec_duplicate.
2840 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2841 * config/mips/mips.c (mips_gen_const_int_vector): Use
2842 gen_const_vec_duplicate.
2843 (mips_expand_vector_init): Use CONST0_RTX.
2844 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2845 (define_split): Use gen_const_vec_duplicate.
2846 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2847 (define_split): Use gen_const_vec_duplicate.
2848 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2849 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2850 * config/spu/spu.c (spu_const): Likewise.
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 * combine.c (can_change_dest_mode): Reject changes in
2857 REGMODE_NATURAL_SIZE.
2859 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2861 * configure.ac (--enable-libssp): New.
2862 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2863 trying to determine target-specific default. Adjust indentation.
2864 * configure: Regenerated.
2865 * doc/install.texi (Configuration): Expand --disable-libssp
2868 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2870 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2873 2017-10-31 Martin Jambor <mjambor@suse.cz>
2876 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2878 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2880 * auto-profile.c (autofdo_source_profile::read): Use
2881 UNKNOWN_LOCATION rather than 0.
2882 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2883 (warning_at): ...this overload.
2884 (warning_at_rich_loc_n): Rename to...
2885 (warning_n): ...this overload.
2886 (error_at_rich_loc): Rename to...
2887 (error_at): ...this overload.
2888 (pedwarn_at_rich_loc): Rename to...
2889 (pedwarn): ...this overload.
2890 (permerror_at_rich_loc): Rename to...
2891 (permerror): ...this overload.
2892 (inform_at_rich_loc): Rename to...
2893 (inform): ...this overload.
2894 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2895 (diagnostic_n_impl_richloc): Rename to...
2896 (diagnostic_n_impl): ...this rich_location *-based decl.
2897 (inform_at_rich_loc): Rename to...
2898 (inform): ...this, and add an assertion.
2899 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2900 (warning_at_rich_loc): Rename to...
2901 (warning_at): ...this, and add an assertion.
2902 (warning_at_rich_loc_n): Rename to...
2903 (warning_n): ...this, and add an assertion.
2904 (warning_n): Update location_t-based implementation for removal of
2905 location_t-based diagnostic_n_impl.
2906 (pedwarn_at_rich_loc): Rename to...
2907 (pedwarn): ...this, and add an assertion.
2908 (permerror_at_rich_loc): Rename to...
2909 (permerror): ...this, and add an assertion.
2910 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2911 (error_at_rich_loc): Rename to...
2912 (error_at): ...this, and add an assertion.
2913 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2914 (driver::do_spec_on_infiles): Likewise.
2915 * substring-locations.c (format_warning_va): Update for renaming
2916 of inform_at_rich_loc.
2918 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2920 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2921 _Float<N>X built-in functions so that the variant without the
2922 "__builtin_" prefix is only enabled for the GNU C and Objective C
2923 languages when they are in non-strict ANSI/ISO mode.
2924 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2925 * target.def (floatn_builtin_p): Add a target hook to control
2926 whether _Float<N> and _Float<N>X built-in functions without the
2927 "__builtin_" prefix are enabled, and return true for C and
2928 Objective C in the default hook. Include langhooks.h in
2930 * targhooks.h (default_floatn_builtin_p): Likewise.
2931 * targhooks.c (default_floatn_builtin_p): Likewise.
2932 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2933 floatn_builtin_p target hook.
2934 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2936 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2937 Eric Botcazou <ebotcazou@adacore.com>
2939 PR rtl-optimization/81803
2940 * lra-constraints.c (curr_insn_transform): Also reload the whole
2941 register for a strict subreg no wider than a word if this is for
2942 a WORD_REGISTER_OPERATIONS target.
2944 2017-10-31 Jason Merrill <jason@redhat.com>
2946 * gdbinit.in: Skip over inlines from timevar.h.
2948 2017-10-31 Martin Liska <mliska@suse.cz>
2950 * doc/gcov.texi: Document new option.
2951 * gcov.c (print_usage): Likewise print it.
2952 (process_args): Support the argument.
2953 (format_count): New function.
2954 (format_gcov): Use the function.
2956 2017-10-31 Martin Liska <mliska@suse.cz>
2958 * gcov.c (struct name_map): do not use typedef.
2959 Define operator== and operator<.
2960 (name_search): Remove.
2961 (name_sort): Remove.
2962 (main): Do not allocate names.
2963 (process_file): Add vertical space.
2964 (generate_results): Use std::find.
2965 (release_structures): Do not release memory.
2966 (find_source): Use std::find.
2968 2017-10-31 Martin Liska <mliska@suse.cz>
2970 * gcov.c (struct line_info): Remove it's typedef.
2971 (line_info::line_info): Add proper ctor.
2972 (line_info::has_block): Do not use a typedef.
2973 (struct source_info): Do not use typedef.
2974 (circuit): Likewise.
2975 (get_cycles_count): Likewise.
2976 (output_intermediate_file): Iterate via vector iterator.
2977 (add_line_counts): Use std::vector methods.
2978 (accumulate_line_counts): Likewise.
2979 (output_lines): Likewise.
2981 2017-10-31 Martin Liska <mliska@suse.cz>
2983 * gcov.c (struct source_info): Remove typedef.
2984 (source_info::source_info): Add proper ctor.
2985 (accumulate_line_counts): Use struct, not it's typedef.
2986 (output_gcov_file): Likewise.
2987 (output_lines): Likewise.
2988 (main): Do not allocate an array.
2989 (output_intermediate_file): Use size of vector container.
2990 (process_file): Resize the vector.
2991 (generate_results): Do not preallocate, use newly added vector
2993 (release_structures): Do not release sources.
2994 (find_source): Use vector methods.
2995 (add_line_counts): Do not use typedef.
2997 2017-10-31 Martin Liska <mliska@suse.cz>
2999 * doc/gcov.texi: Document that.
3000 * gcov.c (add_line_counts): Mark lines with a non-executed
3002 (output_line_beginning): Handle such lines.
3003 (output_lines): Pass new argument.
3004 (output_intermediate_file): Print it in intermediate format.
3006 2017-10-31 Martin Liska <mliska@suse.cz>
3008 * color-macros.h: New file.
3009 * diagnostic-color.c: Factor out color related to macros to
3011 * doc/gcov.texi: Document -k option.
3012 * gcov.c (INCLUDE_STRING): Include string.h.
3013 (print_usage): Add -k option.
3014 (process_args): Parse it.
3015 (pad_count_string): New function.
3016 (output_line_beginning): Likewise.
3017 (DEFAULT_LINE_START): New macro.
3018 (output_lines): Support color output.
3020 2017-10-31 Martin Liska <mliska@suse.cz>
3022 PR gcov-profile/82633
3023 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3024 their interaction with GCOV infrastructure.
3025 * configure.ac: Add -fkeep-{inline,static}-functions to
3027 * configure: Regenerate.
3029 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3032 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3034 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3037 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3038 into a register if it does not fit into an immediate offset field.
3040 2017-10-31 Olivier Hainque <hainque@adacore.com>
3042 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3044 2017-10-31 Julia Koval <julia.koval@intel.com>
3046 * config.gcc: Add gfniintrin.h.
3047 * config/i386/gfniintrin.h: New.
3048 * config/i386/i386-builtin-types.def
3049 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3050 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3051 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3052 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3053 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3054 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3055 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3056 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3057 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3058 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3059 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3060 * config/i386/immintrin.h: Include gfniintrin.h.
3061 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3063 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3065 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3067 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3069 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3070 (arm_ashldi3_1bit): Remove pattern.
3071 (ashrdi3): Remove shift by 1 expansion.
3072 (arm_ashrdi3_1bit): Remove pattern.
3073 (lshrdi3): Remove shift by 1 expansion.
3074 (arm_lshrdi3_1bit): Remove pattern.
3075 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3076 cost of ashldi3 by 1.
3077 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3078 (<shift>di3_neon): Likewise.
3080 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3082 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3083 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3084 and (*aarch64_simd_mov<VQ:mode>).
3085 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3086 pattern alternative.
3087 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3088 attributes to match pattern alternative.
3090 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3092 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3093 * config/rs6000/emmintrin.h: New file.
3094 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3096 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3098 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3099 * (movdi_vfp_cortexa8): Remove pattern.
3101 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3103 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3105 (Specific, alpha*-dec-osf5.1): Remove.
3106 (Specific, mips-sgi-irix5): Remove.
3107 (Specific, mips-sgi-irix6): Remove.
3109 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3112 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3113 arguments to clear_bit_region_be.
3115 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3117 * gimplify.c: Include memmodel.h.
3119 2017-10-30 Martin Jambor <mjambor@suse.cz>
3121 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3122 condition whether loop should be executed at all.
3124 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3126 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3127 gimple folding of vec_madd() intrinsics.
3128 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3129 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3130 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3132 2017-10-30 Richard Biener <rguenther@suse.de>
3134 PR tree-optimization/82762
3136 2017-10-23 Richard Biener <rguenther@suse.de>
3138 PR tree-optimization/82129
3140 2017-08-01 Richard Biener <rguenther@suse.de>
3142 PR tree-optimization/81181
3143 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3144 (compute_antic): ... end of iteration here.
3146 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3148 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3149 -std=iso9899:2017 and -std=gnu17.
3150 * doc/standards.texi (C Language): Document C17 support.
3151 * doc/cpp.texi (Overview): Mention -std=c17.
3152 (Standard Predefined Macros): Document C11 and C17 values of
3153 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3154 * doc/extend.texi (Inline): Do not list individual options for
3155 standards newer than C99.
3156 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3158 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3161 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3163 * asan.c (asan_finish_file): Align asan globals array by shadow
3166 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3169 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3170 (struct store_immediate_info): Add bitregion_start and bitregion_end
3172 (store_immediate_info::store_immediate_info): Add brs and bre
3173 arguments and initialize bitregion_{start,end} from those.
3174 (struct merged_store_group): Add bitregion_start, bitregion_end,
3175 align_base and mask fields. Drop unnecessary struct keyword from
3176 struct store_immediate_info. Add do_merge method.
3177 (clear_bit_region_be): Use memset instead of loop storing zeros.
3178 (merged_store_group::do_merge): New method.
3179 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3180 stores as long as the surrounding bitregions have no gaps.
3181 (merged_store_group::merge_overlapping): Use do_merge.
3182 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3183 is byte aligned, rather than requiring that start and width are
3184 byte aligned. Drop unnecessary struct keyword from
3185 struct store_immediate_info. Allocate and populate also mask array.
3186 Make start of the arrays relative to bitregion_start rather than
3187 start and size them according to bitregion_{end,start} difference.
3188 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3189 struct store_immediate_info.
3190 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3191 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3192 struct keyword from struct store_immediate_info.
3193 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3194 between stores as long as the surrounding bitregions have no gaps.
3196 (struct split_store): Add orig non-static data member.
3197 (split_store::split_store): Initialize orig to false.
3198 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3199 if there is exactly a single original stmt. Change stmts argument
3200 to pointer from reference, if NULL, don't push anything to it. Add
3201 first argument, use it to optimize skipping over orig stmts that
3202 are known to be before bitpos already. Simplify.
3203 (split_group): Return unsigned int count how many stores are or
3204 would be needed rather than a bool. Add allow_unaligned argument.
3205 Change split_stores argument from reference to pointer, if NULL,
3206 only do a dry run computing how many stores would be produced.
3207 Rewritten algorithm to use both alignment and misalign if
3208 !allow_unaligned and handle bitfield stores with gaps.
3209 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3210 from bitregion_start instead of start. Compute allow_unaligned
3211 here, if true, do 2 split_group dry runs to compute which one
3212 produces fewer stores and prefer aligned if equal. Punt if
3213 new count is bigger or equal than original before emitting any
3214 statements, rather than during that. Remove no longer needed
3215 new_ssa_names tracking. Replace num_stmts with
3216 split_stores.length (). Use 32-bit stack allocated entries
3217 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3218 if possible. Handle bitfields with gaps.
3219 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3220 Compute bitregion_{start,end} for the stores and construct
3221 store_immediate_info with that. Formatting fixes.
3223 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3226 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3227 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3229 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3231 * gimplify.c: Include tm_p.h.
3233 * common.opt (gcoff): Re-add as ignored option.
3234 (gcoff1, gcoff2, gcoff3): Likewise.
3236 * Makefile.in (OBJS): Delete sdbout.o.
3237 (GTFILES): Delete $(srcdir)/sdbout.c.
3238 * debug.h: Delete sdb_debug_hooks.
3239 * final.c: Delete sdbout.h include.
3240 (final_scan_insn): Delete SDB_DEBUG check.
3241 (rest_of_clean_state): Likewise.
3242 * output.h: Delete sdb_begin_function_line.
3245 * toplev.c: Delete sdbout.h include.
3246 (process_options): Delete SDB_DEBUG check.
3247 * tree-core.h (tree_type_common): Delete pointer field of
3249 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3250 TYPE_SYMTAB_POINTER.
3251 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3252 (TYPE_SYMTAB_IS_POINTER): Delete.
3253 (TYPE_SYMTAB_IS_DIE): Renumber.
3254 * xcoffout.c: Refer to former sdbout.c file.
3255 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3257 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3258 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3259 (Debugging Options): Delete -gcoff.
3260 (-gstabs): Delete SDB reference.
3262 (-gcoff@var{level}): Delete.
3263 * doc/passes.texi (Debugging information output): Delete SDB and
3264 sdbout.c references.
3265 * doc/tm.texi: Regenerate.
3266 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3267 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3269 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3270 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3271 (SDB_DEBUGGING_INFO): Delete.
3272 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3273 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3274 * target.def (output_source_filename): Delete COFF reference.
3276 * common.opt (gcoff): Delete.
3277 (gxcoff+): Update Negative chain.
3278 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3280 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3281 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3282 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3284 (expand_function_start): Change sdb reference to past tense.
3285 (expand_function_end): Change sdb reference to past tense.
3286 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3287 * opts.c (debug_type_names): Delete coff entry.
3288 (common_handle_option): Delete OPT_gcoff case.
3289 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3291 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3292 * config/cris/cris.h: Delete SDB reference in comment.
3293 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3294 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3295 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3296 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3298 (ix86_expand_prologue): Likewise.
3299 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3300 * config/ia64/ia64.h: Likewise.
3301 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3302 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3304 * config/mmix/mmix.h: Likewise.
3305 * config/nds32/nds32.c: Likewise.
3306 * config/stormy/storym16.h: Likewise.
3307 * config/visium/visium.h: Likewise.
3308 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3310 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3312 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3313 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3314 FRAME_POINTER_REGNUM point at high end of local var area.
3316 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3318 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3319 Move comment around. Do not reset best_edge for a copiable
3320 destination if the copy would cause a partition change.
3321 (better_edge_p): Remove redundant check.
3323 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3325 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3327 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3329 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3330 for math functions that have _Float<N> and _Float<N>X variants.
3331 (mathfn_built_in_2): Add support for math functions that have
3332 _Float<N> and _Float<N>X variants.
3333 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3334 (expand_builtin_mathfn_ternary): Add support for fma with
3335 _Float<N> and _Float<N>X variants.
3336 (expand_builtin): Likewise.
3337 (fold_builtin_3): Likewise.
3338 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3339 create math function _Float<N> and _Float<N>X variants as external
3341 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3342 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3343 the __builtin_ prefix and if not strict ansi, without the prefix.
3344 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3345 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3346 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3347 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3348 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3349 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3350 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3351 function signatures for fma _Float<N> and _Float<N>X variants.
3352 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3353 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3354 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3355 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3356 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3357 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3358 * gencfn-macros.c (print_case_cfn): Add support for math functions
3359 that have _Float<N> and _Float<N>X variants.
3360 (print_define_operator_list): Likewise.
3361 (fltfn_suffixes): Likewise.
3363 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3364 for math functions that have _Float<N> and _Float<N>X variants.
3365 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3366 and _Float<N>X variants.
3367 (COPYSIGN): Likewise.
3370 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3371 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3373 (integer_valued_read_call_p): Likewise.
3374 * fold-const-call.c (fold_const_call_ss): Likewise.
3375 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3376 _Float<N> and _Float<N>X variants.
3377 (fold_const_call_ssss): Add support for fma _Float<N> and
3378 _Float<N>X variants.
3379 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3380 support for copysign and fma _Float<N> and _Float<N>X variants.
3381 (backprop::process_builtin_call_use): Likewise.
3382 * tree-call-cdce.c (can_test_argument_range); Add support for
3383 sqrt _Float<N> and _Float<N>X variants.
3384 (edom_only_function): Likewise.
3385 (get_no_error_domain): Likewise.
3386 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3387 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3388 copysign _Float<N> and _Float<N>X variants.
3389 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3390 handled by machine independent code.
3391 (FMAF128): Likewise.
3392 * doc/cpp.texi (Common Predefined Macros): Document defining
3393 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3394 fma _Float<N> and _Float<N>X variants.
3396 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3399 * config/i386/i386-modes.def (CCFPU): Remove definition.
3400 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3401 (ix86_cc_modes_compatible): Ditto.
3402 (ix86_expand_carry_flag_compare): Ditto.
3403 (ix86_expand_int_movcc): Ditto.
3404 (ix86_expand_int_addcc): Ditto.
3405 (ix86_reverse_condition): Ditto.
3406 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3407 Return true/false for unordered/ordered fp comparisons.
3408 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3409 (ix86_prepare_fp_compare_args): Update for rename.
3410 (ix86_expand_fp_compare): Update for rename. Generate unordered
3411 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3412 (ix86_expand_sse_compare_and_jump): Ditto.
3413 * config/i386/predicates.md (fcmov_comparison_operator):
3414 Remove CCFPU mode handling.
3415 (ix86_comparison_operator): Ditto.
3416 (ix86_carry_flag_operator): Ditto.
3417 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3418 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3419 (*cmpu<mode>_cc_i387): Ditto.
3420 (FPCMP): Remove mode iterator.
3421 (unord): Remove mode attribute.
3422 (unord_subst): New define_subst transformation
3423 (unord): New define_subst attribute.
3425 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3426 (*cmpi<unord>xf_i387): Ditto.
3427 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3428 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3429 using unord_subst transformation.
3430 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3431 (round_saeonly): Also handle CCFP mode.
3432 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3433 Remove UNSPEC_SAHF unspec handling.
3435 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3437 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3439 2017-10-27 Jeff Law <law@redhat.com>
3441 * gimple-ssa-sprintf.c: Include domwalk.h.
3442 (class sprintf_dom_walker): New class, derived from dom_walker.
3443 (sprintf_dom_walker::before_dom_children): New function.
3444 (struct call_info): Moved into sprintf_dom_walker class
3445 (compute_formath_length, handle_gimple_call): Likewise.
3446 (sprintf_length::execute): Call the dominator walker rather
3447 than walking the statements.
3449 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3450 gimple statement locations.
3451 (check_array_bounds): Corresponding changes. Get the statement's
3452 location directly from wi->stmt.
3454 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3457 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3459 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3461 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3462 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3464 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3467 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3468 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3469 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3470 maybe_get_pool_constant.
3471 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3474 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3476 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3477 2.26 caveat. Update gas and gld versions.
3478 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3481 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3483 * cgraph.h (set_malloc_flag): Declare.
3484 * cgraph.c (set_malloc_flag_1): New function.
3485 (set_malloc_flag): Likewise.
3486 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3487 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3489 (read_ipa_call_summary): Add support for reading is_return_callee.
3490 (write_ipa_call_summary): Stream is_return_callee.
3491 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3492 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3493 ipa-prop.h, ipa-fnsummary.h.
3494 (pure_const_names): Change to static.
3495 (malloc_state_e): Define.
3496 (malloc_state_names): Define.
3497 (funct_state_d): Add field malloc_state.
3498 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3499 (check_retval_uses): New function.
3500 (malloc_candidate_p): Likewise.
3501 (analyze_function): Add support for malloc attribute.
3502 (pure_const_write_summary): Stream malloc_state.
3503 (pure_const_read_summary): Add support for reading malloc_state.
3504 (dump_malloc_lattice): New function.
3505 (propagate_malloc): New function.
3506 (warn_function_malloc): New function.
3507 (ipa_pure_const::execute): Call propagate_malloc and
3508 ipa_free_fn_summary.
3509 (pass_local_pure_const::execute): Add support for malloc attribute.
3510 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3511 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3513 2017-10-27 Martin Liska <mliska@suse.cz>
3515 PR gcov-profile/82457
3516 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3517 for fork-like functions to be properly instrumented.
3519 2017-10-27 Richard Biener <rguenther@suse.de>
3522 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3523 info when we redirected EH.
3525 2017-10-26 Michael Collison <michael.collison@arm.com>
3527 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3529 (<optab>_trunchf<GPI:mode>2: New pattern.
3530 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3531 * config/aarch64/iterators.md (wv): New mode attribute.
3532 (vf, VF): New mode attributes.
3533 (vgp, VGP): New mode attributes.
3534 (s): Update attribute with SImode and DImode prefixes.
3536 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3538 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3539 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3540 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3541 (nios2_option_overide): Initialize it. Don't allow R0-relative
3542 addressing with PIC.
3543 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3544 (nios2_symbolic_constant_p): Likewise.
3545 (nios2_legitimate_address_p): Likewise.
3546 (nios2_r0rel_section_name_p): New.
3547 (nios2_symbol_ref_in_r0rel_data_p): New.
3548 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3549 (r0rel_constant_p): New.
3550 (nios2_print_operand_address): Handle r0rel_constant_p.
3551 (nios2_cdx_narrow_form_p): Likewise.
3552 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3553 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3554 (Nios II Options): Document -mr0rel-sec.
3556 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3558 * config/nios2/nios2.c: Include xregex.h.
3559 (nios2_gprel_sec_regex): New.
3560 (nios2_option_overide): Initialize it. Don't allow GP-relative
3561 addressing with PIC.
3562 (nios2_small_section_name_p): Check for regex match.
3563 * config/nios2/nios2.opt (mgprel-sec=): New option.
3564 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3565 (Nios II Options): Document -mgprel-sec.
3567 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3569 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3571 2017-10-26 Tom de Vries <tom@codesourcery.com>
3573 PR tree-optimization/82707
3574 * gimple.c (gimple_copy): Fix unsharing of
3575 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3577 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3579 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3580 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3581 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3582 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3583 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3584 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3585 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3586 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3587 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3588 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3589 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3590 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3591 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3592 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3593 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3594 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3595 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3597 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3599 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3601 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3602 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3603 warning to rs6000.c. Remove the Undocumented flag, since it has
3605 (-mabi=ibmlongdouble): Likewise.
3606 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3607 already set, set the default format for long double.
3608 (rs6000_debug_reg_global): Print whether long double is IBM or
3610 (rs6000_option_override_internal): Rework setting long double
3611 format. Only warn if the user is changing the long double default
3612 and they did not use -Wno-psabi.
3613 * doc/invoke.texi (PowerPC options): Update the documentation for
3614 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3616 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3617 Alan Hayward <alan.hayward@arm.com>
3618 David Sherwood <david.sherwood@arm.com>
3620 * rtl.h (wider_subreg_mode): New function.
3621 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3622 rather than an unsigned int *.
3623 * ira-color.c (regno_max_ref_width): Replace with...
3624 (regno_max_ref_mode): ...this new variable.
3625 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3626 Use wider_subreg_mode.
3627 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3628 rather than an unsigned int *.
3629 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3630 (process_alt_operands): Likewise.
3631 (invariant_p): Likewise.
3632 * lra-spills.c (assign_mem_slot): Likewise.
3633 (add_pseudo_to_slot): Likewise.
3634 * lra.c (collect_non_operand_hard_regs): Likewise.
3635 (add_regs_to_insn_regno_info): Likewise.
3636 * reload1.c (regno_max_ref_width): Replace with...
3637 (regno_max_ref_mode): ...this new variable.
3638 (reload): Update accordingly. Update call to
3639 ira_sort_regnos_for_alter_reg.
3640 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3641 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3642 (scan_paradoxical_subregs): Likewise.
3644 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3646 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3647 (aarch64_frame): Add emit_frame_chain boolean.
3648 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3649 Move eh_return case to aarch64_layout_frame.
3650 (aarch64_layout_frame): Initialize emit_frame_chain.
3651 (aarch64_expand_prologue): Use emit_frame_chain.
3653 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3655 * config/aarch64/aarch64.c (aarch64_layout_frame):
3656 Ensure LR is always stored at the bottom of the callee-saves.
3657 Remove rarely used frame layout which saves callee-saves at top of
3658 frame, so the store of LR can be used as a valid probe in all cases.
3660 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3662 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3663 Improve unaligned TImode/TFmode base/offset split.
3665 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3666 Alan Hayward <alan.hayward@arm.com>
3667 David Sherwood <david.sherwood@arm.com>
3669 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3670 * combine.c (find_single_use_1): Likewise.
3671 (expand_field_assignment): Likewise.
3672 (move_deaths): Likewise.
3673 * lra-constraints.c (simplify_operand_subreg): Likewise.
3674 (curr_insn_transform): Likewise.
3675 * lra.c (collect_non_operand_hard_regs): Likewise.
3676 (add_regs_to_insn_regno_info): Likewise.
3677 * rtlanal.c (reg_referenced_p): Likewise.
3678 (covers_regno_no_parallel_p): Likewise.
3680 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3682 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3683 Don't print any bits outside the precision of the value.
3684 * wide-int.cc (test_printing): Add some new tests.
3686 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3688 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3689 supports -xbrace_comment option.
3690 * configure: Regenerate.
3691 * config.in: Regenerate.
3692 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3693 (ASM_CPU_SPEC): Use it.
3695 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3697 * target.def (static_rtx_alignment): New hook.
3698 * targhooks.h (default_static_rtx_alignment): Declare.
3699 * targhooks.c (default_static_rtx_alignment): New function.
3700 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3701 * doc/tm.texi: Regenerate.
3702 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3703 instead of targetm.constant_alignment. Remove call to
3705 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3706 (cris_preferred_mininum_alignment): New function, split out from...
3707 (cris_constant_alignment): ...here.
3708 (cris_static_rtx_alignment): New function.
3709 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3711 (ix86_constant_alignment): ...here.
3712 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3713 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3714 (mmix_static_rtx_alignment): New function.
3715 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3716 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3718 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3721 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3722 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3724 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3726 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3727 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3728 mark as a sign-extending load.
3729 (local_pic_load_u): Define.
3731 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3734 * fold-const.c (operand_equal_for_comparison_p): Also return true
3735 if ARG0 is a simple variant of ARG1 with narrower precision.
3736 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3738 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3740 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3742 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3743 scatter_static, scatter_per_elt.
3744 * x86-tune-costs.h: Add new cost entries.
3746 2017-10-25 Richard Biener <rguenther@suse.de>
3748 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3749 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3750 class pass_fre): Move to ...
3751 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3752 class pass_fre): ... here and adjust for statistics.
3754 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3757 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3758 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3759 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3760 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3763 2017-10-25 Richard Biener <rguenther@suse.de>
3765 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3766 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3767 eliminate_push_avail, eliminate_insert): Move inside...
3768 (class eliminate_dom_walker): ... this class in preparation
3770 (fini_eliminate): Remove by merging with ...
3771 (eliminate): ... this function. Adjust for class changes.
3772 (pass_pre::execute): Remove fini_eliminate call.
3773 (pass_fre::execute): Likewise.
3775 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3778 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3779 (VPERMI2, VPERMI2I): New mode iterators.
3780 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3781 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3783 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3784 mode iterator. Remove 3 old define_insn patterns.
3785 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3786 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3787 VPERMI2 mode iterator, remove the other two expanders.
3788 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3789 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3790 instructions, remove the other two patterns.
3791 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3792 mode iterator, remove the other two patterns.
3793 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3794 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3795 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3796 and adjust argument order accordingly.
3797 (ix86_expand_vec_perm): Adjust caller.
3798 (expand_vec_perm_1): Likewise.
3799 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3800 (expand_vec_perm_vpermt2_vpshub2): ... this.
3801 (ix86_expand_vec_perm_const_1): Adjust caller.
3802 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3805 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3806 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3807 (vec_shl_<mode>): Remove unused expander.
3808 (avx512bw_<shift_insn><mode>3): New define_insn.
3809 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3810 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3812 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3815 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3818 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3820 PR rtl-optimization/82396
3821 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3822 (autopref_multipass_init): Simplify initialization.
3823 (autopref_rank_data): Simplify sort order.
3824 * gcc/sched-int.h (autopref_multipass_data_): Remove
3825 multi_mem_insn_p, min_offset and max_offset.
3827 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3830 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3831 Check special value of flag_omit_frame_pointer.
3832 (aarch64_can_eliminate): Likewise.
3833 (aarch64_override_options_after_change_1): Simplify handling of
3834 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3836 2017-10-24 Richard Biener <rguenther@suse.de>
3838 PR tree-optimization/82697
3839 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3840 zero for conditional load and unconditional store.
3842 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3844 * doc/install.texi: Document bootstrap-cet.
3846 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3849 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3850 ENDBR instruction at function entrance if function is only
3853 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3856 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3857 patterns to better describe from which operation the CF is computed.
3858 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3859 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3860 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3861 is 0, use _0 suffixed expanders instead of emitting a comparison
3864 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3866 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3867 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3869 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3872 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3873 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3874 * loop-iv.c (iv_get_reaching_def): Likewise.
3875 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3876 variable is promoted and the partition contains undefined values.
3878 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3880 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3882 (nios2_address_cost): Define.
3883 (nios2_legitimize_address): Recognize (exp + constant) directly.
3884 (TARGET_ADDRESS_COST): Define.
3886 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3888 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3889 (nios2_symbolic_memory_operand_p): Declare.
3890 (nios2_split_large_constant): Declare.
3891 (nios2_split_symbolic_memory_operand): Declare.
3892 * config/nios2/nios2.c: Adjust includes.
3893 (nios2_symbolic_constant_allowed): New.
3894 (nios2_symbolic_constant_p): New.
3895 (nios2_plus_symbolic_constant_p): New.
3896 (nios2_valid_addr_expr_p): Recognize addresses involving
3898 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3899 (nios2_symbolic_memory_operand_p): New.
3900 (nios2_large_constant_p): New.
3901 (nios2_split_large_constant): New.
3902 (nios2_split_plus_large_constant): New.
3903 (nios2_split_symbolic_memory_operand): New.
3904 (nios2_legitimize_address): Code refactoring. Handle addresses
3905 involving symbolic constants.
3906 (nios2_emit_move_sequence): Likewise.
3907 (nios2_print_operand): Improve error output.
3908 (nios2_print_operand_address): Handle LO_SUM.
3909 (nios2_cdx_narrow_form_p): Likewise.
3910 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3911 operands involving symbolic constants.
3912 (movhi_internal, movsi_internal): Likewise.
3913 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3914 (extendhisi2, extendqi<mode>2): Likewise.
3916 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3918 * tree-pass.h (PROP_rtl_split_insns): Define.
3919 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3921 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3923 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3925 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3928 * target.def (const_not_ok_for_debug_p): Default to
3929 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3930 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3931 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3932 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3933 which targetm.const_not_ok_for_debug_p returned true.
3934 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3936 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3938 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3939 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3940 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3942 (ix86_const_not_ok_for_debug_p): New function.
3943 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3944 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3946 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3949 * system.h: Conditionally include "unique-ptr.h" if
3950 INCLUDE_UNIQUE_PTR is defined.
3951 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3952 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3954 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3956 * config/rl78/rl78.md: New define_expand "subdi3".
3958 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3961 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3962 DF_REF_INSN if DF_REF_INSN_INFO is false.
3964 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3966 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3967 xmm_move instead of sse_move.
3968 (sse_store_index): New function.
3969 (ix86_register_move_cost): Be more sensible about mismatch stall;
3970 model AVX moves correctly; make difference between sse->integer and
3972 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3973 moves; make difference between SSE and AVX.
3974 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3975 and zmm_move. Increase size of sse load and store tables;
3976 add unaligned load and store tables; add ssemmx_to_integer.
3977 * x86-tune-costs.h: Update all entries according to real
3978 move latencies from Agner Fog's manual and chip documentation.
3980 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3983 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3984 * config/i386/constraints.md (Wf): New constraint.
3985 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3986 (cmp<dwi>_doubleword): Removed.
3987 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3988 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3989 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3990 expand with cmp<dwi>_doubleword. For LTU and GEU use
3991 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3993 * common.opt (gcolumn-info): Enable by default.
3994 * doc/invoke.texi (gcolumn-info): Document new default.
3996 2017-10-23 Richard Biener <rguenther@suse.de>
3998 PR tree-optimization/82672
3999 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4000 Fold the stmt if we propagated into it.
4002 2017-10-23 Richard Biener <rguenther@suse.de>
4004 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4005 (bitmap_remove_expr_from_set): ... this. All callers call this
4006 for non-constant values.
4007 (bitmap_set_subtract): Rename to...
4008 (bitmap_set_subtract_expressions): ... this. Adjust and
4010 (bitmap_set_contains_value): Remove superfluous check.
4011 (bitmap_set_replace_value): Inline into single caller ...
4012 (bitmap_value_replace_in_set): ... here and simplify.
4013 (dependent_clean): Merge into ...
4014 (clean): ... this using an overload. Adjust.
4015 (prune_clobbered_mems): Adjust.
4016 (compute_antic_aux): Likewise.
4017 (compute_partial_antic_aux): Likewise.
4019 2017-10-23 Richard Biener <rguenther@suse.de>
4021 PR tree-optimization/82129
4023 2017-08-01 Richard Biener <rguenther@suse.de>
4025 PR tree-optimization/81181
4026 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4027 (compute_antic): ... end of iteration here.
4029 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4031 * target.def (starting_frame_offset): New hook.
4032 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4033 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4034 * doc/tm.texi.in: Regenerate.
4035 * hooks.h (hook_hwi_void_0): Declare.
4036 * hooks.c (hook_hwi_void_0): New function.
4037 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4038 STARTING_FRAME_OFFSET.
4039 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4040 * reload1.c (reload): Likewise.
4041 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4042 instead of STARTING_FRAME_OFFSET.
4043 * function.c (try_fit_stack_local): Likewise.
4044 (assign_stack_local_1): Likewise
4045 (instantiate_virtual_regs): Likewise.
4046 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4047 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4048 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4049 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4050 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4051 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4052 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4053 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4054 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4055 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4056 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4057 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4058 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4059 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4060 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4061 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4062 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4063 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4064 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4065 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4066 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4067 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4068 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4069 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4070 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4071 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4072 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4073 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4074 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4075 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4076 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4077 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4078 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4079 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4080 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4081 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4082 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4083 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4084 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4085 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4086 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4087 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4088 return a HOST_WIDE_INT.
4089 (avr_builtin_setjmp_frame_value): Use it instead of
4090 STARTING_FRAME_OFFSET.
4091 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4092 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4093 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4095 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4096 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4097 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4098 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4099 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4100 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4101 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4102 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4103 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4104 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4105 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4106 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4108 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4109 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4110 * config/mips/mips.c (mips_compute_frame_info): Refer to
4111 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4112 (mips_starting_frame_offset): New function.
4113 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4114 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4115 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4116 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4117 and return a HOST_WIDE_INT.
4118 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4119 (mmix_initial_elimination_offset): Refer to
4120 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4121 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4122 * config/pa/pa.c (pa_starting_frame_offset): New function.
4123 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4124 (pa_expand_prologue): Likewise.
4125 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4126 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4127 !FRAME_GROWS_DOWNWARD handling to...
4128 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4129 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4130 !FRAME_GROWS_DOWNWARD handling to...
4131 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4132 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4133 !FRAME_GROWS_DOWNWARD handling to...
4134 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4135 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4137 (rs6000_starting_frame_offset): New function.
4138 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4139 !FRAME_GROWS_DOWNWARD handling to...
4140 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4141 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4142 !FRAME_GROWS_DOWNWARD handling to...
4143 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4144 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4145 !FRAME_GROWS_DOWNWARD handling to...
4146 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4147 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4148 (rs6000_starting_frame_offset): New function.
4149 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4150 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4151 * config/vax/vax.c (vax_starting_frame_offset): New function.
4152 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4153 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4154 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4155 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4156 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4157 * system.h (STARTING_FRAME_OFFSET): Poison.
4159 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4161 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4162 SCALAR_TYPE_MODE instead of TYPE_MODE.
4164 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4165 Alan Hayward <alan.hayward@arm.com>
4166 David Sherwood <david.sherwood@arm.com>
4168 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4170 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4171 Alan Hayward <alan.hayward@arm.com>
4172 David Sherwood <david.sherwood@arm.com>
4174 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4176 2017-10-23 Richard Biener <rguenther@suse.de>
4178 PR tree-optimization/82129
4179 * tree-ssa-pre.c (bitmap_set_and): Remove.
4180 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4181 canonicalizing expressions in the set to those with lowest
4182 ID rather than taking that from the first edge.
4184 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4186 * combine.c (rtx_equal_for_field_assignment_p): Use
4187 byte_lowpart_offset.
4189 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4190 Alan Hayward <alan.hayward@arm.com>
4191 David Sherwood <david.sherwood@arm.com>
4193 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4194 to a SUBREG_PROMOTED_VAR.
4196 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4197 Alan Hayward <alan.hayward@arm.com>
4198 David Sherwood <david.sherwood@arm.com>
4200 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4201 (expand_debug_source_expr): Likewise.
4202 * combine.c (combine_simplify_rtx): Likewise.
4203 * cse.c (fold_rtx): Likewise.
4204 * optabs.c (expand_float): Likewise.
4205 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4206 (simplify_binary_operation_1): Likewise.
4208 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4209 Alan Hayward <alan.hayward@arm.com>
4210 David Sherwood <david.sherwood@arm.com>
4212 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4213 (record_promoted_value): Likewise.
4214 * expr.c (expand_expr_real_2): Likewise.
4215 * ree.c (update_reg_equal_equiv_notes): Likewise.
4216 (combine_set_extension): Likewise.
4217 * rtlanal.c (low_bitmask_len): Likewise.
4218 * simplify-rtx.c (neg_const_int): Likewise.
4219 (simplify_binary_operation_1): Likewise.
4221 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4222 Alan Hayward <alan.hayward@arm.com>
4223 David Sherwood <david.sherwood@arm.com>
4225 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4226 * regcprop.c (maybe_mode_change): Likewise.
4227 * reload1.c (alter_reg): Likewise.
4229 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4231 * inchash.h (inchash::hash::add_wide_int): New function.
4232 * lto-streamer-out.c (hash_tree): Use it.
4234 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4236 * inchash.h (inchash::hash::add_wide_int): Rename to...
4237 (inchash::hash::add_hwi): ...this.
4238 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4239 (polymorphic_call_target_hasher::hash): Likewise.
4240 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4241 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4242 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4243 * lto-streamer-out.c (hash_tree): Likewise.
4244 * optc-save-gen.awk: Likewise.
4245 * tree.c (add_expr): Likewise.
4247 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4250 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4251 for ordered inequality comparisons even with TARGET_IEEE_FP.
4253 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4256 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4257 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4258 Expand with cmp<dwi>_doubleword.
4260 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4262 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4263 List CET intrinsics.
4264 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4265 specific to -fcf-protection option.
4267 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4269 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4270 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4271 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4272 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4273 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4274 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4275 (extra_objs): Add cet.o for Linux/x86 targets.
4276 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4277 * config/i386/cet.c: New file.
4278 * config/i386/cetintrin.h: Likewise.
4279 * config/i386/t-cet: Likewise.
4280 * config/i386/cpuid.h (bit_SHSTK): New.
4281 (bit_IBT): Likewise.
4282 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4283 pass IBT and SHSTK bits.
4284 * config/i386/i386-builtin-types.def
4285 (VOID_FTYPE_UNSIGNED_PVOID): New.
4286 (VOID_FTYPE_UINT64_PVOID): Likewise.
4287 * config/i386/i386-builtin.def: Add CET intrinsics.
4288 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4289 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4290 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4291 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4293 * config/i386/i386.c (rest_of_insert_endbranch): New.
4294 (pass_data_insert_endbranch): Likewise.
4295 (pass_insert_endbranch): Likewise.
4296 (make_pass_insert_endbranch): Likewise.
4297 (ix86_notrack_prefixed_insn_p): Likewise.
4298 (ix86_target_string): Add -mibt, -mshstk flags.
4299 (ix86_option_override_internal): Add flag_cf_protection
4301 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4302 (ix86_print_operand): Add 'notrack' prefix output.
4303 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4304 (ix86_expand_builtin): Expand CET intrinsics.
4305 (x86_output_mi_thunk): Add 'endbranch' instruction.
4306 * config/i386/i386.h (TARGET_IBT): New.
4307 (TARGET_IBT_P): Likewise.
4308 (TARGET_SHSTK): Likewise.
4309 (TARGET_SHSTK_P): Likewise.
4310 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4311 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4312 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4313 (builtin_setjmp_setup): New pattern.
4314 (builtin_longjmp): Likewise.
4315 (rdssp<mode>): Likewise.
4316 (incssp<mode>): Likewise.
4317 (saveprevssp): Likewise.
4318 (rstorssp): Likewise.
4319 (wrss<mode>): Likewise.
4320 (wruss<mode>): Likewise.
4321 (setssbsy): Likewise.
4322 (clrssbsy): Likewise.
4323 (nop_endbr): Likewise.
4324 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4326 * config/i386/immintrin.h: Include <cetintrin.h>.
4327 * config/i386/linux-common.h
4328 (file_end_indicate_exec_stack_and_cet): New prototype.
4329 (TARGET_ASM_FILE_END): New.
4331 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4333 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4334 latencies instead of having separate table; make difference between
4335 integer and float costs.
4336 * i386.h (processor_costs): Remove scalar_stmt_cost,
4337 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4338 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4340 * x86-tune-costs.h: Remove entries which has been removed in
4341 procesor_costs from all tables; make cond_taken_branch_cost
4342 and cond_not_taken_branch_cost COST_N_INSNS based.
4344 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4346 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4348 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4350 * config/i386/i386.md (isa): Remove fma_avx512f.
4351 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4352 <avx512>_fmadd_<mode>_mask3<round_name>,
4353 <avx512>_fmsub_<mode>_mask<round_name>,
4354 <avx512>_fmsub_<mode>_mask3<round_name>,
4355 <avx512>_fnmadd_<mode>_mask<round_name>,
4356 <avx512>_fnmadd_<mode>_mask3<round_name>,
4357 <avx512>_fnmsub_<mode>_mask<round_name>,
4358 <avx512>_fnmsub_<mode>_mask3<round_name>,
4359 <avx512>_fmaddsub_<mode>_mask<round_name>,
4360 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4361 <avx512>_fmsubadd_<mode>_mask<round_name>,
4362 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4363 (*vec_widen_umult_even_v16si<mask_name>,
4364 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4365 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4367 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4369 * extend.texi: Add 'nocf_check' documentation.
4370 * gimple.texi: Add second parameter to
4371 gimple_build_call_from_tree.
4372 * invoke.texi: Add -fcf-protection documentation.
4373 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4375 2017-10-20 Richard Biener <rguenther@suse.de>
4377 PR tree-optimization/82473
4378 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4379 the largest input type.
4381 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4383 * c-attribs.c (handle_nocf_check_attribute): New function.
4384 (c_common_attribute_table): Add 'nocf_check' handling.
4385 * gimple-parser.c: Add second argument NULL to
4386 gimple_build_call_from_tree.
4387 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4388 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4390 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4391 * common.opt: Add fcf-protection flag.
4392 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4393 * flag-types.h: Add enum cf_protection_level.
4394 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4395 Add 'nocf_check' attribute propagation to gimple call.
4396 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4397 (gimple_build_call_from_tree): Update prototype.
4398 (gimple_call_nocf_check_p): New function.
4399 (gimple_call_set_nocf_check): Likewise.
4400 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4401 * ipa-icf.c: Add nocf_check attribute in statement hash.
4402 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4403 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4404 * toplev.c (process_options): Add flag_cf_protection handling.
4406 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4408 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4410 2017-10-20 Richard Biener <rguenther@suse.de>
4412 PR tree-optimization/82603
4413 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4414 remove false predicated stores.
4416 2017-10-20 Richard Biener <rguenther@suse.de>
4418 * graphite-isl-ast-to-gimple.c
4419 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4420 Remove return value and simplify, dump copied stmt after lhs
4422 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4423 Reduce dump verbosity.
4424 (gsi_insert_earliest): Likewise.
4425 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4426 * graphite.c (print_global_statistics): Adjust dumping.
4427 (print_graphite_scop_statistics): Likewise.
4428 (print_graphite_statistics): Do not dump loops here.
4429 (graphite_transform_loops): But here.
4431 2017-10-20 Nicolas Roche <roche@adacore.com>
4433 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4434 * configure: Regenerate.
4436 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4439 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4440 functions when optimizing replace GIMPLE_RETURN stmts with
4441 calls to __builtin_unreachable ().
4444 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4445 for -fsanitize=thread link of executables.
4446 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4447 link of executables.
4450 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4452 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4453 define_insns for logical vector shifts to use VI248_AVX512BW
4454 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4455 condition, useless isa and prefix attributes. Change the first
4456 2 of these define_insns to ...
4457 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4458 define_insn for avx512vl.
4459 (<shift_insn><mode>3): ... and this, new define_insn without
4460 masking for non-avx512vl.
4463 * config/i386/sse.md (*andnot<mode>3,
4464 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4465 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4466 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4467 not applied use empty suffix even for TARGET_AVX512VL.
4468 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4469 is applied, supply evex,evex or evex,evex,evex instead of just
4472 2017-10-20 Julia Koval <julia.koval@intel.com>
4474 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4475 (OPTION_MASK_ISA_GFNI_UNSET): New.
4476 (ix86_handle_option): Handle OPT_mgfni.
4477 * config/i386/cpuid.h (bit_GFNI): New.
4478 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4479 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4480 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4481 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4482 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4483 * config/i386/i386.opt: Add mgfni.
4485 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4487 * config/msp430/msp430.c (msp430_option_override): Disable
4488 -fdelete-null-pointer-checks.
4489 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4491 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4493 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4494 of x87 and SSE instructions.
4496 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4498 * asan.c (create_cond_insert_point): Do not update edge count.
4499 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4500 (afdo_propagate_circuit): Likewise.
4501 (afdo_calculate_branch_prob): Likewise.
4502 (afdo_annotate_cfg): Likewise.
4503 * basic-block.h (struct edge_def): Remove count.
4504 (edge_def::count): New accessor.
4505 * bb-reorder.c (rotate_loop): Update.
4506 (find_traces_1_round): Update.
4507 (connect_traces): Update.
4508 (sanitize_hot_paths): Update.
4509 * cfg.c (unchecked_make_edge): Update.
4510 (make_single_succ_edge): Update.
4511 (check_bb_profile): Update.
4512 (dump_edge_info): Update.
4513 (update_bb_profile_for_threading): Update.
4514 (scale_bbs_frequencies_int): Update.
4515 (scale_bbs_frequencies_gcov_type): Update.
4516 (scale_bbs_frequencies_profile_count): Update.
4517 (scale_bbs_frequencies): Update.
4518 * cfganal.c (connect_infinite_loops_to_exit): Update.
4519 * cfgbuild.c (compute_outgoing_frequencies): Update.
4520 (find_many_sub_basic_blocks): Update.
4521 * cfgcleanup.c (try_forward_edges): Update.
4522 (try_crossjump_to_edge): Update
4523 * cfgexpand.c (expand_gimple_cond): Update
4524 (expand_gimple_tailcall): Update
4525 (construct_exit_block): Update
4526 * cfghooks.c (verify_flow_info): Update
4527 (redirect_edge_succ_nodup): Update
4528 (split_edge): Update
4529 (make_forwarder_block): Update
4530 (duplicate_block): Update
4531 (account_profile_record): Update
4532 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4533 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4534 * cfgloopmanip.c (scale_loop_profile): Update.
4536 (lv_adjust_loop_entry_edge): Update.
4537 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4538 (force_nonfallthru_and_redirect): Update.
4539 (purge_dead_edges): Update.
4540 (rtl_flow_call_edges_add): Update.
4541 * cgraphunit.c (init_lowered_empty_function): Update.
4542 (cgraph_node::expand_thunk): Update.
4543 * gimple-pretty-print.c (dump_probability): Update.
4544 (dump_edge_probability): Update.
4545 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4546 * haifa-sched.c (sched_create_recovery_edges): Update.
4547 * hsa-gen.c (convert_switch_statements): Update.
4548 * ifcvt.c (dead_or_predicable): Update.
4549 * ipa-inline-transform.c (inline_transform): Update.
4550 * ipa-split.c (split_function): Update.
4551 * ipa-utils.c (ipa_merge_profiles): Update.
4552 * loop-doloop.c (add_test): Update.
4553 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4554 * lto-streamer-in.c (input_cfg): Update.
4555 (input_function): Update.
4556 * lto-streamer-out.c (output_cfg): Update.
4557 * modulo-sched.c (sms_schedule): Update.
4558 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4559 * predict.c (maybe_hot_edge_p): Update.
4560 (unlikely_executed_edge_p): Update.
4561 (probably_never_executed_edge_p): Update.
4562 (dump_prediction): Update.
4563 (drop_profile): Update.
4564 (propagate_unlikely_bbs_forward): Update.
4565 (determine_unlikely_bbs): Update.
4566 (force_edge_cold): Update.
4567 * profile.c (compute_branch_probabilities): Update.
4568 * reg-stack.c (better_edge): Update.
4569 * shrink-wrap.c (handle_simple_exit): Update.
4570 * tracer.c (better_p): Update.
4571 * trans-mem.c (expand_transaction): Update.
4572 (split_bb_make_tm_edge): Update.
4573 * tree-call-cdce.c: Update.
4574 * tree-cfg.c (gimple_find_sub_bbs): Update.
4575 (gimple_split_edge): Update.
4576 (gimple_duplicate_sese_region): Update.
4577 (gimple_duplicate_sese_tail): Update.
4578 (gimple_flow_call_edges_add): Update.
4579 (insert_cond_bb): Update.
4580 (execute_fixup_cfg): Update.
4581 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4582 * tree-complex.c (expand_complex_div_wide): Update.
4583 * tree-eh.c (lower_resx): Update.
4584 (unsplit_eh): Update.
4585 (cleanup_empty_eh_move_lp): Update.
4586 * tree-inline.c (copy_edges_for_bb): Update.
4587 (freqs_to_counts): Update.
4588 (copy_cfg_body): Update.
4589 * tree-ssa-dce.c (remove_dead_stmt): Update.
4590 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4591 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4592 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4593 (unloop_loops): Update.
4594 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4595 * tree-ssa-loop-split.c (connect_loops): Update.
4596 (split_loop): Update.
4597 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4598 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4599 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4600 * tree-ssa-reassoc.c (branch_fixup): Update.
4601 * tree-ssa-tail-merge.c (replace_block_by): Update.
4602 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4603 (compute_path_counts): Update.
4604 (update_profile): Update.
4605 (recompute_probabilities): Update.
4606 (update_joiner_offpath_counts): Update.
4607 (estimated_freqs_path): Update.
4608 (freqs_to_counts_path): Update.
4609 (clear_counts_path): Update.
4610 (ssa_fix_duplicate_block_edges): Update.
4611 (duplicate_thread_path): Update.
4612 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4613 (case_bit_test_cmp): Update.
4614 (collect_switch_conv_info): Update.
4615 (gen_inbound_check): Update.
4616 (do_jump_if_equal): Update.
4617 (emit_cmp_and_jump_insns): Update.
4618 * tree-tailcall.c (decrease_profile): Update.
4619 (eliminate_tail_call): Update.
4620 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4621 (vect_do_peeling): Update.
4622 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4623 * ubsan.c (ubsan_expand_null_ifn): Update.
4624 (ubsan_expand_ptr_ifn): Update.
4625 * value-prof.c (gimple_divmod_fixed_value): Update.
4626 (gimple_mod_pow2): Update.
4627 (gimple_mod_subtract): Update.
4628 (gimple_ic): Update.
4629 (gimple_stringop_fixed_value): Update.
4631 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4634 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4636 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4638 PR rtl-optimization/82395
4639 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4640 based on non_spilled_static_chain_regno_p.
4642 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4644 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4645 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4646 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4648 2017-10-19 Martin Sebor <msebor@redhat.com>
4650 PR tree-optimization/82596
4651 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4653 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4655 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4656 (get_mem_refs_of_builtin_call): Likewise.
4657 * builtins.c (expand_builtin_apply): Adjust call to
4658 allocate_dynamic_stack_space.
4659 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4660 the third argument to allocate_dynamic_stack_space, otherwise -1.
4661 (expand_builtin): Deal with all alloca variants.
4662 (is_inexpensive_builtin): Likewise.
4663 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4664 * calls.c (special_function_p): Deal with all alloca variants.
4665 (initialize_argument_information): Adjust call to
4666 allocate_dynamic_stack_space.
4667 (expand_call): Likewise.
4668 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4669 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4670 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4671 use it for the stack usage computation.
4672 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4673 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4674 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4675 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4676 (in_loop_p): Remove first argument and useless check.
4677 (pass_walloca::execute): Remove useless test and adjust call to above.
4678 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4679 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4680 (gimplify_call_expr): Deal with all alloca variants.
4681 * hsa-gen.c (gen_hsa_alloca): Likewise.
4682 (gen_hsa_insns_for_call): Likewise.
4683 * ipa-pure-const.c (special_builtin_state): Likewise.
4684 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4685 * tree-object-size.c (alloc_object_size): Likewise.
4686 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4687 (call_may_clobber_ref_p_1): Likewise.
4688 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4689 (ccp_fold_stmt): Likewise.
4690 (optimize_stack_restore): Likewise.
4691 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4692 (mark_all_reaching_defs_necessary_1): Likewise.
4693 (propagate_necessity): Likewise.
4694 (eliminate_unnecessary_stmts): Likewise.
4695 * tree.c (build_common_builtin_nodes): Build
4696 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4697 (build_alloca_call_expr): New function.
4698 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4699 (CASE_BUILT_IN_ALLOCA): Likewise.
4700 (build_alloca_call_expr): Declare.
4701 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4703 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4706 * dwarf2out.c (new_die_raw): New static inline function.
4707 (new_die): Use it to create the DIE.
4708 (add_AT_external_die_ref): Likewise.
4709 (clone_die): Likewise.
4710 (clone_as_declaration): Likewise.
4711 (dwarf2out_vms_debug_main_pointer): Likewise.
4712 (base_type_die): Likewise. Remove early return for corner cases.
4713 Do not call add_pubtype on the DIE here.
4714 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4715 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4716 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4717 native order exists for base types, attach the DIE manually and call
4718 add_pubtype on it. Do not equate a reverse order DIE to the type.
4720 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4722 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4723 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4725 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4727 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4729 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4730 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4731 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4732 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4733 * opts.c (sanitizer_opts): Add builtin.
4734 * ubsan.c (instrument_builtin): New function.
4735 (pass_ubsan::execute): Call it.
4736 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4737 * doc/invoke.texi: Document -fsanitize=builtin.
4739 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4740 builtins, store max (log2 (align), 0) into uchar field instead of
4741 align into uptr field.
4742 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4743 store uchar 0 field instead of uptr 0 field.
4744 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4745 instead of passing one address of struct with 2 locations pass
4746 two addresses of structs with 1 location each.
4747 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4748 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4749 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4750 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4751 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4752 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4753 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4754 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4756 2017-10-19 Martin Liska <mliska@suse.cz>
4759 * file-find.c (remove_prefix): Remove.
4760 * file-find.h (remove_prefix): Likewise.
4761 * gcc-ar.c: Remove smartness of lookup.
4763 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4765 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4766 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4767 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4769 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4772 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4773 (setcc + and to xor + setcc): New peephole2.
4775 2017-10-19 Tom de Vries <tom@codesourcery.com>
4777 * doc/sourcebuild.texi (Test Directives, Variants of
4778 dg-require-support): Add dg-require-stack-size.
4780 2017-10-19 Martin Liska <mliska@suse.cz>
4783 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4784 that have a large alignment.
4785 (gimplify_target_expr): Likewise.
4787 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4789 PR rtl-optimization/82602
4790 * ira.c (rtx_moveable_p): Return false for volatile asm.
4792 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4795 * config/i386/i386-modes.def (CCGZ): New CC mode.
4796 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4797 * config/i386/predicates.md (ix86_comparison_operator):
4799 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4800 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4801 with double-word subtraction.
4802 (put_condition_code): Handle CCGZmode.
4804 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4806 * wide-int.cc (debug (const wide_int &)): New.
4807 (debug (const wide_int *)): New.
4808 (debug (const widest_int &)): New.
4809 (debug (const widest_int *)): New.
4811 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4814 * lra-constraints.c (curr_insn_transform): Use non-input operand
4815 instead of output one for matched reload.
4817 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4819 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4820 (tree-ssa-loop-ivopts.h): New header file.
4821 (struct builtin_info): New fields.
4822 (classify_builtin_1): Compute and record base and offset parts for
4823 memset builtin partition by calling strip_offset.
4824 (offset_cmp, fuse_memset_builtins): New functions.
4825 (finalize_partitions): Fuse adjacent memset partitions by calling
4827 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4828 Expose the interface.
4829 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4831 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4833 PR tree-optimization/82574
4834 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4835 that data reference must be executed exactly once per iteration
4836 against the outermost loop in nest.
4837 (classify_partition): Update call to above function.
4839 2017-10-18 Richard Biener <rguenther@suse.de>
4841 PR tree-optimization/82591
4842 * graphite.c (graphite_transform_loops): Move code gen message
4844 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4845 Here. Handle scop_to_isl_ast failing.
4846 (scop_to_isl_ast): Limit the number of ISL operations.
4848 2017-10-18 Richard Biener <rguenther@suse.de>
4850 * graphite-isl-ast-to-gimple.c
4851 (translate_isl_ast_to_gimple::set_rename): Simplify.
4852 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4853 (graphite_copy_stmts_from_block): ... here.
4854 (copy_bb_and_scalar_dependences): Simplify.
4855 (add_parameters_to_ivs_params): Canonicalize.
4856 (generate_entry_out_of_ssa_copies): Simplify.
4857 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4858 by passing in ISL dimension.
4859 (parameter_index_in_region_1): Rename to ...
4860 (parameter_index_in_region): ... this.
4861 (extract_affine): Adjust assert, pass down parameter index.
4862 (add_param_constraints): Use range-info when available.
4863 (build_scop_context): Adjust.
4864 * sese.c (new_sese_info): Adjust.
4865 (free_sese_info): Likewise.
4866 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4867 Remove unused typedefs.
4868 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4870 2017-10-18 Martin Liska <mliska@suse.cz>
4872 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4874 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4876 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4877 (s390_sched_init): Do not reset s390_sched_state if we entered the
4878 current basic block via a fallthru edge and all others are unlikely.
4880 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4882 * config/s390/s390.c (NUM_SIDES): New variable.
4883 (LONGRUNNING_THRESHOLD): New variable.
4884 (LATENCY_FACTOR): New variable.
4885 (s390_sched_score): Decrease score for long-running instructions on
4887 (s390_sched_variable_issue): Perform bookkeeping for long-running
4890 2017-10-18 Richard Biener <rguenther@suse.de>
4892 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4893 Simplify with removal of the parameter rename map.
4894 (set_rename): Likewise.
4895 (should_copy_to_new_region): Likewise.
4896 (graphite_copy_stmts_from_block): Likewise.
4897 (copy_bb_and_scalar_dependences): Remove initialization of
4898 unused copied_bb_map.
4900 (copy_internal_parameters): Likewise.
4901 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4902 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4903 Use INTEGRAL_TYPE_P.
4904 (parameter_index_in_region_1): Rename to ...
4905 (assign_parameter_index_in_region): ... this. Assert we have
4906 a parameter we handle.
4907 (scan_tree_for_params): Adjust.
4908 * sese.h (parameter_rename_map_t): Remove.
4909 (struct sese_info_t): Remove unused parameter_rename_map and
4910 copied_bb_map members.
4911 * sese.c (new_sese_info): Adjust.
4912 (free_sese_info): Likewise.
4914 2017-10-18 Martin Liska <mliska@suse.cz>
4917 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4918 on an abnormal edge.
4920 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4922 * doc/invoke.texi (ffunction-sections and fdata-sections):
4925 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4927 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4928 the use statement can throw internally.
4930 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4932 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4933 any RTX present on the RHS of a SET.
4934 * compare-elim.c (try_eliminate_compare): Restore comment.
4936 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4938 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4939 may be also called on tcc_type nodes.
4940 * langhooks.c (lhd_tree_size): Likewise.
4942 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4944 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4945 format_warning_at_substring.
4946 (maybe_warn): Convert source_range * param to a location_t. Pass
4947 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4948 (format_directive): Remove code to extract source_ranges and
4949 source_range * in favor of just a location_t.
4950 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4952 * substring-locations.c (format_warning_va): Convert
4953 source_range * param to a location_t.
4954 (format_warning_at_substring): Likewise.
4955 * substring-locations.h (format_warning_va): Likewise.
4956 (format_warning_at_substring): Likewise.
4958 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4960 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4962 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4963 and scatter/gather ops.
4965 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4966 vec_gather_load and vec_scatter_store.
4967 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4968 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4969 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4970 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4971 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4972 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4974 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4976 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4977 Detect FTST instruction and handle its register pops. Only pop
4978 second operand if can_pop_second_op is true.
4979 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4980 set can_pop_second_op to false in the compare_for_stack_reg call.
4982 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4983 output_fp_compare for stack register operands.
4984 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4985 instructions here. Do not emit stack register pops here. Assert
4986 that FCOMPP pops next to top stack register. Rewrite function.
4988 2017-10-17 Nathan Sidwell <nathan@acm.org>
4991 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4992 use DECL_ASSEMBLER_NAME_RAW.
4995 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4998 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4999 Wilco Dijkstra <wilco.dijkstra@arm.com>
5001 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5002 converstion to Pmode from the buf_addr.
5004 2017-10-17 Richard Biener <rguenther@suse.de>
5006 * graphite-dependences.c (scop_get_reads_and_writes): Change
5007 output parameters to references.
5009 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5011 PR 71026/tree-optimization
5012 * fold-const.c (distribute_real_division): Removed.
5013 (fold_binary_loc): Remove calls to distribute_real_divison.
5015 2017-10-17 Richard Biener <rguenther@suse.de>
5017 * graphite-scop-detection.c
5018 (scop_detection::stmt_has_simple_data_refs_p): Always use
5019 the full nest as region.
5020 (try_generate_gimple_bb): Likewise.
5021 * sese.c (scalar_evolution_in_region): Simplify now that
5022 SCEV can handle instantiation in regions.
5023 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5024 in the non-loop part of a function if requested.
5026 2017-10-17 Richard Biener <rguenther@suse.de>
5028 PR tree-optimization/82563
5029 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5031 (graphite_regenerate_ast_isl): Call it.
5032 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5034 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5036 PR tree-optimization/82549
5037 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5038 Formatting fixes. Instead of calling make_bit_field_ref with negative
5041 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5043 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5044 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5045 _mm_maskz_reduce_ss): New.
5046 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5047 __builtin_ia32_reducess_mask): Ditto..
5048 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5049 * config/i386/sse.md (reduces<mode>): Renamed to ...
5050 (reduces<mode><mask_scalar_name>): ... this.
5051 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5053 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5054 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5056 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5058 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5059 * selftest-run-tests.c (selftest::run_tests): Call
5060 selftest::unique_ptr_tests_cc_tests.
5061 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5062 * unique-ptr-tests.cc: New file.
5064 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5067 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5069 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5070 (make_hard_regno_born, make_hard_regno_dead): Update
5071 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5073 2017-10-16 Jeff Law <law@redhat.com>
5075 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5077 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5079 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5081 2017-10-16 Olivier Hainque <hainque@adacore.com>
5083 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5084 with_cpu if we were configured for an e500v2 target cpu name.
5086 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5088 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5089 * doc/invoke.texi: Document +nodsp as a valid extension for
5092 2017-10-16 Martin Liska <mliska@suse.cz>
5094 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5095 (test_set_range): Likewise.
5096 (test_range_functions): Rename to ...
5097 (test_bit_in_range): ... this.
5098 (sbitmap_c_tests): Add new test.
5100 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5102 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5104 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5105 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5107 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5109 * config/aarch64/aarch64-builtins.c
5110 (aarch64_types_quadopu_lane_qualifiers): New.
5111 (TYPES_QUADOPU_LANE): New.
5112 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5113 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5114 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5115 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5116 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5117 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5118 (Vdottype, DOTPROD): New.
5119 (sur): Add SDOT and UDOT.
5121 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5123 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5124 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5125 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5127 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5128 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5129 Enable TARGET_DOTPROD.
5130 (cortex-a75.cortex-a55): Likewise.
5131 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5133 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5135 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5136 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5138 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5140 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5141 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5142 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5143 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5144 * config/arm/types.md (neon_dot, neon_dot_q): New.
5145 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5147 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5149 * config/arm/arm.h (TARGET_DOTPROD): New.
5150 * config/arm/arm.c (arm_arch_dotprod): New.
5151 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5152 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5153 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5154 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5155 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5156 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5157 * doc/invoke.texi (armv8.2-a): Document dotprod
5159 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5161 * i386.c (ix86_vec_cost): New function.
5162 (ix86_rtx_costs): Handle vector operations better.
5163 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5164 * x86-tune-costs.h: Add new costs to all tables.
5166 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5168 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5170 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5171 divsd, sqrtss and sqrtsd
5172 * x86-tune-costs.h: Add new entries to all costs.
5173 (znver1_cost): Fix to match real instruction latencies.
5175 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5176 Michael Collison <michael.collison@arm.com>
5178 * compare-elim.c: Include emit-rtl.h.
5179 (can_merge_compare_into_arith): New function.
5180 (try_validate_parallel): Likewise.
5181 (try_merge_compare): Likewise.
5182 (try_eliminate_compare): Call the above when no previous clobber
5184 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5187 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5191 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5192 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5196 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5197 to be any operand_equal_p operands. For & (B - 1) require
5198 B to be power of 2. Recognize
5199 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5201 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5204 * optabs.c (expand_memory_blockage): Fix call of
5205 targetm.have_memory_blockage.
5207 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5210 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5211 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5212 objects to extra_objs instead of overwriting it.
5214 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5216 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5217 Use any_fp_register_operand as operand[3] predicate. Simplify
5218 equality test for operands[2] and operands[4] memory location.
5219 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5220 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5221 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5222 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5223 any_fp_register_operand as operand[1] predicate. Simplify
5224 equality test for operands[0] and operands[3] memory location.
5225 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5226 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5227 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5229 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5231 * target-insns.def: Add memory_blockage.
5232 * optabs.c (expand_memory_blockage): New function.
5233 (expand_asm_memory_barrier): Rename ...
5234 (expand_asm_memory_blockage): ... to this.
5235 (expand_mem_thread_fence): Call expand_memory_blockage
5236 instead of expand_asm_memory_barrier.
5237 (expand_mem_singnal_fence): Ditto.
5238 (expand_atomic_load): Ditto.
5239 (expand_atomic_store): Ditto.
5240 * doc/md.texi (Standard Pattern Names For Generation):
5241 Document memory_blockage instruction pattern.
5243 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5245 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5246 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5247 * config/rl78/rl78.md: New define_expand "adddi3".
5249 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5251 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5254 2017-10-13 Jeff Law <law@redhat.com>
5256 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5258 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5261 * internal-fn.c (expand_mul_overflow): If both operands have
5262 the same highpart of -1 or 0 and the topmost bit of lowpart
5263 is different, overflow is if res <= 0 rather than res < 0.
5265 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5267 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5268 TARGET_P9_VECTOR code for unaligned_load case.
5270 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5272 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5274 2017-10-13 Nathan Sidwell <nathan@acm.org>
5276 * tree-core.h (tree_contains_struct): Make bool.
5277 * tree.c (tree_contains_struct): Likewise.
5278 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5279 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5280 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5281 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5282 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5284 2017-10-13 Richard Biener <rguenther@suse.de>
5286 * graphite-isl-ast-to-gimple.c
5287 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5288 parameters and dominance check.
5289 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5290 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5291 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5292 Do not update SSA form here or do intermediate IL verification.
5293 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5294 (graphite_initialize): Remove check on the number of loops in
5295 the function and inline into graphite_transform_loops.
5296 (graphite_finalize): Inline into graphite_transform_loops.
5297 (graphite_transform_loops): Perform SSA update and IL verification
5299 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5301 2017-10-13 Richard Biener <rguenther@suse.de>
5303 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5304 graphite_expression_type_precision): Avoid global constructor
5306 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5307 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5308 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5309 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5310 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5311 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5313 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5316 * config/i386/i386.h (ix86_red_zone_size): New.
5317 * config/i386/i386.md (push peephole2s): Replace
5318 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5320 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5321 Alan Hayward <alan.hayward@arm.com>
5322 David Sherwood <david.sherwood@arm.com>
5324 * combine.c (can_change_dest_mode): Reject changes in
5325 REGMODE_NATURAL_SIZE.
5327 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5328 Alan Hayward <alan.hayward@arm.com>
5329 David Sherwood <david.sherwood@arm.com>
5331 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5332 (expand_debug_source_expr): Likewise.
5333 * combine.c (combine_simplify_rtx): Likewise.
5334 * cse.c (fold_rtx): Likewise.
5335 * fwprop.c (canonicalize_address): Likewise.
5336 * targhooks.c (default_shift_truncation_mask): Likewise.
5338 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5339 Alan Hayward <alan.hayward@arm.com>
5340 David Sherwood <david.sherwood@arm.com>
5342 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5343 (widened_mode): Likewise.
5344 (expand_unop): Likewise.
5345 * ree.c (transform_ifelse): Likewise.
5346 (merge_def_and_ext): Likewise.
5347 (combine_reaching_defs): Likewise.
5348 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5350 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5351 Alan Hayward <alan.hayward@arm.com>
5352 David Sherwood <david.sherwood@arm.com>
5354 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5355 * combine.c (gen_lowpart_for_combine): Likewise.
5356 * dwarf2out.c (rtl_for_decl_location): Likewise.
5357 * final.c (alter_subreg): Likewise.
5358 * rtlhooks.c (gen_lowpart_general): Likewise.
5359 (gen_lowpart_if_possible): Likewise.
5361 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5362 Alan Hayward <alan.hayward@arm.com>
5363 David Sherwood <david.sherwood@arm.com>
5365 * calls.c (expand_call): Use subreg_lowpart_offset.
5366 * cse.c (cse_insn): Likewise.
5367 * regcprop.c (copy_value): Likewise.
5368 (copyprop_hardreg_forward_1): Likewise.
5370 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5373 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5374 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5375 =Q constraints to +Q and into insn condition add check
5376 that operands[0] and operands[1] are equal.
5377 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5378 =Q constraints to +Q and into insn condition add check
5379 that operands[0] is equal to either operands[1] or operands[2].
5382 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5383 instead of handling MINUS_EXPR twice (once for each argument),
5384 canonicalize operand order and handle just once, use rtype where
5385 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5388 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5389 any values of __C while still being pattern recognizable as a simple
5392 2017-10-13 Richard Biener <rguenther@suse.de>
5394 PR tree-optimization/82451
5396 2017-10-02 Richard Biener <rguenther@suse.de>
5398 PR tree-optimization/82355
5399 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5400 a mapping for the enclosing loop but avoid generating one for
5402 (copy_bb_and_scalar_dependences): Remove premature codegen
5403 error on PHIs in blocks duplicated into multiple places.
5404 * graphite-scop-detection.c
5405 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5406 in the region use it as loop and nest to analyze the DR in.
5407 (try_generate_gimple_bb): Likewise.
5408 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5409 (add_loop_constraints): For blocks in a loop not in the region
5410 create a dimension with a single iteration.
5411 * sese.h (gbb_loop_at_index): Remove assert.
5413 * cfgloop.c (loop_preheader_edge): For the loop tree root
5414 return the single successor of the entry block.
5415 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5416 Reset the SCEV hashtable and niters.
5417 * graphite-scop-detection.c
5418 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5419 assert that we only have POLYNOMIAL_CHREC that vary in loops
5420 contained in the region.
5421 (scop_detection::graphite_can_represent_expr): Adjust.
5422 (scop_detection::stmt_has_simple_data_refs_p): For loops
5423 not in the region set loop to NULL. The nest is now the
5424 entry edge to the region.
5425 (try_generate_gimple_bb): Likewise.
5426 * sese.c (scalar_evolution_in_region): Adjust for
5427 instantiate_scev change.
5428 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5429 Make nest parameter the edge into the region.
5430 (create_data_ref): Likewise.
5431 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5432 entry edge into a region and adjust instantiate_scev calls.
5433 (create_data_ref): Likewise.
5434 (graphite_find_data_references_in_stmt): Likewise.
5435 (find_data_references_in_stmt): Pass the loop preheader edge
5436 from the nest argument.
5437 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5438 parameter the edge into the region.
5439 (instantiate_parameters): Use the loop preheader edge as entry.
5440 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5442 (get_instantiated_value_entry): Make instantiate_below parameter
5443 the edge into the region.
5444 (instantiate_scev_name): Likewise. Adjust dominance checks,
5445 when we cannot use loop-based instantiation instantiate by
5446 walking use-def chains.
5447 (instantiate_scev_poly): Adjust.
5448 (instantiate_scev_binary): Likewise.
5449 (instantiate_scev_convert): Likewise.
5450 (instantiate_scev_not): Likewise.
5451 (instantiate_array_ref): Remove.
5452 (instantiate_scev_3): Likewise.
5453 (instantiate_scev_2): Likewise.
5454 (instantiate_scev_1): Likewise.
5455 (instantiate_scev_r): Do not blindly handle N-operand trees.
5456 Do not instantiate array-refs. Handle all constants and invariants.
5457 (instantiate_scev): Make instantiate_below parameter
5458 the edge into the region.
5459 (resolve_mixers): Use the loop preheader edge for the region
5460 parameter to instantiate_scev_r.
5461 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5463 2017-10-13 Richard Biener <rguenther@suse.de>
5465 PR tree-optimization/82525
5466 * graphite-isl-ast-to-gimple.c
5467 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5469 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5470 Fail code generation when we cannot represent the isl integer.
5471 (binary_op_to_tree): Elide modulo operations that are no-ops
5472 in the type we code generate. Remove now superfluous code
5475 2017-10-13 Richard Biener <rguenther@suse.de>
5477 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5478 (scop_detection::harmful_loop_in_region): Remove premature
5479 IV type restriction.
5480 (scop_detection::graphite_can_represent_scev): We can handle
5481 pointer IVs just fine.
5483 2017-10-13 Alan Modra <amodra@gmail.com>
5485 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5486 "Clobbers and Scratch Registers". Add paragraph on
5487 alternative to clobbers for scratch registers and OpenBLAS
5490 2017-10-13 Alan Modra <amodra@gmail.com>
5492 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5493 example of a memory input for a string of known length. Move
5494 commentary out of table. Add a number of new examples
5495 covering array memory inputs.
5497 2017-10-12 Martin Liska <mliska@suse.cz>
5499 PR tree-optimization/82493
5500 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5501 (test_range_functions): New function.
5502 (sbitmap_c_tests): Likewise.
5503 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5504 * selftest.h (sbitmap_c_tests): New function.
5506 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5508 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5510 * config/rs6000/amo.h: Fix spacing issue.
5512 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5515 * config/i386/i386.md (*ashl<mode>3_mask_1,
5516 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5517 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5520 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5522 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5523 (profile_probability): Set max_probability
5524 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5526 * profile-count.c (profile_probability::differs_from_p): Do not
5527 rely on max_probaiblity == 10000
5529 2017-10-12 Jeff Law <law@redhat.com>
5531 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5534 2017-10-12 Martin Sebor <msebor@redhat.com>
5538 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5539 (handle_alias_pairs): Call it.
5540 * common.opt (-Wattribute-alias): New option.
5541 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5542 * doc/invoke.texi (-Wattribute-alias): Document.
5544 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5547 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5549 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5551 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5552 (make_hard_regno_born, make_hard_regno_dead): Update
5553 bb_killed_pseudos and bb_gen_pseudos.
5555 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5557 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5559 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5561 * config/alpha/alpha.c (alpha_split_conditional_move):
5562 Use std::swap instead of manually swapping.
5563 (alpha_stdarg_optimize_hook): Ditto.
5564 (alpha_canonicalize_comparison): Ditto.
5566 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5568 * tree-loop-distribution.c (struct builtin_info): New struct.
5569 (struct partition): Refactor fields into struct builtin_info.
5570 (partition_free): Free struct builtin_info.
5571 (build_size_arg_loc, build_addr_arg_loc): Delete.
5572 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5573 information from struct builtin_info.
5574 (find_single_drs): New function refactored from classify_partition.
5575 Also moved builtin validity checks to this function.
5576 (compute_access_range, alloc_builtin): New functions.
5577 (classify_builtin_st, classify_builtin_ldst): New functions.
5578 (classify_partition): Refactor code into functions find_single_drs,
5579 classify_builtin_st and classify_builtin_ldst.
5580 (distribute_loop): Don't do runtime alias check when distributing
5582 (find_seed_stmts_for_distribution): New function.
5583 (pass_loop_distribution::execute): Refactor code finding seed
5584 stmts into above function. Support distribution for the innermost
5585 two-level loop nest. Adjust dump information.
5587 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5589 * tree-loop-distribution.c: Adjust the general comment.
5590 (NUM_PARTITION_THRESHOLD): New macro.
5591 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5592 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5593 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5594 in call to build_partition_graph.
5595 (finalize_partitions): New parameter. Make loop distribution more
5596 conservative by fusing more partitions.
5597 (distribute_loop): Don't do runtime alias check in case of loop nest
5599 (find_seed_stmts_for_distribution): New function.
5600 (prepare_perfect_loop_nest): New function.
5601 (pass_loop_distribution::execute): Refactor code finding seed stmts
5602 and loop nest into above functions. Support loop nest distribution.
5603 Adjust dump information accordingly.
5605 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5607 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5608 and set PTYPE_SEQUENTIAL for merged partition.
5610 2017-10-12 Richard Biener <rguenther@suse.de>
5612 PR tree-optimization/69728
5614 2017-09-19 Richard Biener <rguenther@suse.de>
5616 PR tree-optimization/69728
5617 * graphite-sese-to-poly.c (schedule_error): New global.
5618 (add_loop_schedule): Handle empty domain by failing the
5620 (build_original_schedule): Handle schedule_error.
5622 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5623 domain by returning an unchanged schedule.
5625 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5627 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5628 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5630 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5632 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5635 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5638 * expr.c (store_field): Don't optimize away bitsize == 0 store
5639 from CALL_EXPR with addressable return type.
5641 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5643 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5644 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5645 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5646 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5647 TARGET_ISEL instead of TARGET_ISEL<sel>.
5649 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5651 * config/rs6000/rs6000.c
5652 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5654 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5656 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5658 (reg_or_cint_operand): Fix comment.
5659 (reg_or_zero_operand): New predicate.
5660 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5661 * config/rs6000/rs6000.c (output_isel): Delete.
5662 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5663 instead of reg_or_cint_operand. Output instruction directly (not via
5665 (isel_unsigned_<mode>): Ditto.
5666 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5667 gpc_reg_operand. Add an instruction alternative for this. Output
5668 instruction directly.
5669 (*isel_reversed_unsigned_<mode>): Ditto.
5671 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5673 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5674 (TARGET_CANONICALIZE_COMPARISON): Define.
5676 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5679 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5680 Check whether the dest is REG before adding REG_EQUIV note.
5682 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5685 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5687 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5688 (make_hard_regno_born, make_hard_regno_dead): Update
5689 bb_killed_pseudos and bb_gen_pseudos.
5691 2017-10-11 Nathan Sidwell <nathan@acm.org>
5693 * incpath.h (enum incpath_kind): Name enum, prefix values.
5694 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5695 * incpath.c (heads, tails): Use INC_MAX.
5696 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5697 (merge_include_chains, split_quote_chain,
5698 register_include_chains): Update incpath_kind names.
5699 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5700 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5702 (add_framework_path, darwin_register_objc_includes): Likewise.
5703 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5705 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5707 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5708 Do not use float_operator operator predicate.
5709 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5710 * config/i386/predicates.md (float_operator): Remove predicate.
5712 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5714 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5715 (*jccxf_i387): Ditto.
5716 (*jcc<mode>_i387): Ditto.
5717 (*jccu<mode>_i387): Ditto.
5718 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5719 (*jcc_*_i387 splitters): Remove.
5720 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5721 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5722 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5725 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5727 * profile-count.h (slow_safe_scale_64bit): New function.
5728 (safe_scale_64bit): New inline.
5729 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5730 * profile-count.c: Include wide-int.h
5731 (slow_safe_scale_64bit): New.
5733 2017-10-11 Nathan Sidwell <nathan@acm.org>
5735 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5736 HAS_DECL_ASSEMBLER_NAME_P.
5737 * gimple-expr.c (gimple_decl_printable_name: Check
5738 HAS_DECL_ASSEMBLER_NAME_P too.
5739 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5740 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5741 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5742 * passes.c (rest_of_decl_compilation): Check
5743 HAS_DECL_ASSEMBLER_NAME_P too.
5744 * recog.c (verify_changes): Likewise.
5745 * tree-pretty-print.c (dump_decl_name): Likewise.
5746 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5748 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5749 (SET_DECL_ASSEMBLER_NAME): Use it.
5750 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5751 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5752 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5754 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5756 * config.gcc (i386, x86_64): Add extra objects.
5757 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5758 (ix86_min_insn_size): Declare.
5759 (ix86_issue_rate): Declare.
5760 (ix86_adjust_cost): Declare.
5761 (ia32_multipass_dfa_lookahead): Declare.
5762 (ix86_macro_fusion_p): Declare.
5763 (ix86_macro_fusion_pair_p): Declare.
5764 (ix86_bd_has_dispatch): Declare.
5765 (ix86_bd_do_dispatch): Declare.
5766 (ix86_core2i7_init_hooks): Declare.
5767 (ix86_atom_sched_reorder): Declare.
5768 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5769 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5770 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5771 (rip_relative_addr_p): Rename to ...
5772 (ix86_rip_relative_addr_p): ... this one; export.
5773 (memory_address_length): Update.
5774 (ix86_issue_rate): Move to x86-tune-sched.c.
5775 (ix86_flags_dependent): Move to x86-tune-sched.c.
5776 (ix86_agi_dependent): Move to x86-tune-sched.c.
5777 (exact_dependency_1): Move to x86-tune-sched.c.
5778 (exact_store_load_dependency): Move to x86-tune-sched.c.
5779 (ix86_adjust_cost): Move to x86-tune-sched.c.
5780 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5781 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5782 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5783 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5784 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5785 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5786 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5787 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5788 (min_insn_size): Rename to ...
5789 (ix86_min_insn_size): ... this one; export.
5790 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5791 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5792 (core2i7_first_cycle_multipass_backtrack): Move to
5793 x86-tune-sched-core.c.
5794 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5795 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5796 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5797 (ix86_avoid_jump_mispredicts): Update.
5798 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5799 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5800 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5801 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5802 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5803 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5804 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5805 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5806 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5807 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5808 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5809 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5810 (BIG): Move to ix86-tune-sched-bd.c.
5811 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5812 (enum insn_path): Move to ix86-tune-sched-bd.c.
5813 (get_mem_group): Move to ix86-tune-sched-bd.c.
5814 (is_cmp): Move to ix86-tune-sched-bd.c.
5815 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5816 (is_branch): Move to ix86-tune-sched-bd.c.
5817 (is_prefetch): Move to ix86-tune-sched-bd.c.
5818 (init_window): Move to ix86-tune-sched-bd.c.
5819 (allocate_window): Move to ix86-tune-sched-bd.c.
5820 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5821 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5822 (process_end_window): Move to ix86-tune-sched-bd.c.
5823 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5824 (find_constant): Move to ix86-tune-sched-bd.c.
5825 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5826 (has_immediate): Move to ix86-tune-sched-bd.c.
5827 (get_insn_path): Move to ix86-tune-sched-bd.c.
5828 (get_insn_group): Move to ix86-tune-sched-bd.c.
5829 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5830 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5831 (add_insn_window): Move to ix86-tune-sched-bd.c.
5832 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5833 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5834 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5835 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5836 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5837 (do_dispatch): Move to ix86-tune-sched-bd.c.
5838 (has_dispatch): Move to ix86-tune-sched-bd.c.
5839 * config/i386/t-i386: Add new object files.
5840 * config/i386/x86-tune-costs.h: New file.
5841 * config/i386/x86-tune-sched-atom.c: New file.
5842 * config/i386/x86-tune-sched-bd.c: New file.
5843 * config/i386/x86-tune-sched-core.c: New file.
5844 * config/i386/x86-tune-sched.c: New file.
5846 2017-10-11 Liu Hao <lh_mouse@126.com>
5848 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5849 the generic version below instead.
5850 (should_colorize): Recognize Windows consoles as terminals
5852 * pretty-print.c [__MINGW32__] (write_all): New function.
5853 [__MINGW32__] (find_esc_head): Likewise.
5854 [__MINGW32__] (find_esc_terminator): Likewise.
5855 [__MINGW32__] (eat_esc_sequence): Likewise.
5856 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5858 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5861 2017-10-11 Richard Biener <rguenther@suse.de>
5863 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5864 Properly call analyze_scalar_evolution with the loop of the stmt.
5866 2017-10-11 Richard Biener <rguenther@suse.de>
5868 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5869 * tree-core.h (tree_base): Add chrec_var union member.
5870 * tree.h (CHREC_VAR): Remove.
5871 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5872 * tree-chrec.h (build_polynomial_chrec): Adjust.
5873 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5874 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5876 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5878 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5879 * match.pd: ... here.
5880 ((T) X == (T) Y): Relax condition.
5882 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5884 PR tree-optimization/82472
5885 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5887 (break_alias_scc_partitions): Update postorder number.
5889 2017-10-11 Martin Liska <mliska@suse.cz>
5892 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5894 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5896 2017-10-11 Martin Liska <mliska@suse.cz>
5898 * print-rtl.c (print_insn): Move declaration of idbuf
5899 to same scope as name.
5901 2017-10-11 Martin Liska <mliska@suse.cz>
5906 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5909 2017-10-11 Martin Liska <mliska@suse.cz>
5912 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5915 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5917 * config.gcc (default_gnu_indirect_function): Default to yes for
5918 arm*-*-linux* with glibc.
5920 2017-10-11 Richard Biener <rguenther@suse.de>
5922 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5923 default-defs and types we do not want to analyze.
5924 (interpret_loop_phi): Replace unreachable code with an assert.
5925 (compute_scalar_evolution_in_loop): Remove and inline ...
5926 (analyze_scalar_evolution_1): ... here, replacing condition with
5927 what makes the intent clearer. Remove handling of cases
5928 get_scalar_evolution now handles.
5930 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5932 PR rtl-optimization/81434
5933 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5934 comment for main loop. In sched_group_found if, also add checks for
5935 pass and min_cost_group.
5937 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5939 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5940 (rs6000_insn_cost): New function.
5941 * config/rs6000/rs6000.md (cost): New attribute.
5943 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5944 H.J. Lu <hongjiu.lu@intel.com>
5948 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5949 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5950 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5951 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5952 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5953 OPTION_MASK_ISA_MMX and some other bit set require both
5954 mmx and the ISAs without the mmx bit.
5955 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5956 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5957 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5958 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5959 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5960 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5961 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5962 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5963 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5964 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5965 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5966 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5967 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5968 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5969 Add OPTION_MASK_ISA_MMX.
5971 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5973 * config.gcc (armv7*-*-freebsd*): New target.
5974 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5976 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5978 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5979 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5982 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5984 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5985 a template parameter.
5986 (WIDE_INT_REF_FOR): Update accordingly.
5987 * tree.h (wi::int_traits <const_tree>): Delete.
5988 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5989 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5990 (wi::tree_to_wide_ref): New typedef.
5991 (wi::to_wide): New function.
5992 * calls.c (get_size_range): Use wi::to_wide when operating on
5994 * cgraph.c (cgraph_node::create_thunk): Likewise.
5995 * config/i386/i386.c (ix86_data_alignment): Likewise.
5996 (ix86_local_alignment): Likewise.
5997 * dbxout.c (stabstr_O): Likewise.
5998 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5999 * expr.c (const_vector_from_tree): Likewise.
6000 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6001 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6002 (fold_negate_expr_1, int_const_binop_1, const_binop)
6003 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6004 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6005 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6006 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6007 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6008 (fold_not_const, round_up_loc): Likewise.
6009 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6010 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6011 (alloca_call_type): Likewise.
6012 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6013 * godump.c (go_output_typedef): Likewise.
6014 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6015 * internal-fn.c (get_min_precision): Likewise.
6016 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6017 * ipa-polymorphic-call.c
6018 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6019 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6020 (ipa_modify_call_arguments): Likewise.
6021 * match.pd: Likewise.
6022 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6023 * print-tree.c (print_node_brief, print_node): Likewise.
6024 * stmt.c (expand_case): Likewise.
6025 * stor-layout.c (layout_type): Likewise.
6026 * tree-affine.c (tree_to_aff_combination): Likewise.
6027 * tree-cfg.c (group_case_labels_stmt): Likewise.
6028 * tree-data-ref.c (dr_analyze_indices): Likewise.
6029 (prune_runtime_alias_test_list): Likewise.
6030 * tree-dump.c (dequeue_and_dump): Likewise.
6031 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6032 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6033 * tree-pretty-print.c (dump_generic_node): Likewise.
6034 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6035 (simple_iv_with_niters): Likewise.
6036 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6037 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6038 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6039 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6040 (refine_value_range_using_guard, number_of_iterations_ne_max)
6041 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6042 (get_cst_init_from_scev, record_nonwrapping_iv)
6043 (scev_var_range_cant_overflow): Likewise.
6044 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6045 * tree-ssa-pre.c (compute_avail): Likewise.
6046 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6047 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6048 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6049 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6050 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6051 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6052 (dump_case_nodes, try_switch_expansion): Likewise.
6053 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6054 (vect_do_peeling): Likewise.
6055 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6056 * tree-vect-stmts.c (vectorizable_load): Likewise.
6057 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6058 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6059 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6060 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6061 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6062 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6063 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6064 (evrp_dom_walker::before_dom_children): Likewise.
6065 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6066 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6067 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6068 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6069 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6070 (get_range_pos_neg): Likewise.
6071 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6072 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6073 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6074 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6075 * config/avr/avr.c (avr_fold_builtin): Likewise.
6076 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6077 * config/msp430/msp430.c (msp430_attr): Likewise.
6078 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6079 * config/powerpcspe/powerpcspe-c.c
6080 (altivec_resolve_overloaded_builtin): Likewise.
6081 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6082 (rs6000_expand_ternop_builtin): Likewise.
6083 * config/rs6000/rs6000-c.c
6084 (altivec_resolve_overloaded_builtin): Likewise.
6085 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6086 (rs6000_expand_ternop_builtin): Likewise.
6087 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6089 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6091 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6092 when copying loop nest with only one inner loop.
6094 2017-10-10 Richard Biener <rguenther@suse.de>
6096 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6097 blocks if SCEV is active.
6098 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6100 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6101 (scev_initialize): Assert we are not yet initialized.
6103 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6105 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6106 inner loop's exit stmt by making it always exit the loop, otherwise
6107 we would generate an infinite empty loop.
6109 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6111 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6112 renaming variables in new preheader if it's deleted.
6114 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6116 * tree-loop-distribution.c (struct partition): Remove unused field
6117 loops of the structure.
6118 (partition_alloc, partition_free): Ditto.
6119 (build_rdg_partition_for_vertex): Ditto.
6121 2017-10-09 Jeff Law <law@redhat.com>
6123 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6124 return type to match prototype and documentation.
6126 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6128 * config/rs6000/rs6000.c (processor_costs): Move to ...
6129 * config/rs6000/rs6000.h: ... here.
6130 (rs6000_cost): Declare.
6132 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6134 * except.c (setjmp_fn): New global variable.
6135 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6136 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6137 if DONT_USE_BUILTIN_SETJMP is defined.
6139 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6141 * target.def (insn_cost): New hook.
6142 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6143 * doc/tm.texi: Regenerate.
6144 * rtlanal.c (insn_cost): Use the new hook.
6146 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6148 * combine.c (combine_validate_cost): Compute the new insn_cost,
6149 not just pattern_cost.
6150 (try_combine): Adjust comment.
6152 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6154 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6156 * combine.c (uid_insn_cost): Adjust comment.
6157 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6159 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6160 * dse.c (find_shift_sequence): Ditto.
6161 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6162 (bb_valid_for_noce_process_p): Use pattern_cost.
6163 * rtl.h (insn_rtx_cost): Delete.
6164 (pattern_cost): New prototype.
6165 (insn_cost): New prototype.
6166 * rtlanal.c (insn_rtx_cost): Rename to...
6167 (pattern_cost): ... this.
6170 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6172 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6173 (*jcc<mode>_0_r_i387): Ditto.
6174 (*jccxf_r_i387): Ditto.
6175 (*jcc<mode>_r_i387): Ditto.
6176 (*jccu<mode>_r_i387): Ditto.
6177 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6178 (*jcc): Rename from *jcc_1.
6180 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6182 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6183 deferred rescans after the lvx/stvx recombination pre-pass.
6185 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6187 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6188 memory operation instruction support.
6189 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6190 (rs6000-ibm-aix[789]*): Likewise.
6191 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6192 Document new functions.
6194 2017-10-09 Richard Biener <rguenther@suse.de>
6196 PR tree-optimization/82397
6197 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6198 equality only for semantically equal trees.
6200 2017-10-09 Richard Biener <rguenther@suse.de>
6202 PR tree-optimization/82449
6203 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6204 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6205 allow constant addresses.
6206 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6209 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6211 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6214 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6217 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6220 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6223 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6225 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6228 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6229 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6231 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6233 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6234 (WI_BINARY_PREDICATE_RESULT): Likewise.
6235 (wi::binary_traits::operator_result): New type.
6236 (wi::binary_traits::predicate_result): Likewise.
6237 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6238 (generic_wide_int::operator==, generic_wide_int::operator!=)
6239 (generic_wide_int::operator&, generic_wide_int::and_not)
6240 (generic_wide_int::operator|, generic_wide_int::or_not)
6241 (generic_wide_int::operator^, generic_wide_int::operator+
6242 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6244 (operator~, unary operator-, operator==, operator!=, operator&)
6245 (operator|, operator^, operator+, binary operator-, operator*): New
6247 * expr.c (get_inner_reference): Use wi::bit_and_not.
6248 * fold-const.c (fold_binary_loc): Likewise.
6249 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6250 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6251 (bit_value_binop): Likewise.
6252 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6253 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6254 (extract_range_from_binary_expr_1): Likewise.
6255 (masked_increment): Likewise.
6256 (simplify_bit_ops_using_ranges): Likewise.
6258 2017-10-09 Martin Jambor <mjambor@suse.cz>
6261 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6262 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6263 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6265 (hsa_fixup_mov_insn_type): New function.
6266 (hsa_op_with_type::get_in_type): Use it.
6267 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6268 immediates in an assert.
6269 (hsa_op_with_type::extend_int_to_32bit): New method.
6270 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6271 types. Convert to dest type if necessary.
6272 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6273 (reg_for_gimple_ssa): Pass false as min32int to
6274 hsa_type_for_scalar_tree_type.
6275 (gen_hsa_addr): Fixup type when creating addresable temporary.
6276 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6277 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6278 necessary. Call hsa_fixup_mov_insn_type.
6279 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6280 extend operands and convert to dest type if necessary.
6281 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6282 to dest type if necessary.
6283 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6284 if conversion nt necessary and size matches.
6285 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6286 to dest type if necessary.
6287 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6288 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6290 (gen_hsa_clrsb): Likewise.
6291 (gen_hsa_ffs): Likewise.
6292 (gen_hsa_divmod): Extend operands and convert to dest type if
6294 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6296 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6298 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6299 Remove empty default arguments. Use a brace block as output
6301 (conditional return): Ditto.
6303 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6304 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6306 (group_ending_nop): Ditto.
6307 (doloop_end): Ditto.
6308 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6309 (splitters for those): Ditto.
6311 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6313 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6314 a conditional jump (and the compare for it) so that pc_rtx is the
6316 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6317 for the deleted and renamed ctr<mode>_internal[234] patterns.
6318 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6319 Delete second conditional return pattern.
6320 (ctr<mode>_internal2): Delete this second bdnz pattern.
6321 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6322 (ctr<mode>_internal4): Delete this second bdz pattern.
6324 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6326 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6327 (always_initialized_rtx_for_ssa_name_p): New predicate.
6328 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6329 (finish_out_of_ssa): Free new field of SA.
6330 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6331 * tree-ssa-coalesce.c: Include tree-ssa.h.
6332 (get_parm_default_def_partitions): Remove extern keyword.
6333 (get_undefined_value_partitions): New function.
6334 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6335 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6337 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6339 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6341 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6343 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6345 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6346 for targets that preffer 128bit.
6348 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6350 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6352 2017-10-08 Olivier Hainque <hainque@adacore.com>
6354 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6355 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6357 (thumb_set_return_address): Likewise.
6359 2017-10-08 Olivier Hainque <hainque@adacore.com>
6361 * common/config/arm/arm-common.c (arm_except_unwind_info):
6362 Handle DWARF2_UNWIND_INFO.
6364 2017-10-07 Michael Collison <michael.collison@arm.com>
6366 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6369 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6371 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6372 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6373 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6374 defined, force the creation of a new block for a dispatch label.
6376 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6378 * invoke.texi (Wsuggest-attribute=cold): Document.
6379 * common.opt (Wsuggest-attribute=cold): New
6380 * ipa-pure-const.c (warn_function_cold): New function.
6381 * predict.c (compute_function_frequency): Use it.
6382 * predict.h (warn_function_cold): Declare.
6384 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6386 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6389 2017-10-06 Martin Liska <mliska@suse.cz>
6391 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6392 keyword for member functions.
6393 (struct sanopt_tree_couple): New struct.
6394 (struct sanopt_tree_couple_hash): New function.
6395 (struct sanopt_ctx): Add new hash_map.
6396 (has_dominating_ubsan_ptr_check): New function.
6397 (record_ubsan_ptr_check_stmt): Likewise.
6398 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6399 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6400 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6402 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6405 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6406 aarch64_simd_valid_immediate on CONST_VECTORs.
6407 (aarch64_reg_or_bic_imm): Likewise.
6409 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6411 PR rtl-optimization/82396
6412 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6414 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6416 * graphite-dependences.c (scop_get_reads): Move code to...
6417 (scop_get_must_writes): Move code to...
6418 (scop_get_may_writes): Move code to...
6419 (scop_get_reads_and_writes): ... here.
6420 (scop_get_dependences): Call scop_get_reads_and_writes.
6422 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6424 PR tree-optimization/82434
6425 * fold-const.h (can_native_encode_type_p,
6426 can_native_encode_string_p): Remove.
6427 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6428 don't encode anything, just return what would be otherwise returned.
6429 (native_encode_fixed, native_encode_complex, native_encode_vector):
6431 (native_encode_string): Likewise. Inline by hand
6432 can_native_encode_string_p.
6433 (can_native_encode_type_p): Remove.
6434 (can_native_encode_string_p): Remove.
6435 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6436 STRING_CSTs using can_native_encode_string_p, test all
6437 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6438 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6439 argument from native_encode_expr.
6440 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6441 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6444 2017-10-06 Richard Biener <rguenther@suse.de>
6446 PR tree-optimization/82397
6447 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6448 operand_equal_p but rely on data_ref_compare_tree for detecting
6450 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6451 to match up with dr_group_sort_cmp.
6453 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6456 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6458 * config/s390/s390-builtin-types.def: Regenerate.
6460 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6463 * config/s390/s390-builtin-types.def: Regenerate.
6464 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6465 Change flag from B_VXE to B_VX.
6466 (s390_vec_min_dbl): Remove B_VXE flag.
6468 2017-10-06 Richard Biener <rguenther@suse.de>
6470 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6471 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6472 translate_isl_ast_to_gimple::is_valid_rename,
6473 translate_isl_ast_to_gimple::get_rename,
6474 translate_isl_ast_to_gimple::get_def_bb_for_const,
6475 translate_isl_ast_to_gimple::get_new_name,
6476 translate_isl_ast_to_gimple::collect_all_ssa_names,
6477 translate_isl_ast_to_gimple::copy_loop_phi_args,
6478 translate_isl_ast_to_gimple::collect_all_ssa_names,
6479 translate_isl_ast_to_gimple::copy_loop_phi_args,
6480 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6481 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6482 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6483 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6484 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6485 translate_isl_ast_to_gimple::copy_cond_phi_args,
6486 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6487 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6488 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6489 translate_isl_ast_to_gimple::rename_uses,
6490 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6491 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6492 (set_rename_for_each_def): Likewise.
6493 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6494 here. Handle rewriting SCEV analyzable uses here.
6495 (copy_bb_and_scalar_dependences): Generate code for PHI
6497 (graphite_regenerate_ast_isl): Adjust.
6498 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6499 (add_write, add_read): New functions.
6500 (build_cross_bb_scalars_def): Use it and simplify.
6501 (build_cross_bb_scalars_use): Likewise.
6502 (graphite_find_cross_bb_scalar_vars): Inline into...
6503 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6504 simulating out-of-SSA. Compute liveout and add dependencies.
6505 (build_scops): Force an empty entry block.
6506 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6508 (sese_build_liveouts): Declare.
6509 (sese_trivially_empty_bb_p): Likewise.
6510 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6511 compute liveout and debug_liveout.
6512 (sese_bad_liveouts_use): Remove.
6513 (sese_reset_debug_liveouts_bb): Likewise.
6514 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6515 (sese_build_liveouts): Build liveout and debug_liveout and store
6517 (new_sese_info): Adjust.
6518 (free_sese_info): Likewise.
6519 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6520 do not build liveout here.
6521 (move_sese_in_condition): Adjust region entry.
6522 (scev_analyzable_p): Match up with chrec_apply requirements.
6523 (sese_trivially_empty_bb_p): New.
6524 * tree-into-ssa.c (get_reaching_def): Properly support generating
6525 default-defs for incremental rewrite of anonymous names.
6527 2017-10-06 Richard Biener <rguenther@suse.de>
6529 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6530 precision do not perform modulo reduction.
6532 2017-10-06 Richard Biener <rguenther@suse.de>
6534 PR tree-optimization/82436
6535 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6536 conservatively choose the vectorization factor when checking
6537 whether we can perform the required load permutation.
6538 (vect_transform_slp_perm_load): Assert when we may not fail.
6540 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6542 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6543 message for incompatible -msdata=* and -mcall-* options.
6545 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6547 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6548 rate for post-reload scheduling.
6550 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6552 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6554 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6556 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6557 to improve monte carlo in scimark.
6559 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6561 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6562 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6563 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6564 pentium4_cost, nocona_cost): Set reassociation width to 1.
6565 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6566 width to 2 for fp operations and 1 otherwise.
6567 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6569 (atom_cost): Set reassociation width to 2.
6570 (slm_cost, generic_cost): Set fp reassociation width
6571 to 2 and 1 otherwise.
6572 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6573 (core_cost): Set fp reassociation width to 4 and vector to 2.
6574 (ix86_reassociation_width): Rewrite using cost table; special case
6575 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6576 and TARGET_AVX128_OPTIMAL.
6577 * config/i386/i386.h (processor_costs): Add
6578 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6579 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6580 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6581 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6582 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6583 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6585 2017-10-05 Nathan Sidwell <nathan@acm.org>
6587 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6589 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6591 * config/arm/arm.c (arm_test_fpu_data): New.
6592 (arm_run_selftests): Call arm_test_fpu_data.
6594 2017-10-04 Nathan Sidwell <nathan@acm.org>
6596 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6598 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6600 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6602 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6603 check type for aarch64_simd_valid_immediate.
6604 (aarch64_output_simd_mov_immediate): Update prototype.
6605 (aarch64_simd_valid_immediate): Update prototype.
6606 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6607 support for ORR-immediate.
6608 (and<mode>3): modified pattern to add support for BIC-immediate.
6609 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6610 now checks for valid immediate for BIC and ORR based on new enum
6612 (aarch64_output_simd_mov_immediate): Function now used to output
6613 BIC/ORR imm as well based on new enum argument.
6614 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6616 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6617 (aarch64_reg_or_bic_imm): Likewise.
6619 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6621 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6622 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6624 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6628 PR rtl-optimization/82396
6629 * haifa-sched.c (autopref_multipass_init): Simplify
6631 (autopref_rank_data): Simplify sort order.
6632 * sched-int.h (autopref_multipass_data_): Remove
6633 multi_mem_insn_p, min_offset and max_offset.
6635 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6637 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6639 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6641 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6644 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6646 * doc/sourcebuild.texi: Document vect_long_mult.
6648 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6650 PR tree-optimization/82413
6651 * fold-const.c (build_range_check): Use widest_int when comparing
6652 the maximum ETYPE value with HIGH.
6654 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6656 PR rtl-optimization/82396
6657 * haifa-sched.c (autopref_multipass_init): Simplify
6659 (autopref_rank_data): Simplify sort order.
6660 * sched-int.h (autopref_multipass_data_): Remove
6661 multi_mem_insn_p, min_offset and max_offset.
6663 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6665 PR tree-optimization/82381
6666 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6667 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6670 PR tree-optimization/82374
6671 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6672 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6673 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6674 current_function_decl to the new decl.
6676 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6678 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6679 helper macro for IEEE float128 hardware built-in functions.
6680 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6682 (TRUNCF128_ODD): Likewise.
6683 (ADDF128_ODD): Likewise.
6684 (SUBF128_ODD): Likewise.
6685 (MULF128_ODD): Likewise.
6686 (DIVF128_ODD): Likewise.
6687 (FMAF128_ODD): Likewise.
6688 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6689 UNSPEC_TRUNC_ROUND_TO_ODD.
6690 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6691 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6692 floating point round to odd instructions.
6693 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6694 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6695 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6696 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6697 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6698 (trunc<mode>sf2_hw): Change the truncate with round to odd
6699 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6700 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6701 to odd hardware instructions.
6702 (sub<mode>3_odd): Likewise.
6703 (mul<mode>3_odd): Likewise.
6704 (div<mode>3_odd): Likewise.
6705 (sqrt<mode>2_odd): Likewise.
6706 (fma<mode>4_odd): Likewise.
6707 (fms<mode>4_odd): Likewise.
6708 (nfma<mode>4_odd): Likewise.
6709 (nfms<mode>4_odd): Likewise.
6710 (trunc<mode>df2_odd): Change the truncate with round to odd
6711 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6713 * doc/extend.texi (PowerPC built-in functions): Update documentation
6714 for existing IEEE float128-bit built-in functions. Add built-in
6715 functions that generate the IEEE 128-bit floating point round to
6718 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6720 PR rtl-optimization/77729
6721 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6722 to (X&(C1&~C2))|C2 transformations.
6724 2017-10-03 Martin Jambor <mjambor@suse.cz>
6726 PR tree-optimization/82363
6727 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6728 mismatch, mark lacc written regardless of racc.
6730 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6732 PR tree-optimization/82381
6733 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6734 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6735 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6736 fallthrough into reassoc_stmt_dominates_stmt_p.
6739 * combine.c (combine_instructions): Don't combine in unreachable
6742 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6745 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6746 function to not use the have_cpu variable. Do not set cpu_index,
6747 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6749 (rs6000_valid_attribute_p): Remove duplicate initializations of
6750 old_optimize and func_optimize.
6751 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6752 (rs6000_activate_target_options): Make global.
6753 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6756 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6758 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6759 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6760 Set *poffset to 0 and *psize and *pmax_size to -1 if
6761 *poffset + *psize overflows in HOST_WIDE_INT.
6763 PR tree-optimization/82387
6764 PR tree-optimization/82388
6765 PR tree-optimization/82389
6766 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6767 instead of live_bytes non-NULL.
6769 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6772 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6775 2017-10-02 Richard Biener <rguenther@suse.de>
6777 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6778 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6779 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6781 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6783 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6784 requested precision matches the type's.
6785 * calls.c (alloc_max_size): Calculate the new candidate size as
6786 a widest_int and use wi::to_widest when comparing it with the
6787 current candidate size.
6788 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6789 zero rather than integer_zero_node.
6790 * match.pd: Check for a no-op conversion before using wi::add
6791 rather than after. Use tree_to_uhwi when summing small shift
6792 counts into an unsigned int.
6794 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6795 Alan Hayward <alan.hayward@arm.com>
6796 David Sherwood <david.sherwood@arm.com>
6799 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6800 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6801 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6802 POINTER_AND_FP_REGS.
6804 2017-10-02 Richard Biener <rguenther@suse.de>
6806 PR tree-optimization/82355
6807 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6808 a mapping for the enclosing loop but avoid generating one for
6810 (copy_bb_and_scalar_dependences): Remove premature codegen
6811 error on PHIs in blocks duplicated into multiple places.
6812 * graphite-scop-detection.c
6813 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6814 in the region use it as loop and nest to analyze the DR in.
6815 (try_generate_gimple_bb): Likewise.
6816 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6817 (add_loop_constraints): For blocks in a loop not in the region
6818 create a dimension with a single iteration.
6819 * sese.h (gbb_loop_at_index): Remove assert.
6821 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6823 * omp-expand.c (adjust_context_scope): New function.
6824 (expand_parallel_call): Call adjust_context_scope.
6826 2017-10-01 Jeff Law <law@redhat.com>
6828 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6829 dom_opt_dom_walker class with direct access to private members.
6830 Add comments. Call test_for_singularity.
6831 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6832 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6833 m_dummy_cond anymore.
6834 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6836 (pass_dominator:execute): Build the dummy_cond here and pass it
6837 to the dom_opt_dom_walker ctor.
6838 (test_for_singularity): New function.
6840 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6841 Maya Rashish <coypu@sdf.org>
6843 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6844 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6845 (alpha*-*-netbsd*) Use nbsd_tm_file.
6846 (arm*-*-netbsdelf*) Likewise.
6847 (i[34567]86-*-netbsdelf*) Likewise.
6848 (x86_64-*-netbsd*) Likewise.
6849 (mips*-*-netbsd*) Likewise.
6850 (powerpc-*-netbsd*) Likewise.
6851 (sh*-*-netbsd*) Likewise.
6852 (sparc-*-netbsdelf*) Likewise.
6853 (sparc64-*-netbsd*) Likewise.
6854 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6856 (vax-*-netbsdelf*) Likewise.
6857 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6858 (UINT_FAST8_TYPE) Likewise.
6859 (INT_FAST16_TYPE) Check CHAR_FAST16.
6860 (UINT_FAST16_TYPE) Likewise.
6862 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6865 * config/i386/i386.md
6866 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6867 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6868 *divmodsi4_zext_2): New define_insn_and_split.
6869 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6870 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6871 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6872 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6873 New define_insn_and_split.
6874 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6875 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6876 operands[1] having DImode when mode is SImode.
6878 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6879 always SImode for DIV and MOD in REG_EQUAL notes.
6881 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6884 * match.pd: Fix handling of NaNs in pattern.
6886 2017-09-29 Jeff Law <law@redhat.com>
6888 * sbitmap.c (bitmap_bit_in_range_p): New function.
6889 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6890 * tree-ssa-dse.c (live_bytes_read): New function.
6891 (dse_classify_store): Ignore reads of dead bytes.
6893 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6894 typos and whitespace errors.
6895 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6896 * config/s390/s390.c (s390_emit_prologue): Likewise.
6898 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6901 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6902 with a symbol for LRA.
6904 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6906 PR rtl-optimization/82338
6907 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6909 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6911 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6912 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6913 (qsort_chk): Declare.
6914 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6915 (qsort_chk): New function.
6917 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6919 PR tree-optimization/82337
6920 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6921 phi definition if the PHI result appears in an abnormal PHI.
6922 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6923 basis appears in an abnormal PHI.
6925 2017-09-29 Richard Biener <rguenther@suse.de>
6927 * graphite-isl-ast-to-gimple.c
6928 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6929 (binary_op_to_tree): Use it.
6930 (get_rename_from_scev): Likewise.
6931 (copy_loop_phi_nodes): Likewise.
6932 (copy_bb_and_scalar_dependences): Likewise.
6933 (translate_pending_phi_nodes): Likewise.
6935 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6938 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6939 for movabsq $(i32 << shift), r64.
6941 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6943 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6944 index when encoding %esp as %rsp to avoid 0x67 prefix.
6946 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6948 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6949 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6951 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6953 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6954 Extensions with more than 16 double VFP registers.
6955 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6956 to_clear_mask and all code related to it. Replace the remaining
6957 entry by a sbitmap and adapt code accordingly.
6959 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6961 * brig-builtins.def: Change pure attributes to const.
6963 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6965 * config.gcc (default_gnu_indirect_function): Default to yes for
6966 sparc*-*-linux* with glibc.
6968 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6970 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6971 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6972 when creating .init_array and .fini_array sections with priority
6975 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6978 * config/aarch64/aarch64.c
6979 (aarch64_builtin_support_vector_misalignment): Always return false
6980 when misalignment is unknown.
6982 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6984 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6985 this function to return false if the definition used by the swap
6986 instruction is artificial, or if the memory address from which the
6987 constant value is loaded is not represented by a base address held
6988 in a register or if the base address register is a frame or stack
6989 pointer. Additionally, return false if the base address of the
6990 loaded constant is a SYMBOL_REF but is not considered to be a
6992 (replace_swapped_load_constant): New function.
6993 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6994 loaded constant vector with a load of a swapped constant vector.
6996 2017-09-27 Carl Love <cel@us.ibm.com>
6998 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6999 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7000 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7003 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7005 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7006 first, always call autopref_rank_data otherwise.
7008 2017-09-27 Richard Biener <rguenther@suse.de>
7010 * graphite-scop-detection.c (find_scop_parameters): Move
7011 loop bound handling ...
7012 (gather_bbs::before_dom_children): ... here, avoiding the need
7013 to build scop_info->loop_nest.
7014 (record_loop_in_sese): Remove.
7015 * sese.h (sese_info_t::loop_nest): Remove.
7016 * sese.c (new_sese_info): Do not allocate loop_nest.
7017 (free_sese_info): Do not free loop_nest.
7019 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7022 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7023 lhs from calls if the lhs has addressable type.
7025 2017-09-27 Richard Biener <rguenther@suse.de>
7027 * graphite.h (scop::max_alias_set): New member.
7028 * graphite-scop-detection.c: Remove references to non-existing
7029 --param in comments.
7030 (build_alias_sets): Record the maximum alias set used for drs.
7031 (build_scops): Support zero as unlimited for
7032 --param graphite-max-arrays-per-scop.
7033 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7035 (build_poly_sr_1): ... here. Compute alias set based on the
7036 maximum alias set used for drs rather than
7037 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7039 2017-09-27 Richard Biener <rguenther@suse.de>
7041 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7042 --param loop-block-tile-size=0 to disable tiling.
7044 2017-09-27 Richard Biener <rguenther@suse.de>
7046 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7047 (graphite-max-nb-scop-params): Document special value zero.
7048 * domwalk.h (dom_walker::STOP): New symbolical constant.
7049 (dom_walker::dom_walker): Add optional parameter for bb to
7051 (dom_walker::~dom_walker): Declare.
7052 (dom_walker::before_dom_children): Document STOP return value.
7053 (dom_walker::m_user_bb_to_rpo): New member.
7054 (dom_walker::m_bb_to_rpo): Likewise.
7055 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7056 mapping here if not provided by the user.
7057 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7058 provided by the user.
7059 (dom_walker::STOP): Define.
7060 (dom_walker::walk): Do not compute bb to RPO mapping here.
7061 Support STOP return value from before_dom_children to stop
7063 * graphite-optimize-isl.c (optimize_isl): If the schedule
7064 is the same still generate code if -fgraphite-identity
7065 or -floop-parallelize-all are given.
7066 * graphite-scop-detection.c: Include cfganal.h.
7067 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7069 (gather_bbs::before_dom_children): Return STOP for BBs
7071 (build_scops): Compute bb to RPO mapping and pass it to
7072 the domwalk. Treat --param graphite-max-nb-scop-params=0
7073 as not limiting the number of params.
7074 * graphite.c (graphite_initialize): Remove limit on the
7075 number of basic-blocks in a function.
7076 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7077 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7078 default value of 10.
7080 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7082 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7083 Adjust code to eliminate needing to do the shift right 32-bits
7084 operation after XSCVDPSPN.
7086 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7088 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7089 ((X / Y) != 0 -> X >= Y): Likewise.
7091 2017-09-26 Carl Love <cel@us.ibm.com>
7093 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7094 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7095 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7096 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7097 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7098 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7099 definitions and overloading.
7100 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7101 statement for P9V_BUILTIN_XST_LEN_R.
7102 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7103 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7104 define_expand and define_insn for the instructions and builtins.
7105 * doc/extend.texi: Update the built-in documentation file for the new
7107 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7108 define_insn for the instructions
7110 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7113 * gcc/config/netbsd-protos.h: New file.
7114 * gcc/config/netbsd.c: New file.
7115 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7116 * gcc/config/t-netbsd: New file.
7117 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7118 (tmake_file) Add t-netbsd.
7119 (extra_objs) Add netbsd.o.
7121 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7125 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7127 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7129 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7130 sign extension from a vector register to a GPR by doing a 32-bit
7131 direct move and then an EXTSW.
7132 (extendsi<mode>2 splitter): Likewise.
7133 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7134 right or vector extract after doing XSCVDPSPN. Use
7135 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7137 (movdi_from_sf_zero_ext): Likewise.
7138 (reload_gpr_from_vsxsf): Likewise.
7139 (p8_mfvsrd_4_disf): Delete, no longer used.
7140 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7141 then needing to move the SFmode to a GPR to use the XSCVDPSP
7142 instruction instead of FRSP and XSCVDPSPN.
7143 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7144 it is adjacent to the other XSCVSPDP insns.
7145 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7146 SFmode to be in traditional Altivec registers.
7147 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7148 (vsx_xscvspdpn): Likewise.
7149 (vsx_xscvspdpn_scalar): Likewise.
7151 2017-09-26 Martin Jambor <mjambor@suse.cz>
7153 * tree-sra.c (compare_access_positions): Put integral types first,
7154 stabilize sorting of integral types, remove conditions putting
7155 non-full-precision integers last.
7156 (sort_and_splice_var_accesses): Disable scalarization if a
7157 non-integert would be represented by a non-full-precision integer.
7159 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7161 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7162 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7163 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7164 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7165 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7166 conditionals inside the function instead of around it. Call
7167 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7168 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7170 2017-09-26 Richard Biener <rguenther@suse.de>
7172 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7174 (scop_detection::build_scop_breadth): ... this. Removed.
7175 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7176 (scop_detection::harmful_stmt_in_bb): Likewise.
7177 (scop_detection::graphite_can_represent_stmt): Likewise.
7178 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7179 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7180 (scop_detection::can_represent_loop_1): ... this. Removed.
7181 (scop_detection::harmful_loop_in_region): Simplify after inlining
7182 the above and remove more quadraticness.
7183 (build_scops): Adjust.
7184 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7187 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7190 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7191 REGNO (base) == SP_REG if base is a REG.
7194 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7195 if it is different SSA_NAME.
7196 (optimize_range_tests_cmp_bitwise): New function.
7197 (optimize_range_tests): Call it.
7199 2017-09-26 Richard Biener <rguenther@suse.de>
7201 PR tree-optimization/82321
7202 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7203 for the def being inside the loop.
7205 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7207 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7209 * config/s390/s390-builtins.def: Fix constraint on op4.
7211 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7213 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7214 independent expanders.
7215 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7216 ("vec_ordered", "vec_unordered"): New expanders.
7218 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7220 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7223 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7225 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7226 vec_unpacks_lo_v16qi.
7227 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7229 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7231 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7232 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7233 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7235 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7237 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7239 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7241 ("*vec_slb<mode>"): New insn pattern.
7242 ("vec_shr_<mode>"): New expander.
7243 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7244 and force the shift count operand to V16QImode.
7245 ("vec_srb<mode>"): Set shift count mode to V16QI.
7247 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7249 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7250 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7251 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7253 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7256 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7258 2017-09-26 Richard Biener <rguenther@suse.de>
7260 PR tree-optimization/82320
7261 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7264 2017-09-25 Jeff Law <law@redhat.com>
7266 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7267 prototype for new argument.
7268 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7269 mostly extracted from rs6000_emit_allocate_stack.
7270 (rs6000_emit_probe_stack_range_stack_clash): New function.
7271 (rs6000_emit_allocate_stack): Call
7272 rs6000_emit_probe_stack_range_stack_clash as needed.
7273 (rs6000_emit_probe_stack_range): Add additional argument
7274 to call to gen_probe_stack_range{si,di}.
7275 (output_probe_stack_range): New.
7276 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7277 (output_probe_stack_range_stack_clash): New.
7278 (rs6000_emit_prologue): Emit notes into dump file as requested.
7279 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7280 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7281 Add additional operand and pass it to output_probe_stack_range.
7283 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7285 PR tree-optimization/82163
7286 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7287 (checking_verify_loop_closed_ssa): New parameter.
7288 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7289 (check_loop_closed_ssa_stmt): Delete.
7290 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7291 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7292 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7295 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7297 * brig-builtins.def: Treat HSAIL barrier builtins as
7298 setjmp/longjump style functions.
7300 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7302 * target.def (constant_alignment): New hook.
7303 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7304 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7305 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7306 * doc/tm.texi: Regenerate.
7307 * targhooks.h (default_constant_alignment): Declare.
7308 (constant_alignment_word_strings): Likewise.
7309 * targhooks.c (default_constant_alignment): New function.
7310 (constant_alignment_word_strings): Likewise.
7311 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7312 instead of CONSTANT_ALIGNMENT.
7313 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7314 (force_const_mem): Likewise.
7315 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7316 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7317 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7318 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7319 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7321 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7322 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7323 constant_alignment_word_strings.
7324 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7325 (CONSTANT_ALIGNMENT): Likewise.
7326 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7327 (arm_constant_alignment): New function.
7328 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7329 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7330 constant_alignment_word_strings.
7331 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7332 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7333 constant_alignment_word_strings.
7334 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7335 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7336 (cris_constant_alignment): New function.
7337 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7338 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7339 (epiphany_constant_alignment): New function.
7340 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7341 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7342 constant_alignment_word_strings.
7343 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7344 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7345 constant_alignment_word_strings.
7346 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7347 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7348 constant_alignment_word_strings.
7349 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7350 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7351 * config/i386/i386.c (ix86_constant_alignment): Make static.
7352 Use the same interface as the target hook.
7353 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7354 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7355 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7356 constant_alignment_word_strings.
7357 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7358 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7359 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7360 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7361 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7362 constant_alignment_word_strings.
7363 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7364 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7365 constant_alignment_word_strings.
7366 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7367 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7368 constant_alignment_word_strings.
7369 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7370 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7372 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7373 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7374 * config/mips/mips.c (mips_constant_alignment): New function.
7375 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7376 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7377 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7378 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7379 (mmix_constant_alignment): Make static. Use the same interface
7381 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7382 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7383 constant_alignment_word_strings.
7384 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7385 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7386 constant_alignment_word_strings.
7387 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7388 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7389 constant_alignment_word_strings.
7390 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7391 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7392 (rs6000_constant_alignment): New function.
7393 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7394 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7395 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7396 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7397 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7398 (rs6000_constant_alignment): New function.
7399 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7400 * config/s390/s390.c (s390_constant_alignment): New function.
7401 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7402 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7403 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7404 constant_alignment_word_strings.
7405 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7406 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7407 (sparc_constant_alignment): New function.
7408 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7409 * config/spu/spu.c (spu_constant_alignment): New function.
7410 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7411 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7412 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7413 constant_alignment_word_strings.
7414 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7415 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7416 constant_alignment_word_strings.
7417 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7418 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7419 constant_alignment_word_strings.
7420 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7421 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7422 (visium_constant_alignment): New function.
7423 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7424 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7425 (xtensa_constant_alignment): New function.
7426 * system.h (CONSTANT_ALIGNMENT): Poison.
7428 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7430 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7431 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7432 (rs6000_builtin_valid_without_lhs): New helper function.
7433 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7434 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7436 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7438 * target.h (vec_perm_indices): Use unsigned short rather than
7440 (auto_vec_perm_indices): Likewise.
7441 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7442 Use unsigned int rather than unsigned char.
7443 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7445 2017-09-25 Richard Biener <rguenther@suse.de>
7447 * cfgloop.h (sort_sibling_loops): Declare.
7448 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7449 (sort_sibling_loops): New function sorting the sibling loop list
7451 * graphite.c (graphite_transform_loops): Sort sibling loops.
7453 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7455 * target.def (vec_perm_const_ok): Change sel parameter to
7457 * optabs-query.c (can_vec_perm_p): Update accordingly.
7458 * doc/tm.texi: Regenerate.
7459 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7460 auto_vec_perm_indices and remove separate nelt field.
7461 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7462 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7463 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7464 (aarch64_expand_vec_perm_const): Update accordingly.
7465 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7466 to vec_perm_indices.
7467 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7468 auto_vec_perm_indices and remove separate nelt field.
7469 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7470 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7471 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7473 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7474 to vec_perm_indices.
7475 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7476 sel to vec_perm_indices.
7477 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7478 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7479 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7481 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7484 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7487 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7488 on the FUNCTION_DECL function context if it has a DIE that is a
7491 2017-09-25 Richard Biener <rguenther@suse.de>
7493 PR tree-optimization/82285
7494 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7497 2017-09-25 Tom de Vries <tom@codesourcery.com>
7501 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7504 2017-09-25 Richard Biener <rguenther@suse.de>
7506 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7507 ISL errors other than isl_error_quota happen. Dump if the
7508 schedule is the same.
7509 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7510 errors instead of aborting inside ISL.
7512 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7515 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7516 of libgcc_eh for m64.
7517 * config/i386/darwin64.h: Likewise.
7519 2017-09-25 Richard Biener <rguenther@suse.de>
7522 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7523 attribute for incomplete types nor twice for complete ones.
7525 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7528 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7529 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7532 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7535 * config/i386/i386.opt (mprefer-avx256): Use
7536 ix86_target_flags variable.
7537 * config/i386/i386.c (ix86_target_string): Move
7538 -mprefer-avx256 to flag2_opts.
7540 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7543 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7544 and x != -1 | y != -1 into (x & y) != -1.
7546 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7548 * config.gcc: Add new case statement to set
7549 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7550 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7551 s390x-*-linux* case statements. Added aarch64 to the list of
7552 supported architectures.
7554 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7556 PR tree-optimization/82289
7557 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7558 STMT_VINFO_RELEVANT_P.
7560 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7561 Alan Hayward <alan.hayward@arm.com>
7562 David Sherwood <david.sherwood@arm.com>
7564 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7565 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7566 (extract_range_from_binary_expr_1): Don't call
7567 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7569 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7570 Alan Hayward <alan.hayward@arm.com>
7571 David Sherwood <david.sherwood@arm.com>
7573 * target.def (preferred_vector_alignment): New hook.
7574 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7576 * doc/tm.texi: Regenerate.
7577 * targhooks.h (default_preferred_vector_alignment): Declare.
7578 * targhooks.c (default_preferred_vector_alignment): New function.
7579 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7581 (DR_TARGET_ALIGNMENT): New macro.
7582 (aligned_access_p): Update commentary.
7583 (vect_known_alignment_in_bytes): New function.
7584 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7586 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7587 Calculate the misalignment based on the target alignment rather than
7589 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7590 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7591 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7592 the target alignment, rather than masking the element misalignment
7593 with the number of elements in a vector. Also use the target
7594 alignment when calculating the maximum number of peels.
7595 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7596 instead of TYPE_ALIGN_UNIT.
7597 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7598 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7599 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7600 (vect_create_data_ref_ptr): Likewise.
7601 (vect_setup_realignment): Realign by ANDing with
7602 -DR_TARGET_MISALIGNMENT.
7603 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7604 the number of peels based on DR_TARGET_ALIGNMENT.
7605 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7606 with the guaranteed alignment boundary when deciding whether
7608 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7609 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7610 (ensure_base_align): Remove stmt_info parameter. Get the
7611 target base alignment from DR_TARGET_ALIGNMENT.
7612 (vectorizable_store): Update call accordingly. Interpret
7613 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7615 (vectorizable_load): Likewise.
7617 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7618 Alan Hayward <alan.hayward@arm.com>
7619 David Sherwood <david.sherwood@arm.com>
7621 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7622 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7623 (vect_enhance_data_refs_alignment): Likewise.
7625 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7627 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7628 error. Only quit immediately if parsing is complete.
7629 (BEGIN): Initialize fatal_err and parse_done.
7630 (begin fpu, end fpu): Check number of arguments.
7631 (begin arch, end arch): Likewise.
7632 (begin cpu, end cpu): Likewise.
7633 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7634 (optalias): Likewise.
7636 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7638 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7639 * config/arm/arm-isa.h: Delete. Move definitions to ...
7640 * arm-cpus.in: ... here. Use new feature and fgroup values.
7641 * config/arm/arm.c (arm_option_override): Use lower case for feature
7643 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7644 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7645 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7647 (print_isa_bits_for): New function.
7648 (gen_isa): New function.
7649 (gen_comm_data): Use print_isa_bits_for.
7650 (define feature): New keyword.
7651 (define fgroup): New keyword.
7652 * config/arm/t-arm (TM_H): Remove.
7653 (GTM_H): Add arm-isa.h.
7654 (arm-isa.h): Add rule to generate file.
7655 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7656 case for feature bit names.
7658 2017-09-22 Richard Biener <rguenther@suse.de>
7660 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7662 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7663 print of no dependency loops ...
7664 * graphite.c (graphite_transform_loops): ... here.
7665 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7667 (same_close_phi_node, remove_duplicate_close_phi,
7668 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7669 (canonicalize_loop_closed_ssa): ... here and simplify.
7670 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7671 (optimize_isl): Use dump_printf_loc to tell when we stopped
7672 optimizing because of an ISL timeout.
7674 2017-09-22 Richard Biener <rguenther@suse.de>
7676 PR tree-optimization/82291
7677 * tree-if-conv.c (predicate_mem_writes): Make sure to
7678 remove writes in blocks predicated with false.
7680 2017-09-22 Richard Biener <rguenther@suse.de>
7682 * sese.c: Include cfganal.h.
7683 (if_region_set_false_region): Remove.
7684 (create_if_region_on_edge): Likewise.
7685 (move_sese_in_condition): Re-implement without destroying
7688 2017-09-22 Richard Biener <rguenther@suse.de>
7690 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7691 Verify both BBs contain loop PHI nodes before dispatching to
7693 (graphite_regenerate_ast_isl): Do not recompute dominators,
7694 do not verify three times. Restructure for clarity.
7695 * graphite-scop-detection.c (same_close_phi_node,
7696 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7697 defined_in_loop_p, canonicalize_loop_closed_ssa,
7698 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7699 checking and SSA rewrite, move to ...
7700 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7701 (graphite_initialize): Do not pass in ctx, do not reset the
7702 SCEV cache, compute only dominators.
7703 (graphite_transform_loops): Allocate ISL ctx after
7704 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7705 Maintain post-dominators only around build_scops.
7706 * sese.c (if_region_set_false_region): Make static. Free
7707 and recompute dominators.
7708 (move_sese_in_condition): Assert we don't get called with
7709 post-dominators computed.
7710 * sese.h (if_region_set_false_region): Remove.
7712 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7714 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7715 mode attribute for TARGET_AVX512VL.
7717 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7719 * config/i386/i386.opt (mprefer-avx256): New option.
7720 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7722 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7723 for TARGET_PREFER_AVX256.
7724 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7726 2017-09-21 Jeff Law <law@redhat.com>
7728 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7729 Fix dump output if the only stack space is for pushed registers.
7731 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7733 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7736 2017-09-21 Martin Sebor <msebor@redhat.com>
7739 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7740 code (in C++) or code that triggers warnings.
7742 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7744 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7746 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7748 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7749 * sched-rgn.c: Ditto.
7750 * sel-sched-ir.c: Ditto.
7752 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7754 * toplev.h (set_random_seed): Adjust return type.
7755 * toplev.c (init_local_tick): Move eager initialization of random_seed
7756 to get_random_seed. Adjust comment.
7757 (init_random_seed): Inline to get_random_seed, delete.
7758 (get_random_seed): Initialize random_seed lazily.
7759 (set_random_seed): Do not return previous value.
7760 (print_switch_value): Do not call get_random_seed.
7762 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7764 * cgraph.c (delete_function_version): New, broken out from...
7765 (cgraph_node::delete_function_version): ...here. Rename to
7766 cgraph_node::delete_function_version_by_decl. Update all uses.
7767 (cgraph_node::remove): Call delete_function_version.
7769 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7772 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7773 VAR_DECLs to which addressable non-volatile parameters are mapped
7774 and for id->retvar after the return value assignment. Clear
7775 id->retval and id->retbnd after inlining.
7777 2017-09-21 Richard Biener <rguenther@suse.de>
7779 PR tree-optimization/82276
7780 PR tree-optimization/82244
7781 * tree-vrp.c (build_assert_expr_for): Set
7782 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7784 (remove_range_assertions): Revert earlier change.
7786 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7789 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7791 2017-09-21 Richard Biener <rguenther@suse.de>
7793 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7794 Restore valid IL after code generation errors.
7795 * graphite.c (graphite_transform_loops): Diagnose code
7796 generation issues as MSG_MISSED_OPTIMIZATION and continue
7797 with processing SCOPs.
7799 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7800 Alan Hayward <alan.hayward@arm.com>
7801 David Sherwood <david.sherwood@arm.com>
7803 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7804 rather than choosing between plus_constant and gen_rtx_<CODE>.
7805 * expr.c (emit_push_insn): Likewise.
7806 (expand_expr_real_2): Likewise.
7808 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7809 Alan Hayward <alan.hayward@arm.com>
7810 David Sherwood <david.sherwood@arm.com>
7812 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7814 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7815 Alan Hayward <alan.hayward@arm.com>
7816 David Sherwood <david.sherwood@arm.com>
7818 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7819 calling tree_to_uhwi.
7821 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7822 Alan Hayward <alan.hayward@arm.com>
7823 David Sherwood <david.sherwood@arm.com>
7825 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7826 INTEGER_CST rather than a negative test for ADDR_EXPR.
7828 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7829 Alan Hayward <alan.hayward@arm.com>
7830 David Sherwood <david.sherwood@arm.com>
7832 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7833 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7835 2017-09-21 Richard Biener <rguenther@suse.de>
7837 PR tree-optimization/71351
7838 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7839 graphite_create_new_loop_guard): Remove, fold remaining parts
7841 (translate_isl_ast_node_for): ... here and simplify.
7843 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7846 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7847 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7848 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7849 alternative always use QI mode, for -Os imov (=R,R) alternative
7850 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7853 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7854 Jeff Law <law@redhat.com>
7856 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7857 (allocate_stack_space): New function, partially extracted from
7859 (s390_emit_prologue): Track offset to most recent stack probe.
7860 Code to allocate space moved into allocate_stack_space.
7861 Dump actions when no stack is allocated.
7862 (s390_prologue_plus_offset): New function.
7863 (s390_emit_stack_probe): Likewise.
7865 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7867 * common.opt (Wa, Wl, Wp, g, gz=): Add
7869 (gno-column-info): Remove.
7870 (gcolumn-info): Drop RejectNegative.
7872 (gno-record-gcc-switches): Remove.
7873 (grecord-gcc-switches): Drop RejectNegative.
7874 (gno-split-dwarf): Remove.
7875 (gsplit-dwarf): Drop RejectNegative.
7876 (gno-strict-dwarf): Remove.
7877 (gstrict-dwarf): Drop RejectNegative.
7878 * config/darwin.opt (gfull, gused): Add RejectNegative.
7879 * dwarf2out.c (gen_producer_string): Drop
7880 gno-record-gcc-switches handler.
7881 * optc-gen.awk: Add g to prefixes with negative forms.
7882 * opts-common.c (remapping_prefix_p): New.
7883 (find_opt): Check it.
7884 (generate_canonical_option): Test g prefix.
7885 (option_map): Add -gno- mapping.
7886 (add_misspelling_candidates): Check remapping_prefix_p.
7888 2017-09-20 Jeff Law <law@redhat.com>
7890 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7891 thinko in stack clash protection support.
7893 * explow.c (compute_stack_clash_protection_loop_data): Use
7894 CONST_INT_P instead of explicit test. Verify object is a
7895 CONST_INT_P before looking at INTVAL.
7896 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7897 instead of explicit test.
7899 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7902 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7903 address instead of to r1 and r11.
7905 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7907 * config.gcc: Support "knm".
7908 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7909 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7911 * config/i386/i386.c (m_KNM): Define.
7912 (processor_target_table): Add "knm".
7914 (ix86_option_override_internal): Add "knm".
7915 (ix86_issue_rate): Add PROCESSOR_KNM.
7916 (ix86_adjust_cost): Ditto.
7917 (ia32_multipass_dfa_lookahead): Ditto.
7918 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7919 (fold_builtin_cpu): Add M_INTEL_KNM.
7920 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7921 (processor_type): Add PROCESSOR_KNM.
7922 * config/i386/x86-tune.def: Add m_KNM.
7923 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7925 2017-09-20 Richard Biener <rguenther@suse.de>
7927 PR tree-optimization/80213
7928 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7929 are allowed in empty BBs as well.
7930 (canonicalize_loop_closed_ssa): Also look for other complex
7932 (scop_detection::get_sese): Include the loop-closed PHI block
7934 (scop_detection::merge_sese): Remove code adding extra blocks.
7935 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7936 (build_scops): Assert the final returned scop is invalid.
7938 2017-09-20 Richard Biener <rguenther@suse.de>
7940 PR tree-optimization/82264
7941 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7943 (vn_phi_lookup): Likewise.
7944 (vn_phi_insert): Likewise.
7946 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7948 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7949 that fits into uhwi or shwi, add DW_AT_const_value regardless
7950 of early_dwarf without going through RTL, using add_AT_unsigned
7953 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7954 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7955 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7956 (DEBUG_MACRO_SECTION): Likewise.
7957 (DEBUG_LTO_MACRO_SECTION): Likewise.
7958 (DEBUG_STR_DWO_SECTION): Likewise.
7959 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7960 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7961 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7962 (DEBUG_LTO_LINE_STR_SECTION): Define.
7963 (init_sections_and_labels): Initialize debug_line_str_section
7964 variable. Initialize debug_loc_section for -gdwarf-5 to
7965 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7967 2017-09-20 Richard Biener <rguenther@suse.de>
7969 * graphite-sese-to-poly.c (extract_affine): Properly handle
7970 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7972 2017-09-20 Richard Biener <rguenther@suse.de>
7974 PR tree-optimization/81373
7975 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7976 Force SESE live-out defs to be handled even if they are
7979 2017-09-19 Jeff Law <law@redhat.com>
7981 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7982 nothing for stack adjustments with REG_STACK_CHECK.
7983 * sched-deps.c (parse_add_or_inc): Reject insns with
7984 REG_STACK_CHECK from dependency breaking.
7985 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7986 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7987 * reg-notes.def (STACK_CHECK): New note.
7989 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7990 (ix86_expand_prologue): Dump stack clash info as needed.
7991 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7993 * function.c (dump_stack_clash_frame_info): New function.
7994 * function.h (dump_stack_clash_frame_info): Prototype.
7995 (enum stack_clash_probes): New enum.
7997 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7998 flag_stack_clash_protection.
7999 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8000 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8001 (arm_frame_pointer_required): Likewise.
8002 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8003 (ia64_expand_prologue): Likewise.
8004 * config/mips/mips.c (mips_expand_prologue): Likewise.
8005 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8006 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8007 (sparc_flat_expand_prologue): Likewise.
8008 * config/spu/spu.c (spu_expand_prologue): Likewise.
8010 * explow.c: Include "params.h".
8011 (anti_adjust_stack_and_probe_stack_clash): New function.
8012 (get_stack_check_protect): Likewise.
8013 (compute_stack_clash_protection_loop_data): Likewise.
8014 (emit_stack_clash_protection_loop_start): Likewise.
8015 (emit_stack_clash_protection_loop_end): Likewise.
8016 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8017 Use anti_adjust_stack_and_probe_stack_clash.
8018 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8019 (emit_stack_clash_protection_loop_start): Likewise.
8020 (emit_stack_clash_protection_loop_end): Likewise.
8021 * rtl.h (get_stack_check_protect): Prototype.
8022 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8023 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8024 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8026 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8028 * doc/tm.texi: Rebuilt.
8029 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8030 get_stack_check_protect.
8031 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8032 * config/arm/arm.c (arm_expand_prologue): Likewise.
8033 (arm_frame_pointer_required): Likewise.
8034 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8035 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8036 * config/mips/mips.c (mips_expand_prologue): Likewise.
8037 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8038 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8039 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8040 (sparc_flat_expand_prologue): Likewise.
8042 * common.opt (-fstack-clash-protection): New option.
8043 * flag-types.h (enum stack_check_type): Note difference between
8044 -fstack-check= and -fstack-clash-protection.
8045 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8046 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8047 * toplev.c (process_options): Issue warnings/errors for cases
8048 not handled with -fstack-clash-protection.
8049 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8050 (-fstack-check): Note additional problem with -fstack-check=generic.
8051 Note that -fstack-check is primarily for Ada and refer users
8052 to -fstack-clash-protection for stack-clash-protection.
8053 Document new params for stack clash protection.
8055 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8057 * config/i386/i386.c (ix86_split_long_move): Do not handle
8058 address used for LEA in a special way.
8060 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8062 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8064 2017-09-19 Martin Sebor <msebor@redhat.com>
8067 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8068 of incompatible types.
8070 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8072 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8073 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8074 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8075 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8077 2017-09-19 Richard Biener <rguenther@suse.de>
8079 PR tree-optimization/82244
8080 * tree-vrp.c (remove_range_assertions): Do not propagate
8081 a constant to abnormals but replace the assert with a copy.
8083 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8085 PR rtl-optimization/57878
8086 PR rtl-optimization/68988
8087 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8088 avoidance test involving non_reload_pseudos. Move frequency test
8089 below the general fragmentation avoidance test.
8091 2017-09-19 Richard Biener <rguenther@suse.de>
8093 PR tree-optimization/69728
8094 * graphite-sese-to-poly.c (schedule_error): New global.
8095 (add_loop_schedule): Handle empty domain by failing the
8097 (build_original_schedule): Handle schedule_error.
8099 2017-09-19 Richard Biener <rguenther@suse.de>
8101 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8102 Do not iterate to sibling loops but only to siblings of inner
8105 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8108 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8109 that the registers are different.
8111 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8113 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8114 to processor_model and "amdfam17h" to arch_names_table.
8115 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8117 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8120 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8122 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8123 Alan Hayward <alan.hayward@arm.com>
8124 David Sherwood <david.sherwood@arm.com>
8126 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8128 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8130 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8131 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8132 vect_schedule_slp_instance.
8133 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8134 Update call to vect_slp_analyze_node_operations. Simplify return
8136 (vect_slp_analyze_bb_1): Update call accordingly.
8137 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8138 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8139 (vect_schedule_slp): Update call accordingly.
8141 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8142 Alan Hayward <alan.hayward@arm.com>
8143 David Sherwood <david.sherwood@arm.com>
8145 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8146 with types that aren't in fact scalar.
8148 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8150 * tree-vect-slp.c (vect_record_max_nunits): New function,
8152 (vect_build_slp_tree_1): ...here.
8153 (vect_build_slp_tree_2): Call it for phis too.
8155 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8157 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8158 to vect_get_vec_def_for_operand when getting the mask operand.
8160 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8161 Alan Hayward <alan.hayward@arm.com>
8162 David Sherwood <david.sherwood@arm.com>
8164 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8167 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8168 Alan Hayward <alan.hayward@arm.com>
8169 David Sherwood <david.sherwood@arm.com>
8171 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8172 calculation for vector booleans.
8174 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8175 Alan Hayward <alan.hayward@arm.com>
8176 David Sherwood <david.sherwood@arm.com>
8178 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8180 (vect_transform_stmt): ...here.
8181 (vect_analyze_stmt): Use it instead of calling
8182 vectorizable_live_operation directly.
8184 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8186 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8187 non-SIMT targets in acc vector loops.
8189 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8191 * configure.ac: Add arc and check if assembler supports gdwarf2.
8192 * configure: Regenerate.
8194 2017-09-18 Richard Biener <rguenther@suse.de>
8196 PR tree-optimization/82220
8197 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8198 epilogue niters from the min_profitable_iters compute.
8200 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8203 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8204 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8205 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8207 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8210 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8211 switching to a new text section.
8213 2017-09-18 Richard Biener <rguenther@suse.de>
8215 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8217 (build_alias_set): Reject aliases with no access function.
8219 2017-09-18 Richard Biener <rguenther@suse.de>
8221 PR tree-optimization/79622
8222 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8224 (build_cross_bb_scalars_use): Likewise.
8226 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8228 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8230 2017-09-18 Alan Modra <amodra@gmail.com>
8233 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8234 stack_pointer_rtx for count 0. Update comments. Break up
8235 large rtl expression.
8237 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8239 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8240 Increase to 20 bytes.
8241 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8242 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8243 or avx version of the stub.
8245 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8248 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8249 compute the minimum stack alignment. Also update preferred stack
8250 boundary for leaf functions.
8252 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8254 PR tree-optimization/82228
8255 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8258 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8260 * common/config/nds32/nds32-common.c
8261 (nds32_option_optimization_table): Refine formatting.
8262 (nds32_option_optimization_table): Use -fsched-pressure and
8263 -fomit-frame-pointer for specific optimization level.
8265 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8267 * config/nds32/nds32.c: Refine formatting and comments.
8268 * config/nds32/nds32.h: Likewise.
8269 * config/nds32/nds32.md: Likewise.
8270 * config/nds32/nds32-cost.c: Likewise.
8271 * config/nds32/nds32-isr.c: Likewise.
8272 * config/nds32/nds32-md-auxiliary.c: Likewise.
8273 * config/nds32/nds32-multiple.md: Likewise.
8274 * config/nds32/nds32-predicates.c: Likewise.
8276 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8277 Jakub Jelinek <jakub@redhat.com>
8279 Add support for -std=c++2a.
8280 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8282 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8284 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8287 * doc/extend.texi (Common Function Attributes): Add
8288 references to ARM, AArch64, and S/390 specific attributes.
8289 (Function Specific Option Pragmas): Add AArch64 and S/390
8290 to list of back ends that support the target pragma.
8292 2017-09-15 Nathan Sidwell <nathan@acm.org>
8294 * doc/standards.texi: Fix C++17 description. Update URLs for
8297 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8299 * common.opt (Wcast-align=strict): New warning option.
8300 * doc/invoke.texi: Document -Wcast-align=strict.
8302 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8304 * cgraph.h (cgraph_thunk_info): Add comments.
8305 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8306 assert for VIRTUAL_* arguments stricter.
8308 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8310 PR tree-optimization/71026
8311 * match.pd: Move RDIV patterns from fold-const.c
8312 * fold-const.c (distribute_real_division): Removed.
8313 (fold_binary_loc): Remove calls to distribute_real_divison.
8315 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8317 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8318 c++1z and gnu++1z as deprecated. Change other references to
8319 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8320 Change -Wc++1z-compat to -Wc++17-compat.
8321 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8322 * dwarf2out.c (highest_c_language): Handle C++17.
8323 (gen_compile_unit_die): Likewise.
8325 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8327 PR rtl-optimization/82192
8328 * combine.c (make_extraction): Don't look through non-paradoxical
8329 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8332 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8333 Alan Hayward <alan.hayward@arm.com>
8334 David Sherwood <david.sherwood@arm.com>
8336 * target.def (function_arg_offset): New hook.
8337 * targhooks.h (default_function_arg_offset): Declare.
8338 * targhooks.c (default_function_arg_offset): New function.
8339 * function.c (locate_and_pad_parm): Use
8340 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8341 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8342 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8343 * doc/tm.texi: Regenerate.
8344 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8345 * config/spu/spu.c (spu_function_arg_offset): New function.
8346 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8347 * system.h (FUNCTION_ARG_OFFSET): Poison.
8349 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8350 Alan Hayard <alan.hayward@arm.com>
8351 David Sherwood <david.sherwood@arm.com>
8353 * target.def (truly_noop_truncation): New hook.
8354 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8355 than TRULY_NOOP_TRUNCATION.
8356 * hooks.h (hook_bool_uint_uint_true): Declare.
8357 * hooks.c (hook_bool_uint_uint_true): New function.
8358 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8359 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8360 * doc/tm.texi: Regenerate.
8361 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8362 rather than TRULY_NOOP_TRUNCATION in comments.
8363 (simplify_comparison): Likewise.
8364 (record_truncated_value): Likewise.
8365 * expmed.c (extract_bit_field_1): Likewise.
8366 (extract_split_bit_field): Likewise.
8367 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8368 instead of TRULY_NOOP_TRUNCATION.
8369 * function.c (assign_parm_setup_block): Likewise.
8370 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8371 * rtlhooks.c: Include target.h.
8372 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8373 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8374 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8375 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8376 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8377 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8378 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8379 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8380 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8381 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8382 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8383 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8384 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8385 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8386 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8387 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8388 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8389 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8390 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8391 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8392 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8393 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8394 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8395 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8396 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8397 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8398 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8399 rather than TRULY_NOOP_TRUNCATION in comments.
8400 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8401 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8402 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8403 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8404 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8405 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8406 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8407 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8408 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8409 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8410 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8411 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8412 rather than TRULY_NOOP_TRUNCATION in comments.
8413 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8414 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8415 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8416 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8417 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8418 TRULY_NOOP_TRUNCATION condition.
8419 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8420 (TRULY_NOOP_TRUNCATION): Delete.
8421 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8422 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8423 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8424 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8425 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8426 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8427 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8428 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8429 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8430 rather than TRULY_NOOP_TRUNCATION in comments.
8431 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8432 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8433 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8434 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8435 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8436 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8438 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8441 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8442 (*cmp_ior): Likewise.
8443 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8444 (*ior_scc_scc_cmp): Likewise.
8445 (*and_scc_scc): Likewise.
8446 (*and_scc_scc_cmp): Likewise.
8448 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8449 Alan Hayard <alan.hayward@arm.com>
8450 David Sherwood <david.sherwood@arm.com>
8452 * target.def (can_change_mode_class): New hook.
8453 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8454 (hard_regno_nregs): Likewise.
8455 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8456 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8457 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8458 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8459 (LOAD_EXTEND_OP): Update accordingly.
8460 * doc/tm.texi: Regenerate.
8461 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8462 CANNOT_CHANGE_MODE_CLASS.
8463 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8464 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8465 * combine.c (simplify_set): Update accordingly.
8466 * emit-rtl.c (validate_subreg): Likewise.
8467 * recog.c (general_operand): Likewise.
8468 * regcprop.c (mode_change_ok): Likewise.
8469 * reload1.c (choose_reload_regs): Likewise.
8470 (inherit_piecemeal_p): Likewise.
8471 * rtlanal.c (simplify_subreg_regno): Likewise.
8472 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8473 instead of CANNOT_CHANGE_MODE_CLASS.
8474 (reload_cse_simplify_operands): Likewise.
8475 * reload.c (push_reload): Use targetm.can_change_mode_class
8476 instead of CANNOT_CHANGE_MODE_CLASS.
8477 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8478 REG_CANNOT_CHANGE_MODE_P.
8479 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8480 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8481 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8482 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8483 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8484 (arm_can_change_mode_class): New function.
8485 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8486 than CANNOT_CHANGE_MODE_CLASS in comments.
8487 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8488 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8489 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8490 (ix86_can_change_mode_class): ...this new function, inverting the
8491 sense of the return value.
8492 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8493 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8494 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8495 (ia64_can_change_mode_class): New function.
8496 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8497 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8498 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8499 (m32c_can_change_mode_class): ...this new function, inverting the
8500 sense of the return value.
8501 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8502 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8503 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8504 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8505 (mips_can_change_mode_class): ...this new function, inverting the
8506 sense of the return value.
8507 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8508 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8509 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8510 (msp430_can_change_mode_class): New function.
8511 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8512 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8513 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8514 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8515 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8516 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8517 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8518 (pa_cannot_change_mode_class): Replace with...
8519 (pa_can_change_mode_class): ...this new function, inverting the
8520 sense of the return value.
8521 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8522 than CANNOT_CHANGE_MODE_CLASS in comments.
8523 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8524 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8525 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8526 (pdp11_cannot_change_mode_class): Replace with...
8527 (pdp11_can_change_mode_class): ...this new function, inverting the
8528 sense of the return value.
8529 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8530 * config/powerpcspe/powerpcspe-protos.h
8531 (rs6000_cannot_change_mode_class_ptr): Delete.
8532 * config/powerpcspe/powerpcspe.c
8533 (rs6000_cannot_change_mode_class_ptr): Delete.
8534 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8535 (rs6000_option_override_internal): Assign to
8536 targetm.can_change_mode_class instead of
8537 rs6000_cannot_change_mode_class_ptr.
8538 (rs6000_cannot_change_mode_class): Replace with...
8539 (rs6000_can_change_mode_class): ...this new function, inverting the
8540 sense of the return value.
8541 (rs6000_debug_cannot_change_mode_class): Replace with...
8542 (rs6000_debug_can_change_mode_class): ...this new function.
8543 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8544 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8545 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8546 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8547 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8549 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8550 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8551 (rs6000_option_override_internal): Assign to
8552 targetm.can_change_mode_class instead of
8553 rs6000_cannot_change_mode_class_ptr.
8554 (rs6000_cannot_change_mode_class): Replace with...
8555 (rs6000_can_change_mode_class): ...this new function, inverting the
8556 sense of the return value.
8557 (rs6000_debug_cannot_change_mode_class): Replace with...
8558 (rs6000_debug_can_change_mode_class): ...this new function.
8559 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8560 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8561 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8562 (s390_can_change_mode_class): ...this new function, inverting the
8563 sense of the return value.
8564 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8565 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8566 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8567 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8568 (sh_cannot_change_mode_class): Replace with...
8569 (sh_can_change_mode_class): ...this new function, inverting the
8570 sense of the return value.
8571 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8572 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8573 (sparc_can_change_mode_class): New function.
8574 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8575 * config/spu/spu.c (spu_can_change_mode_class): New function.
8576 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8577 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8578 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8579 (visium_can_change_mode_class): New function.
8580 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8582 2017-09-15 Richard Biener <rguenther@suse.de>
8584 PR tree-optimization/82217
8585 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8586 but not undefined case.
8588 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8591 * postreload.c (reload_cse_simplify_operands): Skip
8592 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8594 2017-09-15 Richard Biener <rguenther@suse.de>
8596 PR tree-optimization/68823
8597 * graphite-scop-detection.c (build_alias_set): If we have a
8598 possible dependence check whether we can handle them by just
8599 looking at the DRs DR_ACCESS_FNs.
8600 (build_scops): If build_alias_set fails, fail the SCOP.
8602 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8604 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8605 to support float128 built-in functions that require the ISA 3.0
8607 (BU_FLOAT128_3_HW): Likewise.
8608 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8610 (FMAF128): Likewise.
8612 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8613 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8614 floating point instructions.
8615 (rs6000_invalid_builtin): Likewise.
8616 (rs6000_builtin_mask_names): Likewise.
8617 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8618 (RS6000_BTM_FLOAT128_HW): Likewise.
8619 (RS6000_BTM_COMMON): Likewise.
8620 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8622 * doc/extend.texi (RS/6000 built-in functions): Document the
8623 IEEE 128-bit floating point square root and fused multiply-add
8626 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8628 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8629 reg (r2) isn't in the set of registers defined in the prologue.
8631 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8632 Alan Hayward <alan.hayward@arm.com>
8633 David Sherwood <david.sherwood@arm.com>
8635 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8636 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8637 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8638 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8639 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8641 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8642 max_vectorization_factor.
8643 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8645 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8646 Alan Hayward <alan.hayward@arm.com>
8647 David Sherwood <david.sherwood@arm.com>
8649 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8650 (vect_worthwhile_without_simd_p): Declare.
8651 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8652 (vectorizable_reduction): Use it.
8653 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8654 (vectorizable_operation): Likewise.
8656 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8657 Alan Hayward <alan.hayward@arm.com>
8658 David Sherwood <david.sherwood@arm.com>
8660 * tree-vectorizer.h (vect_get_num_copies): New function.
8661 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8662 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8663 (vectorizable_induction): Likewise.
8664 (vectorizable_live_operation): Likewise.
8665 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8666 (vectorizable_bswap): Likewise.
8667 (vectorizable_call): Likewise.
8668 (vectorizable_conversion): Likewise.
8669 (vectorizable_assignment): Likewise.
8670 (vectorizable_shift): Likewise.
8671 (vectorizable_operation): Likewise.
8672 (vectorizable_store): Likewise.
8673 (vectorizable_load): Likewise.
8674 (vectorizable_condition): Likewise.
8675 (vectorizable_comparison): Likewise.
8676 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8678 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8679 Alan Hayward <alan.hayward@arm.com>
8680 David Sherwood <david.sherwood@arm.com>
8682 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8683 of vect_init_vector.
8685 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8686 Alan Hayward <alan.hayward@arm.com>
8687 David Sherwood <david.sherwood@arm.com>
8689 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8690 an inline wrapper that provides a location.
8691 (gimple_build_vector): Likewise.
8692 * gimple-fold.c (gimple_build_vector_from_val): New function.
8693 (gimple_build_vector): Likewise.
8694 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8695 functions to build the initial value. Always return a gimple value.
8696 (get_initial_defs_for_reduction): Likewise. Only compute
8698 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8699 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8700 (vectorizable_induction): Use gimple_build_vector rather than
8703 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8704 Alan Hayward <alan.hayward@arm.com>
8705 David Sherwood <david.sherwood@arm.com>
8707 * target.h (vec_perm_indices): New typedef.
8708 (auto_vec_perm_indices): Likewise.
8709 * optabs-query.h: Include target.h
8710 (can_vec_perm_p): Take a vec_perm_indices *.
8711 * optabs-query.c (can_vec_perm_p): Likewise.
8712 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8713 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8714 * tree-vect-generic.c (lower_vec_perm): Likewise.
8715 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8716 (vect_grouped_load_supported): Likewise.
8717 (vect_shift_permute_load_chain): Likewise.
8718 (vect_permute_store_chain): Use auto_vec_perm_indices.
8719 (vect_permute_load_chain): Likewise.
8720 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8721 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8722 Update uses of can_vec_perm_p.
8723 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8724 mode with a number of elements. Take a vec_perm_indices *.
8725 (vect_create_epilog_for_reduction): Update accordingly.
8726 Use auto_vec_perm_indices.
8727 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8728 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8729 (vect_transform_slp_perm_load): Likewise.
8730 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8731 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8732 (vect_gen_perm_mask_checked): Likewise.
8733 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8734 (vect_gen_perm_mask_checked): Likewise.
8735 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8736 (vectorizable_store): Likewise.
8737 (vectorizable_load): Likewise.
8738 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8739 (vectorizable_bswap): Likewise.
8741 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8742 Alan Hayward <alan.hayward@arm.com>
8743 David Sherwood <david.sherwood@arm.com>
8745 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8746 * tree.c (build_vector): Likewise.
8747 (build_vector_from_ctor): Update accordingly.
8748 (build_vector_from_val): Likewise.
8749 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8750 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8751 * tree-vect-generic.c (add_rshift): Likewise.
8752 (expand_vector_divmod): Likewise.
8753 (optimize_vector_constructor): Likewise.
8754 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8755 (vect_transform_slp_perm_load): Likewise.
8756 (vect_schedule_slp_instance): Likewise.
8757 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8758 (vectorizable_call): Likewise.
8759 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8760 * expmed.c (make_tree): Likewise.
8761 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8762 a vector passed to build_vector.
8763 (fold_convert_const): Likewise.
8764 (exact_inverse): Likewise.
8765 (fold_ternary_loc): Likewise.
8766 (fold_relational_const): Likewise.
8767 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8768 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8769 (const_unop): Likewise. Store the reduction accumulator in a
8770 variable rather than an array.
8771 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8772 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8773 the new vector, rather than constructing it after the input arrays.
8774 (native_interpret_vector): Use auto_vec<tree> when building
8775 a vector passed to build_vector. Add elements in order.
8776 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8777 auto_vec<tree> when building a vector passed to build_vector.
8778 (vect_create_epilog_for_reduction): Likewise.
8779 (vectorizable_induction): Likewise.
8780 (get_initial_def_for_reduction): Likewise. Fix indentation of
8782 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8784 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8785 passed to build_vector.
8787 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8788 Alan Hayward <alan.hayward@arm.com>
8789 David Sherwood <david.sherwood@arm.com>
8791 * tree-core.h (tree_base::u): Add an "nelts" field.
8792 (tree_vector): Use VECTOR_CST_NELTS as the length.
8793 * tree.c (tree_size): Likewise.
8794 (make_vector): Initialize VECTOR_CST_NELTS.
8795 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8796 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8797 TYPE_VECTOR_SUBPARTS.
8798 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8799 as the number of units, setting it from VECTOR_CST_NELTS.
8800 (const_vector_from_tree): Likewise.
8801 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8802 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8803 (fold_negate_expr_1): Likewise.
8804 (fold_convert_const): Likewise.
8805 (const_binop): Likewise. Differentiate the number of output and
8807 (const_unop): Likewise.
8808 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8809 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8810 in cases that did the opposite.
8812 2017-09-14 Richard Biener <rguenther@suse.de>
8814 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8817 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8819 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8821 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8824 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8825 if and where to split a bb, except for splitting before debug insn
8826 sequences followed by non-label real insn. Delete debug insns
8827 in between basic blocks.
8829 * combine.c (make_compound_operation_int): Formatting fixes.
8831 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8832 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8833 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8834 * config/sol2.h (LINK_EH_SPEC): Likewise.
8835 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8836 * config/s390/linux.h (LINK_SPEC): Likewise.
8837 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8838 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8839 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8840 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8841 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8842 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8843 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8844 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8846 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8848 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8850 (ENDFILE_LINUX_SPEC): Likewise.
8851 (LINK_EH_SPEC): Likewise.
8852 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8853 (LINK_OS_LINUX_SPEC32): Likewise.
8854 (LINK_OS_LINUX_SPEC64): Likewise.
8855 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8856 (LINK_OS_LINUX_SPEC): Likewise.
8858 2017-09-13 Martin Liska <mliska@suse.cz>
8861 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8862 CASE_HIGH is NULL_TREE.
8864 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8865 Alan Hayward <alan.hayward@arm.com>
8866 David Sherwood <david.sherwood@arm.com>
8868 * target.def (secondary_memory_needed): New hook.
8869 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8870 instead of SECONDARY_MEMORY_NEEDED.
8871 (secondary_memory_needed_mode): Likewise.
8872 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8873 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8874 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8875 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8876 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8877 * doc/tm.texi: Regenerate.
8878 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8879 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8880 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8881 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8882 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8883 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8884 mode argument first and change the reg_class arguments to reg_class_t.
8885 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8886 Make static. Update the call to inline_secondary_memory_needed.
8887 (ix86_register_move_cost): Update the call to
8888 inline_secondary_memory_needed.
8889 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8890 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8892 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8893 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8895 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8896 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8897 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8898 and match hook interface. Add comment from mips.h.
8899 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8900 * config/mmix/mmix.md (truncdfsf2): Refer to
8901 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8903 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8904 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8905 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8906 (pa_secondary_memory_needed): New function.
8907 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8908 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8909 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8910 (pdp11_secondary_memory_needed): Make static and match hook interface.
8911 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8912 * config/powerpcspe/powerpcspe-protos.h
8913 (rs6000_secondary_memory_needed_ptr): Delete.
8914 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8916 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8917 (rs6000_option_override_internal): Assign to
8918 targetm.secondary_memory_needed rather than
8919 rs6000_secondary_memory_needed_ptr.
8920 (rs6000_secondary_memory_needed): Match hook interface.
8921 (rs6000_debug_secondary_memory_needed): Likewise.
8922 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8923 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8924 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8925 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8926 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8927 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8929 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8930 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8931 (rs6000_option_override_internal): Assign to
8932 targetm.secondary_memory_needed rather than
8933 rs6000_secondary_memory_needed_ptr.
8934 (rs6000_secondary_memory_needed): Match hook interface.
8935 (rs6000_debug_secondary_memory_needed): Likewise.
8936 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8937 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8938 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8939 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8940 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8941 (sparc_secondary_memory_needed): New function.
8942 * lra-constraints.c (check_and_process_move): Refer to
8943 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8945 (curr_insn_transform): Likewise.
8946 (process_alt_operands): Use targetm.secondary_memory_needed
8947 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8948 (check_secondary_memory_needed_p): Likewise.
8949 (choose_split_class): Likewise.
8950 * reload.c: Unconditionally include code that was previously
8951 conditional on SECONDARY_MEMORY_NEEDED.
8952 (push_secondary_reload): Use targetm.secondary_memory_needed
8953 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8954 (push_reload): Likewise.
8955 * reload1.c: Unconditionally include code that was previously
8956 conditional on SECONDARY_MEMORY_NEEDED.
8957 (choose_reload_regs): Use targetm.secondary_memory_needed
8958 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8959 (gen_reload): Likewise.
8960 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8962 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8963 Alan Hayward <alan.hayward@arm.com>
8964 David Sherwood <david.sherwood@arm.com>
8966 * target.def (secondary_memory_needed_mode): New hook:
8967 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8968 * targhooks.h (default_secondary_memory_needed_mode): New function.
8969 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8970 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8971 * doc/tm.texi: Regenerate.
8972 * lra-constraints.c (check_and_process_move): Use
8973 targetm.secondary_memory_needed_mode instead of
8974 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8975 (curr_insn_transform): Likewise.
8976 * reload.c (get_secondary_mem): Likewise.
8977 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8978 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8980 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8981 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8982 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8983 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8984 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8986 * config/powerpcspe/powerpcspe-protos.h
8987 (rs6000_secondary_memory_needed_mode): Delete.
8988 * config/powerpcspe/powerpcspe.c
8989 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8990 (rs6000_secondary_memory_needed_mode): Make static.
8991 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8992 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8994 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8996 (rs6000_secondary_memory_needed_mode): Make static.
8997 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8998 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8999 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9000 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9001 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9003 (sparc_secondary_memory_needed_mode): New function.
9004 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9006 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9008 * config/aarch64/constraints.md (Umq): New constraint.
9009 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9011 (mov<mode>): Update condition.
9013 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9015 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9016 when bitposition is the same.
9018 2017-09-13 Richard Biener <rguenther@suse.de>
9020 * dwarf2out.c (output_die_symbol): Remove.
9021 (output_die): Do not output a DIEs symbol.
9023 2017-09-13 Richard Biener <rguenther@suse.de>
9026 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9027 default-def to avoid breaking iterator update with the weird
9028 interaction with cgraph_update_edges_for_call_stmt_node.
9030 2017-09-13 Richard Biener <rguenther@suse.de>
9032 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9033 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9034 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9035 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9037 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9039 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9040 Disable pc relative literal load irrespective of
9041 TARGET_FIX_ERR_A53_84341 for default.
9043 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9045 * config/sparc/sparc.c (output_return): Output the source location of
9046 the insn in the delay slot, if any.
9047 (output_sibcall): Likewise.
9049 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9052 * common.opt (-static-pie): New alias.
9053 (shared): Negate static-pie.
9054 (-no-pie): Update help text.
9056 (static-pie): New option.
9057 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9058 -static-pie support.
9059 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9060 (LINK_EH_SPEC): Likewise.
9061 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9062 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9063 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9064 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9065 (init_gcc_specs): Likewise.
9066 (init_spec): Likewise.
9067 (display_help): Update help message for -pie.
9068 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9071 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9073 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9074 (movdi_aarch64): Likewise.
9075 (movti_aarch64): Likewise.
9077 2017-09-12 Simon Wright <simon@pushface.org>
9080 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9081 calculation of the minor version, always output as 0.
9083 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9086 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9087 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9088 on it early, rather than manual conversion late. For
9089 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9090 instead of performing manual conversion.
9092 2017-09-12 Carl Love <cel@us.ibm.com>
9094 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9095 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9097 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9098 VMULOSW): Add definitions.
9099 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9100 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9101 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9102 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9103 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9105 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9107 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9109 (movti_aarch64): Likewise.
9110 (movdf_aarch64): Likewise.
9111 (movtf_aarch64): Likewise.
9112 (load_pairdi): Likewise.
9113 (store_pairdi): Likewise.
9114 (load_pairdf): Likewise.
9115 (store_pairdf): Likewise.
9116 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9117 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9118 (ldr_got_small_<mode>): Likewise.
9119 (ldr_got_small_28k_<mode>): Likewise.
9120 (ldr_got_tiny): Likewise.
9121 * config/aarch64/iterators.md (ldst_sz): New.
9122 (ldpstp_sz): Likewise.
9123 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9125 (thunderx_load): Split load_8 to load_16.
9126 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9128 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9129 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9130 (xgene1_store_pair): Split store_8 to store_16.
9131 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9132 (falkor_st_0_st_sd): Split store_8 to store_16.
9134 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9136 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9137 and store1/2/3/4 to store_4/8/12/16.
9138 * config/aarch64/aarch64.md: Update for rename.
9139 * config/arm/arm.md: Likewise.: Likewise.
9140 * config/arm/arm.c: Likewise.
9141 * config/arm/thumb1.md: Likewise.
9142 * config/arm/thumb2.md: Likewise.
9143 * config/arm/vfp.md: Likewise.
9144 * config/arm/arm-generic.md: Likewise.
9145 * config/arm/arm1020e.md: Likewise.
9146 * config/arm/arm1026ejs.md: Likewise.
9147 * config/arm/arm1136jfs.md: Likewise.
9148 * config/arm/arm926ejs.md: Likewise.
9149 * config/arm/cortex-a15.md: Likewise.
9150 * config/arm/cortex-a17.md: Likewise.
9151 * config/arm/cortex-a5.md: Likewise.
9152 * config/arm/cortex-a53.md: Likewise.
9153 * config/arm/cortex-a57.md: Likewise.
9154 * config/arm/cortex-a7.md: Likewise.
9155 * config/arm/cortex-a8.md: Likewise.
9156 * config/arm/cortex-a9.md: Likewise.
9157 * config/arm/cortex-m4.md: Likewise.
9158 * config/arm/cortex-m7.md: Likewise.
9159 * config/arm/cortex-r4.md: Likewise.
9160 * config/arm/exynos-m1.md: Likewise.
9161 * config/arm/fa526.md: Likewise.
9162 * config/arm/fa606te.md: Likewise.
9163 * config/arm/fa626te.md: Likewise.
9164 * config/arm/fa726te.md: Likewise.
9165 * config/arm/fmp626.md: Likewise.
9166 * config/arm/iwmmxt.md: Likewise.
9167 * config/arm/ldmstm.md: Likewise.
9168 * config/arm/marvell-pj4.md: Likewise.
9169 * config/arm/xgene1.md: Likewise.
9170 * config/aarch64/thunderx.md: Likewise.
9171 * config/aarch64/thunderx2t99.md: Likewise.
9172 * config/aarch64/falkor.md: Likewise.
9174 2017-09-12 Martin Liska <mliska@suse.cz>
9176 * attribs.c (private_lookup_attribute): New function.
9177 * attribs.h (private_lookup_attribute): Declared here.
9178 (lookup_attribute): Called from this place.
9180 2017-09-12 Richard Biener <rguenther@suse.de>
9182 PR tree-optimization/82157
9183 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9184 stmts with side-effects.
9186 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9187 Alan Hayward <alan.hayward@arm.com>
9188 David Sherwood <david.sherwood@arm.com>
9190 * target.def (hard_regno_nregs): New hook.
9191 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9192 * targhooks.h (default_hard_regno_nregs): Declare.
9193 * targhooks.c (default_hard_regno_nregs): New function.
9194 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9195 (TARGET_HARD_REGNO_NREGS): ...this hook.
9196 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9197 (CLASS_MAX_NREGS): Likewise.
9198 * doc/tm.texi: Regenerate.
9199 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9200 instead of HARD_REGNO_NREGS.
9201 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9202 HARD_REGNO_NREGS in the comment.
9203 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9204 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9205 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9206 Return an unsigned int.
9207 (TARGET_HARD_REGNO_NREGS): Redefine.
9208 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9209 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9210 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9211 (arc_hard_regno_nregs): New function.
9212 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9213 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9214 (arm_hard_regno_nregs): New function.
9215 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9216 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9217 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9218 (TARGET_HARD_REGNO_NREGS): Redefine.
9219 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9220 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9221 (HARD_REGNO_NREGS): Delete.
9222 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9223 (cr16_hard_regno_nregs): New function.
9224 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9225 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9226 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9227 (cris_hard_regno_nregs): New function.
9228 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9229 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9230 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9231 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9232 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9233 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9234 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9235 (frv_hard_regno_nregs): Make static. Take and return an
9237 (frv_class_max_nregs): Remove outdated copy of documentation.
9238 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9239 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9240 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9241 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9242 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9243 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9244 (TARGET_HARD_REGNO_NREGS): Redefine.
9245 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9246 (CLASS_MAX_NREGS): Update comment.
9247 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9248 (ia64_hard_regno_nregs): New function.
9249 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9250 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9251 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9252 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9253 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9255 (m32c_hard_regno_nregs): Likewise. Make static.
9256 (TARGET_HARD_REGNO_NREGS): Redefine.
9257 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9258 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9259 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9260 (m68k_hard_regno_nregs): New function.
9261 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9262 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9263 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9264 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9265 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9266 Take and return an unsigned int.
9267 (TARGET_HARD_REGNO_NREGS): Redefine.
9268 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9269 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9270 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9271 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9272 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9273 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9274 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9275 (msp430_hard_regno_nregs): Make static. Take and return an
9277 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9278 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9279 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9280 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9281 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9282 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9283 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9284 (TARGET_HARD_REGNO_NREGS): Redefine.
9285 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9286 (PA_HARD_REGNO_NREGS): ...this.
9287 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9288 (PA_HARD_REGNO_NREGS): ...this.
9289 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9290 (pa_hard_regno_nregs): New function.
9291 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9292 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9293 (pdp11_hard_regno_nregs): New function.
9294 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9295 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9296 (rs6000_hard_regno_nregs_hook): New function.
9297 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9298 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9299 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9300 Take and return an unsigned int. Move earlier in file.
9301 (TARGET_HARD_REGNO_NREGS): Redefine.
9302 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9303 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9304 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9305 (rl78_hard_regno_nregs): Make static. Take and return an
9307 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9308 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9309 (rs6000_hard_regno_nregs_hook): New function.
9310 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9311 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9312 (TARGET_HARD_REGNO_NREGS): Redefine.
9313 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9314 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9315 instead of HARD_REGNO_NREGS.
9316 (s390_hard_regno_nregs): New function.
9317 (s390_hard_regno_mode_ok): Add comment from s390.h.
9318 (TARGET_HARD_REGNO_NREGS): Redefine.
9319 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9320 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9321 (sh_hard_regno_nregs): New function.
9322 (sh_pass_in_reg_p): Use it.
9323 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9324 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9325 (sparc_hard_regno_nregs): New function.
9326 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9327 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9328 (spu_function_arg_advance): Use it, supplying a valid register number.
9329 (TARGET_HARD_REGNO_NREGS): Redefine.
9330 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9331 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9332 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9333 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9334 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9335 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9336 (CLASS_MAX_NREGS): Remove copy of old documentation.
9337 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9338 (visium_hard_regno_nregs): New function.
9339 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9340 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9341 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9342 xtensa_hard_regno_nregs): New function.
9343 * system.h (HARD_REGNO_NREGS): Poison.
9345 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9347 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9348 hard_regno_nregs instead of HARD_REGNO_NREGS.
9349 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9350 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9351 (c6x_expand_epilogue): Likewise.
9352 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9353 (frv_read_iacc_argument): Likewise.
9354 * config/sh/sh.c: Include regs.h.
9355 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9356 (regs_used): Likewise.
9357 (output_stack_adjust): Likewise.
9358 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9359 * expmed.c: Include regs.h.
9360 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9361 * ree.c: Include regs.h.
9362 (combine_reaching_defs): Use hard_regno_nregs instead of
9364 (add_removable_extension): Likewise.
9366 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9368 * regs.h (hard_regno_nregs): Turn into a function.
9369 (end_hard_regno): Update accordingly.
9370 * caller-save.c (setup_save_areas): Likewise.
9371 (save_call_clobbered_regs): Likewise.
9372 (replace_reg_with_saved_mem): Likewise.
9373 (insert_restore): Likewise.
9374 (insert_save): Likewise.
9375 * combine.c (can_change_dest_mode): Likewise.
9376 (move_deaths): Likewise.
9377 (distribute_notes): Likewise.
9378 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9379 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9380 (rs6000_split_multireg_move): Likewise.
9381 (rs6000_register_move_cost): Likewise.
9382 (rs6000_memory_move_cost): Likewise.
9383 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9384 (rs6000_split_multireg_move): Likewise.
9385 (rs6000_register_move_cost): Likewise.
9386 (rs6000_memory_move_cost): Likewise.
9387 * cselib.c (cselib_reset_table): Likewise.
9388 (cselib_lookup_1): Likewise.
9389 * emit-rtl.c (set_mode_and_regno): Likewise.
9390 * function.c (aggregate_value_p): Likewise.
9391 * ira-color.c (setup_profitable_hard_regs): Likewise.
9392 (check_hard_reg_p): Likewise.
9393 (calculate_saved_nregs): Likewise.
9394 (assign_hard_reg): Likewise.
9395 (improve_allocation): Likewise.
9396 (calculate_spill_cost): Likewise.
9397 * ira-emit.c (modify_move_list): Likewise.
9398 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9399 (ira_hard_reg_in_set_p): Likewise.
9400 * ira.c (setup_reg_mode_hard_regset): Likewise.
9401 (clarify_prohibited_class_mode_regs): Likewise.
9402 (check_allocation): Likewise.
9403 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9404 (lra_setup_reg_renumber): Likewise.
9405 (setup_try_hard_regno_pseudos): Likewise.
9406 (spill_for): Likewise.
9407 (assign_hard_regno): Likewise.
9408 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9409 * lra-constraints.c (in_class_p): Likewise.
9410 (lra_constraint_offset): Likewise.
9411 (simplify_operand_subreg): Likewise.
9412 (lra_constraints): Likewise.
9413 (split_reg): Likewise.
9414 (split_if_necessary): Likewise.
9415 (invariant_p): Likewise.
9416 (inherit_in_ebb): Likewise.
9417 * lra-lives.c (process_bb_lives): Likewise.
9418 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9419 (get_hard_regs): Likewise.
9420 (do_remat): Likewise.
9421 * lra-spills.c (assign_spill_hard_regs): Likewise.
9422 * mode-switching.c (create_pre_exit): Likewise.
9423 * postreload.c (reload_combine_recognize_pattern): Likewise.
9424 * recog.c (peep2_find_free_register): Likewise.
9425 * regcprop.c (kill_value_regno): Likewise.
9426 (set_value_regno): Likewise.
9427 (copy_value): Likewise.
9428 (maybe_mode_change): Likewise.
9429 (find_oldest_value_reg): Likewise.
9430 (copyprop_hardreg_forward_1): Likewise.
9431 * regrename.c (check_new_reg_p): Likewise.
9432 (regrename_do_replace): Likewise.
9433 * reload.c (push_reload): Likewise.
9434 (combine_reloads): Likewise.
9435 (find_dummy_reload): Likewise.
9436 (operands_match_p): Likewise.
9437 (find_reloads): Likewise.
9438 (find_equiv_reg): Likewise.
9439 (reload_adjust_reg_for_mode): Likewise.
9440 * reload1.c (count_pseudo): Likewise.
9441 (count_spilled_pseudo): Likewise.
9442 (find_reg): Likewise.
9443 (clear_reload_reg_in_use): Likewise.
9444 (free_for_value_p): Likewise.
9445 (allocate_reload_reg): Likewise.
9446 (choose_reload_regs): Likewise.
9447 (reload_adjust_reg_for_temp): Likewise.
9448 (emit_reload_insns): Likewise.
9449 (delete_output_reload): Likewise.
9450 * rtlanal.c (subreg_get_info): Likewise.
9451 * sched-deps.c (sched_analyze_reg): Likewise.
9452 * sel-sched.c (init_regs_for_mode): Likewise.
9453 (mark_unavailable_hard_regs): Likewise.
9454 (choose_best_reg_1): Likewise.
9455 (verify_target_availability): Likewise.
9456 * valtrack.c (dead_debug_insert_temp): Likewise.
9457 * var-tracking.c (track_loc_p): Likewise.
9458 (emit_note_insn_var_location): Likewise.
9459 * varasm.c (make_decl_rtl): Likewise.
9460 * reginfo.c (choose_hard_reg_mode): Likewise.
9461 (init_reg_modes_target): Refer directly to
9462 this_target_regs->x_hard_regno_nregs.
9464 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9466 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9467 instead of hard_regno_nregs.
9469 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9471 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9472 end_hard_regno instead of hard_regno_nregs.
9473 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9474 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9475 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9476 * ira-color.c (improve_allocation): Likewise.
9477 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9478 * lra-lives.c (mark_regno_live): Likewise.
9479 (mark_regno_dead): Likewise.
9480 * lra-remat.c (operand_to_remat): Likewise.
9481 * lra.c (collect_non_operand_hard_regs): Likewise.
9482 * postreload.c (reload_combine_note_store): Likewise.
9483 (move2add_valid_value_p): Likewise.
9484 * reload.c (regno_clobbered_p): Likewise.
9486 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9488 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9490 * config/v850/v850.c (v850_reorg): Likewise.
9491 * reload.c (refers_to_regno_for_reload_p): Likewise.
9492 (find_equiv_reg): Likewise.
9493 * reload1.c (reload_reg_reaches_end_p): Likewise.
9495 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9497 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9499 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9500 * config/arm/arm.c (output_move_neon): Likewise.
9501 (arm_attr_length_move_neon): Likewise.
9502 (neon_split_vcombine): Likewise.
9503 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9504 (c6x_mark_reg_written): Likewise.
9505 (c6x_dwarf_register_span): Likewise.
9506 * config/i386/i386.c (ix86_save_reg): Likewise.
9507 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9508 (rws_access_reg): Likewise.
9509 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9510 * mode-switching.c (create_pre_exit): Likewise.
9511 * ree.c (combine_reaching_defs): Likewise.
9512 (add_removable_extension): Likewise.
9513 * regcprop.c (find_oldest_value_reg): Likewise.
9514 (copyprop_hardreg_forward_1): Likewise.
9515 * reload.c (reload_inner_reg_of_subreg): Likewise.
9516 (push_reload): Likewise.
9517 (combine_reloads): Likewise.
9518 (find_dummy_reload): Likewise.
9519 (reload_adjust_reg_for_mode): Likewise.
9520 * reload1.c (find_reload_regs): Likewise.
9521 (forget_old_reloads_1): Likewise.
9522 (reload_reg_free_for_value_p): Likewise.
9523 (reload_adjust_reg_for_temp): Likewise.
9524 (emit_reload_insns): Likewise.
9525 (delete_output_reload): Likewise.
9526 * sel-sched.c (choose_best_reg_1): Likewise.
9527 (choose_best_pseudo_reg): Likewise.
9529 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9530 Alan Hayward <alan.hayward@arm.com>
9531 David Sherwood <david.sherwood@arm.com>
9533 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9534 * target.def (slow_unaligned_access): New hook.
9535 * targhooks.h (default_slow_unaligned_access): Declare.
9536 * targhooks.c (default_slow_unaligned_access): New function.
9537 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9538 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9539 * doc/tm.texi: Regenerate.
9540 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9541 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9542 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9544 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9545 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9547 (rs6000_slow_unaligned_access): New function.
9548 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9549 (expand_block_compare): Likewise.
9550 (expand_strn_compare): Likewise.
9551 (rs6000_rtx_costs): Likewise.
9552 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9553 (riscv_slow_unaligned_access): Likewise.
9554 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9555 (riscv_slow_unaligned_access_p): ...this and make static.
9556 (riscv_option_override): Update accordingly.
9557 (riscv_slow_unaligned_access): New function.
9558 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9559 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9560 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9561 (rs6000_slow_unaligned_access): New function.
9562 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9563 (rs6000_rtx_costs): Likewise.
9564 * config/rs6000/rs6000-string.c (expand_block_compare)
9565 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9566 of SLOW_UNALIGNED_ACCESS.
9567 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9568 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9569 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9570 of SLOW_UNALIGNED_ACCESS.
9571 * expmed.c (simple_mem_bitfield_p): Likewise.
9572 * expr.c (alignment_for_piecewise_move): Likewise.
9573 (emit_group_load_1): Likewise.
9574 (emit_group_store): Likewise.
9575 (copy_blkmode_from_reg): Likewise.
9576 (emit_push_insn): Likewise.
9577 (expand_assignment): Likewise.
9578 (store_field): Likewise.
9579 (expand_expr_real_1): Likewise.
9580 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9581 * lra-constraints.c (simplify_operand_subreg): Likewise.
9582 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9583 * gimple-ssa-store-merging.c: Likewise in block comment at start
9585 * tree-ssa-strlen.c: Include target.h.
9586 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9587 of SLOW_UNALIGNED_ACCESS.
9588 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9590 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9592 PR rtl-optimization/82185
9593 * expmed.c (emit_store_flag_int): Only test tem if it has been
9596 2017-09-12 Richard Biener <rguenther@suse.de>
9599 * match.pd ((FTYPE) N CMP CST): Fix typo.
9601 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9603 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9605 (mips_near_type_p): Add 'short_call' attribute as a synonym
9607 * doc/extend.texi (short_call): Document new function attribute.
9609 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9612 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9613 assertion check that in the condition.
9614 (get_atomic_generic_size): Likewise. Before testing if parameter
9615 has pointer type, if it has array type, call for C++
9616 default_conversion to perform array-to-pointer conversion.
9618 2017-09-12 Richard Biener <rguenther@suse.de>
9620 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9621 for operations we cannot scalarize.
9623 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9625 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9626 vectors heap vectors. Clean up comments.
9627 Make visited_bbs a reference.
9628 (profitable_jump_thread_path): Make GC
9629 vectors heap vectors. Clean up comments.
9631 (convert_and_register_jump_thread_path): Make GC vectors heap
9633 (check_subpath_and_update_thread_path): Same. Clean up comments.
9634 Make visited_bbs a reference.
9635 (handle_phi): Abstract common code to to
9636 register_jump_thread_path_if_profitable.
9637 Rename VAR_BB to DEF_BB.
9639 Make GC vectors heap vectors.
9640 Make visited_bbs a reference.
9641 (handle_assignment): Same.
9642 (register_jump_thread_path_if_profitable): New.
9643 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9645 Make GC vectors heap vectors. Clean up comments.
9646 Make visited_bbs a reference.
9647 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9648 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9651 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9654 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9655 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9657 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9659 Revert r251800 and r251799.
9661 2017-09-11 Martin Jambor <mjambor@suse.cz>
9664 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9665 arguments in advance.
9666 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9667 use it to find predecessor edges.
9668 (naive_outof_ssa): Collect vector of predecessors.
9670 2017-09-08 Jason Merrill <jason@redhat.com>
9672 PR c++/70029 - ICE with ref-qualifier and -flto
9673 * langhooks.h (struct lang_hooks_for_types): Add
9674 copy_lang_qualifiers.
9675 * attribs.c (build_type_attribute_qual_variant): Use it.
9676 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9678 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9679 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9681 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9684 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9685 (*mulsi3_sp64): New instruction.
9686 (mulsi3): New expander.
9688 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9690 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9692 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9694 * sancov.c: Include memmodel.h.
9696 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9699 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9702 2017-09-07 Carl Love <cel@us.ibm.com>
9704 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9705 the sldi instruction.
9707 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9709 * sancov.c: Include tm_p.h.
9711 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9714 * output.h (switch_to_other_text_partition): New declaration.
9715 * varasm.c (switch_to_other_text_partition): New function.
9716 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9717 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9718 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9719 to the other text partition before emitting LCL label and switch back
9720 after emitting the word after it.
9722 2017-09-07 Richard Biener <rguenther@suse.de>
9724 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9725 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9726 critical edges here, after loop init.
9727 (pass_data_pre): Remove PROP_no_crit_edges flags.
9728 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9729 for valueization of call args to avoid leaking VN_TOP.
9730 (visit_use): Assert we do not visit default defs.
9731 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9732 Use error_mark_node to more easily detect leaking VN_TOP.
9733 All default-defs are varying, not VN_TOP. Mark them visited.
9734 (run_scc_vn): Make code match comment.
9736 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9738 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9739 OPTION_MASK_FLOAT128_KEYWORD.
9740 (POWERPC_MASKS): Likewise.
9741 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9742 support for the -mfloat128-type option, and make -mfloat128
9743 default on PowerPC Linux systems. Define or undefine
9744 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9745 Define __float128 to be __ieee128 if IEEE 128-bit support is
9746 enabled, or undefine it.
9747 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9748 Delete defining __FLOAT128_TYPE__.
9749 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9750 -mfloat128-type option and make -mfloat128 default on PowerPC
9752 (TARGET_FLOAT128_TYPE): Likewise.
9753 (-mfloat128-type): Likewise.
9754 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9755 Delete the -mfloat128-type option and make -mfloat128 default on
9756 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9757 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9758 128-bit floating point is enabled. Change tests from using
9759 -mfloat128-type to -mfloat128.
9760 (rs6000_mangle_type): Use the correct mangling for the __float128
9761 type even if normal long double is restricted to 64-bits.
9762 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9764 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9765 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9766 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9768 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9769 documentation for -mfloat128.
9771 2017-09-06 Olivier Hainque <hainque@adacore.com>
9773 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9775 2017-09-06 Wish Wu <wishwu007@gmail.com>
9776 Jakub Jelinek <jakub@redhat.com>
9778 * asan.c (initialize_sanitizer_builtins): Add
9779 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9780 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9781 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9782 BT_FN_VOID_UINT64_PTR variables.
9783 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9784 (BT_FN_VOID_UINT16_UINT16): Likewise.
9785 (BT_FN_VOID_UINT32_UINT32): Likewise.
9786 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9787 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9788 (BT_FN_VOID_UINT64_PTR): Likewise.
9789 * common.opt (flag_sanitize_coverage): New variable.
9790 (fsanitize-coverage=trace-pc): Remove.
9791 (fsanitize-coverage=): Add.
9792 * flag-types.h (enum sanitize_coverage_code): New enum.
9793 * fold-const.c (fold_range_test): Disable non-short-circuit
9794 optimization if flag_sanitize_coverage.
9795 (fold_truth_andor): Likewise.
9796 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9797 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9798 (coverage_sanitizer_opts): New array.
9799 (get_closest_sanitizer_option): Add OPTS argument, handle also
9800 OPT_fsanitize_coverage_.
9801 (parse_sanitizer_options): Adjusted to also handle
9802 OPT_fsanitize_coverage_.
9803 (common_handle_option): Add OPT_fsanitize_coverage_.
9804 * sancov.c (instrument_comparison, instrument_switch): New function.
9805 (sancov_pass): Add trace-cmp support.
9806 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9807 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9808 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9809 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9810 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9811 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9812 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9813 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9814 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9815 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9817 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9819 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9820 error. Only quit immediately if parsing is complete.
9821 (BEGIN): Initialize fatal_err and parse_done.
9822 (begin fpu, end fpu): Check number of arguments.
9823 (begin arch, end arch): Likewise.
9824 (begin cpu, end cpu): Likewise.
9825 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9826 (optalias): Likewise.
9828 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9830 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9831 * config/arm/arm-isa.h: Delete. Move definitions to ...
9832 * arm-cpus.in: ... here. Use new feature and fgroup values.
9833 * config/arm/arm.c (arm_option_override): Use lower case for feature
9835 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9836 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9837 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9839 (print_isa_bits_for): New function.
9840 (gen_isa): New function.
9841 (gen_comm_data): Use print_isa_bits_for.
9842 (define feature): New keyword.
9843 (define fgroup): New keyword.
9844 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9845 (arm-isa.h): Add rule to generate file.
9846 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9847 case for feature bit names.
9849 2017-09-06 Richard Biener <rguenther@suse.de>
9851 * tree-ssa-pre.c (NECESSARY): Remove.
9852 (create_expression_by_pieces): Do not touch pass-local flags.
9853 (insert_into_preds_of_block): Likewise.
9854 (do_pre_regular_insertion): Likewise.
9855 (eliminate_insert): Likewise.
9856 (eliminate_dom_walker::before_dom_children): Likewise.
9857 (fini_eliminate): Do not look at inserted_exprs.
9858 (mark_operand_necessary): Remove.
9859 (remove_dead_inserted_code): Replace with simple work-list
9860 algorithm based on inserted_exprs and SSA uses.
9861 (pass_pre::execute): Re-order fini_eliminate and
9862 remove_dead_inserted_code.
9864 2017-09-06 Olivier Hainque <hainque@adacore.com>
9866 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9867 for VxWorks 7. Adjust surrounding comments.
9869 2017-09-06 Richard Biener <rguenther@suse.de>
9871 * gimple-ssa-strength-reduction.c
9872 (find_candidates_dom_walker::before_dom_children): Also allow
9875 2017-09-06 Richard Biener <rguenther@suse.de>
9877 PR tree-optimization/82108
9878 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9879 for gap in the non-permutation SLP case.
9881 2017-09-06 Martin Jambor <mjambor@suse.cz>
9883 PR tree-optimization/82078
9884 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9885 add_access_to_work_queue...
9886 (build_accesses_from_assign): ...here.
9887 (propagate_all_subaccesses): Make sure racc is the group
9888 representative, if there is one.
9890 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9893 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9896 2017-09-06 Richard Biener <rguenther@suse.de>
9898 * gimple-ssa-strength-reduction.c
9899 (find_candidates_dom_walker::before_doom_children): Use a
9900 type and not a mode check.
9902 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9905 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9907 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9909 2017-09-05 Jeff Law <law@redhat.com>
9911 PR tree-optimization/64910
9912 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9913 cases where we have 3 or more operands.
9915 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9918 * omp-low.c (lower_omp_for): Recompute tree invariant if
9919 gimple_omp_for_initial/final is ADDR_EXPR.
9922 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9923 into gimple val before gimplification fo the COND_EXPR.
9925 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9927 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9928 REGION_COPY argument.
9929 (thread_through_all_blocks): Remove unused argument to
9930 duplicate_thread_path.
9932 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9933 Alan Hayward <alan.hayward@arm.com>
9934 David Sherwood <david.sherwood@arm.com>
9936 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9937 Take a scalar_mode rather than a machine_mode.
9938 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9939 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9940 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9941 (aarch64_gen_adjusted_ldpstp): Likewise.
9942 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9944 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9945 Alan Hayward <alan.hayward@arm.com>
9946 David Sherwood <david.sherwood@arm.com>
9948 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9949 Take a scalar_int_mode instead of a machine_mode.
9950 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9951 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9952 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9953 (aarch64_simd_attr_length_rglist): Delete.
9954 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9955 a scalar_int_mode instead of a machine_mode.
9956 (aarch64_add_offset): Likewise.
9957 (aarch64_internal_mov_immediate): Likewise
9958 (aarch64_add_constant_internal): Likewise.
9959 (aarch64_add_constant): Likewise.
9960 (aarch64_movw_imm): Likewise.
9961 (aarch64_rtx_arith_op_extract_p): Likewise.
9962 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9963 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9964 Remove assert that the mode isn't a vector.
9965 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9966 (aarch64_expand_mov_immediate): Update calls after above changes.
9967 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9968 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9969 (aarch64_move_imm): Likewise.
9970 (aarch64_can_const_movi_rtx_p): Likewise.
9971 (aarch64_strip_extend): Likewise.
9972 (aarch64_extr_rtx_p): Likewise.
9973 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9974 a CONST_INT when the mode parameter is VOIDmode.
9975 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9977 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9979 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9980 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9981 (bitwise_type_for_mode): Update accordingly.
9983 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9985 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9986 * stor-layout.c (mode_for_size_tree): Likewise.
9987 (mode_for_array): Update accordingly.
9988 (layout_decl): Likewise.
9989 (compute_record_mode): Likewise. Only set the mode once.
9991 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9993 * target.def (get_mask_mode): Change return type to opt_mode.
9995 * doc/tm.texi: Regenerate.
9996 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9997 * targhooks.c (default_get_mask_mode): Likewise.
9998 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9999 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10000 targetm.get_mask_mode.
10001 * tree.c (build_truth_vector_type): Likewise.
10003 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10005 * machmode.h (mode_for_vector): Return an opt_mode.
10006 * stor-layout.c (mode_for_vector): Likewise.
10007 (mode_for_int_vector): Update accordingly.
10008 (layout_type): Likewise.
10009 * config/i386/i386.c (emit_memmov): Likewise.
10010 (ix86_expand_set_or_movmem): Likewise.
10011 (ix86_expand_vector_init): Likewise.
10012 (ix86_get_mask_mode): Likewise.
10013 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10015 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10016 * expmed.c (extract_bit_field_1): Likewise.
10017 * expr.c (expand_expr_real_2): Likewise.
10018 * optabs-query.c (can_vec_perm_p): Likewise.
10019 (can_vec_mask_load_store_p): Likewise.
10020 * optabs.c (expand_vec_perm): Likewise.
10021 * targhooks.c (default_get_mask_mode): Likewise.
10022 * tree-vect-stmts.c (vectorizable_store): Likewise.
10023 (vectorizable_load): Likewise.
10024 (get_vectype_for_scalar_type_and_size): Likewise.
10026 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10028 * machmode.h (mode_for_int_vector): New function.
10029 * stor-layout.c (mode_for_int_vector): Likewise.
10030 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10031 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10032 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10033 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10034 (s390_expand_vcond): Likewise.
10036 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10038 * machmode.h (opt_machine_mode): New type.
10039 (opt_mode<T>): Allow construction from anything that can be
10041 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10042 (mode_for_size): Return an opt_machine_mode.
10043 * stor-layout.c (mode_for_size): Likewise.
10044 (mode_for_size_tree): Update call accordingly.
10045 (bitwise_mode_for_mode): Likewise.
10046 (make_fract_type): Likewise.
10047 (make_accum_type): Likewise.
10048 * caller-save.c (replace_reg_with_saved_mem): Update call
10050 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10051 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10052 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10053 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10054 * expmed.c (extract_bit_field_1): Likewise.
10055 * reload.c (get_secondary_mem): Likewise.
10056 * varasm.c (assemble_integer): Likewise.
10057 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10060 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10062 * machmode.h (decimal_float_mode_for_size): New function.
10063 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10064 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10065 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10066 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10067 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10068 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10070 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10072 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10073 (get_builtin_sync_mode): Likewise.
10074 (expand_ifn_atomic_compare_exchange): Likewise.
10075 (expand_builtin_atomic_clear): Likewise.
10076 (expand_builtin_atomic_test_and_set): Likewise.
10077 (fold_builtin_atomic_always_lock_free): Likewise.
10078 * calls.c (compute_argument_addresses): Likewise.
10079 (emit_library_call_value_1): Likewise.
10080 (store_one_arg): Likewise.
10081 * combine.c (combine_instructions): Likewise.
10082 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10083 * config/arm/arm.c (arm_function_value): Likewise.
10084 (aapcs_allocate_return_reg): Likewise.
10085 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10086 * config/i386/i386.c (construct_container): Likewise.
10087 (ix86_gimplify_va_arg): Likewise.
10088 (ix86_expand_sse_cmp): Likewise.
10089 (emit_memmov): Likewise.
10090 (emit_memset): Likewise.
10091 (expand_small_movmem_or_setmem): Likewise.
10092 (ix86_expand_pextr): Likewise.
10093 (ix86_expand_pinsr): Likewise.
10094 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10095 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10097 * config/mips/mips.c (mips_function_value_1) Likewise.
10098 (mips_block_move_straight): Likewise.
10099 (mips_expand_ins_as_unaligned_store): Likewise.
10100 * config/powerpcspe/powerpcspe.c
10101 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10102 (rs6000_darwin64_record_arg_flush): Likewise.
10103 * config/rs6000/rs6000.c
10104 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10105 (rs6000_darwin64_record_arg_flush): Likewise.
10106 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10107 (sparc_function_value_1): Likewise.
10108 * config/spu/spu.c (adjust_operand): Likewise.
10109 (spu_emit_branch_or_set): Likewise.
10110 (arith_immediate_p): Likewise.
10111 * emit-rtl.c (gen_lowpart_common): Likewise.
10112 * expr.c (expand_expr_real_1): Likewise.
10113 * function.c (assign_parm_setup_block): Likewise.
10114 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10115 * reload1.c (alter_reg): Likewise.
10116 * stor-layout.c (mode_for_vector): Likewise.
10117 (layout_type): Likewise.
10119 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10121 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10122 (spu_convert_move): Likewise.
10123 * lower-subreg.c (resolve_simple_move): Likewise.
10125 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10128 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10129 define_insn to a define_expand.
10130 (altivec_vsum2sws_direct): New define_insn.
10131 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10133 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10135 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10138 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10143 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10144 realign stack if stack alignment needed is less than incoming
10147 2017-09-05 Marek Polacek <polacek@redhat.com>
10150 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10153 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10155 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10157 2017-09-05 Richard Biener <rguenther@suse.de>
10159 PR tree-optimization/82084
10160 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10162 2017-09-05 Richard Biener <rguenther@suse.de>
10164 PR tree-optimization/82102
10165 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10167 2017-09-05 Martin Liska <mliska@suse.cz>
10169 PR tree-optimization/82032
10170 * tree-cfg.c (generate_range_test): New function.
10171 * tree-cfg.h (generate_range_test): Declared here.
10172 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10173 (cleanup_control_expr_graph): Use it.
10174 * tree-switch-conversion.c (try_switch_expansion): Remove
10176 (emit_case_nodes): Use generate_range_test.
10178 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10181 * config/i386/i386.md (*<btsc><mode>_mask): Add
10182 TARGET_USE_BT to insn constraint.
10183 (*btr<mode>_mask): Ditto.
10185 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10187 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10188 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10190 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10193 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10194 TARGET_NEON and TARGET_IWMMXT.
10195 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10196 TARGET_NEON and TARGET_IWMMXT.
10197 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10199 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10201 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10202 (ix86_rewrite_tls_address): Ditto.
10203 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10204 (ix86_rewrite_tls_address_1): Ditto.
10205 (ix86_rewrite_tls_address): Ditto.
10206 * config/i386/predicates.md (tls_address_pattern): New predicate.
10207 * config/i386/i386.md (TLS address splitter): New splitter.
10209 2017-09-04 Richard Biener <rguenther@suse.de>
10211 PR tree-optimization/82084
10212 * fold-const.h (can_native_encode_string_p): Declare.
10213 * fold-const.c (can_native_encode_string_p): Factor out from ...
10214 (native_encode_string): ... here.
10215 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10216 vectorizing stores from constants we later cannot handle.
10218 2017-09-04 Marek Polacek <polacek@redhat.com>
10221 * doc/invoke.texi: Update -Wtautological-compare documentation.
10223 2017-09-04 Jeff Law <law@redhat.com>
10225 PR tree-optimization/64910
10226 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10227 swap the first and last operand if the last is a constant.
10229 2017-09-04 Marek Polacek <polacek@redhat.com>
10232 * convert.c (do_narrow): When sanitizing signed integer overflows,
10233 bail out for signed types.
10234 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10236 2017-09-04 Richard Biener <rguenther@suse.de>
10238 PR tree-optimization/82060
10239 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10240 Move devirtualization after stmt folding and before EH/AB/noreturn
10241 cleanup to get the stmt refs canonicalized. Use a bool instead
10242 of gimple_modified_p since that doesn't work for NOPs. Schedule
10243 NOPs generated by folding for removal.
10245 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10246 Alan Hayward <alan.hayward@arm.com>
10247 David Sherwood <david.sherwood@arm.com>
10249 * coretypes.h (pad_direction): New enum.
10250 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10251 (FUNCTION_ARG_PADDING): Likewise.
10252 * target.def (function_arg_padding): New hook.
10253 * targhooks.h (default_function_arg_padding): Declare.
10254 * targhooks.c (default_function_arg_padding): New function.
10255 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10256 (TARGET_FUNCTION_ARG_PADDING): ...this.
10257 * doc/tm.texi: Regenerate.
10258 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10259 instead of direction.
10260 (compute_argument_addresses): Likewise.
10261 (load_register_parameters): Likewise.
10262 (emit_library_call_value_1): Likewise.
10263 (store_one_arg): Use targetm.calls.function_arg_padding instead
10264 of FUNCTION_ARG_PADDING.
10265 (must_pass_in_stack_var_size_or_pad): Likewise.
10266 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10267 (emit_group_store): Likewise.
10268 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10269 instead of FUNCTION_ARG_PADDING.
10270 (emit_push_insn): Likewise, and propagate enum change throughout
10272 * function.h (direction): Delete.
10273 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10275 * function.c (assign_parm_find_stack_rtl): Likewise.
10276 (assign_parm_setup_block_p): Likewise.
10277 (assign_parm_setup_block): Likewise.
10278 (gimplify_parameters): Likewise.
10279 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10280 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10282 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10283 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10284 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10285 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10286 (aarch64_function_arg_padding): ...this new function.
10287 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10288 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10289 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10290 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10291 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10292 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10293 (arm_pad_arg_upward): Replace with...
10294 (arm_function_arg_padding): ...this new function.
10295 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10297 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10298 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10299 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10300 (ia64_hpux_function_arg_padding): Replace with...
10301 (ia64_function_arg_padding): ...this new function. Use pad_direction
10302 instead of direction. Check for TARGET_HPUX.
10303 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10304 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10305 (iq2000_function_arg_padding): New function.
10306 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10307 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10308 (mips_function_arg_padding): ...this new function.
10309 (mips_pad_reg_upward): Update accordingly.
10310 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10311 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10312 targetm.calls.function_arg_padding.
10313 (FUNCTION_ARG_PADDING): Delete.
10314 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10315 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10316 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10317 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10318 (nios2_block_reg_padding): Return pad_direction instead of direction.
10319 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10320 instead of direction.
10321 (nios2_function_arg_padding): Likewise. Make static.
10322 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10323 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10324 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10325 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10326 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10327 (pa_function_arg_padding): Make static. Return pad_direction instead
10329 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10330 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10331 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10332 instead of direction. Use targetm.calls.function_arg_padding.
10333 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10334 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10335 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10336 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10337 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10339 (function_arg_padding): Rename to...
10340 (rs6000_function_arg_padding): ...this. Make static. Return
10341 pad_direction instead of direction.
10342 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10343 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10344 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10345 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10346 instead of direction. Use targetm.calls.function_arg_padding.
10347 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10348 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10349 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10350 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10351 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10352 (function_arg_padding): Rename to...
10353 (rs6000_function_arg_padding): ...this. Make static. Return
10354 pad_direction instead of direction.
10355 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10356 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10357 * config/s390/s390.c (s390_function_arg_padding): New function.
10358 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10359 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10360 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10361 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10362 (function_arg_padding): Rename to...
10363 (sparc_function_arg_padding): ...this. Make static. Return
10364 pad_direction instead of direction.
10365 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10366 * config/spu/spu.c (spu_function_arg_padding): New function.
10367 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10368 * system.h (FUNCTION_ARG_PADDING): Poison.
10370 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10371 Alan Hayward <alan.hayward@arm.com>
10372 David Sherwood <david.sherwood@arm.com>
10374 * target.def (modes_tieable_p): New hook.
10375 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10376 (TARGET_MODES_TIEABLE_P): ...this.
10377 * doc/tm.texi.in: Regenerate.
10378 * hooks.h (hook_bool_mode_mode_true): Declare.
10379 * hooks.c (hook_bool_mode_mode_true): New function.
10380 * combine.c (subst): Use targetm.modes_tieable_p instead of
10382 * dse.c (find_shift_sequence): Likewise.
10383 * expmed.c (extract_low_bits): Likewise.
10384 * lower-subreg.c: Include target.h.
10385 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10387 * rtlanal.c (rtx_cost): Likewise.
10388 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10389 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10390 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10391 (TARGET_MODES_TIEABLE_P): Redefine.
10392 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10393 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10394 (TARGET_MODES_TIEABLE_P): Redefine.
10395 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10396 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10397 (arc_modes_tieable_p): New function.
10398 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10399 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10400 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10401 (arm_modes_tieable_p): Make static.
10402 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10403 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10404 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10405 (TARGET_MODES_TIEABLE_P): Redefine.
10406 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10407 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10408 (TARGET_MODES_TIEABLE_P): Redefine.
10409 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10410 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10411 (cr16_modes_tieable_p): New function.
10412 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10413 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10414 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10415 (TRULY_NOOP_TRUNCATION): Update comment.
10416 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10417 (TRULY_NOOP_TRUNCATION): Update comment.
10418 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10419 (frv_modes_tieable_p): New function.
10420 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10421 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10422 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10423 (TARGET_MODES_TIEABLE_P): Redefine.
10424 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10425 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10426 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10427 (TARGET_MODES_TIEABLE_P): Redefine.
10428 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10429 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10430 (ia64_modes_tieable_p): New function.
10431 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10432 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10433 (iq2000_modes_tieable_p): New function.
10434 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10435 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10436 (lm32_modes_tieable_p): New function.
10437 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10438 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10439 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10440 (TARGET_MODES_TIEABLE_P): Redefine.
10441 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10442 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10443 (m32r_modes_tieable_p): New function.
10444 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10445 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10446 (m68k_modes_tieable_p): New function.
10447 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10448 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10449 (mcore_modes_tieable_p): New function.
10450 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10451 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10453 (TARGET_MODES_TIEABLE_P): Redefine.
10454 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10455 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10456 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10457 (TARGET_MODES_TIEABLE_P): Redefine.
10458 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10459 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10460 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10461 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10462 (mn10300_modes_tieable_p): ...this and make static.
10463 (TARGET_MODES_TIEABLE_P): Redefine.
10464 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10465 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10466 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10467 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10468 (msp430_modes_tieable_p): Make static.
10469 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10470 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10471 (TARGET_MODES_TIEABLE_P): Redefine.
10472 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10473 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10474 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10475 (TARGET_MODES_TIEABLE_P): Redefine.
10476 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10477 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10478 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10479 (TARGET_MODES_TIEABLE_P): Redefine.
10480 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10481 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10482 (pdp11_modes_tieable_p): New function.
10483 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10484 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10485 (rs6000_modes_tieable_p): New function.
10486 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10487 * config/powerpcspe/powerpcspe.md: Update comment.
10488 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10489 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10490 (TARGET_MODES_TIEABLE_P): Redefine.
10491 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10492 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10493 (rl78_modes_tieable_p): New function.
10494 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10495 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10496 (rs6000_modes_tieable_p): New function.
10497 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10498 * config/rs6000/rs6000.md: Update comment.
10499 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10500 * config/rx/rx.c (rx_modes_tieable_p): New function.
10501 (TARGET_MODES_TIEABLE_P): Redefine.
10502 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10503 * config/s390/s390.c (s390_modes_tieable_p): New function.
10504 (TARGET_MODES_TIEABLE_P): Redefine.
10505 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10506 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10507 (sh_modes_tieable_p): New function.
10508 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10509 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10510 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10511 (sparc_modes_tieable_p): Make static.
10512 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10513 * config/spu/spu.c (spu_modes_tieable_p): New function.
10514 (TARGET_MODES_TIEABLE_P): Redefine.
10515 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10516 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10517 (TARGET_MODES_TIEABLE_P): Redefine.
10518 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10519 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10520 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10521 * config/v850/v850.c (v850_modes_tieable_p): New function.
10522 (TARGET_MODES_TIEABLE_P): Redefine.
10523 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10524 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10525 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10526 (visium_modes_tieable_p): New function.
10527 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10528 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10529 (xtensa_modes_tieable_p): New function.
10530 * system.h (MODES_TIEABLE_P): Poison.
10532 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10533 Alan Hayward <alan.hayward@arm.com>
10534 David Sherwood <david.sherwood@arm.com>
10536 * target.def (hard_regno_mode_ok): New hook.
10537 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10538 (TARGET_HARD_REGNO_MODE_OK): ...this.
10539 * doc/tm.texi.in: Regenerate.
10540 * hooks.h (hook_bool_uint_mode_true): Declare.
10541 * hooks.c (hook_bool_uint_mode_true): New function.
10542 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10543 HARD_REGNO_MODE_OK.
10544 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10545 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10546 instead of HARD_REGNO_MODE_OK.
10547 * caller-save.c: Include target.h.
10548 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10549 HARD_REGNO_MODE_OK.
10550 * combine.c (can_combine_p): Likewise.
10551 (combinable_i3pat): Likewise.
10552 (can_change_dest_mode): Likewise.
10553 * expr.c (init_expr_target): Likewise.
10554 (convert_move): Likewise.
10555 (convert_modes): Likewise.
10556 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10557 (setup_prohibited_mode_move_regs): Likewise.
10558 * ira.h (target_ira): Likewise.
10559 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10560 * lra-constraints.c (process_alt_operands): Likewise.
10561 (split_reg): Likewise.
10562 * recog.c (peep2_find_free_register): Likewise.
10563 * ree.c (combine_reaching_defs): Likewise.
10564 * regcprop.c (maybe_mode_change): Likewise.
10565 * reginfo.c (init_reg_sets_1): Likewise.
10566 (choose_hard_reg_mode): Likewise.
10567 (simplifiable_subregs): Likewise.
10568 * regrename.c (check_new_reg_p): Likewise.
10569 * reload.c (find_valid_class): Likewise.
10570 (find_valid_class_1): Likewise.
10571 (reload_inner_reg_of_subreg): Likewise.
10572 (push_reload): Likewise.
10573 (combine_reloads): Likewise.
10574 (find_dummy_reload): Likewise.
10575 (find_reloads): Likewise.
10576 * reload1.c (find_reg): Likewise.
10577 (set_reload_reg): Likewise.
10578 (allocate_reload_reg): Likewise.
10579 (choose_reload_regs): Likewise.
10580 (reload_adjust_reg_for_temp): Likewise.
10581 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10582 (simplify_subreg_regno): Likewise.
10583 * sel-sched.c (init_regs_for_mode): Likewise.
10584 * varasm.c (make_decl_rtl): Likewise.
10585 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10586 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10587 HARD_REGNO_MODE_OK.
10588 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10589 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10590 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10591 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10592 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10593 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10594 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10595 (arc_mode_class): Delete.
10596 (HARD_REGNO_MODE_OK): Delete.
10597 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10598 (arc_hard_regno_mode_ok): Rename old array to...
10599 (arc_hard_regno_mode_ok_modes): ...this.
10600 (arc_conditional_register_usage): Update accordingly.
10601 (arc_mode_class): Make static.
10602 (arc_hard_regno_mode_ok): New function.
10603 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10604 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10605 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10606 (arm_hard_regno_mode_ok): Make static.
10607 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10608 HARD_REGNO_MODE_OK.
10609 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10610 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10611 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10613 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10614 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10615 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10616 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10617 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10618 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10619 * config/bfin/predicates.md (valid_reg_operand): Use
10620 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10621 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10622 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10623 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10624 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10625 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10626 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10627 (cr16_hard_regno_mode_ok): Make static and return a bool.
10628 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10629 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10630 (cris_hard_regno_mode_ok): New function.
10631 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10632 (epiphany_mode_class): Delete.
10633 (HARD_REGNO_MODE_OK): Delete.
10634 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10635 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10636 (hard_regno_mode_ok): Rename to...
10637 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10638 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10639 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10640 HARD_REGNO_MODE_OK.
10641 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10642 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10643 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10644 (frv_hard_regno_mode_ok): Make static and return a bool.
10645 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10646 HARD_REGNO_MODE_OK.
10647 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10648 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10649 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10650 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10652 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10653 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10654 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10655 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10657 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10658 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10659 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10660 (ia64_hard_regno_mode_ok): New function.
10661 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10662 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10663 (iq2000_hard_regno_mode_ok): New function.
10664 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10665 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10666 (lm32_hard_regno_mode_ok): New function.
10667 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10668 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10669 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10670 instead of HARD_REGNO_MODE_OK.
10671 (m32c_hard_regno_ok): Rename to...
10672 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10673 (m32c_cannot_change_mode_class): Update accordingly.
10674 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10675 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10676 (m32r_mode_class): Delete.
10677 (HARD_REGNO_MODE_OK): Delete.
10678 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10679 (m32r_hard_regno_mode_ok): Rename to...
10680 (m32r_hard_regno_modes): ...this.
10681 (m32r_mode_class): Make static.
10682 (m32r_hard_regno_mode_ok): New function.
10683 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10684 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10685 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10686 (m68k_hard_regno_mode_ok): Make static.
10687 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10688 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10689 (mcore_hard_regno_mode_ok): New function.
10690 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10691 (HARD_REGNO_MODE_OK): Delete.
10692 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10694 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10695 (microblaze_hard_regno_mode_ok): New function.
10696 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10697 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10698 (mips_hard_regno_mode_ok): Delete.
10699 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10700 (mips_hard_regno_mode_ok_p): ...this and make static.
10701 (mips_hard_regno_mode_ok_p): Rename to...
10702 (mips_hard_regno_mode_ok_uncached): ...this.
10703 (mips_hard_regno_mode_ok): New function.
10704 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10705 of HARD_REGNO_MODE_OK.
10706 (mips_option_override): Update after above name changes.
10707 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10708 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10709 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10710 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10711 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10712 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10713 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10714 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10715 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10716 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10717 (msp430_hard_regno_mode_ok): Make static and return a bool.
10718 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10719 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10720 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10722 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10723 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10724 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10725 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10726 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10727 (PA_HARD_REGNO_MODE_OK): ...this
10728 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10729 (PA_HARD_REGNO_MODE_OK): ...this.
10730 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10731 (pa_hard_regno_mode_ok): New function.
10732 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10733 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10734 (pdp11_hard_regno_mode_ok): New function.
10735 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10736 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10738 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10740 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10741 (rs6000_hard_regno_mode_ok): Rename to...
10742 (rs6000_hard_regno_mode_ok_uncached): ...this.
10743 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10744 (rs6000_hard_regno_mode_ok): New function.
10745 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10746 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10747 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10748 (riscv_hard_regno_mode_ok): ...this and make static.
10749 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10750 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10751 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10752 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10753 (rl78_hard_regno_mode_ok): Make static and return bool.
10754 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10755 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10757 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10758 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10759 (rs6000_hard_regno_mode_ok): Rename to...
10760 (rs6000_hard_regno_mode_ok_uncached): ...this.
10761 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10762 (rs6000_hard_regno_mode_ok): New function.
10763 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10764 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10765 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10766 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10767 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10768 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10769 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10770 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10771 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10772 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10773 (sh_hard_regno_mode_ok): Make static.
10774 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10775 instead of HARD_REGNO_MODE_OK.
10776 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10777 (sparc_mode_class): Delete.
10778 (HARD_REGNO_MODE_OK): Delete.
10779 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10780 (hard_regno_mode_classes): Make static.
10781 (sparc_mode_class): Likewise.
10782 (sparc_hard_regno_mode_ok): New function.
10783 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10784 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10785 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10787 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10788 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10789 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10790 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10791 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10793 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10794 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10795 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10796 (visium_hard_regno_mode_ok): New function.
10797 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10798 instead of HARD_REGNO_MODE_OK.
10799 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10800 (HARD_REGNO_MODE_OK): Delete.
10801 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10802 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10803 (xtensa_option_override): Update accordingly.
10804 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10805 (xtensa_hard_regno_mode_ok): New function.
10806 * system.h (HARD_REGNO_MODE_OK): Poison.
10808 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10809 Alan Hayward <alan.hayward@arm.com>
10810 David Sherwood <david.sherwood@arm.com>
10812 * target.def (hard_regno_call_part_clobbered): New hook.
10813 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10814 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10815 * doc/tm.texi: Regenerate.
10816 * hooks.h (hook_bool_uint_mode_false): Declare.
10817 * hooks.c (hook_bool_uint_mode_false): New function.
10818 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10819 * cselib.c (cselib_process_insn): Use
10820 targetm.hard_regno_call_part_clobbered instead of
10821 HARD_REGNO_CALL_PART_CLOBBERED.
10822 * ira-conflicts.c (ira_build_conflicts): Likewise.
10823 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10824 * lra-constraints.c (need_for_call_save_p): Likewise.
10825 * lra-lives.c: Include target.h.
10826 (check_pseudos_live_through_calls): Use
10827 targetm.hard_regno_call_part_clobbered instead of
10828 HARD_REGNO_CALL_PART_CLOBBERED.
10829 * regcprop.c: Include target.h.
10830 (copyprop_hardreg_forward_1): Use
10831 targetm.hard_regno_call_part_clobbered instead of
10832 HARD_REGNO_CALL_PART_CLOBBERED.
10833 * reginfo.c (choose_hard_reg_mode): Likewise.
10834 * regrename.c (check_new_reg_p): Likewise.
10835 * reload.c (find_equiv_reg): Likewise.
10836 * reload1.c (emit_reload_insns): Likewise.
10837 * sched-deps.c (deps_analyze_insn): Likewise.
10838 * sel-sched.c (init_regs_for_mode): Likewise.
10839 (mark_unavailable_hard_regs): Likewise.
10840 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10841 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10842 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10844 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10845 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10846 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10848 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10850 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10851 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10852 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10854 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10855 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10856 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10858 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10859 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10861 * config/powerpcspe/powerpcspe.c
10862 (rs6000_hard_regno_call_part_clobbered): New function.
10863 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10864 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10865 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10867 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10868 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10869 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10871 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10872 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10873 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10875 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10876 Alan Hayward <alan.hayward@arm.com>
10877 David Sherwood <david.sherwood@arm.com>
10879 * rtl.h (subreg_memory_offset): Declare.
10880 * emit-rtl.c (subreg_memory_offset): New function.
10881 * expmed.c (store_bit_field_1): Use it.
10882 * expr.c (undefined_operand_subword_p): Likewise.
10883 * simplify-rtx.c (simplify_subreg): Likewise.
10885 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10887 PR rtl-optimization/57448
10890 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10891 using atomic_load pattern.
10892 (expand_atomic_store): Likewise.
10894 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10897 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10898 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10899 handling. Use replace_call_with_value with NULL instead of
10900 gsi_replace, unlink_stmt_vdef and release_defs.
10902 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10905 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10907 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10910 * rtl.h (emit_library_call_value_1): Declare.
10911 (emit_library_call): Replace declaration with a series of overloads.
10912 Remove the parameter count argument.
10913 (emit_library_call_value): Likewise.
10914 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10915 with an "rtx_mode_t *".
10916 (emit_library_call_value): Delete.
10917 (emit_library_call): Likewise.
10918 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10919 (asan_emit_allocas_unpoison): Likewise.
10920 * builtins.c (expand_builtin_powi): Likewise.
10921 (expand_asan_emit_allocas_unpoison): Likewise.
10922 * cfgexpand.c (expand_main_function): Likewise.
10923 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10924 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10925 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10926 * config/arm/arm.c (arm_trampoline_init): Likewise.
10927 (arm_call_tls_get_addr): Likewise.
10928 (arm_expand_divmod_libfunc): Likewise.
10929 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10930 (smulsi3_highpart): Likewise.
10931 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10932 (c6x_expand_compare): Likewise.
10933 (c6x_expand_movmem): Likewise.
10934 * config/frv/frv.c (frv_trampoline_init): Likewise.
10935 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10936 (ix86_expand_divmod_libfunc): Likewise.
10937 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10938 (ia64_expand_compare): Likewise.
10939 (ia64_profile_hook): Likewise.
10940 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10941 (nonlocal_goto): Likewise.
10942 (restore_stack_nonlocal): Likewise.
10943 * config/m32r/m32r.c (block_move_call): Likewise.
10944 (m32r_trampoline_init): Likewise.
10945 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10946 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10947 (m68k_call_m68k_read_tp): Likewise.
10948 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10949 (microblaze_expand_divide): Likewise.
10950 * config/mips/mips.h (mips_args): Likewise.
10951 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10952 (MIPS_ICACHE_SYNC): Likewise.
10953 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10954 (nios2_trampoline_init): Likewise.
10955 * config/pa/pa.c (hppa_tls_call): Likewise.
10956 (pa_trampoline_init): Likewise.
10957 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10958 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10959 (expand_strn_compare): Likewise.
10960 (rs6000_generate_compare): Likewise.
10961 (rs6000_expand_float128_convert): Likewise.
10962 (output_profile_hook): Likewise.
10963 (rs6000_trampoline_init): Likewise.
10964 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10965 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10966 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10967 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10968 (rs6000_generate_compare): Likewise.
10969 (rs6000_expand_float128_convert): Likewise.
10970 (output_profile_hook): Likewise.
10971 (rs6000_trampoline_init): Likewise.
10972 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10973 * config/sh/sh.c (sh_trampoline_init): Likewise.
10974 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10975 (sparc_emit_float_lib_cmp): Likewise.
10976 (sparc32_initialize_trampoline): Likewise.
10977 (sparc64_initialize_trampoline): Likewise.
10978 (sparc_profile_hook): Likewise.
10979 * config/spu/spu.c (ea_load_store): Likewise.
10980 * config/spu/spu.md (floatunssidf2): Likewise.
10981 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10982 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10983 * config/visium/visium.c (expand_block_move_4): Likewise.
10984 (expand_block_move_2): Likewise.
10985 (expand_block_move_1): Likewise.
10986 (expand_block_set_4): Likewise.
10987 (expand_block_set_2): Likewise.
10988 (expand_block_set_1): Likewise.
10989 (visium_trampoline_init): Likewise.
10990 (visium_profile_hook): Likewise.
10991 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10992 (xtensa_setup_frame_addresses): Likewise.
10993 (xtensa_trampoline_init): Likewise.
10994 * except.c (sjlj_emit_function_enter): Likewise.
10995 (sjlj_emit_function_exit): Likewise.
10996 * explow.c (allocate_dynamic_stack_space): Likewise.
10997 (probe_stack_range): Likewise.
10998 * expr.c (convert_mode_scalar): Likewise.
10999 * optabs.c (expand_binop): Likewise.
11000 (expand_twoval_binop_libfunc): Likewise.
11001 (expand_unop): Likewise.
11002 (prepare_cmp_insn): Likewise.
11003 (prepare_float_lib_cmp): Likewise.
11004 (expand_float): Likewise.
11005 (expand_fix): Likewise.
11006 (expand_fixed_convert): Likewise.
11007 (maybe_emit_sync_lock_test_and_set): Likewise.
11008 (expand_atomic_compare_and_swap): Likewise.
11009 (expand_mem_thread_fence): Likewise.
11010 (expand_atomic_fetch_op): Likewise.
11012 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11014 * doc/generic.texi (OpenACC): Adjust URL.
11015 * doc/invoke.texi (C Dialect Options): Ditto.
11017 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11019 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11020 predicate for operand 1. Add (m,<S>) constraint.
11021 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11022 Prevent memory operand 1 with register operand 2.
11024 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11026 PR rtl-optimization/82024
11027 * combine.c (try_combine): If the combination result is a PARALLEL,
11028 and we only need to retain the SET in there that would be placed
11029 at I2, check that we can place that at I3 instead, before doing so.
11031 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11034 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11036 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11037 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11040 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11041 Jeff Law <law@redhat.com>
11043 * varasm.c (bss_initializer_p): Do not put constants into .bss
11044 (categorize_decl_for_section): Handle bss_initializer_p returning
11045 false when DECL_INITIAL is NULL.
11047 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11050 * config/s390/s390.c (s390_can_inline_p): New function.
11052 2017-09-01 Jeff Law <law@redhat.com>
11054 PR tree-optimization/82052
11055 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11056 Always initialize the returned slot after a hash table miss
11057 when INSERT is true.
11059 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11061 * config/s390/s390.md (mem_signal_fence): Remove.
11062 * doc/md.texi (mem_signal_fence): Remove.
11063 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11065 * target-insns.def (mem_signal_fence): Remove.
11067 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11070 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11073 * asan.c (create_odr_indicator): Strip name encoding from assembler
11074 name before appending it after __odr_asan_.
11076 2017-09-01 Martin Liska <mliska@suse.cz>
11078 PR tree-optimization/82059
11079 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11080 frequency only when an edge is redirected.
11082 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11084 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11085 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11086 (arc_conditional_register_usage): Remove ARC600 lp_count
11088 (arc_file_start): Emit Tag_ARC_CPU_variation.
11089 (arc_can_use_doloop_p): New conditions to use ZOLs.
11090 (hwloop_fail): New function.
11091 (hwloop_optimize): Likewise.
11092 (hwloop_pattern_reg): Likewise.
11093 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11094 (arc_reorg_loops): New function, calls reorg_loops.
11095 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11096 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11098 (arc_loop_hazard): Remove function, functionality moved into
11100 (arc_hazard): Remove arc_loop_hazard call.
11101 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11102 into hwloop_optimize.
11103 (arc_label_align): Remove ZOL handling.
11104 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11105 * config/arc/arc.md (doloop_begin): Remove pattern.
11106 (doloop_begin_i): Likewise.
11107 (doloop_end_i): Likewise.
11108 (doloop_fallback): Likewise.
11109 (doloop_fallback_m): Likewise.
11110 (doloop_end): Reimplement expand.
11111 (arc_lp): New pattern for LP instruction.
11112 (loop_end): New pattern.
11113 (loop_fail): Likewise.
11114 (decrement_and_branch_until_zero): Likewise.
11115 * config/arc/arc.opt (mlpc-width): New option.
11116 * doc/invoke.texi (mlpc-width): Document option.
11118 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11120 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11121 (arc_ccfsm_advance): Fix checking for delay slots.
11122 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11124 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11126 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11128 (movhi_insn): Update store instruction constraint which are saving
11129 6-bit short immediates.
11130 (movsi_insn): Consider also short scaled load operations.
11131 (zero_extendhisi2_i): Use Usd constraint instead of T.
11132 (extendhisi2_i): Add q constraint.
11133 (arc_clzsi2): Add type and length attributes.
11134 (arc_ctzsi2): Likewise.
11135 * config/arc/constraints.md (Usc): Update constraint, the
11136 assembler can parse two relocations for a single instruction.
11138 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11140 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11141 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11143 2017-08-31 Olivier Hainque <hainque@adacore.com>
11145 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11146 match as powerpc-wrs-vxworks*.
11148 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11150 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11151 register constraint for by-element operand.
11152 (aarch64_mls_elt_merge<mode>): Likewise.
11154 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11156 * config/arc/arc.c (arc_can_follow_jump): Check for short
11159 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11161 * config.gcc: Use g.opt for arc.
11162 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11163 functionality moved to ...
11164 (legitimate_scaled_address_p): New function, ...here.
11165 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11166 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11167 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11169 (arc_override_options): Handle G option.
11170 (arc_output_pic_addr_const): Correct function definition.
11171 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11172 (arc_decl_anon_ns_mem_p): Delete.
11173 (arc_in_small_data_p): Overhaul this function to take into
11174 consideration the value given via G option.
11175 (arc_rewrite_small_data_1): Renamed and corrected old
11176 arc_rewrite_small_data function.
11177 (arc_rewrite_small_data): New function.
11178 (small_data_pattern): Don't use pic_offset_table_rtx.
11179 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11180 * config/arc/simdext.md (movmisalignv2hi): Use
11181 prepare_move_operands function.
11183 (movmisalign*): Likewise.
11184 * doc/invoke.texi (ARC options): Document -G option.
11186 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11188 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11190 * config/arc/arc.c (arc_print_operand): Output scalled address for
11191 sdata whenever is possible.
11192 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11193 load/stores are available.
11194 (compact_sda_memory_operand): Check for the alignment required by
11195 code density instructions.
11196 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11198 * config/arc/constraints.md (Usd): Update constraint.
11199 (Us0): New constraint.
11200 (Usc): Update constraint.
11202 2017-08-31 Richard Biener <rguenther@suse.de>
11204 PR middle-end/82054
11205 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11206 function only once.
11208 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11210 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11211 Resize type_signature.
11213 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11214 Alan Hayward <alan.hayward@arm.com>
11215 David Sherwood <david.sherwood@arm.com>
11217 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11218 subregs whose inner modes can be stored in GPRs.
11219 (aarch64_classify_index): Likewise.
11221 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11222 Alan Hayward <alan.hayward@arm.com>
11223 David Sherwood <david.sherwood@arm.com>
11225 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11226 (V_INT_EQUIV): ...this.
11227 (v_cmp_result): Rename to...
11228 (v_int_equiv): ...this.
11229 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11230 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11231 (copysign<mode>3): Likewise.
11232 (aarch64_simd_bsl<mode>_internal): Likewise.
11233 (aarch64_simd_bsl<mode>): Likewise.
11234 (vec_cmp<mode><mode>): Likewise.
11235 (vcond<mode><mode>): Likewise.
11236 (vcond<v_cmp_mixed><mode>): Likewise.
11237 (vcondu<mode><v_cmp_mixed>): Likewise.
11238 (aarch64_cm<optab><mode>): Likewise.
11239 (aarch64_cmtst<mode>): Likewise.
11240 (aarch64_fac<optab><mode>): Likewise.
11241 (vec_perm_const<mode>): Likewise.
11242 (vcond_mask_<mode><v_cmp_result>): Rename to...
11243 (vcond_mask_<mode><v_int_equiv>): ...this.
11244 (vec_cmp<mode><v_cmp_result>): Rename to...
11245 (vec_cmp<mode><v_int_equiv>): ...this.
11247 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11248 Alan Hayward <alan.hayward@arm.com>
11249 David Sherwood <david.sherwood@arm.com>
11251 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11253 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11254 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11255 (UNSPEC_LD4_DREG): New unspecs.
11256 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11257 (aarch64_ld2<mode>_dreg_be): Replace with...
11258 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11260 (aarch64_ld3<mode>_dreg_le)
11261 (aarch64_ld3<mode>_dreg_be): Replace with...
11262 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11264 (aarch64_ld4<mode>_dreg_le)
11265 (aarch64_ld4<mode>_dreg_be): Replace with...
11266 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11269 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11271 PR tree-optimization/81987
11272 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11273 insert an initializer in a location not dominated by the stride
11276 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11278 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11279 on the entire header of the finally block in the fallthru case.
11281 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11283 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11285 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11287 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11288 rs6000_emit_move_from_cr and call renamed function.
11289 (rs6000_emit_prologue): Call renamed functions.
11290 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11291 movesi_from_cr, remove volatile CRs.
11293 2017-08-30 Jon Beniston <jon@beniston.com>
11294 Richard Biener <rguenther@suse.de>
11296 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11297 of VECTOR_MODE_P check.
11298 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11299 element vector types.
11301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11303 * df.h (df_read_modify_subreg_p): Remove in favor of...
11304 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11305 const_rtx instead of an rtx.
11306 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11307 * df-problems.c (df_word_lr_mark_ref): Likewise.
11308 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11309 (mark_pseudo_reg_dead): Likewise.
11310 (mark_ref_dead): Likewise.
11311 * reginfo.c (init_subregs_of_mode): Likewise.
11312 * sched-deps.c (sched_analyze_1): Likewise.
11313 * df-scan.c (df_def_record_1): Likewise.
11314 (df_uses_record): Likewise.
11315 (df_read_modify_subreg_p): Remove in favor of...
11316 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11317 const_rtx instead of an rtx.
11319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11320 Alan Hayward <alan.hayward@arm.com>
11321 David Sherwood <david.sherwood@arm.com>
11323 * rtl.h (partial_subreg_p): New function.
11324 * caller-save.c (save_call_clobbered_regs): Use it.
11325 * calls.c (expand_call): Likewise.
11326 * combine.c (combinable_i3pat): Likewise.
11327 (simplify_set): Likewise.
11328 (make_extraction): Likewise.
11329 (make_compound_operation_int): Likewise.
11330 (gen_lowpart_or_truncate): Likewise.
11331 (force_to_mode): Likewise.
11332 (make_field_assignment): Likewise.
11333 (reg_truncated_to_mode): Likewise.
11334 (record_truncated_value): Likewise.
11335 (move_deaths): Likewise.
11336 * cse.c (record_jump_cond): Likewise.
11337 (cse_insn): Likewise.
11338 * cselib.c (cselib_lookup_1): Likewise.
11339 * expmed.c (extract_bit_field_using_extv): Likewise.
11340 * function.c (assign_parm_setup_reg): Likewise.
11341 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11342 * ira-build.c (create_insn_allocnos): Likewise.
11343 * lra-coalesce.c (merge_pseudos): Likewise.
11344 * lra-constraints.c (match_reload): Likewise.
11345 (simplify_operand_subreg): Likewise.
11346 (curr_insn_transform): Likewise.
11347 * lra-lives.c (process_bb_lives): Likewise.
11348 * lra.c (new_insn_reg): Likewise.
11349 (lra_substitute_pseudo): Likewise.
11350 * regcprop.c (mode_change_ok): Likewise.
11351 (maybe_mode_change): Likewise.
11352 (copyprop_hardreg_forward_1): Likewise.
11353 * reload.c (push_reload): Likewise.
11354 (find_reloads): Likewise.
11355 (find_reloads_subreg_address): Likewise.
11356 * reload1.c (alter_reg): Likewise.
11357 (eliminate_regs_1): Likewise.
11358 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11360 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11362 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11363 back to if statements, including unpack.
11365 2017-08-30 Martin Liska <mliska@suse.cz>
11367 PR inline-asm/82001
11368 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11370 (func_checker::compare_asm_inputs_outputs): ... this function.
11371 (func_checker::compare_gimple_asm): Use the function to compare
11372 also ASM constrains.
11373 * ipa-icf-gimple.h: Rename the function.
11375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11376 Alan Hayward <alan.hayward@arm.com>
11377 David Sherwood <david.sherwood@arm.com>
11379 * coretypes.h (complex_mode): New type.
11380 * gdbhooks.py (build_pretty_printer): Handle it.
11381 * machmode.h (complex_mode): New class.
11382 (complex_mode::includes_p): New function.
11383 (is_complex_int_mode): Likewise.
11384 (is_complex_float_mode): Likewise.
11385 * genmodes.c (get_mode_class): Handle complex mode classes.
11386 * function.c (expand_function_end): Use is_complex_int_mode.
11388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11389 Alan Hayward <alan.hayward@arm.com>
11390 David Sherwood <david.sherwood@arm.com>
11392 * coretypes.h (scalar_mode_pod): New typedef.
11393 * gdbhooks.py (build_pretty_printer): Handle it.
11394 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11395 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11396 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11397 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11398 as_a <scalar_mode>.
11400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11401 Alan Hayward <alan.hayward@arm.com>
11402 David Sherwood <david.sherwood@arm.com>
11404 * machmode.h (mode_for_vector): Take a scalar_mode instead
11406 * stor-layout.c (mode_for_vector): Likewise.
11407 * explow.c (promote_mode): Use as_a <scalar_mode>.
11408 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11411 Alan Hayward <alan.hayward@arm.com>
11412 David Sherwood <david.sherwood@arm.com>
11414 * target.def (preferred_simd_mode): Take a scalar_mode
11415 instead of a machine_mode.
11416 * targhooks.h (default_preferred_simd_mode): Likewise.
11417 * targhooks.c (default_preferred_simd_mode): Likewise.
11418 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11419 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11420 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11421 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11422 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11423 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11424 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11425 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11427 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11428 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11429 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11430 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11431 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11432 * doc/tm.texi: Regenerate.
11433 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11437 Alan Hayward <alan.hayward@arm.com>
11438 David Sherwood <david.sherwood@arm.com>
11440 * target.def (scalar_mode_supported_p): Take a scalar_mode
11441 instead of a machine_mode.
11442 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11443 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11444 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11445 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11446 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11447 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11448 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11449 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11450 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11451 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11452 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11453 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11454 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11455 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11457 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11458 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11459 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11460 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11461 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11463 * doc/tm.texi: Regenerate.
11465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11466 Alan Hayward <alan.hayward@arm.com>
11467 David Sherwood <david.sherwood@arm.com>
11469 * coretypes.h (opt_scalar_mode): New typedef.
11470 * gdbhooks.py (build_pretty_printers): Handle it.
11471 * machmode.h (mode_iterator::get_2xwider): Add overload for
11473 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11475 * expr.c (convert_mode_scalar): Likewise.
11476 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11477 * optabs.c (expand_float): Likewise.
11478 (expand_fix): Likewise.
11479 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11482 Alan Hayward <alan.hayward@arm.com>
11483 David Sherwood <david.sherwood@arm.com>
11485 * optabs.c (expand_float): Explicitly check for scalars before
11486 using a branching expansion.
11487 (expand_fix): Likewise.
11489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11490 Alan Hayward <alan.hayward@arm.com>
11491 David Sherwood <david.sherwood@arm.com>
11493 * expr.c (convert_mode): Split scalar handling out into...
11494 (convert_mode_scalar): ...this new function. Treat the modes
11497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11498 Alan Hayward <alan.hayward@arm.com>
11499 David Sherwood <david.sherwood@arm.com>
11501 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11503 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11506 Alan Hayward <alan.hayward@arm.com>
11507 David Sherwood <david.sherwood@arm.com>
11509 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11510 rather than a machine_mode.
11511 (fixed_from_string): Likewise.
11512 (fixed_convert): Likewise.
11513 (fixed_convert_from_int): Likewise.
11514 (fixed_convert_from_real): Likewise.
11515 (real_convert_from_fixed): Likewise.
11516 * fixed-value.c (fixed_from_double_int): Likewise.
11517 (fixed_from_string): Likewise.
11518 (fixed_convert): Likewise.
11519 (fixed_convert_from_int): Likewise.
11520 (fixed_convert_from_real): Likewise.
11521 (real_convert_from_fixed): Likewise.
11522 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
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 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11529 of separate mode class checks. Do not allow vector modes here.
11530 (immed_wide_int_const): Use as_a <scalar_mode>.
11531 * explow.c (trunc_int_for_mode): Likewise.
11532 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11533 (wi::shwi): Likewise.
11534 (wi::min_value): Likewise.
11535 (wi::max_value): Likewise.
11536 * dwarf2out.c (loc_descriptor): Likewise.
11537 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11538 for CONST_WIDE_INT.
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 * tree.h (SCALAR_TYPE_MODE): New macro.
11545 * expr.c (expand_expr_addr_expr_1): Use it.
11546 (expand_expr_real_2): Likewise.
11547 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11548 (fold_convert_const_fixed_from_int): Likewise.
11549 (fold_convert_const_fixed_from_real): Likewise.
11550 (native_encode_fixed): Likewise
11551 (native_encode_complex): Likewise
11552 (native_encode_vector): Likewise.
11553 (native_interpret_fixed): Likewise.
11554 (native_interpret_real): Likewise.
11555 (native_interpret_complex): Likewise.
11556 (native_interpret_vector): Likewise.
11557 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11558 (simd_clone_adjust_argument_types): Likewise.
11559 (simd_clone_init_simd_arrays): Likewise.
11560 (simd_clone_adjust): Likewise.
11561 * stor-layout.c (layout_type): Likewise.
11562 * tree.c (build_minus_one_cst): Likewise.
11563 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11564 * tree-inline.c (estimate_move_cost): Likewise.
11565 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11566 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11567 (vectorizable_reduction): Likewise.
11568 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11569 (vect_recog_mixed_size_cond_pattern): Likewise.
11570 (check_bool_pattern): Likewise.
11571 (adjust_bool_pattern): Likewise.
11572 (search_type_for_mask_1): Likewise.
11573 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11574 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11575 (vectorizable_load): Likewise.
11576 (vectorizable_store): Likewise.
11577 * ubsan.c (ubsan_encode_value): Likewise.
11578 * varasm.c (output_constant): Likewise.
11580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11581 Alan Hayward <alan.hayward@arm.com>
11582 David Sherwood <david.sherwood@arm.com>
11584 * coretypes.h (scalar_mode): New class.
11585 * machmode.h (scalar_mode): Likewise.
11586 (scalar_mode::includes_p): New function.
11587 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11588 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11589 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11590 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11591 * expmed.c (store_bit_field_1): Likewise.
11592 (extract_bit_field_1): Likewise.
11593 * expr.c (write_complex_part): Likewise.
11594 (read_complex_part): Likewise.
11595 (emit_move_complex_push): Likewise.
11596 (expand_expr_real_2): Likewise.
11597 * function.c (assign_parm_setup_reg): Likewise.
11598 (assign_parms_unsplit_complex): Likewise.
11599 * optabs.c (expand_binop): Likewise.
11600 * rtlanal.c (subreg_get_info): Likewise.
11601 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11602 * varasm.c (output_constant_pool_2): Likewise.
11604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11605 Alan Hayward <alan.hayward@arm.com>
11606 David Sherwood <david.sherwood@arm.com>
11608 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11610 (expmed_mult_highpart_optab): Likewise.
11611 (expmed_mult_highpart): Likewise.
11613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11614 Alan Hayward <alan.hayward@arm.com>
11615 David Sherwood <david.sherwood@arm.com>
11617 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11618 instead of a machine_mode.
11619 (builtin_memset_read_str): Likewise.
11620 * builtins.c (c_readstr): Likewise.
11621 (builtin_memcpy_read_str): Likewise.
11622 (builtin_strncpy_read_str): Likewise.
11623 (builtin_memset_read_str): Likewise.
11624 (builtin_memset_gen_str): Likewise.
11625 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11626 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11627 instead of a machine_mode.
11628 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11630 (make_extraction): Likewise.
11631 (try_widen_shift_mode): Take and return scalar_int_modes instead
11633 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11634 a scalar_int_mode instead of a machine_mode.
11635 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11636 (avr_addr_space_pointer_mode): Likewise.
11637 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11638 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11639 (msp430_unwind_word_mode): Likewise.
11640 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11641 (spu_addr_space_pointer_mode): Likewise.
11642 (spu_addr_space_address_mode): Likewise.
11643 (spu_libgcc_cmp_return_mode): Likewise.
11644 (spu_libgcc_shift_count_mode): Likewise.
11645 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11646 (rl78_addr_space_pointer_mode): Likewise.
11647 (fl78_unwind_word_mode): Likewise.
11648 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11650 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11651 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11652 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11653 (mips_valid_pointer_mode): Likewise.
11654 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11655 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11656 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11658 (ft32_addr_space_address_mode): Likewise.
11659 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11660 scalar_int_mode instead of a machine_mode.
11661 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11663 (m32c_addr_space_address_mode): Likewise.
11664 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11665 (rs6000_eh_return_filter_mode): Likewise.
11666 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11667 (rs6000_eh_return_filter_mode): Likewise.
11668 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11669 (s390_libgcc_shift_count_mode): Likewise.
11670 (s390_unwind_word_mode): Likewise.
11671 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11673 * target.def (mode_rep_extended): Likewise.
11674 (valid_pointer_mode): Likewise.
11675 (addr_space.valid_pointer_mode): Likewise.
11676 (eh_return_filter_mode): Return a scalar_int_mode rather than
11678 (libgcc_cmp_return_mode): Likewise.
11679 (libgcc_shift_count_mode): Likewise.
11680 (unwind_word_mode): Likewise.
11681 (addr_space.pointer_mode): Likewise.
11682 (addr_space.address_mode): Likewise.
11683 * doc/tm.texi: Regenerate.
11684 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11686 (do_jump): Use scalar_int_mode for local variables.
11687 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11688 rather than a machine_mode.
11689 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11690 (scompare_loc_descriptor_wide): Likewise.
11691 (scompare_loc_descriptor_narrow): Likewise.
11692 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11694 * except.c (sjlj_emit_dispatch_table): Likewise.
11695 (expand_builtin_eh_copy_values): Likewise.
11696 * explow.c (convert_memory_address_addr_space_1): Likewise.
11697 Take a scalar_int_mode rather than a machine_mode.
11698 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11699 than a machine_mode.
11700 (memory_address_addr_space): Use scalar_int_mode for local variables.
11701 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11702 rather than a machine_mode.
11703 * expmed.c (mask_rtx): Likewise.
11704 (init_expmed_one_conv): Likewise.
11705 (expand_mult_highpart_adjust): Likewise.
11706 (extract_high_half): Likewise.
11707 (expmed_mult_highpart_optab): Likewise.
11708 (expmed_mult_highpart): Likewise.
11709 (expand_smod_pow2): Likewise.
11710 (expand_sdiv_pow2): Likewise.
11711 (emit_store_flag_int): Likewise.
11712 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11714 (extract_low_bits): Likewise.
11715 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11717 * expr.c (pieces_addr::adjust): Likewise.
11718 (can_store_by_pieces): Likewise.
11719 (store_by_pieces): Likewise.
11720 (clear_by_pieces_1): Likewise.
11721 (expand_expr_addr_expr_1): Likewise.
11722 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11723 (expand_expr_real_1): Likewise.
11724 (try_casesi): Likewise.
11725 * final.c (shorten_branches): Likewise.
11726 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11727 type of "mode" to machine_mode.
11728 * internal-fn.c (expand_arith_overflow_result_store): Take a
11729 scalar_int_mode rather than a machine_mode.
11730 (expand_mul_overflow): Use scalar_int_mode for local variables.
11731 * loop-doloop.c (doloop_modify): Likewise.
11732 (doloop_optimize): Likewise.
11733 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11734 than a machine_mode.
11735 (expand_doubleword_shift_condmove): Likewise.
11736 (expand_doubleword_shift): Likewise.
11737 (expand_doubleword_clz): Likewise.
11738 (expand_doubleword_popcount): Likewise.
11739 (expand_doubleword_parity): Likewise.
11740 (expand_absneg_bit): Use scalar_int_mode for local variables.
11741 (prepare_float_lib_cmp): Likewise.
11742 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11743 rather than a machine_mode.
11744 (convert_memory_address_addr_space): Likewise.
11745 (get_mode_bounds): Likewise.
11746 (get_address_mode): Return a scalar_int_mode rather than a
11748 * rtlanal.c (get_address_mode): Likewise.
11749 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11750 than a machine_mode.
11751 * targhooks.c (default_mode_rep_extended): Likewise.
11752 (default_valid_pointer_mode): Likewise.
11753 (default_addr_space_valid_pointer_mode): Likewise.
11754 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11755 than a machine_mode.
11756 (default_libgcc_cmp_return_mode): Likewise.
11757 (default_libgcc_shift_count_mode): Likewise.
11758 (default_unwind_word_mode): Likewise.
11759 (default_addr_space_pointer_mode): Likewise.
11760 (default_addr_space_address_mode): Likewise.
11761 * targhooks.h (default_eh_return_filter_mode): Likewise.
11762 (default_libgcc_cmp_return_mode): Likewise.
11763 (default_libgcc_shift_count_mode): Likewise.
11764 (default_unwind_word_mode): Likewise.
11765 (default_addr_space_pointer_mode): Likewise.
11766 (default_addr_space_address_mode): Likewise.
11767 (default_mode_rep_extended): Take a scalar_int_mode rather than
11769 (default_valid_pointer_mode): Likewise.
11770 (default_addr_space_valid_pointer_mode): Likewise.
11771 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11773 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11774 rather than a machine_mode.
11775 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11776 for local variables.
11777 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11778 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11779 than a machine_mode.
11781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11782 Alan Hayward <alan.hayward@arm.com>
11783 David Sherwood <david.sherwood@arm.com>
11785 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11786 the mode argument to scalar_int_mode.
11787 (do_jump_by_parts_zero_rtx): Likewise.
11788 (do_jump_by_parts_equality_rtx): Likewise.
11789 (do_jump_by_parts_greater): Take a mode argument.
11790 (do_jump_by_parts_equality): Likewise.
11791 (do_jump_1): Update calls accordingly.
11793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11794 Alan Hayward <alan.hayward@arm.com>
11795 David Sherwood <david.sherwood@arm.com>
11797 * is-a.h (safe_dyn_cast): New function.
11798 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11799 (jump_table_for_label): Likewise.
11800 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11801 instead of an rtx_insn *.
11802 (shorten_branches): Use dyn_cast instead of LABEL_P and
11803 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11804 rtx_jump_table_data::get_data_mode.
11805 (final_scan_insn): Likewise.
11807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11808 Alan Hayward <alan.hayward@arm.com>
11809 David Sherwood <david.sherwood@arm.com>
11811 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11812 trying to combine a full-register integer set with a subreg
11815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11816 Alan Hayward <alan.hayward@arm.com>
11817 David Sherwood <david.sherwood@arm.com>
11819 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11820 that is always either address_mode or pointer_mode.
11822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11823 Alan Hayward <alan.hayward@arm.com>
11824 David Sherwood <david.sherwood@arm.com>
11826 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11827 when the two are known to be equal.
11829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11830 Alan Hayward <alan.hayward@arm.com>
11831 David Sherwood <david.sherwood@arm.com>
11833 * simplify-rtx.c (simplify_const_unary_operation): Use
11834 is_a <scalar_int_mode> instead of checking for a nonzero
11835 precision. Forcibly convert op_mode to a scalar_int_mode
11836 in that case. More clearly differentiate the operand and
11837 result modes and use the former when deciding what the value
11838 of a count-bits operation should be. Use is_int_mode instead
11839 of checking for a MODE_INT. Remove redundant check for whether
11840 this mode has a zero precision.
11842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11843 Alan Hayward <alan.hayward@arm.com>
11844 David Sherwood <david.sherwood@arm.com>
11846 * optabs.c (widen_leading): Change the type of the mode argument
11847 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11848 (widen_bswap): Likewise.
11849 (expand_parity): Likewise.
11850 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11851 (expand_ffs): Likewise.
11852 (epand_unop): Check for scalar integer modes before calling the
11855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11856 Alan Hayward <alan.hayward@arm.com>
11857 David Sherwood <david.sherwood@arm.com>
11859 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11861 (expand_expr_real_1): Update call accordingly.
11863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11864 Alan Hayward <alan.hayward@arm.com>
11865 David Sherwood <david.sherwood@arm.com>
11867 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11868 value_mode arguments. Use scalar_int_mode internally.
11869 (store_bit_field_1): Rename the new integer mode from imode
11870 to op0_mode and use it instead of GET_MODE (op0). Update calls
11871 to store_split_bit_field, store_bit_field_using_insv and
11872 store_fixed_bit_field.
11873 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11874 Use scalar_int_mode internally. Use a bit count rather than a mode
11875 when calculating the largest bit size for get_best_mode.
11876 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11877 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11878 Remove assertion that OP0 has a scalar integer mode.
11879 (store_split_bit_field): Add op0_mode and value_mode arguments.
11880 Update calls to extract_fixed_bit_field.
11881 (extract_bit_field_using_extv): Add an op0_mode argument.
11882 Use scalar_int_mode internally.
11883 (extract_bit_field_1): Rename the new integer mode from imode to
11884 op0_mode and use it instead of GET_MODE (op0). Update calls to
11885 extract_split_bit_field, extract_bit_field_using_extv and
11886 extract_fixed_bit_field.
11887 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11888 to extract_split_bit_field and extract_fixed_bit_field_1.
11889 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11890 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11891 on the target mode.
11892 (extract_split_bit_field): Add an op0_mode argument. Update call
11893 to extract_fixed_bit_field.
11895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11896 Alan Hayward <alan.hayward@arm.com>
11897 David Sherwood <david.sherwood@arm.com>
11899 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11900 * explow.c (hard_function_value): Likewise.
11901 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11902 convert_to_mode call outside the loop.
11903 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11904 for the mode iterator. Require the mode specified by max_pieces
11906 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11908 (copy_blkmode_to_reg): Likewise.
11909 (set_storage_via_setmem): Likewise.
11910 * optabs.c (prepare_cmp_insn): Likewise.
11911 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11912 * stor-layout.c (finish_bitfield_representative): Likewise.
11914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11915 Alan Hayward <alan.hayward@arm.com>
11916 David Sherwood <david.sherwood@arm.com>
11918 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11919 * expr.c (convert_move): Use them.
11920 (convert_modes): Likewise.
11921 (store_expr_with_bounds): Likewise.
11923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11924 Alan Hayward <alan.hayward@arm.com>
11925 David Sherwood <david.sherwood@arm.com>
11927 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11928 parameter for the mode of "x". Remove the "known_x", "known_mode"
11929 and "known_ret" arguments. Change the type of the mode argument
11930 to scalar_int_mode.
11931 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11932 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11933 (reg_num_sign_bit_copies_for_combine): Likewise.
11934 * rtlanal.c (nonzero_bits1): Likewise.
11935 (num_sign_bit_copies1): Likewise.
11936 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11937 (reg_num_sign_bit_copies_general): Likewise.
11938 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11939 (reg_nonzero_bits_general): Likewise.
11941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11942 Alan Hayward <alan.hayward@arm.com>
11943 David Sherwood <david.sherwood@arm.com>
11945 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11946 than in subroutines. Return 1 for non-integer modes.
11947 (cached_num_sign_bit_copies): Change the type of the mode parameter
11948 to scalar_int_mode.
11949 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11950 classes. Handle CONST_INT_P first and then check whether X also
11951 has a scalar integer mode. Check the same thing for inner registers
11952 of a SUBREG and for values that are being extended or truncated.
11954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11955 Alan Hayward <alan.hayward@arm.com>
11956 David Sherwood <david.sherwood@arm.com>
11958 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11959 in subroutines. Return the mode mask for non-integer modes.
11960 (cached_nonzero_bits): Change the type of the mode parameter
11961 to scalar_int_mode.
11962 (nonzero_bits1): Likewise. Remove early exit for other mode
11963 classes. Handle CONST_INT_P first and then check whether X
11964 also has a scalar integer mode.
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 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11971 Return a scalar_int_mode and assert that the size is greater than
11973 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11974 (op_by_pieces_d::op_by_pieces_d): Likewise.
11975 (op_by_pieces_d::run): Likewise.
11976 (can_store_by_pieces): Likewise.
11978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11979 Alan Hayward <alan.hayward@arm.com>
11980 David Sherwood <david.sherwood@arm.com>
11982 * combine.c (extract_left_shift): Add a mode argument and update
11984 (make_compound_operation_int): Change the type of the mode parameter
11985 to scalar_int_mode and update the call to extract_left_shift.
11987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11988 Alan Hayward <alan.hayward@arm.com>
11989 David Sherwood <david.sherwood@arm.com>
11991 * combine.c (simplify_and_const_int): Change the type of the mode
11992 parameter to scalar_int_mode.
11993 (simplify_and_const_int_1): Likewise. Update recursive call.
11995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11996 Alan Hayward <alan.hayward@arm.com>
11997 David Sherwood <david.sherwood@arm.com>
11999 * combine.c (simplify_compare_const): Check that the mode is a
12000 scalar_int_mode (rather than VOIDmode) before testing its
12002 (simplify_comparison): Move COMPARISON_P handling out of the
12003 loop and restrict the latter part of the loop to scalar_int_modes.
12004 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12005 and when considering SUBREG_REGs. Use is_int_mode instead of
12006 checking GET_MODE_CLASS against MODE_INT.
12008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12009 Alan Hayward <alan.hayward@arm.com>
12010 David Sherwood <david.sherwood@arm.com>
12012 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12013 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12014 shift_unit_mode and for modes involved in scalar shifts.
12016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12017 Alan Hayward <alan.hayward@arm.com>
12018 David Sherwood <david.sherwood@arm.com>
12020 * combine.c (force_int_to_mode): New function, split out from...
12021 (force_to_mode): ...here. Keep xmode up-to-date and use it
12022 instead of GET_MODE (x).
12024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12025 Alan Hayward <alan.hayward@arm.com>
12026 David Sherwood <david.sherwood@arm.com>
12028 * optabs-query.h (extraction_insn::struct_mode): Change type to
12029 opt_scalar_int_mode and update comment.
12030 (extraction_insn::field_mode): Change type to scalar_int_mode.
12031 (extraction_insn::pos_mode): Likewise.
12032 * combine.c (make_extraction): Update accordingly.
12033 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12034 (get_optab_extraction_insn): Likewise.
12035 * recog.c (simplify_while_replacing): Likewise.
12036 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12037 parameter to opt_scalar_int_mode.
12039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12040 Alan Hayward <alan.hayward@arm.com>
12041 David Sherwood <david.sherwood@arm.com>
12043 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12044 to a scalar_int_mode instead of a machine_mode.
12045 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12046 (get_best_mode): Return a boolean and use a pointer argument to store
12047 the selected mode. Replace the limit mode parameter with a bit limit.
12048 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12049 for the values returned by bit_field_mode_iterator::next_mode.
12050 (store_bit_field): Update call to get_best_mode.
12051 (store_fixed_bit_field): Likewise.
12052 (extract_fixed_bit_field): Likewise.
12053 * expr.c (optimize_bitfield_assignment_op): Likewise.
12054 * fold-const.c (optimize_bit_field_compare): Likewise.
12055 (fold_truth_andor_1): Likewise.
12056 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12057 Update for new type of m_mode.
12058 (get_best_mode): As above.
12060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12061 Alan Hayward <alan.hayward@arm.com>
12062 David Sherwood <david.sherwood@arm.com>
12064 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12065 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12066 (store_bit_field): Check is_a <scalar_int_mode> before calling
12067 strict_volatile_bitfield_p.
12068 (extract_bit_field): Likewise.
12070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12071 Alan Hayward <alan.hayward@arm.com>
12072 David Sherwood <david.sherwood@arm.com>
12074 * target.def (cstore_mode): Return a scalar_int_mode.
12075 * doc/tm.texi: Regenerate.
12076 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12077 * targhooks.h (default_cstore_mode): Likewise.
12078 * targhooks.c (default_cstore_mode): Likewise, using a forced
12080 * expmed.c (emit_cstore): Expect the target of the cstore to be
12083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12084 Alan Hayward <alan.hayward@arm.com>
12085 David Sherwood <david.sherwood@arm.com>
12087 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12089 (niter_desc): Likewise mode.
12090 (iv_analyze): Add a mode parameter.
12092 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12093 and change its type to scalar_int_mode.
12094 * loop-iv.c: Update commentary at head of file.
12095 (iv_constant): Pass the mode paraeter before the rtx it describes
12096 and change its type to scalar_int_mode. Remove VOIDmode handling.
12097 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12098 (iv_extend): Likewise.
12099 (shorten_into_mode): Likewise.
12100 (iv_add): Use scalar_int_mode.
12101 (iv_mult): Likewise.
12102 (iv_shift): Likewise.
12103 (canonicalize_iv_subregs): Likewise.
12104 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12105 it describes and change its mode to scalar_int_mode. Also change
12106 the type of the returned inner_mode to scalar_int_mode.
12107 (get_biv_step): Likewise, turning outer_mode from a pointer
12108 into a direct parameter. Update call to get_biv_step_1.
12109 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12110 iv_constant and get_biv_step.
12111 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12112 and change its type to scalar_int_mode. Don't initialise iv->mode
12113 to VOIDmode and remove later checks for its still being VOIDmode.
12114 Update calls to iv_analyze_op and iv_analyze_expr. Check
12115 is_a <scalar_int_mode> when changing the mode under consideration.
12116 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12117 Update call to iv_analyze_expr.
12118 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12119 inner register is not also a scalar_int_mode. Update call to
12121 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12122 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12123 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12124 separate mode class checks. Update calls to iv_analyze. Remove
12125 fix-up of VOIDmodes after iv_analyze_biv.
12126 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12127 don't have a scalar_int_mode. Update call to biv_p.
12129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12130 Alan Hayward <alan.hayward@arm.com>
12131 David Sherwood <david.sherwood@arm.com>
12133 * cfgexpand.c (convert_debug_memory_address): Use
12134 as_a <scalar_int_mode>.
12135 * combine.c (expand_compound_operation): Likewise.
12136 (make_extraction): Likewise.
12137 (change_zero_ext): Likewise.
12138 (simplify_comparison): Likewise.
12139 * cse.c (cse_insn): Likewise.
12140 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12141 (mem_loc_descriptor): Likewise.
12142 (loc_descriptor): Likewise.
12143 * expmed.c (init_expmed_one_mode): Likewise.
12144 (synth_mult): Likewise.
12145 (emit_store_flag_1): Likewise.
12146 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12147 of a comparison with size.
12148 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12149 (reduce_to_bit_field_precision): Likewise.
12150 * function.c (expand_function_end): Likewise.
12151 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12152 * loop-doloop.c (doloop_modify): Likewise.
12153 * optabs.c (expand_binop): Likewise.
12154 (expand_unop): Likewise.
12155 (expand_copysign_absneg): Likewise.
12156 (prepare_cmp_insn): Likewise.
12157 (maybe_legitimize_operand): Likewise.
12158 * recog.c (const_scalar_int_operand): Likewise.
12159 * rtlanal.c (get_address_mode): Likewise.
12160 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12161 (simplify_cond_clz_ctz): Likewise.
12162 * tree-nested.c (get_nl_goto_field): Likewise.
12163 * tree.c (build_vector_type_for_mode): Likewise.
12164 * var-tracking.c (use_narrower_mode): Likewise.
12166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12167 Alan Hayward <alan.hayward@arm.com>
12168 David Sherwood <david.sherwood@arm.com>
12170 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12171 * builtins.c (expand_builtin_signbit): Use it.
12172 * cfgexpand.c (expand_debug_expr): Likewise.
12173 * dojump.c (do_jump): Likewise.
12174 (do_compare_and_jump): Likewise.
12175 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12176 * expmed.c (make_tree): Likewise.
12177 * expr.c (expand_expr_real_2): Likewise.
12178 (expand_expr_real_1): Likewise.
12179 (try_casesi): Likewise.
12180 * fold-const-call.c (fold_const_call_ss): Likewise.
12181 * fold-const.c (unextend): Likewise.
12182 (extract_muldiv_1): Likewise.
12183 (fold_single_bit_test): Likewise.
12184 (native_encode_int): Likewise.
12185 (native_encode_string): Likewise.
12186 (native_interpret_int): Likewise.
12187 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12188 * internal-fn.c (expand_addsub_overflow): Likewise.
12189 (expand_neg_overflow): Likewise.
12190 (expand_mul_overflow): Likewise.
12191 (expand_arith_overflow): Likewise.
12192 * match.pd: Likewise.
12193 * stor-layout.c (layout_type): Likewise.
12194 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12195 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12196 * tree-ssanames.c (get_range_info): Likewise.
12197 * tree-switch-conversion.c (array_value_type) Likewise.
12198 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12199 (vect_recog_divmod_pattern): Likewise.
12200 (vect_recog_mixed_size_cond_pattern): Likewise.
12201 * tree-vrp.c (extract_range_basic): Likewise.
12202 (simplify_float_conversion_using_ranges): Likewise.
12203 * tree.c (int_fits_type_p): Likewise.
12204 * ubsan.c (instrument_bool_enum_load): Likewise.
12205 * varasm.c (mergeable_string_section): Likewise.
12206 (narrowing_initializer_constant_valid_p): Likewise.
12207 (output_constant): 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 * machmode.h (NARROWEST_INT_MODE): New macro.
12214 * expr.c (alignment_for_piecewise_move): Use it instead of
12215 GET_CLASS_NARROWEST_MODE (MODE_INT).
12216 (push_block): Likewise.
12217 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12219 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12222 Alan Hayward <alan.hayward@arm.com>
12223 David Sherwood <david.sherwood@arm.com>
12225 * postreload.c (move2add_valid_value_p): Change the type of the
12226 mode parameter to scalar_int_mode.
12227 (move2add_use_add2_insn): Add a mode parameter and use it instead
12229 (move2add_use_add3_insn): Likewise.
12230 (reload_cse_move2add): Update accordingly.
12232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12233 Alan Hayward <alan.hayward@arm.com>
12234 David Sherwood <david.sherwood@arm.com>
12236 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12238 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12239 * optabs.c (expand_unop): Likewise.
12241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12242 Alan Hayward <alan.hayward@arm.com>
12243 David Sherwood <david.sherwood@arm.com>
12245 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12246 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12247 (popcount_loc_descriptor): Likewise.
12248 (bswap_loc_descriptor): Likewise.
12249 (rotate_loc_descriptor): Likewise.
12250 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12251 calling the functions above.
12253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12254 Alan Hayward <alan.hayward@arm.com>
12255 David Sherwood <david.sherwood@arm.com>
12257 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12259 (try_combine): Likewise.
12260 (simplify_if_then_else): Likewise.
12261 * cse.c (cse_insn): Likewise.
12262 * dwarf2out.c (mem_loc_descriptor): Likewise.
12263 * emit-rtl.c (gen_lowpart_common): Likewise.
12264 * simplify-rtx.c (simplify_truncation): Likewise.
12265 (simplify_binary_operation_1): Likewise.
12266 (simplify_const_relational_operation): Likewise.
12267 (simplify_ternary_operation): Likewise.
12268 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12271 Alan Hayward <alan.hayward@arm.com>
12272 David Sherwood <david.sherwood@arm.com>
12274 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12275 * reload.c (push_reload): Likewise.
12276 (find_reloads): Likewise.
12278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12279 Alan Hayward <alan.hayward@arm.com>
12280 David Sherwood <david.sherwood@arm.com>
12282 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12283 (make_compound_operation_int): Likewise.
12284 (change_zero_ext): Likewise.
12285 * expr.c (convert_move): Likewise.
12286 (convert_modes): Likewise.
12287 * fwprop.c (forward_propagate_subreg): Likewise.
12288 * loop-iv.c (get_biv_step_1): Likewise.
12289 * optabs.c (widen_operand): Likewise.
12290 * postreload.c (move2add_valid_value_p): Likewise.
12291 * recog.c (simplify_while_replacing): Likewise.
12292 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12293 (simplify_binary_operation_1): Likewise. Remove redundant
12294 mode equality check.
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 (combine_simplify_rtx): Add checks for
12301 is_a <scalar_int_mode>.
12302 (simplify_if_then_else): Likewise.
12303 (make_field_assignment): Likewise.
12304 (simplify_comparison): Likewise.
12305 * ifcvt.c (noce_try_bitop): Likewise.
12306 * loop-invariant.c (canonicalize_address_mult): Likewise.
12307 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12310 Alan Hayward <alan.hayward@arm.com>
12311 David Sherwood <david.sherwood@arm.com>
12313 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12314 is_a <scalar_int_mode> instead of != BLKmode.
12316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12317 Alan Hayward <alan.hayward@arm.com>
12318 David Sherwood <david.sherwood@arm.com>
12320 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12321 instead of != VOIDmode.
12322 * combine.c (if_then_else_cond): Likewise.
12323 (change_zero_ext): Likewise.
12324 * dwarf2out.c (mem_loc_descriptor): Likewise.
12325 (loc_descriptor): Likewise.
12326 * rtlanal.c (canonicalize_condition): Likewise.
12327 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12330 Alan Hayward <alan.hayward@arm.com>
12331 David Sherwood <david.sherwood@arm.com>
12333 * simplify-rtx.c (simplify_binary_operation_1): Use
12334 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12337 Alan Hayward <alan.hayward@arm.com>
12338 David Sherwood <david.sherwood@arm.com>
12340 * wide-int.h (int_traits<unsigned char>) New class.
12341 (int_traits<unsigned short>) Likewise.
12342 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12343 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12345 * combine.c (set_nonzero_bits_and_sign_copies): Use
12346 is_a <scalar_int_mode>.
12347 (find_split_point): Likewise.
12348 (combine_simplify_rtx): Likewise.
12349 (simplify_logical): Likewise.
12350 (expand_compound_operation): Likewise.
12351 (expand_field_assignment): Likewise.
12352 (make_compound_operation): Likewise.
12353 (extended_count): Likewise.
12354 (change_zero_ext): Likewise.
12355 (simplify_comparison): Likewise.
12356 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12357 (ucompare_loc_descriptor): Likewise.
12358 (minmax_loc_descriptor): Likewise.
12359 (mem_loc_descriptor): Likewise.
12360 (loc_descriptor): Likewise.
12361 * expmed.c (init_expmed_one_mode): Likewise.
12362 * lra-constraints.c (lra_constraint_offset): Likewise.
12363 * optabs.c (prepare_libcall_arg): Likewise.
12364 * postreload.c (move2add_note_store): Likewise.
12365 * reload.c (operands_match_p): Likewise.
12366 * rtl.h (load_extend_op): Likewise.
12367 * rtlhooks.c (gen_lowpart_general): Likewise.
12368 * simplify-rtx.c (simplify_truncation): Likewise.
12369 (simplify_unary_operation_1): Likewise.
12370 (simplify_binary_operation_1): Likewise.
12371 (simplify_const_binary_operation): Likewise.
12372 (simplify_const_relational_operation): Likewise.
12373 (simplify_subreg): Likewise.
12374 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12375 * var-tracking.c (adjust_mems): Likewise.
12376 (prepare_call_arguments): Likewise.
12378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12379 Alan Hayward <alan.hayward@arm.com>
12380 David Sherwood <david.sherwood@arm.com>
12382 * machmode.h (is_int_mode): New fuction.
12383 * combine.c (find_split_point): Use it.
12384 (combine_simplify_rtx): Likewise.
12385 (simplify_if_then_else): Likewise.
12386 (simplify_set): Likewise.
12387 (simplify_shift_const_1): Likewise.
12388 (simplify_comparison): Likewise.
12389 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12390 * cse.c (notreg_cost): Likewise.
12391 (cse_insn): Likewise.
12392 * cselib.c (cselib_lookup_1): Likewise.
12393 * dojump.c (do_jump_1): Likewise.
12394 (do_compare_rtx_and_jump): Likewise.
12395 * dse.c (get_call_args): Likewise.
12396 * dwarf2out.c (rtl_for_decl_init): Likewise.
12397 (native_encode_initializer): Likewise.
12398 * expmed.c (emit_store_flag_1): Likewise.
12399 (emit_store_flag): Likewise.
12400 * expr.c (convert_modes): Likewise.
12401 (store_field): Likewise.
12402 (expand_expr_real_1): Likewise.
12403 * fold-const.c (fold_read_from_constant_string): Likewise.
12404 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12405 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12406 * optabs.c (expand_binop): Likewise.
12407 (expand_unop): Likewise.
12408 (expand_abs_nojump): Likewise.
12409 (expand_one_cmpl_abs_nojump): Likewise.
12410 * simplify-rtx.c (mode_signbit_p): Likewise.
12411 (val_signbit_p): Likewise.
12412 (val_signbit_known_set_p): Likewise.
12413 (val_signbit_known_clear_p): Likewise.
12414 (simplify_relational_operation_1): Likewise.
12415 * tree.c (vector_type_mode): Likewise.
12417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12418 Alan Hayward <alan.hayward@arm.com>
12419 David Sherwood <david.sherwood@arm.com>
12421 * machmode.h (smallest_mode_for_size): Fix formatting.
12422 (smallest_int_mode_for_size): New function.
12423 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12424 instead of smallest_mode_for_size.
12425 * combine.c (make_extraction): Likewise.
12426 * config/arc/arc.c (arc_expand_movmem): Likewise.
12427 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12428 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12429 * config/s390/s390.c (s390_expand_insv): Likewise.
12430 * config/sparc/sparc.c (assign_int_registers): Likewise.
12431 * config/spu/spu.c (spu_function_value): Likewise.
12432 (spu_function_arg): Likewise.
12433 * coverage.c (get_gcov_type): Likewise.
12434 (get_gcov_unsigned_t): Likewise.
12435 * dse.c (find_shift_sequence): Likewise.
12436 * expmed.c (store_bit_field_1): Likewise.
12437 * expr.c (convert_move): Likewise.
12438 (store_field): Likewise.
12439 * internal-fn.c (expand_arith_overflow): Likewise.
12440 * optabs-query.c (get_best_extraction_insn): Likewise.
12441 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12442 * stor-layout.c (layout_type): Likewise.
12443 (initialize_sizetypes): Likewise.
12444 * targhooks.c (default_get_mask_mode): Likewise.
12445 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12448 Alan Hayward <alan.hayward@arm.com>
12449 David Sherwood <david.sherwood@arm.com>
12451 * machmode.h (opt_mode::else_blk): New function.
12452 (int_mode_for_mode): Declare.
12453 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12454 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12456 * cfgexpand.c (expand_debug_expr): Likewise.
12457 * combine.c (gen_lowpart_or_truncate): Likewise.
12458 (gen_lowpart_for_combine): Likewise.
12459 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12460 * config/avr/avr.c (avr_to_int_mode): Likewise.
12461 (avr_out_plus_1): Likewise.
12462 (avr_out_plus): Likewise.
12463 (avr_out_round): Likewise.
12464 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12465 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12466 (s390_expand_vcond): Likewise.
12467 * config/spu/spu.c (spu_split_immediate): Likewise.
12468 (spu_expand_mov): Likewise.
12469 * dse.c (get_stored_val): Likewise.
12470 * expmed.c (store_bit_field_1): Likewise.
12471 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12473 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12474 (extract_low_bits): Likewise.
12475 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12476 handling rather than repeating the check.
12477 (emit_group_store): Likewise.
12478 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12479 * optabs.c (expand_absneg_bit): Likewise.
12480 (expand_copysign_absneg): Likewise.
12481 (expand_copysign_bit): Likewise.
12482 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12483 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12484 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12485 * var-tracking.c (prepare_call_arguments): Likewise.
12486 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12487 int_mode_for_mode instead of mode_for_size.
12488 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): 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 (int_mode_for_size): New function.
12495 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12496 instead of mode_for_size.
12497 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12499 * combine.c (expand_field_assignment): Use int_mode_for_size
12500 instead of mode_for_size.
12501 (make_extraction): Likewise.
12502 (simplify_shift_const_1): Likewise.
12503 (simplify_comparison): Likewise.
12504 * dojump.c (do_jump): Likewise.
12505 * dwarf2out.c (mem_loc_descriptor): Likewise.
12506 * emit-rtl.c (init_derived_machine_modes): Likewise.
12507 * expmed.c (flip_storage_order): Likewise.
12508 (convert_extracted_bit_field): Likewise.
12509 * expr.c (copy_blkmode_from_reg): Likewise.
12510 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12511 * internal-fn.c (expand_mul_overflow): Likewise.
12512 * lower-subreg.c (simple_move): Likewise.
12513 * optabs-libfuncs.c (init_optabs): Likewise.
12514 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12515 * tree.c (vector_type_mode): Likewise.
12516 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12517 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12518 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12519 * tree-vect-stmts.c (vectorizable_load): Likewise.
12520 (vectorizable_store): Likewise.
12522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12523 Alan Hayward <alan.hayward@arm.com>
12524 David Sherwood <david.sherwood@arm.com>
12526 * coretypes.h (pod_mode): New type.
12527 (scalar_int_mode_pod): New typedef.
12528 * machmode.h (pod_mode): New class.
12529 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12530 * genmodes.c (emit_mode_int_n): Update accordingly.
12531 * lower-subreg.h (target_lower_subreg): Change type to
12532 scalar_int_mode_pod.
12533 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12534 scalar_int_mode_pod.
12536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12537 Alan Hayward <alan.hayward@arm.com>
12538 David Sherwood <david.sherwood@arm.com>
12540 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12541 machine_mode to scalar_int_mode.
12542 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12543 (rs6000_option_override_internal): Remove cast to int.
12544 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12545 machine_mode to scalar_int_mode.
12546 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12547 (rs6000_option_override_internal): Remove cast to int.
12548 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12549 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12551 * config/s390/s390.c (s390_expand_builtin): Likewise.
12552 * coretypes.h (scalar_int_mode): New type.
12553 (opt_scalar_int_mode): New typedef.
12554 * machmode.h (scalar_int_mode): New class.
12555 (scalar_int_mode::includes_p): New function.
12556 (byte_mode): Change type to scalar_int_mode.
12557 (word_mode): Likewise.
12558 (ptr_mode): Likewise.
12559 * emit-rtl.c (byte_mode): Likewise.
12560 (word_mode): Likewise.
12561 (ptr_mode): Likewise.
12562 (init_derived_machine_modes): Update accordingly.
12563 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12564 and MODE_PARTIAL_INT.
12565 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12566 opt_scalar_int_mode.
12568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12569 Alan Hayward <alan.hayward@arm.com>
12570 David Sherwood <david.sherwood@arm.com>
12572 * target.def (libgcc_floating_mode_supported_p): Take a
12574 * doc/tm.texi: Regenerate.
12575 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12577 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12578 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12582 Alan Hayward <alan.hayward@arm.com>
12583 David Sherwood <david.sherwood@arm.com>
12585 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12586 * doc/tm.texi: Regenerate.
12587 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12588 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12589 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12590 * targhooks.h (default_floatn_mode): Likewise.
12591 * targhooks.c (default_floatn_mode): Likewise.
12592 * tree.c (build_common_tree_nodes): Update accordingly.
12594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12595 Alan Hayward <alan.hayward@arm.com>
12596 David Sherwood <david.sherwood@arm.com>
12598 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12599 (mode_iterator::iterate_p): Likewise.
12600 (mode_iterator::get_wider): Likewise.
12601 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12603 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12604 Alan Hayward <alan.hayward@arm.com>
12605 David Sherwood <david.sherwood@arm.com>
12607 * coretypes.h (opt_scalar_float_mode): New typedef.
12608 * machmode.h (float_mode_for_size): New function.
12609 * emit-rtl.c (double_mode): Delete.
12610 (init_emit_once): Use float_mode_for_size.
12611 * stor-layout.c (layout_type): Likewise.
12612 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12615 Alan Hayward <alan.hayward@arm.com>
12616 David Sherwood <david.sherwood@arm.com>
12618 * output.h (assemble_real): Take a scalar_float_mode.
12619 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12620 * config/arm/arm.md (consttable_4): Likewise.
12621 (consttable_8): Likewise.
12622 (consttable_16): Likewise.
12623 * config/mips/mips.md (consttable_float): Likewise.
12624 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12625 * varasm.c (assemble_real): Take a scalar_float_mode.
12626 (output_constant_pool_2): Update accordingly.
12627 (output_constant): Likewise.
12629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12630 Alan Hayward <alan.hayward@arm.com>
12631 David Sherwood <david.sherwood@arm.com>
12633 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12634 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12635 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12636 (native_encode_real): Likewise.
12637 (native_interpret_real): Likewise.
12638 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12639 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12642 Alan Hayward <alan.hayward@arm.com>
12643 David Sherwood <david.sherwood@arm.com>
12645 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12646 <scalar_float_mode>. Simplify.
12647 (gen_extend_conv_libfunc): Likewise.
12649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12650 Alan Hayward <alan.hayward@arm.com>
12651 David Sherwood <david.sherwood@arm.com>
12653 * coretypes.h (scalar_float_mode): New type.
12654 * machmode.h (mode_traits::from_int): Use machine_mode if
12655 USE_ENUM_MODES is defined.
12656 (is_a): New function.
12658 (dyn_cast): Likewise.
12659 (scalar_float_mode): New class.
12660 (scalar_float_mode::includes_p): New function.
12661 (is_float_mode): Likewise.
12662 * gdbhooks.py (MachineModePrinter): New class.
12663 (build_pretty_printer): Use it for scalar_float_mode.
12664 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12665 (format_helper::format_helper): Turn into a template.
12666 * genmodes.c (get_mode_class): New function.
12667 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12668 or machine_mode if none.
12669 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12670 as_a <scalar_float_mode>.
12671 * dwarf2out.c (mem_loc_descriptor): Likewise.
12672 (insert_float): Likewise.
12673 (add_const_value_attribute): Likewise.
12674 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12675 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12676 (expand_unop): Update accordingly.
12677 (expand_abs_nojump): Likewise.
12678 (expand_copysign_absneg): Take a scalar_float_mode.
12679 (expand_copysign_bit): Likewise.
12680 (expand_copysign): Update accordingly.
12682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12683 Alan Hayward <alan.hayward@arm.com>
12684 David Sherwood <david.sherwood@arm.com>
12686 * coretypes.h (opt_mode): New class.
12687 * machmode.h (opt_mode): Likewise.
12688 (opt_mode::else_void): New function.
12689 (opt_mode::require): Likewise.
12690 (opt_mode::exists): Likewise.
12691 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12692 (GET_MODE_2XWIDER_MODE): Likewise.
12693 (mode_iterator::get_wider): Update accordingly.
12694 (mode_iterator::get_2xwider): Likewise.
12695 (mode_iterator::get_known_wider): Likewise, turning into a template.
12696 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12697 forcing a wider mode to exist.
12698 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12699 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12700 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12701 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12702 * lower-subreg.c (init_lower_subreg): Likewise.
12703 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12704 on the final iteration.
12705 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12706 a wider mode exists before asking for a move pattern.
12707 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12708 forcing a wider mode to exist.
12709 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12710 returning false if no such mode exists.
12711 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12712 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12713 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12714 Avoid checking for a MODE_INT if we already know the mode is not a
12716 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12717 forcing a wider mode to exist.
12718 (expmed_mult_highpart_optab): Likewise.
12719 (expmed_mult_highpart): Likewise.
12720 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12722 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12723 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12724 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12725 * internal-fn.c (expand_mul_overflow): Update use of
12726 GET_MODE_2XWIDER_MODE.
12727 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12728 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12729 GET_MODE_WIDER_MODE.
12730 (convert_plusminus_to_widen): Likewise.
12731 * tree-switch-conversion.c (array_value_type): Likewise.
12732 * var-tracking.c (emit_note_insn_var_location): Likewise.
12733 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12734 Return false inside rather than outside the loop if no wider mode
12736 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12737 and GET_MODE_2XWIDER_MODE
12738 (can_compare_p): Use else_void.
12739 * gdbhooks.py (OptMachineModePrinter): New class.
12740 (build_pretty_printer): Use it for opt_mode.
12742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12743 Alan Hayward <alan.hayward@arm.com>
12744 David Sherwood <david.sherwood@arm.com>
12746 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12747 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12750 Alan Hayward <alan.hayward@arm.com>
12751 David Sherwood <david.sherwood@arm.com>
12753 * machmode.h (mode_traits): New structure.
12754 (get_narrowest_mode): New function.
12755 (mode_iterator::start): Likewise.
12756 (mode_iterator::iterate_p): Likewise.
12757 (mode_iterator::get_wider): Likewise.
12758 (mode_iterator::get_known_wider): Likewise.
12759 (mode_iterator::get_2xwider): Likewise.
12760 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12761 (FOR_EACH_MODE): Likewise.
12762 (FOR_EACH_MODE_FROM): Likewise.
12763 (FOR_EACH_MODE_UNTIL): Likewise.
12764 (FOR_EACH_WIDER_MODE): Likewise.
12765 (FOR_EACH_2XWIDER_MODE): Likewise.
12766 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12767 * combine.c (simplify_comparison): Likewise
12768 * config/i386/i386.c (type_natural_mode): Likewise.
12769 * cse.c (cse_insn): Likewise.
12770 * dse.c (find_shift_sequence): Likewise.
12771 * emit-rtl.c (init_derived_machine_modes): Likewise.
12772 (init_emit_once): Likewise.
12773 * explow.c (hard_function_value): Likewise.
12774 * expmed.c (extract_fixed_bit_field_1): Likewise.
12775 (extract_bit_field_1): Likewise.
12776 (expand_divmod): Likewise.
12777 (emit_store_flag_1): Likewise.
12778 * expr.c (init_expr_target): Likewise.
12779 (convert_move): Likewise.
12780 (alignment_for_piecewise_move): Likewise.
12781 (widest_int_mode_for_size): Likewise.
12782 (emit_block_move_via_movmem): Likewise.
12783 (copy_blkmode_to_reg): Likewise.
12784 (set_storage_via_setmem): Likewise.
12785 (compress_float_constant): Likewise.
12786 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12787 * optabs-query.c (get_best_extraction_insn): Likewise.
12788 * optabs.c (expand_binop): Likewise.
12789 (expand_twoval_unop): Likewise.
12790 (expand_twoval_binop): Likewise.
12791 (widen_leading): Likewise.
12792 (widen_bswap): Likewise.
12793 (expand_parity): Likewise.
12794 (expand_unop): Likewise.
12795 (prepare_cmp_insn): Likewise.
12796 (prepare_float_lib_cmp): Likewise.
12797 (expand_float): Likewise.
12798 (expand_fix): Likewise.
12799 (expand_sfix_optab): Likewise.
12800 * postreload.c (move2add_use_add2_insn): Likewise.
12801 * reg-stack.c (reg_to_stack): Likewise.
12802 * reginfo.c (choose_hard_reg_mode): Likewise.
12803 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12804 * stor-layout.c (mode_for_size): Likewise.
12805 (smallest_mode_for_size): Likewise.
12806 (mode_for_vector): Likewise.
12807 (finish_bitfield_representative): Likewise.
12808 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12809 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12810 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12811 * var-tracking.c (prepare_call_arguments): Likewise.
12813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12814 Alan Hayward <alan.hayward@arm.com>
12815 David Sherwood <david.sherwood@arm.com>
12817 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12818 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12819 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12820 * machmode.h (mode_size): Move earlier in file.
12821 (mode_precision): Likewise.
12822 (mode_inner): Likewise.
12823 (mode_nunits): Likewise.
12824 (mode_unit_size): Likewise.
12825 (unit_unit_precision): Likewise.
12826 (mode_wider): Likewise.
12827 (mode_2xwider): Likewise.
12828 (machine_mode): New class.
12829 (mode_to_bytes): New function.
12830 (mode_to_bits): Likewise.
12831 (mode_to_precision): Likewise.
12832 (mode_to_inner): Likewise.
12833 (mode_to_unit_size): Likewise.
12834 (mode_to_unit_precision): Likewise.
12835 (mode_to_nunits): Likewise.
12836 (GET_MODE_SIZE): Use mode_to_bytes.
12837 (GET_MODE_BITSIZE): Use mode_to_bits.
12838 (GET_MODE_PRECISION): Use mode_to_precision.
12839 (GET_MODE_INNER): Use mode_to_inner.
12840 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12841 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12842 (GET_MODE_NUNITS): Use mode_to_nunits.
12843 * system.h (ALWAYS_INLINE): New macro.
12844 * config/powerpcspe/powerpcspe-c.c
12845 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12846 int for arg1_mode and arg2_mode.
12848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12849 Alan Hayward <alan.hayward@arm.com>
12850 David Sherwood <david.sherwood@arm.com>
12852 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12853 Prefix mode names with E_ in case statements.
12854 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12855 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12856 (aarch64_split_simd_move): Likewise.
12857 (aarch64_gen_storewb_pair): Likewise.
12858 (aarch64_gen_loadwb_pair): Likewise.
12859 (aarch64_gen_store_pair): Likewise.
12860 (aarch64_gen_load_pair): Likewise.
12861 (aarch64_get_condition_code_1): Likewise.
12862 (aarch64_constant_pool_reload_icode): Likewise.
12863 (get_rsqrte_type): Likewise.
12864 (get_rsqrts_type): Likewise.
12865 (get_recpe_type): Likewise.
12866 (get_recps_type): Likewise.
12867 (aarch64_gimplify_va_arg_expr): Likewise.
12868 (aarch64_simd_container_mode): Likewise.
12869 (aarch64_emit_load_exclusive): Likewise.
12870 (aarch64_emit_store_exclusive): Likewise.
12871 (aarch64_expand_compare_and_swap): Likewise.
12872 (aarch64_gen_atomic_cas): Likewise.
12873 (aarch64_emit_bic): Likewise.
12874 (aarch64_emit_atomic_swap): Likewise.
12875 (aarch64_emit_atomic_load_op): Likewise.
12876 (aarch64_evpc_trn): Likewise.
12877 (aarch64_evpc_uzp): Likewise.
12878 (aarch64_evpc_zip): Likewise.
12879 (aarch64_evpc_ext): Likewise.
12880 (aarch64_evpc_rev): Likewise.
12881 (aarch64_evpc_dup): Likewise.
12882 (aarch64_gen_ccmp_first): Likewise.
12883 (aarch64_gen_ccmp_next): Likewise.
12884 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12885 (alpha_emit_xfloating_libcall): Likewise.
12886 (emit_insxl): Likewise.
12887 (alpha_arg_type): Likewise.
12888 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12889 (arc_preferred_simd_mode): Likewise.
12890 (arc_secondary_reload): Likewise.
12891 (get_arc_condition_code): Likewise.
12892 (arc_print_operand): Likewise.
12893 (arc_legitimate_constant_p): Likewise.
12894 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12895 * config/arc/arc.md (casesi_load): Likewise.
12896 (casesi_compact_jump): Likewise.
12897 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12898 (cc_use_register): Likewise.
12899 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12900 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12901 (arm_init_iwmmxt_builtins): Likewise.
12902 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12903 (neon_expand_vector_init): Likewise.
12904 (arm_attr_length_move_neon): Likewise.
12905 (maybe_get_arm_condition_code): Likewise.
12906 (arm_emit_vector_const): Likewise.
12907 (arm_preferred_simd_mode): Likewise.
12908 (arm_output_iwmmxt_tinsr): Likewise.
12909 (thumb1_output_casesi): Likewise.
12910 (thumb2_output_casesi): Likewise.
12911 (arm_emit_load_exclusive): Likewise.
12912 (arm_emit_store_exclusive): Likewise.
12913 (arm_expand_compare_and_swap): Likewise.
12914 (arm_evpc_neon_vuzp): Likewise.
12915 (arm_evpc_neon_vzip): Likewise.
12916 (arm_evpc_neon_vrev): Likewise.
12917 (arm_evpc_neon_vtrn): Likewise.
12918 (arm_evpc_neon_vext): Likewise.
12919 (arm_validize_comparison): Likewise.
12920 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12921 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12922 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12923 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12924 (c6x_preferred_simd_mode): Likewise.
12925 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12926 (epiphany_rtx_costs): Likewise.
12927 * config/epiphany/predicates.md (proper_comparison_operator):
12929 * config/frv/frv.c (condexec_memory_operand): Likewise.
12930 (frv_emit_move): Likewise.
12931 (output_move_single): Likewise.
12932 (output_condmove_single): Likewise.
12933 (frv_hard_regno_mode_ok): Likewise.
12934 (frv_matching_accg_mode): Likewise.
12935 * config/h8300/h8300.c (split_adds_subs): Likewise.
12936 (h8300_rtx_costs): Likewise.
12937 (h8300_print_operand): Likewise.
12938 (compute_mov_length): Likewise.
12939 (output_logical_op): Likewise.
12940 (compute_logical_op_length): Likewise.
12941 (compute_logical_op_cc): Likewise.
12942 (h8300_shift_needs_scratch_p): Likewise.
12943 (output_a_shift): Likewise.
12944 (compute_a_shift_length): Likewise.
12945 (compute_a_shift_cc): Likewise.
12946 (expand_a_rotate): Likewise.
12947 (output_a_rotate): Likewise.
12948 * config/i386/i386.c (classify_argument): Likewise.
12949 (function_arg_advance_32): Likewise.
12950 (function_arg_32): Likewise.
12951 (function_arg_64): Likewise.
12952 (function_value_64): Likewise.
12953 (ix86_gimplify_va_arg): Likewise.
12954 (ix86_legitimate_constant_p): Likewise.
12955 (put_condition_code): Likewise.
12956 (split_double_mode): Likewise.
12957 (ix86_avx256_split_vector_move_misalign): Likewise.
12958 (ix86_expand_vector_logical_operator): Likewise.
12959 (ix86_split_idivmod): Likewise.
12960 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12961 (ix86_build_const_vector): Likewise.
12962 (ix86_build_signbit_mask): Likewise.
12963 (ix86_match_ccmode): Likewise.
12964 (ix86_cc_modes_compatible): Likewise.
12965 (ix86_expand_branch): Likewise.
12966 (ix86_expand_sse_cmp): Likewise.
12967 (ix86_expand_sse_movcc): Likewise.
12968 (ix86_expand_int_sse_cmp): Likewise.
12969 (ix86_expand_vec_perm_vpermi2): Likewise.
12970 (ix86_expand_vec_perm): Likewise.
12971 (ix86_expand_sse_unpack): Likewise.
12972 (ix86_expand_int_addcc): Likewise.
12973 (ix86_split_to_parts): Likewise.
12974 (ix86_vectorize_builtin_gather): Likewise.
12975 (ix86_vectorize_builtin_scatter): Likewise.
12976 (avx_vpermilp_parallel): Likewise.
12977 (inline_memory_move_cost): Likewise.
12978 (ix86_tieable_integer_mode_p): Likewise.
12979 (x86_maybe_negate_const_int): Likewise.
12980 (ix86_expand_vector_init_duplicate): Likewise.
12981 (ix86_expand_vector_init_one_nonzero): Likewise.
12982 (ix86_expand_vector_init_one_var): Likewise.
12983 (ix86_expand_vector_init_concat): Likewise.
12984 (ix86_expand_vector_init_interleave): Likewise.
12985 (ix86_expand_vector_init_general): Likewise.
12986 (ix86_expand_vector_set): Likewise.
12987 (ix86_expand_vector_extract): Likewise.
12988 (emit_reduc_half): Likewise.
12989 (ix86_emit_i387_round): Likewise.
12990 (ix86_mangle_type): Likewise.
12991 (ix86_expand_round_sse4): Likewise.
12992 (expand_vec_perm_blend): Likewise.
12993 (canonicalize_vector_int_perm): Likewise.
12994 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12995 (expand_vec_perm_1): Likewise.
12996 (expand_vec_perm_interleave3): Likewise.
12997 (expand_vec_perm_even_odd_pack): Likewise.
12998 (expand_vec_perm_even_odd_1): Likewise.
12999 (expand_vec_perm_broadcast_1): Likewise.
13000 (ix86_vectorize_vec_perm_const_ok): Likewise.
13001 (ix86_expand_vecop_qihi): Likewise.
13002 (ix86_expand_mul_widen_hilo): Likewise.
13003 (ix86_expand_sse2_abs): Likewise.
13004 (ix86_expand_pextr): Likewise.
13005 (ix86_expand_pinsr): Likewise.
13006 (ix86_preferred_simd_mode): Likewise.
13007 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13008 * config/i386/sse.md (*andnot<mode>3): Likewise.
13009 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13010 (*<code><mode>3): Likewise.
13011 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13012 (ia64_expand_atomic_op): Likewise.
13013 (ia64_arg_type): Likewise.
13014 (ia64_mode_to_int): Likewise.
13015 (ia64_scalar_mode_supported_p): Likewise.
13016 (ia64_vector_mode_supported_p): Likewise.
13017 (expand_vec_perm_broadcast): Likewise.
13018 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13019 (iq2000_function_arg_advance): Likewise.
13020 (iq2000_function_arg): Likewise.
13021 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13022 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13023 (m68k_libcall_value): Likewise.
13024 (m68k_function_value): Likewise.
13025 (sched_attr_op_type): Likewise.
13026 * config/mcore/mcore.c (mcore_output_move): Likewise.
13027 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13029 (microblaze_function_arg): Likewise.
13030 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13031 (mips_print_operand): Likewise.
13032 (mips_mode_ok_for_mov_fmt_p): Likewise.
13033 (mips_vector_mode_supported_p): Likewise.
13034 (mips_preferred_simd_mode): Likewise.
13035 (mips_expand_vpc_loongson_even_odd): Likewise.
13036 (mips_expand_vec_unpack): Likewise.
13037 (mips_expand_vi_broadcast): Likewise.
13038 (mips_expand_vector_init): Likewise.
13039 (mips_expand_vec_reduc): Likewise.
13040 (mips_expand_msa_cmp): Likewise.
13041 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13042 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13043 (cc_flags_for_mode): Likewise.
13044 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13045 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13046 (nds32_output_casesi_pc_relative): Likewise.
13047 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13048 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13049 (nvptx_gen_unpack): Likewise.
13050 (nvptx_gen_pack): Likewise.
13051 (nvptx_gen_shuffle): Likewise.
13052 (nvptx_gen_wcast): Likewise.
13053 (nvptx_preferred_simd_mode): Likewise.
13054 * config/pa/pa.c (pa_secondary_reload): Likewise.
13055 * config/pa/predicates.md (base14_operand): Likewise.
13056 * config/powerpcspe/powerpcspe-c.c
13057 (altivec_resolve_overloaded_builtin): Likewise.
13058 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13060 (rs6000_preferred_simd_mode): Likewise.
13061 (output_vec_const_move): Likewise.
13062 (rs6000_expand_vector_extract): Likewise.
13063 (rs6000_split_vec_extract_var): Likewise.
13064 (reg_offset_addressing_ok_p): Likewise.
13065 (rs6000_legitimate_offset_address_p): Likewise.
13066 (rs6000_legitimize_address): Likewise.
13067 (rs6000_emit_set_const): Likewise.
13068 (rs6000_const_vec): Likewise.
13069 (rs6000_emit_move): Likewise.
13070 (spe_build_register_parallel): Likewise.
13071 (rs6000_darwin64_record_arg_recurse): Likewise.
13072 (swap_selector_for_mode): Likewise.
13073 (spe_init_builtins): Likewise.
13074 (paired_init_builtins): Likewise.
13075 (altivec_init_builtins): Likewise.
13076 (do_load_for_compare): Likewise.
13077 (rs6000_generate_compare): Likewise.
13078 (rs6000_expand_float128_convert): Likewise.
13079 (emit_load_locked): Likewise.
13080 (emit_store_conditional): Likewise.
13081 (rs6000_output_function_epilogue): Likewise.
13082 (rs6000_handle_altivec_attribute): Likewise.
13083 (rs6000_function_value): Likewise.
13084 (emit_fusion_gpr_load): Likewise.
13085 (emit_fusion_p9_load): Likewise.
13086 (emit_fusion_p9_store): Likewise.
13087 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13088 (fusion_gpr_mem_load): Likewise.
13089 (fusion_addis_mem_combo_load): Likewise.
13090 (fusion_addis_mem_combo_store): Likewise.
13091 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13092 (fusion_gpr_mem_load): Likewise.
13093 (fusion_addis_mem_combo_load): Likewise.
13094 (fusion_addis_mem_combo_store): Likewise.
13095 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13097 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13098 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13099 (rs6000_preferred_simd_mode): Likewise.
13100 (output_vec_const_move): Likewise.
13101 (rs6000_expand_vector_extract): Likewise.
13102 (rs6000_split_vec_extract_var): Likewise.
13103 (reg_offset_addressing_ok_p): Likewise.
13104 (rs6000_legitimate_offset_address_p): Likewise.
13105 (rs6000_legitimize_address): Likewise.
13106 (rs6000_emit_set_const): Likewise.
13107 (rs6000_const_vec): Likewise.
13108 (rs6000_emit_move): Likewise.
13109 (rs6000_darwin64_record_arg_recurse): Likewise.
13110 (swap_selector_for_mode): Likewise.
13111 (paired_init_builtins): Likewise.
13112 (altivec_init_builtins): Likewise.
13113 (rs6000_expand_float128_convert): Likewise.
13114 (emit_load_locked): Likewise.
13115 (emit_store_conditional): Likewise.
13116 (rs6000_output_function_epilogue): Likewise.
13117 (rs6000_handle_altivec_attribute): Likewise.
13118 (rs6000_function_value): Likewise.
13119 (emit_fusion_gpr_load): Likewise.
13120 (emit_fusion_p9_load): Likewise.
13121 (emit_fusion_p9_store): Likewise.
13122 * config/rx/rx.c (rx_gen_move_template): Likewise.
13123 (flags_from_mode): Likewise.
13124 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13125 (s390_slb_comparison): Likewise.
13126 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13127 (s390_vector_mode_supported_p): Likewise.
13128 (s390_cc_modes_compatible): Likewise.
13129 (s390_match_ccmode_set): Likewise.
13130 (s390_canonicalize_comparison): Likewise.
13131 (s390_emit_compare_and_swap): Likewise.
13132 (s390_branch_condition_mask): Likewise.
13133 (s390_rtx_costs): Likewise.
13134 (s390_secondary_reload): Likewise.
13135 (__SECONDARY_RELOAD_CASE): Likewise.
13136 (s390_expand_cs): Likewise.
13137 (s390_preferred_simd_mode): Likewise.
13138 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13139 * config/sh/sh.c (sh_print_operand): Likewise.
13140 (dump_table): Likewise.
13141 (sh_secondary_reload): Likewise.
13142 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13143 * config/sh/sh.md (casesi_worker_1): Likewise.
13144 (casesi_worker_2): Likewise.
13145 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13146 (fcc_comparison_operator): Likewise.
13147 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13148 (emit_soft_tfmode_cvt): Likewise.
13149 (sparc_preferred_simd_mode): Likewise.
13150 (output_cbranch): Likewise.
13151 (sparc_print_operand): Likewise.
13152 (sparc_expand_vec_perm_bmask): Likewise.
13153 (vector_init_bshuffle): Likewise.
13154 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13155 (spu_vector_mode_supported_p): Likewise.
13156 (spu_expand_insv): Likewise.
13157 (spu_emit_branch_or_set): Likewise.
13158 (spu_handle_vector_attribute): Likewise.
13159 (spu_builtin_splats): Likewise.
13160 (spu_builtin_extract): Likewise.
13161 (spu_builtin_promote): Likewise.
13162 (spu_expand_sign_extend): Likewise.
13163 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13164 (tilegx_simd_int): Likewise.
13165 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13166 (tilepro_simd_int): Likewise.
13167 * config/v850/v850.c (const_double_split): Likewise.
13168 (v850_print_operand): Likewise.
13169 (ep_memory_offset): Likewise.
13170 * config/vax/vax.c (vax_rtx_costs): Likewise.
13171 (vax_output_int_move): Likewise.
13172 (vax_output_int_add): Likewise.
13173 (vax_output_int_subtract): Likewise.
13174 * config/visium/predicates.md (visium_branch_operator): Likewise.
13175 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13176 (visium_print_operand_address): Likewise.
13177 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13178 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13179 (xtensa_expand_conditional_branch): Likewise.
13180 (xtensa_copy_incoming_a7): Likewise.
13181 (xtensa_output_literal): Likewise.
13182 * dfp.c (decimal_real_maxval): Likewise.
13183 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13186 Alan Hayward <alan.hayward@arm.com>
13187 David Sherwood <david.sherwood@arm.com>
13189 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13190 (mode_nunits_inline): Likewise.
13191 (mode_inner_inline): Likewise.
13192 (mode_unit_size_inline): Likewise.
13193 (mode_unit_precision_inline): Likewise.
13194 (emit_insn_modes_h): Likewise. Also emit a #define of the
13196 (emit_mode_wider): Add an E_ prefix to mode names.
13197 (emit_mode_complex): Likewise.
13198 (emit_mode_inner): Likewise.
13199 (emit_mode_adjustments): Likewise.
13200 (emit_mode_int_n): Likewise.
13201 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13202 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13203 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13204 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13205 (CRC32_BUILTIN, ENTRY): Likewise.
13206 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13207 (aarch64_pop_regs): Likewise.
13208 (aarch64_process_components): Likewise.
13209 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13210 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13211 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13212 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13213 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13214 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13216 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13217 (pcmpistr): Likewise.
13218 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13219 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13220 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13222 * config/rl78/rl78.c (mduc_regs): Likewise.
13223 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13224 (htm_expand_builtin): Likewise.
13225 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13226 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13227 * config/xtensa/xtensa.c (print_operand): Likewise.
13228 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13229 (NUM_MODE_VECTOR_INT): Likewise.
13230 * genoutput.c (null_operand): Likewise.
13231 (output_operand_data): Likewise.
13232 * genrecog.c (print_parameter_value): Likewise.
13233 * lra.c (debug_operand_data): Likewise.
13235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13237 * dwarf2out.c (scompare_loc_descriptor_wide)
13238 (scompare_loc_descriptor_narrow): New functions, split out from...
13239 (scompare_loc_descriptor): ...here.
13240 * expmed.c (emit_store_flag_int): New function, split out from...
13241 (emit_store_flag): ...here.
13243 2017-08-30 Richard Biener <rguenther@suse.de>
13245 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13246 (dwarf2out_early_finish): Move setting of AT_pubnames from
13247 early debug output to early finish.
13249 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13251 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13252 and -mdata-region to the assembler.
13254 2017-08-30 Richard Biener <rguenther@suse.de>
13256 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13258 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13259 (dwarf2out_early_global_decl): Only generate a DIE for the
13260 abstract origin if it doesn't already exist or is a declaration DIE.
13261 (resolve_addr): Do not add the linkage name twice when
13262 generating a stub DIE for the DW_TAG_GNU_call_site target.
13264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13266 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13267 Use machine_mode rather than int for arg1_mode.
13269 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13272 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13273 that the second argument of the built-in functions to unpack
13274 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13275 switch statement instead a lot of if statements.
13276 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13277 Allow 64-bit values to be in Altivec registers as well as
13278 traditional floating point registers.
13279 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13281 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13283 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13284 MAX_REGS_PER_ADDRESS == 1.
13286 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13288 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13289 * config/i386/i386.c (ix86_option_override_internal): Simplify
13290 setting of opts->x_flag_entry.
13292 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13293 Jakub Jelinek <jakub@redhat.com>
13294 Richard Biener <rguenther@suse.de>
13296 PR tree-optimization/81503
13297 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13298 folded constant fits in the target type; reorder tests for clarity.
13300 2017-08-29 Martin Liska <mliska@suse.cz>
13302 * passes.def: Include pass_lower_switch.
13303 * stmt.c (dump_case_nodes): Remove and move to
13304 tree-switch-conversion.
13305 (case_values_threshold): Likewise.
13306 (expand_switch_as_decision_tree_p): Likewise.
13307 (emit_case_decision_tree): Likewise.
13308 (expand_case): Likewise.
13309 (balance_case_nodes): Likewise.
13310 (node_has_low_bound): Likewise.
13311 (node_has_high_bound): Likewise.
13312 (node_is_bounded): Likewise.
13313 (emit_case_nodes): Likewise.
13314 (struct simple_case_node): New struct.
13315 (add_case_node): Remove.
13316 (emit_case_dispatch_table): Use vector instead of case_list.
13317 (reset_out_edges_aux): Remove.
13318 (compute_cases_per_edge): Likewise.
13319 (expand_case): Build list of simple_case_node.
13320 (expand_sjlj_dispatch_table): Use it.
13321 * tree-switch-conversion.c (struct case_node): Moved from
13322 stmt.c and adjusted.
13323 (emit_case_nodes): Likewise.
13324 (node_has_low_bound): Likewise.
13325 (node_has_high_bound): Likewise.
13326 (node_is_bounded): Likewise.
13327 (case_values_threshold): Likewise.
13328 (reset_out_edges_aux): Likewise.
13329 (compute_cases_per_edge): Likewise.
13330 (add_case_node): Likewise.
13331 (dump_case_nodes): Likewise.
13332 (balance_case_nodes): Likewise.
13333 (expand_switch_as_decision_tree_p): Likewise.
13334 (emit_jump): Likewise.
13335 (emit_case_decision_tree): Likewise.
13336 (try_switch_expansion): Likewise.
13337 (do_jump_if_equal): Likewise.
13338 (emit_cmp_and_jump_insns): Likewise.
13339 (fix_phi_operands_for_edge): New function.
13340 (record_phi_operand_mapping): Likewise.
13341 (class pass_lower_switch): New pass.
13342 (pass_lower_switch::execute): New function.
13343 (make_pass_lower_switch): Likewise.
13344 (conditional_probability):
13345 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13346 * tree-pass.h: Add make_pass_lower_switch.
13348 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13351 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13354 2017-08-29 Richard Biener <rguenther@suse.de>
13356 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13357 we do not add a DW_AT_inline attribute twice.
13358 (gen_subprogram_die): Remove code setting DW_AT_inline on
13359 DECL_ABSTRACT_P nodes.
13361 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13363 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13364 calls to internal functions.
13365 (gimplify_modify_expr): Likewise.
13366 * tree-call-cdce.c (use_internal_fn): Likewise.
13367 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13368 (convert_to_divmod): Set the nothrow flag.
13369 * tree-if-conv.c (predicate_mem_writes): Likewise.
13370 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13371 (vectorizable_call): Likewise.
13372 (vectorizable_store): Likewise.
13373 (vectorizable_load): Likewise.
13374 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13375 (vect_recog_mask_conversion_pattern): Likewise.
13377 2017-08-29 Martin Liska <mliska@suse.cz>
13380 * gcc.c (driver_handle_option): Add new argument.
13381 * opts-common.c (handle_option): Pass
13382 target_option_override_hook.
13383 * opts-global.c (lang_handle_option): Add new option.
13384 (set_default_handlers): Add new argument.
13385 (decode_options): Likewise.
13386 * opts.c (target_handle_option): Likewise.
13387 (common_handle_option): Call target_option_override_hook.
13388 * opts.h (struct cl_option_handler_func): Add hook for
13389 target option override.
13390 (struct cl_option_handlers): Likewise.
13391 (set_default_handlers): Add new argument.
13392 (decode_options): Likewise.
13393 (common_handle_option): Likewise.
13394 (target_handle_option): Likewise.
13395 * toplev.c (toplev::main): Pass targetm.target_option.override
13398 2017-08-29 Richard Biener <rguenther@suse.de>
13399 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13401 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13402 life to the active subtree.
13404 2017-08-28 Jeff Law <law@redhat.com>
13406 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13407 derive_equivalences.
13408 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13409 Code moved into....
13410 (edge_info::derive_equivalences): New private member function
13412 * tree-ssa-dom.c (class edge_info): Changed from a struct
13413 to a class. Add ctor/dtor, methods and data members.
13414 (edge_info::edge_info): Renamed from allocate_edge_info.
13415 Initialize additional members.
13416 (edge_info::~edge_info): New.
13417 (free_dom_edge_info): Delete the edge info.
13418 (record_edge_info): Use new class & associated member functions.
13419 Tighten forms for testing for edge equivalences.
13420 (record_temporary_equivalences): Iterate over the simple
13421 equivalences rather than assuming there's only one per edge.
13422 (cprop_into_successor_phis): Iterate over the simple
13423 equivalences rather than assuming there's only one per edge.
13424 (optimize_stmt): Use operand_equal_p rather than pointer
13425 equality for mini-DSE code.
13427 2017-08-28 Nathan Sidwell <nathan@acm.org>
13429 * gcc.c (execute): Fold SIGPIPE handling into switch
13430 statement. Adjust internal error message.
13432 2017-08-28 Richard Biener <rguenther@suse.de>
13435 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13436 Do nothing for removed DIEs.
13438 2017-08-28 Richard Biener <rguenther@suse.de>
13440 PR tree-optimization/81977
13441 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13444 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13447 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13448 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13449 using targetm.gen_mem_thread_fence.
13451 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13454 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13455 predicate to register_operand. Reorder operands.
13456 (*btr<mode>): Ditto.
13457 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13458 (*btr<mode>_mask): Ditto.
13460 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13462 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13463 * config/rs6000/xmmintrin.h: New file.
13464 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13466 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13469 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13470 parameter and_insn and return it.
13471 (recombine_lvx_pattern): Insert a copy to ensure availability of
13472 the base register of the copied masking operation at the point of
13473 the instruction replacement.
13474 (recombine_stvx_pattern): Likewise.
13476 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13478 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13479 undocumented switches.
13480 (-mpower9-dform-vector): Likewise.
13481 (-mpower9-dform): Likewise.
13482 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13483 comments to delete references to -mpower9-dform* switches.
13484 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13485 Delete reference to -mpower9-dform* switches, test for
13486 -mpower9-vector instead.
13487 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13488 (OTHER_P9_VECTOR_MASKS): Likewise.
13489 (POWERPC_MASKS): Likewise.
13490 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13491 tests against -mpower9-dform* to -mpower9-vector. Delete code
13492 that checked for -mpower9-dform* consistancy with other options.
13493 Add test for -mpower9-misc to enable other power9 switches.
13494 (rs6000_init_hard_regno_mode_ok): Likewise.
13495 (rs6000_option_override_internal): Likewise.
13496 (rs6000_emit_prologue): Likewise.
13497 (rs6000_emit_epilogue): Likewise.
13498 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13499 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13500 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13502 (emit_fusion_p9_store): Likewise.
13503 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13504 resetting these macros if the assembler does not support ISA 3.0
13506 (TARGET_P9_DFORM_VECTOR): Likewise.
13507 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13508 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13510 2017-08-25 Alan Modra <amodra@gmail.com>
13512 PR rtl-optimization/81747
13513 * cse.c (cse_extended_basic_block): Don't attempt to record
13514 equivalences for degenerate conditional branches that branch
13515 to their fall-through.
13517 2017-08-24 Martin Sebor <msebor@redhat.com>
13519 PR middle-end/81908
13520 * gimple-fold.c (size_must_be_zero_p): New function.
13521 (gimple_fold_builtin_memory_op): Call it.
13523 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13525 * config/rs6000/mm_malloc.h: New file.
13527 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13529 PR tree-optimization/81913
13530 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13531 analysis when either IVs in condition can wrap.
13533 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13535 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13536 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13538 2017-08-24 Richard Biener <rguenther@suse.de>
13541 * targhooks.c (default_target_can_inline_p): Properly
13542 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13543 is present and always compare.
13544 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13545 infer -mfpmath=sse from TARGET_SSE_P.
13546 (ix86_can_inline_p): Properly use target_option_default_node when
13547 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13549 2017-08-24 Richard Biener <rguenther@suse.de>
13552 * dwarf2out.c (output_die): Handle flag_generate_offload like
13554 (output_comp_unit): Likewise.
13555 (gen_array_type_die): Likewise.
13556 (dwarf2out_early_finish): Likewise.
13557 (note_variable_value_in_expr): Likewise.
13558 (dwarf2out_finish): Likewise. Adjust assert.
13559 * cgraphunit.c (symbol_table::compile): Move setting of
13560 flag_generate_offload earlier ...
13561 (symbol_table::finalize_compilation_unit): ... here, before
13562 early debug finalization.
13564 2017-08-24 Richard Biener <rguenther@suse.de>
13566 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13567 and ipa-fnsummary.h.
13568 (ix86_can_inline_p): When ix86_fpmath flags do not match
13569 check whether the callee uses FP math at all.
13571 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13573 PR middle-end/81931
13574 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13575 instead of TYPE_PRECISION.
13577 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13578 Alan Hayward <alan.hayward@arm.com>
13579 David Sherwood <david.sherwood@arm.com>
13581 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13583 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13585 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13587 (ix86_frame::stack_realign_allocate): New field.
13588 (struct machine_frame_state): Modify comments.
13589 (machine_frame_state::sp_realigned_fp_end): New field.
13590 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13591 layout calculation.
13592 (sp_valid_at): Add assertion to assure no attempt to access invalid
13593 offset of a realigned stack.
13594 (fp_valid_at): Likewise.
13595 (choose_baseaddr): Modify comments.
13596 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13597 ix86_expand_prologue.
13598 (ix86_expand_prologue): Modify stack realignment and allocation.
13599 (ix86_expand_epilogue): Modify comments.
13600 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13601 avx2_runtime, avx512f, and avx512f_runtime.
13603 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13605 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13606 (mstackrealign): Do not init to -1.
13607 * config/i386/i386.c (ix86_option_override_internal):
13608 Check opts_set, not opts when setting default value of
13609 opts->x_ix86_force_align_arg_pointer.
13611 2017-08-23 Richard Biener <rguenther@suse.de>
13613 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13614 lang_hooks.decl_printable_name.
13615 * print-rtl-function.c (print_rtx_function): Likewise.
13616 * tree-pretty-print.c (dump_function_header): Likewise.
13618 2017-08-23 Richard Biener <rguenther@suse.de>
13621 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13622 -g0 at compile-time.
13624 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13626 PR middle-end/19706
13627 * doc/sourcebuild.texi (Other hardware attributes):
13630 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13632 PR middle-end/19706
13633 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13634 Fix single-use check.
13636 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13638 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13640 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13642 * doc/install.texi: Modify to add more details on running selected
13645 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13647 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13648 is combined with -mabi=ms.
13649 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13652 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13654 PR tree-optimization/81488
13655 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13656 and cached_basis fields.
13657 (MAX_SPREAD): New constant.
13658 (alloc_cand_and_find_basis): Initialize new fields.
13659 (clear_visited): New function.
13660 (create_phi_basis_1): Rename from create_phi_basis, set visited
13661 and cached_basis fields.
13662 (create_phi_basis): New wrapper function.
13663 (phi_add_costs_1): Rename from phi_add_costs, add spread
13664 parameter, set visited field, short-circuit when limits reached.
13665 (phi_add_costs): New wrapper function.
13666 (record_phi_increments_1): Rename from record_phi_increments, set
13668 (record_phi_increments): New wrapper function.
13669 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13670 (phi_incr_cost): New wrapper function.
13671 (all_phi_incrs_profitable_1): Rename from
13672 all_phi_incrs_profitable, set visited field.
13673 (all_phi_incrs_profitable): New wrapper function.
13675 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13676 Alan Hayward <alan.hayward@arm.com>
13677 David Sherwood <david.sherwood@arm.com>
13679 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13680 that takes the outer and inner modes.
13681 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13682 comparison as the canonical test for a paradoxical subreg.
13683 * combine.c (simplify_set): Use paradoxical_subreg_p.
13684 (make_extraction): Likewise.
13685 (force_to_mode): Likewise.
13686 (rtx_equal_for_field_assignment_p): Likewise.
13687 (gen_lowpart_for_combine): Likewise.
13688 (simplify_comparison): Likewise.
13689 * cse.c (equiv_constant): Likewise.
13690 * expmed.c (store_bit_field_1): Likewise.
13691 * final.c (alter_subreg): Likewise.
13692 * fwprop.c (propagate_rtx): Likewise.
13693 (forward_propagate_subreg): Likewise.
13694 * ira-conflicts.c (ira_build_conflicts): Likewise.
13695 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13696 * lra-constraints.c (curr_insn_transform): Likewise.
13697 (split_reg): Likewise.
13698 * lra-eliminations.c (move_plus_up): Likewise.
13699 (lra_eliminate_regs_1): Likewise.
13700 * recog.c (general_operand): Likewise.
13701 * ree.c (combine_reaching_defs): Likewise.
13702 * reload.c (push_reload): Likewise.
13703 (find_reloads): Likewise.
13704 * reload1.c (elimination_effects): Likewise.
13705 (compute_reload_subreg_offset): Likewise.
13706 (choose_reload_regs): Likewise.
13707 * rtlanal.c (subreg_lsb_1): Likewise.
13708 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13709 (simplify_subreg): Likewise.
13710 * var-tracking.c (track_loc_p): Likewise.
13711 * emit-rtl.c (byte_lowpart_offset): Likewise.
13712 (paradoxical_subreg_p): Delete out-of-line definition.
13714 2017-08-22 Jeff Law <law@redhat.com>
13716 PR tree-optimization/81741
13717 PR tree-optimization/71947
13718 * tree-ssa-dom.c: Include tree-inline.h.
13719 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13720 equivalences if one is more expensive to compute than the other.
13721 * tree-ssa-scopedtables.h (class const_or_copies): Make
13722 record_const_or_copy_raw method private.
13723 (class avail_exprs_stack): New method simplify_binary_operation.
13724 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13725 avail_exprs_stack::simplify_binary_operation as needed.
13726 (avail_exprs_stack::simplify_binary_operation): New function.
13728 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13730 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13731 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13732 (DOT_SYMBOLS): Likewise.
13733 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13734 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13735 (RS6000_ABI_NAME): Likewise.
13736 (TARGET_CMODEL): Likewise.
13737 (TOC_SECTION_ASM_OP): Likewise.
13738 (SET_CMODEL): New macro.
13739 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13741 2017-08-22 Richard Biener <rguenther@suse.de>
13743 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13744 to allow for free-lang-data replacements similar to verify_type_variant.
13746 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13747 Alan Hayward <alan.hayward@arm.com>
13748 David Sherwood <david.sherwood@arm.com>
13750 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13751 VOIDmode for the LABEL_REF.
13753 2017-08-22 Richard Biener <rguenther@suse.de>
13755 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13757 2017-08-22 Richard Biener <rguenther@suse.de>
13759 * common.opt (feliminate-dwarf2-dups): Ignore.
13760 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13761 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13762 same_die_p_wrap, compute_section_prefix,
13763 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13764 (comdat_symbol_id, comdat_symbol_number): Likewise.
13765 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13767 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13768 (output_die): Mark unreachable path unreachable.
13769 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13770 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13771 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13772 (dwarf2out_early_finish): Likewise.
13774 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13776 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13778 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13781 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13782 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13783 (avr_attribute_table) <io, io_low, address>: Initialize
13784 .decl_required with true.
13786 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13788 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13789 undocumented debugging options.
13790 (-mvsx-scalar-double): Likewise.
13791 (-mallow-df-permute): Likewise.
13792 (-mvectorize-builtins): Likewise.
13793 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13794 (rs6000_builtin_vectorized_function): Likewise.
13795 (rs6000_builtin_md_vectorized_function): Likewise.
13796 (rs6000_opt_vars): Likewise.
13798 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13801 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13802 (*btrq_imm): Rename from *btrq.
13803 (*btcq_imm): Rename from *btcq.
13804 (btsc): New code attribute.
13805 (*<btsc><mode>): New insn pattern.
13806 (*btr<mode>): Ditto.
13807 (*<btsc><mode>_mask): New insn_and_split pattern.
13808 (*btr<mode>_mask): Ditto.
13810 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13811 Alan Hayward <alan.hayward@arm.com>
13812 David Sherwood <david.sherwood@arm.com>
13814 * function.c (pad_below): Simplify padding calculation.
13816 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13817 Alan Hayward <alan.hayward@arm.com>
13818 David Sherwood <david.sherwood@arm.com>
13820 * target.def (function_prologue): Remove frame size argument.
13821 (function_epilogue): Likewise.
13822 * doc/tm.texi: Regenerate.
13823 * final.c (final_start_function): Update call to function_prologue.
13824 (final_end_function): Update call to function_epilogue.
13825 (default_function_pro_epilogue): Remove frame size argument.
13826 * output.h (default_function_pro_epilogue): Likewise.
13827 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13828 (arm_output_function_prologue): Likewise.
13829 * config/frv/frv.c (frv_function_prologue): Likewise.
13830 (frv_function_epilogue): Likewise.
13831 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13832 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13833 (ia64_output_function_epilogue): Likewise.
13834 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13835 (m32r_output_function_epilogue): Likewise.
13836 * config/microblaze/microblaze.c (microblaze_function_prologue)
13837 (microblaze_function_epilogue): Likewise.
13838 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13839 (mips_output_function_epilogue): Likewise.
13840 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13841 (mmix_target_asm_function_epilogue): Likewise.
13842 * config/msp430/msp430.c (msp430_start_function): Likewise.
13843 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13844 (nds32_asm_function_epilogue): Likewise.
13845 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13846 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13847 (pa_output_function_epilogue): Likewise.
13848 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13849 (rs6000_output_function_epilogue): Likewise.
13850 * config/rl78/rl78.c (rl78_start_function): Likewise.
13851 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13852 (rs6000_output_function_epilogue): Likewise.
13853 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13854 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13855 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13856 (sparc_asm_function_epilogue): Likewise.
13858 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13860 * tree.h (type_has_mode_precision_p): New function.
13861 * convert.c (convert_to_integer_1): Use it.
13862 * expr.c (expand_expr_real_2): Likewise.
13863 (expand_expr_real_1): Likewise.
13864 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13865 * match.pd: Likewise.
13866 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13867 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13868 * tree-tailcall.c (process_assignment): Likewise.
13869 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13870 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13871 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13872 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13873 (vectorizable_assignment): Likewise.
13874 (vectorizable_shift): Likewise.
13875 (vectorizable_operation): Likewise.
13876 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13878 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13880 * match.pd: Add pow (C, x) simplification.
13882 2017-08-21 Richard Biener <rguenther@suse.de>
13884 PR tree-optimization/81900
13885 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13886 for blocks with abnormal predecessors.
13887 (compute_antic): Do not set visited flag prematurely.
13889 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13892 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13894 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13896 * stor-layout.h (vector_type_mode): Move to...
13897 * tree.h (vector_type_mode): ...here.
13898 * stor-layout.c (vector_type_mode): Move to...
13899 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13901 2017-08-21 Richard Biener <rguenther@suse.de>
13903 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13904 register_external_die hooks.
13905 (debug_false_tree_charstarstar_uhwistar): Declare.
13906 (debug_nothing_tree_charstar_uhwi): Likewise.
13907 * debug.c (do_nothing_debug_hooks): Adjust.
13908 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13909 (debug_nothing_tree_charstar_uhwi): Likewise.
13910 * dbxout.c (dbx_debug_hooks): Adjust.
13911 (xcoff_debug_hooks): Likewise.
13912 * sdbout.c (sdb_debug_hooks): Likewise.
13913 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13914 * dwarf2out.c (macinfo_label_base): New global.
13915 (dwarf2out_register_external_die): New function for the
13916 register_external_die hook.
13917 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13918 (dwarf2_debug_hooks): Use them.
13919 (dwarf2_lineno_debug_hooks): Adjust.
13920 (struct die_struct): Add with_offset flag.
13921 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13922 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13923 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13924 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13925 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13926 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13927 defining section names for the early LTO debug variants.
13928 (reset_indirect_string): New helper.
13929 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13930 (print_dw_val): Add support for offsetted symbol references.
13931 (get_ultimate_context): Split out from is_cxx.
13932 (is_cxx): Use get_ultimate_context.
13933 (is_fortran): Add decl overload.
13934 (compute_comp_unit_symbol): Split out worker from
13935 compute_section_prefix.
13936 (compute_section_prefix): Call compute_comp_unit_symbol and
13937 set comdat_type_p here.
13938 (output_die): Skip DIE symbol output for the LTO added one.
13939 Handle DIE symbol references with offset.
13940 (output_comp_unit): Guard section name mangling properly.
13941 For LTO debug sections emit a symbol at the section beginning
13942 which we use to refer to its DIEs.
13943 (add_abstract_origin_attribute): For DIEs registered via
13944 dwarf2out_register_external_die directly refer to the early
13945 DIE rather than indirectly through the shadow one we created.
13946 Remove obsolete call to dwarf2out_abstract_function for
13947 non-function/block origins.
13948 (gen_array_type_die): When generating early LTO debug do
13949 not emit DW_AT_string_length.
13950 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13951 late when in LTO. As suggested place a gcc_unreachable for
13952 the DECL_ABSTRACT_P case.
13953 (gen_subprogram_die): Avoid another specification DIE
13954 for early built declarations/definitions for the late LTO case.
13955 (gen_variable_die): Add type references for late duplicated VLA dies
13957 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13958 we have the abstract instance already.
13959 (process_scope_var): Adjust decl DIE contexts in LTO which
13960 first puts them in limbo.
13961 (gen_decl_die): Do not generate type DIEs late apart from
13962 types for VLAs or for decls we do not yet have a DIE. Do not
13963 call dwarf2out_abstract_function late.
13964 (dwarf2out_early_global_decl): Make sure to create DIEs
13965 for abstract instances of a decl first.
13966 (dwarf2out_late_global_decl): Adjust comment.
13967 (output_macinfo_op): With multiple macro sections use
13968 macinfo_label_base to distinguish labels.
13969 (output_macinfo): Likewise. Update macinfo_label_base.
13970 Pass in the line info label.
13971 (note_variable_value_in_expr): When generating LTO resolve
13972 all variable values here by generating DIEs as needed.
13973 (init_sections_and_labels): Add early LTO debug flag parameter
13974 and generate different sections and names if set. Add generation
13975 counter for the labels so we can have multiple of them.
13976 (reset_dies): Helper to allow DIEs to be output multiple times.
13977 (dwarf2out_finish): When outputting DIEs to the fat part of an
13978 LTO object first reset DIEs.
13979 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13980 (modified_type_die): Check for decl_ultimate_origin being self
13982 (gen_type_die_with_usage): Likewise.
13983 (gen_typedef_die): Allow decl_ultimate_origin being self.
13984 (set_decl_abstract_flags): Remove.
13985 (set_block_abstract_flags): Likewise.
13986 (dwarf2out_abstract_function): Treat the early generated DIEs
13987 as the abstract copy and only add DW_AT_inline and
13988 DW_AT_artificial here and call set_decl_origin_self.
13989 If the DIE has an abstract origin don't do anything.
13990 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13991 if we have none yet (Go fails to build one, PR78628).
13992 (variably_modified_type_p): Prevent endless recursion for Ada
13993 cyclic pointer types.
13994 * lto-streamer-in.c: Include debug.h.
13995 (dref_queue): New global.
13996 (lto_read_tree_1): Stream in DIE references.
13997 (lto_input_tree): Register DIE references.
13998 (input_function): Stream DECL_DEBUG_ARGS.
13999 * lto-streamer-out.c: Include debug.h.
14000 (lto_write_tree_1): Output DIE references.
14001 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14002 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14003 (output_function): Stream DECL_DEBUG_ARGS.
14004 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14005 Stream DECL_ABSTRACT_ORIGIN.
14006 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14007 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14008 DECL_CONTEXT for file-scope decls.
14009 * lto-streamer.h (struct dref_entry): Declare.
14010 (dref_queue): Likewise.
14011 * cfgexpand.c (pass_expand::execute): Do not call the
14012 outlining_inline_function hook here.
14013 * lto-wrapper.c (debug_obj): New global.
14014 (tool_cleanup): Unlink it if required.
14015 (debug_objcopy): New function.
14016 (run_gcc): Handle early debug sections in the IL files by
14017 extracting them to separate files, partially linkin them and
14018 feeding the result back as result to the linker.
14019 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14020 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14021 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14022 sections into a separate segment.
14023 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14025 (darwin_asm_dwarf_section): Likewise.
14026 (darwin_asm_output_dwarf_offset): Likewise.
14027 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14029 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14030 Alan Hayward <alan.hayward@arm.com>
14031 David Sherwood <david.sherwood@arm.com>
14033 * read-md.h (md_reader::record_potential_iterator_use): Replace
14034 pointer argument with an rtx and an index.
14035 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14036 (apply_mode_iterator): Likewise.
14037 (apply_code_iterator): Likewise.
14038 (apply_int_iterator): Likewise.
14039 (apply_subst_iterator): Likewise.
14040 (record_iterator_use): Likewise.
14041 (record_attribute_use): Likewise.
14042 (md_reader::record_potential_iterator_use): Likewise. Update calls
14043 to record_iterator_use and apply_iterator.
14044 (iterator_use): Replace ptr with x and index.
14045 (attribute_use): Likewise.
14046 (apply_attribute_uses): Update calls to apply_iterator.
14047 (apply_iterators): Likewise. Update initialization of iterator_use.
14048 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14049 and record_potential_iterator_use.
14050 (rtx_reader::read_rtx_operand): Likewise.
14052 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14053 Alan Hayward <alan.hayward@arm.com>
14054 David Sherwood <david.sherwood@arm.com>
14056 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14059 2017-08-21 Richard Biener <rguenther@suse.de>
14061 PR middle-end/81884
14062 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14063 at struct end conservatively when comparing common bases.
14065 2017-08-21 Richard Biener <rguenther@suse.de>
14067 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14068 (mem_ref_in_stmt): Remove.
14069 (determine_max_movement): Use ref index to get at the reference.
14070 (invariantness_dom_walker::before_dom_children): Deal with
14071 lim data already initialized.
14072 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14074 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14076 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14077 (TARGET_ISA_ROUND): Ditto.
14078 (TARGET_ROUND): Ditto.
14079 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14080 * config/i386/i386.md: Ditto.
14081 * config/i386/sse.md: Ditto.
14082 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14083 with OPTION_MASK_ISA_SSE4_1.
14085 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14088 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14089 __builtin_ia32_lzcnt_u16.
14091 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14094 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14095 (rs6000_set_current_function): Rewrite function to use it.
14097 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14100 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14101 and TYPE_WARN_IF_NOT_ALIGN.
14102 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14103 (handle_warn_if_not_align): New.
14104 (place_union_field): Call handle_warn_if_not_align.
14105 (place_field): Call handle_warn_if_not_align.
14106 Copy TYPE_WARN_IF_NOT_ALIGN.
14107 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14108 (layout_type): Likewise.
14109 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14111 (tree_decl_common): Add warn_if_not_align.
14112 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14113 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14114 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14115 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14116 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14117 * doc/extend.texi: Document warn_if_not_aligned attribute.
14118 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14120 2017-08-17 Martin Liska <mliska@suse.cz>
14123 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14124 (get_data_dependence): Use it as pointer type.
14125 (distribute_loop): Likewise.
14127 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14129 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14130 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14131 (p8_vmrgow_<mode>_direct): New define_insn.
14132 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14133 handle endianness for vmrgew and vmrgow permute patterns.
14135 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14137 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14138 * config/rs6000/rs6000-cpus.def: Remove comment.
14139 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14140 (POWERPC_MASKS): Likewise.
14141 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14142 use of TARGET_VSX_TIMODE.
14143 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14144 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14145 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14146 (rs6000_option_override_internal): Remove dead code.
14147 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14148 (rs6000_legitimize_reload_address): Likewise.
14149 (rs6000_legitimate_address_p): Likewise.
14150 (rs6000_opt_masks): Delete "vsx-timode".
14151 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14152 from function comment.
14153 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14154 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14155 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14157 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14158 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14159 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14160 (VSX_TI): Likewise.
14162 (define_peephole2): Likewise.
14164 2017-08-17 Martin Sebor <msebor@redhat.com>
14167 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14168 past the end of an array.
14169 (test_pp_format): Add test cases.
14171 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14173 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14174 missing ECF_NOTHROW flags.
14176 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14179 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14180 operands residing in integer registers.
14181 (*vsx_le_perm_load_<mode>): Likewise.
14182 (*vsx_le_perm_store_<mode>): Likewise.
14183 (define_peephole2): Add peepholes to optimize the above.
14185 2017-08-17 Marek Polacek <polacek@redhat.com>
14187 PR middle-end/81814
14188 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14189 to mimic what shorten_compare did. Change the return type to bool.
14190 (fold_cond_expr_with_comparison): Update call to
14191 operand_equal_for_comparison_p.
14192 (fold_ternary_loc): Likewise.
14194 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14196 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14198 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14200 2017-08-17 Richard Biener <rguenther@suse.de>
14202 * tree-ssa-structalias.c (solve_graph): When propagating
14203 to successors update the graphs succ edges and avoid duplicate work.
14205 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14208 * config/i386/i386.c (ix86_option_override_internal): Save target
14209 specific options after ix86_stack_protector_guard_reg was changed.
14211 2017-08-17 Richard Biener <rguenther@suse.de>
14213 PR tree-optimization/81827
14214 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14215 (new_var_info): Initialize it conservatively.
14216 (get_call_vi): Mark register vars.
14217 (new_scalar_tmp_constraint_exp): Likewise.
14218 (handle_rhs_call): Likewise.
14219 (handle_const_call): Likewise.
14220 (create_function_info_for): Likewise.
14221 (solve_constraints): Sort varinfos to separate register from
14222 non-register vars to pack points-to solution bitmaps during
14225 2017-08-17 Marek Polacek <polacek@redhat.com>
14227 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14229 2017-08-17 Richard Biener <rguenther@suse.de>
14231 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14232 to true when overflow is undefined and we saturated the result.
14234 2017-08-17 Alan Modra <amodra@gmail.com>
14237 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14238 Don't use store multiple if only one reg needs saving.
14239 (interesting_frame_related_regno): New function.
14240 (rs6000_frame_related): Don't emit frame info for regs that
14242 (rs6000_emit_epilogue): Likewise.
14244 2017-08-16 Nathan Sidwell <nathan@acm.org>
14246 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14247 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14248 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14249 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14250 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14251 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14252 (verify_type): Adjust for TYPE_BINFO move.
14253 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14254 process TYPE_BINFO directly.
14255 (hash_tree): Likewise.
14256 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14258 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14261 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14263 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14265 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14268 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14269 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14270 Add ix86_binary_operator_ok to insn constraint.
14271 (*iordi_1_bts): Ditto.
14272 (*xordi_1_btc): Ditto.
14273 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14274 Update corresponding peephole2 pattern.
14278 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14280 PR tree-optimization/81832
14281 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14282 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14284 2017-08-16 Marek Polacek <polacek@redhat.com>
14287 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14288 regarding size_zero_node.
14290 2017-08-16 Martin Liska <mliska@suse.cz>
14293 * config.gcc: Respect previously set extra_objs in case
14296 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14298 PR tree-optimization/81835
14299 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14300 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14301 not depend on the phi.
14303 2017-08-16 Alan Modra <amodra@gmail.com>
14305 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14308 2017-08-16 Alan Modra <amodra@gmail.com>
14310 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14311 (save_reg_p): ..into this. Update all callers.
14312 (first_reg_to_save): Simplify.
14314 2017-08-16 Alan Modra <amodra@gmail.com>
14316 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14319 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14323 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14324 constant count if that count is less than 32.
14326 2017-08-15 Nathan Sidwell <nathan@acm.org>
14328 * gcc.c (execute): Emit friendlier message if inferior is killed
14329 by an external cause.
14331 2017-08-15 Richard Biener <rguenther@suse.de>
14333 PR tree-optimization/81790
14334 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14335 CONSTRUCTORs from simplifying and VN.
14337 2017-08-14 Martin Sebor <msebor@redhat.com>
14339 * builtin-attrs.def: Add comments.
14341 2017-08-14 Martin Sebor <msebor@redhat.com>
14344 * doc/extend.texi (attribute nonstring): Document new attribute.
14346 2017-08-14 Martin Sebor <msebor@redhat.com>
14349 * tree-diagnostic.c (default_tree_printer): Handle %G.
14350 * gimple-pretty-print.h (percent_G_format): Declare new function.
14351 * gimple-pretty-print.c (percent_G_format): Define.
14352 * tree-pretty-print.c (percent_K_format): Add argument.
14354 2017-08-14 Martin Sebor <msebor@redhat.com>
14356 PR translation/79998
14357 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14358 Remove a stray space.
14360 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14363 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14364 (*iordi_1_bts): Ditto.
14365 (*xordi_1_btc): Ditto.
14367 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14370 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14371 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14373 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14374 quoted strings, and make more translator-friendly.
14375 (darwin_rs6000_override_options): Likewise.
14376 (rs6000_option_override_internal): Likewise.
14377 (rs6000_return_in_memory): Fix overlong line.
14378 (init_cmulative_args): Use quoted strings, and make more
14379 translator-friendly.
14380 (rs6000_pass_by_reference): Fix overlong line.
14381 (def_builtin): Use quoted strings.
14382 (altivec_expand_predicate_builtin): Use quoted strings, and make
14383 more translator-friendly.
14384 (htm_expand_builtin): Use quoted strings.
14385 (cpu_expand_builtin): Use quoted strings, and make more
14386 translator-friendly.
14387 (altivec_expand_builtin): Likewise.
14388 (paired_expand_predicate_builtin): Likewise.
14389 (rs6000_invalid_builtin): Likewise.
14390 (builtin_function_type): Use quoted strings.
14391 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14392 more translator-friendly.
14393 (rs6000_trampoline_init): Likewise.
14394 (rs6000_handle_altivec_attribute): Likewise.
14395 (rs6000_inner_target_options): Use quoted strings.
14396 (rs6000_disable_incompatible_switches): Likewise.
14397 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14398 strings, and make more translator-friendly.
14399 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14401 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14403 PR tree-optimization/81799
14404 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14405 cond_expr to simple gimple operand.
14407 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14409 PR middle-end/46932
14410 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14412 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14416 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14417 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14418 TARGET_GASISR_PROLOGUES.
14419 * config/avr/avr.c (avr_option_override): Same.
14420 (avr_pass_pre_proep::execute): Same.
14422 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14425 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14426 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14428 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14430 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14431 define_insn_and_split. Split to a memory load from 0 in
14432 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14433 using PTR mode iterator.
14434 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14435 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14436 (*add_tp_<mode>): Redefine as define_insn_and_split.
14437 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14438 address space. Merge with *add_tp_x32 using PTR mode iterator.
14439 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14440 Split to an add with a memory load from 0 in
14441 DEFAULT_TLS_SEG_REG address space.
14443 2017-08-12 Andrew Pinski <apinski@cavium.com>
14445 * config/aarch64/aarch64-option-extensions.def (rdma):
14446 Fix feature string to what Linux prints out in /proc/cpuinfo.
14448 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14451 * dwarf2out.c (modified_type_die): For C typedef types that have
14452 an ultimate origin, process the ultimate origin instead of the
14454 (gen_typedef_die): Assert that input DECLs have no ultimate
14456 (gen_type_die_with_usage): For typedef variants that have an
14457 ultimate origin, just call gen_decl_die on the original DECL.
14458 (process_scope_var): Avoid creating DIEs for local typedefs and
14459 concrete static variables.
14461 2017-08-12 Alan Modra <amodra@gmail.com>
14465 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14466 match gnu-user.h startfile.
14467 (ENDFILE_LINUX_SPEC): Similarly.
14469 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14472 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14474 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14476 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14477 * config/aarch64/aarch64.md (mov<mode>): Change.
14478 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14479 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14480 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14482 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14484 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14485 for storage order barriers.
14487 2017-08-11 Martin Liska <mliska@suse.cz>
14490 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14491 variables of void type.
14493 2017-08-11 Martin Liska <mliska@suse.cz>
14495 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14496 TARGET_SUPPORTS_ALIASES.
14497 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14498 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14499 (optimize_weakref): Likewise.
14500 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14501 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14502 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14504 2017-08-11 Martin Liska <mliska@suse.cz>
14507 * config/i386/i386.c (make_resolver_func): Do complete
14508 refactoring of the function.
14510 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14513 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14514 * config/i386/i386.c (ix86_stack_protect_guard): Use
14515 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14516 * doc/invoke.texi (x86 Options): Document
14517 -mstack-protector-guard-symbol= option.
14519 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14521 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14522 * config/i386/i386.c (ix86_split_stack_guard): New function.
14523 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14524 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14525 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14526 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14527 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14528 (split_stack_space_check): Call ix86_split_stack_guard.
14530 2017-08-10 Martin Sebor <msebor@redhat.com>
14532 * print-tree.c (print_node): Print location using the established
14534 Replace spaces with colons.
14535 (debug_raw, debug): Ditto.
14537 2017-08-10 Martin Sebor <msebor@redhat.com>
14540 * pretty-print.c (pp_format): Correct the handling of %s precision.
14542 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14545 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14547 (ix86_finalize_stack_frame_flags): This. Also clear
14548 frame_pointer_needed if -fno-omit-frame-pointer is used without
14550 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14551 with ix86_finalize_stack_frame_flags.
14552 (ix86_expand_epilogue): Likewise.
14553 (ix86_expand_split_stack_prologue): Likewise.
14554 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14556 2017-08-10 Martin Liska <mliska@suse.cz>
14559 * c-attribs.c (handle_target_attribute):
14560 Report warning for an empty string argument of target attribute.
14562 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14565 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14567 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14568 or DECL_NONLOCAL labels.
14569 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14570 or DECL_NONLOCAL labels here.
14572 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14574 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14575 to indicate when early gimple folding has been disabled.
14576 (rs6000_gimple_fold_builtin): Add debug content.
14577 (rs6000_invalid_builtin): Fix whitespace.
14578 (rs6000_expand_builtin): Fix whitespace.
14579 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14581 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14584 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14585 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14587 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14589 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14590 (qdf24xx): Likewise.
14591 * config/aarch64/aarch64.md: Include falkor.md.
14592 * config/aarch64/falkor.md: New.
14594 2017-08-09 Marek Polacek <polacek@redhat.com>
14597 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14598 * diagnostic.c (emit_diagnostic): Add a comment.
14599 (emit_diagnostic_valist): New function.
14601 2017-08-09 Marek Polacek <polacek@redhat.com>
14604 * input.c (make_location): New overload.
14605 * input.h (make_location): Declare.
14607 2017-08-08 Alan Modra <amodra@gmail.com>
14608 H.J. Lu <hongjiu.lu@intel.com>
14611 * gcc.c (NO_PIE_SPEC): Delete.
14612 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14614 (LINK_PIE_SPEC): ..to here.
14615 (LINK_COMMAND_SPEC): Support -no-pie.
14616 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14617 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14619 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14620 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14621 (ENDFILE_CRTEND_SPEC): Similarly.
14623 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14626 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14627 (mstack-protector-guard-offset=): Ditto.
14628 * config/i386/i386.c (ix86_option_override): Handle
14629 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14631 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14632 ix86_stack_protect_guard_offset variables.
14633 (TARGET_STACK_PROTECT_GUARD): Always define.
14634 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14635 and -mstack-protector-guard-offset= options.
14637 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14639 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14640 boundary case for the last candidate.
14642 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14644 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14645 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14647 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14649 PR middle-end/19706
14650 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14651 * config/aarch64/aarch64-builtins.c
14652 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14653 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14654 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14656 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14657 Andrew Pinski <pinskia@gmail.com>
14659 PR middle-end/19706
14660 * internal-fn.def (XORSIGN): New.
14661 * optabs.def (xorsign_optab): New.
14662 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14663 (convert_expand_mult_copysign): New.
14664 (pass_optimize_widening_mul::execute): Call
14665 convert_expand_mult_copysign.
14667 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14669 PR tree-optimization/81354
14670 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14671 Insert on edges rather than explicitly creating landing pads.
14672 (analyze_candidates_and_replace): Commit edge inserts.
14674 2017-08-08 Richard Biener <rguenther@suse.de>
14676 PR middle-end/81719
14677 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14678 (expand_simple_operations): Also look through ADDR_EXPRs with
14679 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14681 2017-08-08 Richard Biener <rguenther@suse.de>
14683 PR tree-optimization/81723
14684 * tree-vect-slp.c (struct bst_traits): New hash traits.
14685 (bst_fail): New global.
14686 (vect_build_slp_tree_2): New worker, split out from ...
14687 (vect_build_slp_tree): ... this now wrapping it with using
14688 bst_fail set to cache SLP tree build fails. Properly handle
14690 (vect_analyze_slp_instance): Allocate and free bst_fail.
14692 2017-08-08 Martin Liska <mliska@suse.cz>
14695 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14696 LABEL_DECLs that can be from a different function.
14698 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14700 PR tree-optimization/81744
14701 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14702 loop's number of iterations.
14704 2017-08-08 Martin Liska <mliska@suse.cz>
14706 * asan.c: Include header files.
14707 * attribs.c (build_decl_attribute_variant): New function moved
14709 (build_type_attribute_qual_variant): Likewise.
14710 (cmp_attrib_identifiers): Likewise.
14711 (simple_cst_list_equal): Likewise.
14712 (omp_declare_simd_clauses_equal): Likewise.
14713 (attribute_value_equal): Likewise.
14714 (comp_type_attributes): Likewise.
14715 (build_type_attribute_variant): Likewise.
14716 (lookup_ident_attribute): Likewise.
14717 (remove_attribute): Likewise.
14718 (merge_attributes): Likewise.
14719 (merge_type_attributes): Likewise.
14720 (merge_decl_attributes): Likewise.
14721 (merge_dllimport_decl_attributes): Likewise.
14722 (handle_dll_attribute): Likewise.
14723 (attribute_list_equal): Likewise.
14724 (attribute_list_contained): Likewise.
14725 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14726 (lookup_attribute_by_prefix): Likewise.
14727 * bb-reorder.c: Include header files.
14728 * builtins.c: Likewise.
14729 * calls.c: Likewise.
14730 * cfgexpand.c: Likewise.
14731 * cgraph.c: Likewise.
14732 * cgraphunit.c: Likewise.
14733 * convert.c: Likewise.
14734 * dwarf2out.c: Likewise.
14735 * final.c: Likewise.
14736 * fold-const.c: Likewise.
14737 * function.c: Likewise.
14738 * gimple-expr.c: Likewise.
14739 * gimple-fold.c: Likewise.
14740 * gimple-pretty-print.c: Likewise.
14741 * gimple.c: Likewise.
14742 * gimplify.c: Likewise.
14743 * hsa-common.c: Likewise.
14744 * hsa-gen.c: Likewise.
14745 * internal-fn.c: Likewise.
14746 * ipa-chkp.c: Likewise.
14747 * ipa-cp.c: Likewise.
14748 * ipa-devirt.c: Likewise.
14749 * ipa-fnsummary.c: Likewise.
14750 * ipa-inline.c: Likewise.
14751 * ipa-visibility.c: Likewise.
14753 * lto-cgraph.c: Likewise.
14754 * omp-expand.c: Likewise.
14755 * omp-general.c: Likewise.
14756 * omp-low.c: Likewise.
14757 * omp-offload.c: Likewise.
14758 * omp-simd-clone.c: Likewise.
14759 * opts-global.c: Likewise.
14760 * passes.c: Likewise.
14761 * predict.c: Likewise.
14762 * sancov.c: Likewise.
14763 * sanopt.c: Likewise.
14764 * symtab.c: Likewise.
14765 * toplev.c: Likewise.
14766 * trans-mem.c: Likewise.
14767 * tree-chkp.c: Likewise.
14768 * tree-eh.c: Likewise.
14769 * tree-into-ssa.c: Likewise.
14770 * tree-object-size.c: Likewise.
14771 * tree-parloops.c: Likewise.
14772 * tree-profile.c: Likewise.
14773 * tree-ssa-ccp.c: Likewise.
14774 * tree-ssa-live.c: Likewise.
14775 * tree-ssa-loop.c: Likewise.
14776 * tree-ssa-sccvn.c: Likewise.
14777 * tree-ssa-structalias.c: Likewise.
14778 * tree-ssa.c: Likewise.
14779 * tree-streamer-in.c: Likewise.
14780 * tree-vectorizer.c: Likewise.
14781 * tree-vrp.c: Likewise.
14782 * tsan.c: Likewise.
14783 * ubsan.c: Likewise.
14784 * varasm.c: Likewise.
14785 * varpool.c: Likewise.
14786 * tree.c: Remove functions moved to attribs.[ch].
14787 * tree.h: Likewise.
14788 * config/aarch64/aarch64.c: Add attrs.h header file.
14789 * config/alpha/alpha.c: Likewise.
14790 * config/arc/arc.c: Likewise.
14791 * config/arm/arm.c: Likewise.
14792 * config/avr/avr.c: Likewise.
14793 * config/bfin/bfin.c: Likewise.
14794 * config/c6x/c6x.c: Likewise.
14795 * config/cr16/cr16.c: Likewise.
14796 * config/cris/cris.c: Likewise.
14797 * config/darwin.c: Likewise.
14798 * config/epiphany/epiphany.c: Likewise.
14799 * config/fr30/fr30.c: Likewise.
14800 * config/frv/frv.c: Likewise.
14801 * config/ft32/ft32.c: Likewise.
14802 * config/h8300/h8300.c: Likewise.
14803 * config/i386/winnt.c: Likewise.
14804 * config/ia64/ia64.c: Likewise.
14805 * config/iq2000/iq2000.c: Likewise.
14806 * config/lm32/lm32.c: Likewise.
14807 * config/m32c/m32c.c: Likewise.
14808 * config/m32r/m32r.c: Likewise.
14809 * config/m68k/m68k.c: Likewise.
14810 * config/mcore/mcore.c: Likewise.
14811 * config/microblaze/microblaze.c: Likewise.
14812 * config/mips/mips.c: Likewise.
14813 * config/mmix/mmix.c: Likewise.
14814 * config/mn10300/mn10300.c: Likewise.
14815 * config/moxie/moxie.c: Likewise.
14816 * config/msp430/msp430.c: Likewise.
14817 * config/nds32/nds32-isr.c: Likewise.
14818 * config/nds32/nds32.c: Likewise.
14819 * config/nios2/nios2.c: Likewise.
14820 * config/nvptx/nvptx.c: Likewise.
14821 * config/pa/pa.c: Likewise.
14822 * config/pdp11/pdp11.c: Likewise.
14823 * config/powerpcspe/powerpcspe.c: Likewise.
14824 * config/riscv/riscv.c: Likewise.
14825 * config/rl78/rl78.c: Likewise.
14826 * config/rx/rx.c: Likewise.
14827 * config/s390/s390.c: Likewise.
14828 * config/sh/sh.c: Likewise.
14829 * config/sol2.c: Likewise.
14830 * config/sparc/sparc.c: Likewise.
14831 * config/spu/spu.c: Likewise.
14832 * config/stormy16/stormy16.c: Likewise.
14833 * config/tilegx/tilegx.c: Likewise.
14834 * config/tilepro/tilepro.c: Likewise.
14835 * config/v850/v850.c: Likewise.
14836 * config/vax/vax.c: Likewise.
14837 * config/visium/visium.c: Likewise.
14838 * config/xtensa/xtensa.c: Likewise.
14840 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14843 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14844 constraints since the -mupper-regs-* switches have been
14846 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14847 into a vector from a double word element that was extracted from
14848 another vector, and eliminate extra XXPERMDI instructions.
14849 (vsx_concat_<mode>_2): Likewise.
14850 (vsx_concat_<mode>_3): Likewise.
14851 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14852 concat to allow optimizing inserts from previous extracts.
14854 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14856 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14857 memory reference to a SSP offset in TLS address space.
14858 (ix86_print_operand) <case '@'>: Remove.
14859 (ix86_print_operand_punct_valid_p): Remove '@' code.
14860 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14861 UNSPEC_SP_TLS_TEST.
14862 (stack_tls_protect_set_<mode>): Remove.
14863 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14864 (stack_tls_protect_test_<mode>): Remove.
14865 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14867 2017-08-07 Olivier Hainque <hainque@adacore.com>
14870 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14872 2017-08-07 Douglas Rupp <rupp@adacore.com>
14874 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14875 variable was referenced as multidir in command.
14877 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14880 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14883 2017-08-07 Martin Liska <mliska@suse.cz>
14885 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14886 * config/rl78/rl78.c: Add include of attribs.h.
14887 * config/sh/sh.c: Likewise.
14888 * config/v850/v850.c: Likewise.
14890 2017-08-07 Tom de Vries <tom@codesourcery.com>
14892 PR middle-end/78266
14893 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14895 2017-08-07 Martin Liska <mliska@suse.cz>
14897 * config/mips/mips.c: Include attribs.h.
14899 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14902 * doc/invoke.texi: Document change in behvaior for -Ofast for
14905 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14907 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14909 (aarch64_pop_regs): Likewise.
14910 (aarch64_gen_load_pair): Likewise.
14911 (aarch64_save_callee_saves): Likewise.
14912 (aarch64_restore_callee_saves): Likewise.
14914 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14916 * config/i386/i386.c: Revert the last change.
14918 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14921 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14923 (ix86_finalize_stack_frame_flags): This. Also clear
14924 frame_pointer_needed if -fno-omit-frame-pointer is used without
14926 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14927 with ix86_finalize_stack_frame_flags.
14928 (ix86_expand_epilogue): Likewise.
14929 (ix86_expand_split_stack_prologue): Likewise.
14931 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14934 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14935 to P_AES for Westmere.
14937 2017-08-07 Jonathan Yong <10walls@gmail.com>
14939 * config/i386/mingw.opt (fset-stack-executable): Removed.
14940 * config/i386/cygming.opt (fset-stack-executable): Moved
14942 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14944 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14946 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14948 2017-08-07 Marek Polacek <polacek@redhat.com>
14950 PR middle-end/81737
14951 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14953 2017-08-07 Martin Liska <mliska@suse.cz>
14955 * attribs.h (canonicalize_attr_name): New function.
14956 (cmp_attribs): Move from c-format.c and adjusted.
14957 (is_attribute_p): Moved from tree.h.
14958 * tree-inline.c: Add new includes.
14959 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14960 (private_is_attribute_p): Remove.
14961 (private_lookup_attribute): Likewise.
14962 (private_lookup_attribute_by_prefix): Simplify.
14963 (remove_attribute): Use is_attribute_p.
14964 * tree.h: Remove removed declarations.
14966 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14968 PR middle-end/81698
14969 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14970 instead of computing it in the function. Formatting fix.
14971 (expand_case): Don't rely on default_edge being the first edge,
14972 clear it if removing it, pass default_edge to
14973 emit_case_dispatch_table.
14974 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14977 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14979 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14980 insn in the function, emit NOP after the insn.
14982 2017-08-06 Tom de Vries <tom@codesourcery.com>
14984 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14987 2017-08-06 Tom de Vries <tom@codesourcery.com>
14989 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14992 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14994 PR tree-optimization/57371
14995 * match.pd: New pattern.
14997 2017-08-04 Marek Polacek <polacek@redhat.com>
14999 PR middle-end/81695
15000 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15001 perform the computation in offset_int.
15003 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15005 PR tree-optimization/81136
15006 * tree-vectorizer.h: Include tree-hash-traits.h.
15007 (vec_base_alignments): New typedef.
15008 (vec_info): Add a base_alignments field.
15009 (vect_record_base_alignments): Declare.
15010 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15012 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15013 (create_data_ref): Add an is_conditional_in_stmt argument.
15014 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15015 the is_conditional_in_stmt field.
15016 (data_ref_loc): Add an is_conditional_in_stmt field.
15017 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15018 (find_data_references_in_stmt): Update call to create_data_ref.
15019 (graphite_find_data_references_in_stmt): Likewise.
15020 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15021 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15022 (vect_record_base_alignment): New function.
15023 (vect_record_base_alignments): Likewise.
15024 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15025 for nested statements even if we fail to compute a misalignment.
15026 Use pooled base alignments for unconditional references.
15027 (vect_find_same_alignment_drs): Compare base addresses instead
15029 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15030 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15032 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15034 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15035 Add an explicit name for the enum. Use auto_vec for slp_instances
15036 and grouped_stores.
15037 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15039 (_bb_vec_info): Add a constructor and destructor.
15040 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15041 (destroy_loop_vec_info): Delete.
15042 (vect_destroy_datarefs): Likewise.
15043 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15044 (vec_info::vec_info): New function.
15045 (vec_info::~vec_info): Likewise.
15046 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15047 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15048 destroy_loop_vec_info.
15049 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15050 (_loop_vec_info::_loop_vec_info): ...this.
15051 (destroy_loop_vec_info): Replace with...
15052 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15053 the stmt_vec_infos. Leave handling of vec_info information to its
15054 destructor. Remove explicit vector releases.
15055 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15056 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15057 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15058 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15059 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15060 (destroy_bb_vec_info): Replace with...
15061 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15062 information to its destructor.
15063 (vect_slp_analyze_bb_1): Use new and delete instead of
15064 new_bb_vec_info and destroy_bb_vec_info.
15065 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15068 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15070 * tree-data-ref.h (subscript): Add access_fn field.
15071 (data_dependence_relation): Add could_be_independent_p.
15072 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15073 (same_access_functions): Move to tree-data-ref.c.
15074 * tree-data-ref.c (ref_contains_union_access_p): New function.
15075 (access_fn_component_p): Likewise.
15076 (access_fn_components_comparable_p): Likewise.
15077 (dr_analyze_indices): Add a reference to access_fn_component_p.
15078 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15080 (constant_access_functions): Likewise.
15081 (add_other_self_distances): Likewise.
15082 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15083 (initialize_data_dependence_relation): Use XCNEW and remove
15084 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15085 of access functions that have the same type. Allow the
15086 subsequence to end with different bases in some circumstances.
15087 Record the chosen access functions in SUB_ACCESS_FN.
15088 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15089 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15090 (subscript_dependence_tester_1): Likewise dra and drb.
15091 (build_classic_dist_vector): Update calls accordingly.
15092 (subscript_dependence_tester): Likewise.
15093 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15094 DDR_COULD_BE_INDEPENDENT_P.
15095 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15096 comp_alias_ddrs instead of may_alias_ddrs.
15097 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15099 (vect_analyze_data_ref_dependence): Use it if
15100 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15101 distance vectors if that fails.
15102 (dependence_distance_ge_vf): New function.
15103 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15104 LOOP_VINFO_MAY_ALIAS_DDRS.
15106 2017-08-04 Richard Biener <rguenther@suse.de>
15108 PR middle-end/81705
15109 * fold-const.c (fold_binary_loc): Properly restrict
15110 minus_var0 && minus_var1 case when associating undefined overflow
15113 2017-08-04 Tom de Vries <tom@codesourcery.com>
15115 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15117 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15119 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15120 Don't start diagnostic messages with a capital letter.
15121 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15123 (rs6000_invalid_builtin): Likewise.
15124 (rs6000_trampoline_init): Likewise.
15126 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15129 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15130 after setting changeable df flags.
15132 2017-08-03 Richard Biener <rguenther@suse.de>
15134 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15135 up if the use is in USE - X.
15137 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15139 * toplev.c (dumpfile.h): New include.
15140 (internal_error_reentered): New static function. Use it...
15141 (internal_error_function): ...here to handle reentered internal_error.
15143 2017-08-03 Richard Biener <rguenther@suse.de>
15145 PR middle-end/81148
15146 * fold-const.c (split_tree): Add minus_var and minus_con
15147 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15148 here but always use minus_*.
15149 (associate_trees): Assert we never associate with MINUS_EXPR
15150 and NULL first operand. Do not recurse for PLUS_EXPR operands
15151 when associating as MINUS_EXPR either.
15152 (fold_binary_loc): Track minus_var and minus_con.
15154 2017-08-03 Tom de Vries <tom@codesourcery.com>
15157 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15158 ACCEL_COMPILER, apply finish_options on
15159 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15161 2017-08-03 Tom de Vries <tom@codesourcery.com>
15164 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15165 function_entry_patch_area_size > 0.
15167 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15170 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15171 instead of 10??LU, perform unit multiplication in wide_int,
15172 don't change alloc_object_size_limit if the limit is larger
15175 PR tree-optimization/81655
15176 PR tree-optimization/81588
15177 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15178 the case when ranges[i].low and high are 1 for unsigned type with
15181 PR middle-end/81052
15182 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15183 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15185 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15187 * tree-vrp.h: Add include guard.
15189 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15192 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15193 (ud2): New insn pattern.
15194 * config/i386/i386.c (ix86_expand_epilogue):
15195 For naked functions, generate ud2 instead of trap insn.
15197 2017-08-02 Marek Polacek <polacek@redhat.com>
15200 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15202 2017-08-02 Tom de Vries <tom@codesourcery.com>
15203 Cesar Philippidis <cesar@codesourcery.com>
15205 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15206 Add missing edge probabilities.
15208 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15210 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15211 Correct endianness.
15213 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15215 PR middle-end/79499
15216 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15217 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15218 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15219 prologue_seq sequences - if any.
15221 2017-08-02 Richard Biener <rguenther@suse.de>
15223 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15224 via vectors if supported, integer extracts via punning if supported
15225 or otherwise vector extracts.
15227 2017-08-02 Richard Biener <rguenther@suse.de>
15229 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15231 (bitmap_insert_into_set): ... this.
15233 2017-08-02 Richard Biener <rguenther@suse.de>
15235 PR tree-optimization/81633
15237 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15239 PR tree-optimization/71752
15240 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15242 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15244 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15245 (machine_function::call_ms2sysv_pad_out): Remove field.
15246 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15247 (ix86_compute_frame_layout): Likewise.
15249 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15252 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15253 attribute with interrupt attribute.
15255 2017-08-01 Andrew Pinski <apinski@cavium.com>
15257 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15258 BIT_INSERT_EXPR's operand 1
15259 to see if the types precision matches.
15261 2017-08-01 Martin Liska <mliska@suse.cz>
15263 PR middle-end/70140
15264 * builtins.c (expand_builtin_memcpy_args): Remove.
15265 (expand_builtin_memcpy): Call newly added function
15266 expand_builtin_memory_copy_args.
15267 (expand_builtin_memcpy_with_bounds): Likewise.
15268 (expand_builtin_mempcpy): Remove last argument.
15269 (expand_builtin_mempcpy_with_bounds): Likewise.
15270 (expand_builtin_memory_copy_args): New function created from
15271 expand_builtin_mempcpy_args with small modifications.
15272 (expand_builtin_mempcpy_args): Remove.
15273 (expand_builtin_stpcpy): Remove unused argument.
15274 (expand_builtin): Likewise.
15275 (expand_builtin_with_bounds): Likewise.
15277 2017-08-01 Martin Liska <mliska@suse.cz>
15280 Make mempcpy more optimal (PR middle-end/70140).
15282 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15285 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15286 __builtin_vec_cmpne verify both arguments are compatible vectors
15287 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15288 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15289 move computation of aligned to after checking the argument types.
15293 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15294 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15297 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15298 Doug Rupp <rupp@adacore.com>
15299 Olivier Hainque <hainque@adacore.com>
15301 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15302 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15304 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15305 for TARGET_OS_CPP_BUILTIN.
15306 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15307 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15309 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15310 passing required abi options to the assembler for EABI configurations.
15311 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15312 of .text.hot and .text.unlikely sections for kernel modules when
15313 using ARM style exceptions.
15314 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15315 options. Add EXTRA_CC1_SPEC.
15316 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15318 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15320 (ARM_TARGET2_DWARF_FORMAT): Define.
15321 * config/arm/t-vxworks: Adjust multilib control to removal of the
15322 Diab command line options.
15324 2017-08-01 Martin Liska <mliska@suse.cz>
15326 PR gcov-profile/81561
15327 * gcov.c (unblock): Make unblocking safe as we need to preserve
15328 index correspondence of blocks and block_lists.
15330 2017-08-01 Richard Biener <rguenther@suse.de>
15332 PR tree-optimization/81181
15333 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15334 (compute_antic): ... end of iteration here.
15336 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15338 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15339 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15340 (ftree-slp-vectorize): Likewise.
15341 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15342 can no longer be set independent of flag_tree_loop_vectorize.
15343 * omp-general.c (emp_max_vf): Likewise.
15344 * opts.c (enable_fdo_optimizations): Remove references to
15345 flag_tree_vectorize, these are now implicit.
15346 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15347 and leave it for the options machinery.
15349 2017-08-01 Martin Liska <mliska@suse.cz>
15351 PR middle-end/70140
15352 * builtins.c (expand_builtin_memcpy_args): Remove.
15353 (expand_builtin_memcpy): Call newly added function
15354 expand_builtin_memory_copy_args.
15355 (expand_builtin_memcpy_with_bounds): Likewise.
15356 (expand_builtin_mempcpy): Remove last argument.
15357 (expand_builtin_mempcpy_with_bounds): Likewise.
15358 (expand_builtin_memory_copy_args): New function created from
15359 expand_builtin_mempcpy_args with small modifications.
15360 (expand_builtin_mempcpy_args): Remove.
15361 (expand_builtin_stpcpy): Remove unused argument.
15362 (expand_builtin): Likewise.
15363 (expand_builtin_with_bounds): Likewise.
15365 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15368 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15369 print "ds:" only for immediates in generic address space.
15371 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15374 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15375 (ix86_function_ok_for_sibcall): Return false for naked functions.
15377 2017-08-01 Richard Biener <rguenther@suse.de>
15379 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15380 (compute_antic): Seed worklist with exit block predecessors.
15381 * cfganal.c (dfs_find_deadend): For a cycle return the source
15382 of the edge closing it.
15384 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15386 * config/aarch64/aarch64.c
15387 (aarch64_can_const_movi_rtx_p): Move 0 check.
15389 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15391 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15392 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15394 * match.pd: Ditto in address comparison pattern.
15396 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15398 PR tree-optimization/81627
15399 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15400 closed ssa form for store-store chain.
15402 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15404 PR tree-optimization/81620
15405 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15406 for store-store chain.
15408 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15410 PR tree-optimization/81588
15411 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15412 ranges[i].in_p, invert comparison code ccode. For >/>=,
15413 swap rhs1 and rhs2 and comparison code unconditionally,
15414 for </<= don't do that. Don't swap rhs1/rhs2 again if
15415 ranges[i].in_p, instead invert comparison code ccode if
15416 opcode or oe->rank is BIT_IOR_EXPR.
15419 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15420 a direct optab to conversion optab.
15421 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15422 with GET_MODE_INNER as last argument instead of optab_handler.
15423 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15424 vector extraction if possible and optab is available.
15425 * expr.c (store_constructor): Use convert_optab_handler instead
15426 of optab_handler. Use vector initialization from smaller
15427 vectors if possible and optab is available.
15428 * tree-vect-stmts.c (vectorizable_load): Likewise.
15429 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15430 now have two modes.
15431 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15432 of vec_init from half-sized vectors with the same element mode.
15433 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15434 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15435 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15436 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15437 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15438 after mode in gen_vec_extract* calls.
15439 (vec_extract<mode>): Renamed to ...
15440 (vec_extract<mode><ssescalarmodelower>): ... this.
15441 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15442 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15443 element mode after mode in gen_vec_init* calls.
15444 (VEC_INIT_HALF_MODE): New mode iterator.
15445 (vec_init<mode>): Renamed to ...
15446 (vec_init<mode><ssescalarmodelower>): ... this.
15447 (vec_init<mode><ssehalfvecmodelower>): New expander.
15448 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15449 (vec_extractv2sfsf): ... this.
15450 (vec_initv2sf): Renamed to ...
15451 (vec_initv2sfsf): ... this.
15452 (vec_extractv2si): Renamed to ...
15453 (vec_extractv2sisi): ... this.
15454 (vec_initv2si): Renamed to ...
15455 (vec_initv2sisi): ... this.
15456 (vec_extractv4hi): Renamed to ...
15457 (vec_extractv4hihi): ... this.
15458 (vec_initv4hi): Renamed to ...
15459 (vec_initv4hihi): ... this.
15460 (vec_extractv8qi): Renamed to ...
15461 (vec_extractv8qiqi): ... this.
15462 (vec_initv8qi): Renamed to ...
15463 (vec_initv8qiqi): ... this.
15464 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15465 (vec_init<mode>): Renamed to ...
15466 (vec_init<mode><VEC_base_l>): ... this.
15467 (vec_extract<mode>): Renamed to ...
15468 (vec_extract<mode><VEC_base_l>): ... this.
15469 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15470 (vec_initv2sfsf): ... this.
15471 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15472 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15473 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15474 element mode after mode in gen_vec_init* calls.
15475 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15476 (vec_init<mode><Vel>): ... this.
15477 (vec_extract<mode>): Renamed to ...
15478 (vec_extract<mode><Vel>): ... this.
15479 * config/aarch64/iterators.md (Vel): New mode attribute.
15480 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15481 Add element mode after mode in gen_vec_extract* calls.
15482 * config/s390/vector.md (non_vec_l): New mode attribute.
15483 (vec_extract<mode>): Renamed to ...
15484 (vec_extract<mode><non_vec_l>): ... this.
15485 (vec_init<mode>): Renamed to ...
15486 (vec_init<mode><non_vec_l>): ... this.
15487 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15488 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15490 * config/arm/iterators.md (V_elem_l): New mode attribute.
15491 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15492 (vec_extract<mode><V_elem_l>): ... this.
15493 (vec_extractv2di): Renamed to ...
15494 (vec_extractv2didi): ... this.
15495 (vec_init<mode>): Renamed to ...
15496 (vec_init<mode><V_elem_l>): ... this.
15497 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15498 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15499 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15500 Add element mode after gen_vec_extract* calls.
15501 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15502 (vec_init<mode><unitmode>): ... this.
15503 (vec_extract<mode>): Renamed to ...
15504 (vec_extract<mode><unitmode>): ... this.
15505 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15506 (vec_init<mode><unitmode>): ... this.
15507 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15508 (vec_initv2sfsf): ... this.
15509 (vec_extractv2sf): Renamed to ...
15510 (vec_extractv2sfsf): ... this.
15511 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15512 Add element mode after gen_vec_extract* calls.
15513 * config/mips/mips.md (unitmode): New mode iterator.
15514 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15515 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15516 * config/spu/spu.md (inner_l): New mode attribute.
15517 (vec_init<mode>): Renamed to ...
15518 (vec_init<mode><inner_l>): ... this.
15519 (vec_extract<mode>): Renamed to ...
15520 (vec_extract<mode><inner_l>): ... this.
15521 * config/sparc/sparc.md (veltmode): New mode iterator.
15522 (vec_init<VMALL:mode>): Renamed to ...
15523 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15524 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15525 (vec_initv2sisi): ... this.
15526 (vec_initv2sf): Renamed to ...
15527 (vec_initv2sfsf): ... this.
15528 (vec_extractv2sf): Renamed to ...
15529 (vec_extractv2sfsf): ... this.
15530 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15531 (vec_init<mode>): Renamed to ...
15532 (vec_init<mode><VEC_base_l>): ... this.
15533 (vec_extract<mode>): Renamed to ...
15534 (vec_extract<mode><VEC_base_l>): ... this.
15535 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15536 (vec_initv2sfsf): ... this.
15537 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15538 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15539 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15540 gen_vec_init* calls.
15542 2017-08-01 Richard Biener <rguenther@suse.de>
15544 PR tree-optimization/81297
15545 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15546 TREE_OVERFLOW from INTEGER_CSTs.
15548 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15550 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15552 2017-07-31 Carl Love <cel@us.ibm.com>
15554 * config/rs6000/rs6000-c: Add support for built-in functions
15555 vector signed char vec_xl_be (signed long long, signed char *);
15556 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15557 vector signed int vec_xl_be (signed long long, signed int *);
15558 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15559 vector signed long long vec_xl_be (signed long long, signed long long *);
15560 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15561 vector signed short vec_xl_be (signed long long, signed short *);
15562 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15563 vector double vec_xl_be (signed long long, double *);
15564 vector float vec_xl_be (signed long long, float *);
15565 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15566 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15567 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15569 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15570 (altivec_expand_builtin): Add switch statement to call
15571 altivec_expand_xl_be for each builtin.
15572 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15573 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15574 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15575 __builtin_vsx_le_be_v16qi.
15576 * doc/extend.texi: Update the built-in documentation file for the
15577 new built-in functions.
15579 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15582 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15584 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15586 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15588 * config.gcc: Add z14.
15589 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15590 CPU model numbers for z13s and z14.
15591 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15593 * config/s390/s390-opts.h (enum processor_type): Rename
15594 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15595 * config/s390/s390.c (processor_table): Add field for CPU name to
15596 be passed to Binutils.
15597 (s390_asm_output_machine_for_arch): Use the new field in
15598 processor_table for Binutils.
15599 (s390_expand_builtin): Replace arch12 with z14.
15600 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15601 (s390_get_sched_attrmask): Likewise.
15602 (s390_get_unit_mask): Likewise.
15603 * config/s390/s390.opt: Add z14 to processor_type enum.
15605 2017-07-31 Martin Jambor <mjambor@suse.cz>
15608 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15609 regardless of optimization level.
15611 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15612 Martin Liska <mliska@suse.cz>
15614 * predict.def: Remove old comment and adjust probability.
15615 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15616 PREDICT statements.
15618 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15621 * config/i386/i386.c (ix86_function_naked): New function.
15622 (ix86_can_use_return_insn_p): Return false for naked functions.
15623 (ix86_expand_prologue): Skip prologue for naked functions.
15624 (ix86_expand_epilogue): Skip epilogue for naked functions
15625 and emit trap instruction.
15626 (ix86_warn_func_return): New function.
15627 (ix86_attribute_table): Add "naked" attribute specification.
15628 (TARGET_WARN_FUNC_RETURN): Define.
15629 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15631 2017-07-31 Martin Liska <mliska@suse.cz>
15633 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15634 (dump_gimple_bb_header): Always dump BB info.
15635 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15637 2017-07-31 Martin Liska <mliska@suse.cz>
15640 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15641 also with current_function_decl non-null equality.
15643 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15646 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15647 change type to the element type, instead add eltype variable and
15648 use it where we are interested in the element type.
15650 PR tree-optimization/81603
15651 * ipa-polymorphic-call.c
15652 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15653 offset arithmetic in offset_int, bail out if the resulting bit offset
15654 doesn't fit into shwi.
15656 2017-07-31 Martin Liska <mliska@suse.cz>
15658 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15659 (gimplify_save_expr): Fix comment.
15661 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15664 * config/i386/i386.c (ix86_function_arg): Update arguments for
15666 (ix86_compute_frame_layout): Set the initial stack offset to
15667 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15668 INCOMING_FRAME_SP_OFFSET.
15669 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15670 stack before exception handler returns.
15671 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15672 the 'ERROR_CODE' for exception handler.
15674 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15676 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15677 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15678 (ASM_OUTPUT_REG_POP): Ditto.
15679 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15680 instead of asm_fprintf to output pure string.
15682 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15684 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15685 to imported_module_or_decl hook.
15686 (debug_nothing_tree_tree_tree_bool): Remove.
15687 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15688 * debug.c (do_nothing_debug_hooks): Use
15689 debug_nothing_tree_tree_tree_bool_bool instead of
15690 debug_nothing_tree_tree_tree_bool.
15691 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15692 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15693 * sdbout.c (sdb_debug_hooks): Likewise.
15694 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15695 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15697 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15698 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15699 attribute, don't add anything.
15701 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15703 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15704 (fold_build2_stat_loc): Likewise.
15705 (fold_build3_stat_loc): Likewise.
15706 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15707 (fold_build1_loc): Remove macro.
15708 (fold_build2_loc): Likewise.
15709 (fold_build3_loc): Likewise.
15711 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15713 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15714 (gimple_build_debug_bind_source_stat): Likewise.
15715 * gimple.h (gimple_build_debug_bind): Remove macro.
15716 (gimple_build_debug_bind_source): Likewise.
15718 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15720 * bitmap.c (bitmap_alloc): Adjust.
15721 (bitmap_gc_alloc): Likewise.
15722 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15724 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15726 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15727 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15728 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15729 (bitmap_gc_alloc_stat): Likewise.
15730 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15732 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15734 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15735 * rtl.h (shallow_copy_rtx): Remove macro.
15737 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15739 * emit-rtl.c (gen_raw_REG): Adjust.
15740 * gengenrtl.c (gendef): Likewise.
15741 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15742 * rtl.h (rtx_alloc): Remove macro.
15744 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15746 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15747 (build_tree_list_stat): Likewise.
15748 * tree.h (build_tree_list): Remove macro.
15749 (build_tree_list_vec): Likewise.
15751 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15753 * tree.c (make_vector_stat): Remove _stat from name.
15754 (build_vector_stat): Likewise.
15755 * tree.h (make_vector_stat): Remove macro.
15756 (build_vector_stat): Likewise.
15758 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15760 * tree.h (build_var_debug_value): Remove prototype.
15762 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15764 * tree.c (tree_cons_stat): Remove _stat from name.
15765 * tree.h (tree_cons): Remove macro.
15767 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15769 * tree.c (build_vl_exp_stat): Remove _stat from name.
15770 * tree.h (build_vl_exp): Remove macro.
15772 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15774 * tree.c (build_decl_stat): Remove _stat from name.
15775 * tree.h (build_decl): Remove macro.
15777 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15779 * gimple.c (gimple_build_with_ops_stat): Adjust.
15780 (gimple_alloc_stat): Remove _stat from name.
15781 * gimple.h (gimple_alloc): Remove macro.
15783 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15785 * tree.c (make_tree_vec_stat): Remove _stat from name.
15786 (grow_tree_vec_stat): Likewise.
15787 * tree.h (make_tree_vec_stat): Adjust prototype.
15788 (grow_tree_vec_stat): Likewise.
15789 (make_tree_vec): Remove macro.
15790 (grow_tree_vec): Likewise.
15792 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15794 * fold-const.c (fold_build1_stat_loc): Adjust.
15795 (fold_build2_stat_loc): Likewise.
15796 (fold_build3_stat_loc): Likewise.
15797 * tree.c (build0_stat): Remove _stat from name.
15798 (build1_stat): Likewise.
15799 (build2_stat): Likewise.
15800 (build3_stat): Likewise.
15801 (build4_stat): Likewise.
15802 (build5_stat): Likewise.
15803 * tree.h (build1_loc): Remove macro, and rename _stat function
15805 (build2_loc): Likewise.
15806 (build3_loc): Likewise.
15807 (build4_loc): Likewise.
15808 (build5_loc): Likewise.
15810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15812 * tree.c (make_int_cst_stat): Remove _stat from name.
15813 * tree.h (make_int_cst_stat): Adjust prototype.
15814 (make_int_cst): Remove macro.
15816 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15818 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15819 * tree.h (make_tree_binfo_stat): Adjust prototype.
15820 (make_tree_binfo): Remove.
15822 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15824 * tree.c (copy_node_stat): Rename to copy_node.
15825 (build_distinct_type_copy): Adjust.
15826 * tree.h (copy_node_stat): Adjust prototype.
15827 (copy_node): Remove macro.
15829 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15831 * tree.c (make_node_stat): rename to make_node.
15832 (build_tree_list_stat): Adjust.
15833 (build0_stat): Likewise.
15834 (build2_stat): Likewise.
15835 (build3_stat): Likewise.
15836 (build4_stat): Likewise.
15837 (build5_stat): Likewise.
15838 (build_decl_stat): Likewise.
15839 * tree.h (make_node_stat): Adjust prototype.
15840 (make_node): remove macro.
15842 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15844 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15845 (PPC_FEATURE2_SCV): Likewise.
15846 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15848 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15850 * config/aarch64/aarch64.c
15851 (aarch64_internal_mov_immediate): Add new special pattern.
15852 * config/aarch64/aarch64.md (*movdi_aarch64):
15853 Add reg/32bit const mov case.
15855 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15856 Richard Sandiford <richard.sandiford@linaro.org>
15858 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15859 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15860 Add integer and movi cases.
15861 (movi-split-hf-df-sf split, fp16): New.
15862 (enabled): Added TARGET_FP_F16INST.
15863 * config/aarch64/iterators.md (GPF_HF): New.
15864 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15866 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15868 * config/aarch64/aarch64.c
15869 (aarch64_simd_container_mode): Add prototype.
15870 (aarch64_expand_mov_immediate): Add HI support.
15871 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15872 (aarch64_can_const_movi_rtx_p): New.
15873 (aarch64_preferred_reload_class):
15874 Remove restrictions of using FP registers for certain SIMD operations.
15875 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15876 (aarch64_valid_floating_const): Add integer move validation.
15877 (aarch64_simd_imm_scalar_p): Remove.
15878 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15879 (aarch64_legitimate_constant_p): Expand list of supported cases.
15880 * config/aarch64/aarch64-protos.h
15881 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15882 (aarch64_reinterpret_float_as_int): New.
15883 (aarch64_simd_imm_scalar_p): Remove.
15884 * config/aarch64/constraints.md (Uvi): New.
15885 (Dd): Split into Ds and new Dd.
15886 * config/aarch64/aarch64.md (*movsi_aarch64):
15888 (*movdi_aarch64): Add SIMD mov case.
15890 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15892 * tree-predcom.c: (struct chain): Handle store-store chain in which
15893 stores for elimination only store loop invariant values.
15894 (execute_pred_commoning_chain): Ditto.
15895 (prepare_initializers_chain_store_elim): Ditto.
15896 (prepare_finalizers): Ditto.
15897 (is_inv_store_elimination_chain): New function.
15898 (initialize_root_vars_store_elim_1): New function.
15900 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15902 * tree-predcom.c: Revise general description of the pass.
15903 (enum chain_type): New enum type for store elimination.
15904 (struct chain): New field supporting store elimination.
15905 (struct component): Ditto.
15906 (dump_chain): Dump store-stores chain.
15907 (release_chain): Release resources.
15908 (split_data_refs_to_components): Compute and create component
15909 contains only stores for elimination.
15910 (get_chain_last_ref_at): New function.
15911 (make_invariant_chain): Initialization.
15912 (make_rooted_chain): Specify chain type in parameter and record it.
15913 (add_looparound_copies): Skip for store-stores chain.
15914 (determine_roots_comp): Compute type of chain and pass it to
15916 (initialize_root_vars_store_elim_2): New function.
15917 (finalize_eliminated_stores): New function.
15918 (remove_stmt): Handle store for elimination.
15919 (execute_pred_commoning_chain): Execute predictive commoning on
15920 store-store chains.
15921 (determine_unroll_factor): Skip unroll for store-stores chain.
15922 (prepare_initializers_chain_store_elim): New function.
15923 (prepare_initializers_chain): Hanlde store-store chain.
15924 (prepare_finalizers_chain, prepare_finalizers): New function.
15925 (tree_predictive_commoning_loop): Return integer value indicating
15926 if loop is unrolled or lcssa form is corrupted.
15927 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15929 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15931 * tree-predcom.c (initialize_root): Delete.
15932 (execute_pred_commoning_chain): Initialize root vars and replace
15933 reference of non-combined chain directly, rather than call above
15936 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15938 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15939 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15941 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15943 * tree-predcom.c (struct chain): New field init_seq.
15944 (release_chain): Release init_seq.
15945 (prepare_initializers_chain): Record intialization stmts in above
15947 (insert_init_seqs): New function.
15948 (tree_predictive_commoning_loop): Call insert_init_seqs.
15950 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15952 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15954 2017-07-28 Richard Biener <rguenther@suse.de>
15956 * match.pd: Remove superfluous :c.
15957 * genmatch.c (simplify::id): Add member.
15958 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15960 (current_id): New global.
15961 (dt_node::parent): Move from ...
15962 (dt_operand::parent): ... here. Add for_id member.
15963 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15964 (decision_tree::find_node): Relax order requirement when
15965 merging DT_TRUE nodes to ones inbetween the current simplify
15966 and the one we try to merge with. Add diagnostic whenever
15967 we need to enforce pattern order by not merging.
15968 (decision_tree::insert): Set current_id.
15969 (decision_tree::print_node): Dump parent node and for_id.
15970 (parser::last_id): Add member.
15971 (parser::push_simplify): Assign unique id.
15972 (parser::parser): Initialize last_id.
15974 2017-07-28 Martin Liska <mliska@suse.cz>
15977 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15978 gimple_build_debug_bind.
15980 2017-07-28 Richard Biener <rguenther@suse.de>
15982 PR tree-optimization/81502
15983 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15985 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15987 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15988 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15989 for BIT_FIELD_REF args.
15990 * fold-const.c (make_bit_field_ref): Likewise.
15991 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15993 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15996 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15997 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15998 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15999 Or it into SANITIZER_UNDEFINED.
16000 * ubsan.c: Include gimple-fold.h and varasm.h.
16001 (ubsan_expand_ptr_ifn): New function.
16002 (instrument_pointer_overflow): New function.
16003 (maybe_instrument_pointer_overflow): New function.
16004 (instrument_object_size): Formatting fix.
16005 (pass_ubsan::execute): Call instrument_pointer_overflow
16006 and maybe_instrument_pointer_overflow.
16007 * internal-fn.c (expand_UBSAN_PTR): New function.
16008 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16009 * sanitizer.def (__ubsan_handle_pointer_overflow,
16010 __ubsan_handle_pointer_overflow_abort): New builtins.
16011 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16012 * internal-fn.def (UBSAN_PTR): New internal function.
16013 * opts.c (sanitizer_opts): Add pointer-overflow.
16014 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16015 * fold-const.c (build_range_check): Compute pointer range check in
16016 integral type if pointer arithmetics would be needed. Formatting
16019 2017-07-28 Martin Liska <mliska@suse.cz>
16022 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16023 parameters that are of a variable-length.
16025 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16027 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16029 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16030 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16031 (CRT_CALL_STATIC_FUNCTION): Likewise.
16032 (ASM_DEFAULT_SPEC): New define.
16033 (ASM_SPEC32): Likewise.
16034 (ASM_SPEC64): Likewise.
16035 (ASM_SPEC_COMMON): Likewise.
16036 (ASM_SPEC): Likewise.
16037 (INVALID_64BIT): Likewise.
16038 (LINK_OS_DEFAULT_SPEC): Likewise.
16039 (LINK_OS_SPEC32): Likewise.
16040 (LINK_OS_SPEC64): Likewise.
16041 (POWERPC_LINUX): Likewise.
16042 (PTRDIFF_TYPE): Likewise.
16043 (RESTORE_FP_PREFIX): Likewise.
16044 (RESTORE_FP_SUFFIX): Likewise.
16045 (SAVE_FP_PREFIX): Likewise.
16046 (SAVE_FP_SUFFIX): Likewise.
16047 (SIZE_TYPE): Likewise.
16048 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16049 (TARGET_64BIT): Likewise.
16050 (TARGET_64BIT): Likewise.
16051 (TARGET_AIX): Likewise.
16052 (WCHAR_TYPE_SIZE): Likewise.
16053 (WCHAR_TYPE): Undefine.
16054 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16055 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16056 (CPP_OS_RTEMS_SPEC): Delete.
16057 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16058 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16060 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16062 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16064 PR tree-optimization/81578
16065 * tree-parloops.c (build_new_reduction): Bail out if
16066 reduction_code isn't one of the standard OpenMP reductions.
16067 Move the details printing after that decision.
16069 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16071 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16072 related to reload_in_progress.
16073 (splat_input_operand): Likewise.
16074 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16076 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16078 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16079 (TARGET_INSTANTIATE_DECLS): Likewise.
16080 (legitimate_indexed_address_p): Delete reload_in_progress code.
16081 (rs6000_debug_legitimate_address_p): Likewise.
16082 (rs6000_eliminate_indexed_memrefs): Likewise.
16083 (rs6000_emit_le_vsx_store): Likewise.
16084 (rs6000_emit_move_si_sf_subreg): Likewise.
16085 (rs6000_emit_move): Likewise.
16086 (register_to_reg_type): Likewise.
16087 (rs6000_pre_atomic_barrier): Likewise.
16088 (rs6000_machopic_legitimize_pic_address): Likewise.
16089 (rs6000_allocate_stack_temp): Likewise.
16090 (rs6000_address_for_fpconvert): Likewise.
16091 (rs6000_address_for_altivec): Likewise.
16092 (rs6000_secondary_memory_needed_rtx): Delete function.
16093 (rs6000_check_sdmode): Likewise.
16094 (rs6000_alloc_sdmode_stack_slot): Likewise.
16095 (rs6000_instantiate_decls): Likewise.
16096 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16097 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16098 Delete reload_in_progress.
16099 (*vec_reload_and_plus_<mptrsize>): Likewise.
16100 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16101 (vsx_div_v2di): Likewise.
16102 (vsx_udiv_v2di): Likewise.
16104 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16106 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16107 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16108 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16109 (rs6000_debug_reg_global): Delete print of LRA status.
16110 (rs6000_option_override_internal): Delete dead LRA related code.
16111 (rs6000_lra_p): Delete function.
16112 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16114 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16116 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16117 * config/riscv/rtems.h: New file.
16119 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16120 Sudakshina Das <sudi.das@arm.com>
16122 * config/aarch64/aarch64.md
16123 (define_split for and<mode>3nr_compare): Move
16124 non aarch64_logical_operand to a register.
16125 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16126 register immediate operand to a register.
16127 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16129 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16131 PR middle-end/81564
16132 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16134 2017-07-27 Richard Biener <rguenther@suse.de>
16136 PR tree-optimization/81573
16137 PR tree-optimization/81494
16138 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16139 multi defuse cycle case.
16141 2017-07-27 Richard Biener <rguenther@suse.de>
16143 PR tree-optimization/81571
16144 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16147 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16149 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16150 earlier and only if MASK_FPU is set. Adjust formatting.
16152 2017-07-27 Martin Liska <mliska@suse.cz>
16154 * opt-functions.awk: Add validation of value of Init.
16155 * optc-gen.awk: Pass new argument.
16157 2017-07-27 Martin Liska <mliska@suse.cz>
16159 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16160 Fix wrong condition.
16162 2017-07-27 Martin Liska <mliska@suse.cz>
16164 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16165 BBs and edges seen by autoFDO.
16167 2017-07-27 Richard Biener <rguenther@suse.de>
16169 PR tree-optimization/81502
16170 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16171 with incompatible but same sized type.
16172 (execute_update_addresses_taken): Likewise.
16174 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16176 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16177 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16179 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16182 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16183 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16184 Change s_operand to memory_operand.
16186 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16188 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16189 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16190 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16191 Emit instructions rather than returning an expression. Handle TFmode
16192 and KFmode by casting to TImode.
16193 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16194 (rs6000_emit_le_vsx_store): Likewise.
16195 * config/rs6000/vsx.md (VSX_TI): New iterator.
16196 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16197 (*vsx_le_undo_permute_<mode>): Likewise.
16198 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16199 emit the split sequence.
16200 (*vsx_le_perm_store_<mode>): Likewise.
16202 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16204 PR tree-optimization/81555
16205 PR tree-optimization/81556
16206 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16207 if true, force CHANGED for the recursive invocation.
16208 (reassociate_bb): Remember original length of ops array, pass
16209 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16211 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16212 attributes for noipa attribute. For naked attribute use
16213 lookup_attribute first before lookup_attribute_spec.
16214 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16216 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16217 for functions with noipa attribute.
16218 (cgraph_externally_visible_p): Return true for functions with noipa
16220 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16221 for functions with noipa attribute.
16222 * doc/extend.texi: Document noipa function attribute.
16223 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16224 also for functions with noipa attribute.
16225 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16227 2017-07-26 Andrew Pinski <apinski@cavium.com>
16229 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16230 vec_unalign_load_cost and vec_unalign_store_cost.
16232 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16234 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16235 -mvsx-small-integer option.
16236 (ISA_3_0_MASKS_IEEE): Likewise.
16237 (OTHER_VSX_VECTOR_MASKS): Likewise.
16238 (POWERPC_MASKS): Likewise.
16239 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16240 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16241 code, only testing for DImode being allowed in non-VSX floating
16243 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16244 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16246 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16247 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16248 TARGET_P8_VECTOR test.
16249 (rs6000_secondary_reload_simple_move): Likewise.
16250 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16251 since TARGET_P9_VECTOR was already tested.
16252 (rs6000_opt_masks): Remove -mvsx-small-integer.
16253 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16254 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16256 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16257 test for TARGET_VEXTRACTUB was used, and that uses
16259 (p9 extract splitter): Likewise.
16260 (vsx_extract_<mode>_di_p9): Likewise.
16261 (vsx_extract_<mode>_store_p9): Likewise.
16262 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16263 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16264 the elimination of TARGET_VSX_SMALL_INTEGER.
16265 (vsx_extract_<mode>_p8): Likewise.
16266 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16267 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16268 (vsx_set_<mode>_p9): Likewise.
16269 (vsx_set_v4sf_p9): Likewise.
16270 (vsx_set_v4sf_p9_zero): Likewise.
16271 (vsx_insert_extract_v4sf_p9): Likewise.
16272 (vsx_insert_extract_v4sf_p9_2): Likewise.
16273 * config/rs6000/rs6000.md (sign extend splitter): Change
16274 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16275 (floatsi<mode>2_lfiwax_mem): Likewise.
16276 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16277 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16278 since a test for TARGET_P9_VECTOR was used.
16279 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16280 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16281 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16282 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16283 TARGET_P8_VECTOR test.
16284 (fix_trunc<mode>si2_stfiwx): Likewise.
16285 (fix_trunc<mode>si2_internal): Likewise.
16286 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16287 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16289 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16290 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16291 TARGET_P8_VECTOR test.
16292 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16293 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16294 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16296 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16297 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16298 since a test for TARGET_P9_VECTOR was used.
16299 (splitter for loading small constants): Likewise.
16301 2017-07-26 Andrew Pinski <apinski@cavium.com>
16303 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16306 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16309 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16310 (fp_valid_at): Likewise.
16312 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16314 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16315 (qdf24xx_addrcost_table): Likewise.
16316 (cortexa57_tunings): Update to use generic_branch_cost.
16317 (cortexa72_tunings): Likewise.
16318 (cortexa73_tunings): Likewise.
16319 (qdf24xx_tunings): Likewise.
16321 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16323 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16324 (thunderx2t99_branch_cost): Likewise.
16325 (cortexa35_tunings): Update to use generic_branch_cost.
16326 (cortexa53_tunings): Likewise.
16327 (cortexa57_tunings): Likewise.
16328 (cortexa72_tunings): Likewise.
16329 (cortexa73_tunings): Likewise.
16330 (thunderx2t99_tunings): Likewise.
16332 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16334 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16335 (sparc_option_override): Honour MASK_FSMULD.
16336 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16337 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16338 * config/sparc/sparc.opt (mfsmuld): New option.
16339 * doc/invoke.texi (mfsmuld): Document option.
16341 2017-07-26 Marek Polacek <polacek@redhat.com>
16343 PR middle-end/70992
16344 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16346 2017-07-26 Richard Biener <rguenther@suse.de>
16348 * gimple-match-head.c (do_valueize): Return OP if valueize
16350 (get_def): New helper to get at the def stmt of a SSA name
16351 if valueize allows.
16352 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16353 do_valueize to get at the def stmt.
16354 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16356 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16358 PR middle-end/46932
16359 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16361 2017-07-26 Martin Liska <mliska@suse.cz>
16364 * function.c (expand_function_start): Make expansion of
16365 nonlocal_goto_save_area after parm_birth_insn.
16367 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16369 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16370 from all CPU target flags enable members.
16372 2017-07-26 Richard Biener <rguenther@suse.de>
16374 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16375 (decision_tree::gen): Make 'type' const.
16376 (write_predicate): Likewise.
16378 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16380 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16381 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16382 (rs6000_option_override_internal): Likewise.
16383 (rs6000_expand_vector_set): Likewise.
16384 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16385 (TARGET_UPPER_REGS_SF): Likewise.
16386 (TARGET_UPPER_REGS_DI): Likewise.
16387 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16388 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16389 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16390 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16391 (Splitters for DI constants in Altivec registers): Likewise.
16392 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16393 (vsx_set_v4sf_p9): Likewise.
16394 (vsx_set_v4sf_p9_zero): Likewise.
16395 (vsx_insert_extract_v4sf_p9): Likewise.
16396 (vsx_insert_extract_v4sf_p9_2): Likewise.
16398 2017-07-25 Carl Love <cel@us.ibm.com>
16400 * doc/extend.texi: Update the built-in documentation file for the
16401 existing built-in functions
16402 vector signed char vec_cnttz (vector signed char);
16403 vector unsigned char vec_cnttz (vector unsigned char);
16404 vector signed short vec_cnttz (vector signed short);
16405 vector unsigned short vec_cnttz (vector unsigned short);
16406 vector signed int vec_cnttz (vector signed int);
16407 vector unsigned int vec_cnttz (vector unsigned int);
16408 vector signed long long vec_cnttz (vector signed long long);
16409 vector unsigned long long vec_cnttz (vector unsigned long long);
16411 2017-07-25 Andrew Pinski <apinski@cavium.com>
16413 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16414 accesses where the use is for the first operand of a BIT_INSERT.
16416 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16419 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16420 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16422 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16424 * config/i386/gstabs.h: Delete.
16425 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16427 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16429 * config/i386/i386.c (ix86_decompose_address): Do not check for
16430 register RTX when looking at index_reg or base_reg.
16431 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16433 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16435 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16436 to update EH info here.
16438 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16440 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16442 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16444 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16446 2017-07-25 Torsten Duwe <duwe@suse.de>
16448 * common.opt: Introduce -fpatchable-function-entry
16449 command line option, and its variables function_entry_patch_area_size
16450 and function_entry_patch_area_start.
16451 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16452 including a two-value parser.
16453 * target.def (print_patchable_function_entry): New target hook.
16454 * targhooks.h (default_print_patchable_function_entry): New function.
16455 * targhooks.c (default_print_patchable_function_entry): Likewise.
16456 * toplev.c (process_options): Switch off IPA-RA if
16457 patchable function entries are being generated.
16458 * varasm.c (assemble_start_function): Look at the
16459 patchable-function-entry command line switch and current
16460 function attributes and maybe generate NOP instructions by
16461 calling the print_patchable_function_entry hook.
16462 * doc/extend.texi: Document patchable_function_entry attribute.
16463 * doc/invoke.texi: Document -fpatchable_function_entry
16464 command line option.
16465 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16467 * doc/tm.texi: Re-generate.
16469 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16472 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16473 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16475 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16477 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16479 2017-07-25 Richard Biener <rguenther@suse.de>
16481 PR tree-optimization/81455
16482 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16483 not walk in cycles when looking for guards.
16485 2017-07-25 Richard Biener <rguenther@suse.de>
16487 PR tree-optimization/81529
16488 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16489 when optimizing backedge uses.
16491 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16493 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16495 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16496 to dl_section_ref. On AIX, append an expression to subtract
16497 the size of the section length to dl_section_ref.
16499 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16501 * configure.ac: If any of the config.* scripts fail, exit 1.
16502 * configure: Regenerate.
16504 2017-07-25 Richard Biener <rguenther@suse.de>
16506 PR middle-end/81546
16507 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16508 of immediate uses, be more verbose on errors.
16510 2017-07-25 Richard Biener <rguenther@suse.de>
16512 PR tree-optimization/81510
16513 * tree-vect-loop.c (vect_is_simple_reduction): When the
16514 reduction stmt is not inside the loop bail out.
16516 2017-07-25 Richard Biener <rguenther@suse.de>
16518 PR tree-optimization/81303
16519 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16520 profitability check against LOOP_VINFO_NITERSM1.
16522 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16524 * domwalk.c (cmp_bb_postorder): Simplify.
16525 (sort_bbs_postorder): New function. Use it...
16526 (dom_walker::walk): ...here to optimize common cases.
16528 2017-07-25 Martin Liska <mliska@suse.cz>
16531 * ipa-visibility.c (function_and_variable_visibility): Make the
16532 redirection just on target that supports aliasing.
16533 Fix GNU coding style.
16535 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16538 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16540 * config/arm/rtems.h: Likewise.
16541 * config/bfin/rtems.h: Likewise.
16542 * config/i386/rtemself.h: Likewise.
16543 * config/lm32/rtems.h: Likewise.
16544 * config/m32c/rtems.h: Likewise.
16545 * config/m68k/rtemself.h: Likewise.
16546 * config/microblaze/rtems.h: Likewise.
16547 * config/mips/rtems.h: Likewise.
16548 * config/moxie/rtems.h: Likewise.
16549 * config/nios2/rtems.h: Likewise.
16550 * config/powerpcspe/rtems.h: Likewise.
16551 * config/rs6000/rtems.h: Likewise.
16552 * config/rtems.h: Likewise.
16553 * config/sh/rtems.h: Likewise.
16554 * config/sh/rtemself.h: Likewise.
16555 * config/sparc/rtemself.h: Likewise.
16557 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16560 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16561 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16562 * tree-ssa-structalias.c (alias_get_name): Same.
16564 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16567 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16568 instructions if no du chain is found.
16570 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16572 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16574 2017-07-25 Richard Biener <rguenther@suse.de>
16576 PR middle-end/81505
16577 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16580 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16582 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16583 upper-regs options.
16584 (ISA_2_7_MASKS_SERVER): Likewise.
16585 (ISA_3_0_MASKS_IEEE): Likewise.
16586 (OTHER_P8_VECTOR_MASKS): Likewise.
16587 (OTHER_VSX_VECTOR_MASKS): Likewise.
16588 (POWERPC_MASKS): Likewise.
16589 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16590 duplicate list of options.
16591 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16592 explicit -mupper-regs options.
16593 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16594 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16595 alias for -mupper-regs-df.
16596 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16597 (rs6000_init_hard_regno_mode_ok): Likewise.
16598 (rs6000_option_override_internal): Likewise.
16599 (rs6000_opt_masks): Likewise.
16600 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16601 options in terms of whether -mvsx or -mpower8-vector was used.
16602 (TARGET_UPPER_REGS_DI): Likewise.
16603 (TARGET_UPPER_REGS_SF): Likewise.
16604 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16605 -mupper-regs-* options.
16607 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16609 * passes.c (emergency_dump_function): Print some empty lines and a
16610 header before the RTL dump.
16612 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16614 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16616 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16619 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16620 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16622 2017-07-24 Carl Love <cel@us.ibm.com>
16624 * config/rs6000/rs6000-c.c: Add support for built-in functions
16625 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16626 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16627 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16628 vec_extract_fp_from_shortl): Add defines for the two builtins.
16629 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16630 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16632 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16633 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16634 * doc/extend.texi: Update the built-in documentation file for the
16635 new built-in function.
16637 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16640 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16642 * doc/generic.texi: Likewise.
16643 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16644 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16646 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16648 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16649 (aarch64_mls_elt_merge<mode>): Likewise.
16651 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16653 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16654 having __cxa_atexit.
16656 2017-07-23 Michael Collison <michael.collison@arm.com>
16658 * config/arm/arm.c (arm_option_override): Deprecate
16659 use of -mstructure-size-boundary.
16660 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16661 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16663 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16666 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16667 Reduce cost estimate for direct moves.
16669 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16672 * config/i386/i386.c (ix86_option_override_internal): Disable
16673 BMI, BMI2 and TBM instructions for -m16.
16675 2017-07-21 Carl Love <cel@us.ibm.com>
16677 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16678 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16679 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16680 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16681 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16682 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16683 VMULOSW): New enum "unspec" values.
16684 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16685 altivec_vmulosw): New patterns.
16686 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16687 VMULOSW): Add definitions.
16689 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16691 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16692 (qdf24xx): Likewise.
16693 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16694 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16695 (AARCH64_FL_V8_1): Renumber.
16696 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16697 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16698 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16699 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16700 rdma to feature modifiers list.
16702 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16704 PR middle-end/56727
16705 * ipa-visibility (function_and_variable_visibility): Convert
16706 recursive PLT call to direct call if appropriate.
16708 2017-07-21 Andrew Pinski <apinski@cavium.com>
16710 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16711 operand 1 to see if the types precision matches.
16712 * fold-const.c (operand_equal_p): Likewise.
16714 2017-07-21 Richard Biener <rguenther@suse.de>
16716 PR tree-optimization/81303
16717 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16718 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16719 (vect_peeling_hash_get_lowest_cost): Adjust.
16720 (vect_enhance_data_refs_alignment): Likewise. Use
16721 vect_get_peeling_costs_all_drs to compute the penalty for no
16722 peeling to match up costs.
16724 2017-07-21 Richard Biener <rguenther@suse.de>
16726 PR tree-optimization/81500
16727 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16728 we didn't identify a reduction path.
16730 2017-07-21 Tom de Vries <tom@codesourcery.com>
16731 Cesar Philippidis <cesar@codesourcery.com>
16733 PR gcov-profile/81442
16734 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16737 2017-07-21 Tom de Vries <tom@codesourcery.com>
16740 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16742 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16743 nvptx_override_options_after_change.
16745 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16747 * dwarf2out.c (output_file_names): Avoid double testing for
16748 dwarf_version >= 5.
16750 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16752 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16754 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16756 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16758 (try_crossjump_to_edge): Do not punt on partitioned functions.
16760 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16762 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16763 Put all BBs reachable only via paths crossing cold region to cold
16765 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16767 2016-07-21 Richard Biener <rguenther@suse.de>
16769 PR tree-optimization/81303
16770 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16771 into account prologue and epilogue iterations when raising
16772 min_profitable_iters to sth at least covering one vector iteration.
16774 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16776 * config/arm/arm.c (arm_test_cpu_arch_dat):
16779 2017-07-20 Nathan Sidwell <nathan@acm.org>
16781 Remove TYPE_METHODS.
16782 * tree.h (TYPE_METHODS): Delete.
16783 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16784 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16785 (dbxout_type_methods): Scan TYPE_FIELDS.
16786 (dbxout_type): Don't check TYPE_METHODS here.
16787 * function.c (use_register_for_decl): Always ignore register for
16788 class types when not optimizing.
16789 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16790 * tree.c (free_lang_data_in_type): Stitch out member functions and
16791 templates from TYPE_FIELDS.
16792 (build_distinct_type_copy, verify_type_variant,
16793 verify_type): Member fns are on TYPE_FIELDS.
16794 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16795 * tree-pretty-print.c (dump_generic_node): Likewise.
16797 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16800 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16801 V2TImode and V4TImode.
16802 (ix86_expand_vector_extract): Likewise.
16803 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16804 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16805 (ssescalarmode): Handle V4TImode and V2TImode.
16806 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16807 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16808 (VEXTRACTI128_MODE): New mode iterator.
16809 (splitter for *vec_extractv?ti first element): New.
16810 (VEC_INIT_MODE): New mode iterator.
16811 (vec_init<mode>): Consolidate 3 expanders into one using
16812 VEC_INIT_MODE mode iterator.
16814 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16816 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16817 non_spilled_static_chain_regno_p.
16819 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16821 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16823 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16825 * bb-reorder.c (connect_traces): Allow copying of blocks within
16828 2017-07-20 Richard Biener <rguenther@suse.de>
16830 * gimple.h (gimple_phi_result): Add gphi * overload.
16831 (gimple_phi_result_ptr): Likewise.
16832 (gimple_phi_arg): Likewise. Adjust index assert to only
16833 allow actual argument accesses rather than all slots available
16835 (gimple_phi_arg_def): Add gphi * overload.
16836 * tree-phinodes.c (make_phi_node): Initialize only actual
16838 (resize_phi_node): Clear memory not covered by old node,
16839 do not initialize excess argument slots.
16840 (reserve_phi_args_for_new_edge): Initialize new argument slot
16843 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16845 PR tree-optimization/81388
16847 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16849 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16850 by removing computation of may_be_zero.
16852 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16853 Tom de Vries <tom@codesourcery.com>
16855 PR middle-end/81030
16856 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16857 when gimple level profile disagrees with what RTL expander did.
16859 2017-07-20 Richard Biener <rguenther@suse.de>
16861 PR tree-optimization/61171
16862 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16863 (vect_analyze_stmt): Add slp instance parameter.
16864 (vectorizable_reduction): Likewise.
16865 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16866 (vect_is_simple_reduction): Deal with chains not detected
16867 as SLP reduction chain, specifically not properly associated
16868 chains containing a mix of plus/minus.
16869 (get_reduction_op): Remove.
16870 (get_initial_defs_for_reduction): Simplify, pass in whether
16871 this is a reduction chain, pass in the SLP node for the PHIs.
16872 (vect_create_epilog_for_reduction): Get the SLP instance as
16874 (vectorizable_reduction): Get the SLP instance as arg.
16875 During analysis remember the SLP node with the PHIs in the
16876 instance. Simplify getting at the vectorized reduction PHIs.
16877 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16878 through SLP instance.
16879 (vect_slp_analyze_operations): Likewise.
16880 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16881 (vect_transform_stmt): Likewise.
16883 2017-07-20 Tom de Vries <tom@codesourcery.com>
16885 PR tree-optimization/81489
16886 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16887 read of phi arg location to before loop that modifies phi.
16889 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16891 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16894 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16896 PR middle-end/81331
16897 * except.c (execute): Fix ordering issue.
16899 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16901 PR rtl-optimization/81423
16902 * combine.c (make_compound_operation_int): Don't try to optimize
16903 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16905 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16907 PR rtl-optimization/81423
16908 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16909 with a constant that is -1 in the truncated to mode.
16911 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16913 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16914 (determine_unlikely_bbs): ... here.
16915 * predict.h (propagate_unlikely_bbs_forward): Declare.
16916 * cfgexpand.c (pass_expand::execute): Use it.
16917 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16919 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16920 propagate_unlikely_bbs_forward.
16922 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16924 PR middle-end/81331
16925 * except.c (maybe_add_nop_after_section_switch): New function.
16928 2017-07-19 Tom de Vries <tom@codesourcery.com>
16930 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16932 2017-07-19 Tom de Vries <tom@codesourcery.com>
16934 * gimple.h (gimple_phi_arg): Make assert more strict.
16936 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16938 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16939 * config/rs6000/mmintrin.h: New file.
16940 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16942 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16944 PR tree-optimization/81346
16945 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16947 2017-07-19 Tom de Vries <tom@codesourcery.com>
16949 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16951 2017-07-19 Tom de Vries <tom@codesourcery.com>
16953 * config/nvptx/nvptx-modes.def: Add V2DImode.
16954 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16955 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16956 (nvptx_output_mov_insn): Handle lack of mov.b128.
16957 (nvptx_print_operand): Handle 'H' and 'L' codes.
16958 (nvptx_vector_mode_supported): Allow V2DImode.
16959 (nvptx_preferred_simd_mode): New function.
16960 (nvptx_data_alignment): New function.
16961 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16962 nvptx_preferred_simd_mode.
16963 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16965 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16967 2017-07-19 Tom de Vries <tom@codesourcery.com>
16969 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16970 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16971 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16972 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16973 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16974 (mov<VECIM>_insn): New define_insn.
16975 (define_expand "mov<VECIM>): New define_expand.
16977 2017-07-19 Tom de Vries <tom@codesourcery.com>
16979 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16981 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16983 PR tree-optimization/81346
16984 * fold-const.h (fold_div_compare, range_check_type): Declare.
16985 * fold-const.c (range_check_type): New function.
16986 (build_range_check): Use range_check_type.
16987 (fold_div_compare): No longer static, rewritten into
16988 a match.pd helper function.
16989 (fold_comparison): Don't call fold_div_compare here.
16990 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16991 as helper function.
16993 2017-07-19 Nathan Sidwell <nathan@acm.org>
16995 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16996 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16997 * tree.c (find_decls_types_r, verify_type): Use
16998 TYPE_{MIN,MAX}_VALUE_RAW.
16999 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17000 (hash_tree): Likewise.
17001 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17003 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17006 2017-07-18 Tom de Vries <tom@codesourcery.com>
17008 PR middle-end/81464
17009 * omp-expand.c (expand_omp_for_static_chunk): Handle
17010 equal-argument loop exit phi.
17012 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17015 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17016 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17017 operand 2 predicate.
17018 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17019 operand 2 predicate.
17020 (ror,rol -> rorx splitters): Use const_int_operand as
17021 operand 2 predicate.
17023 2017-06-18 Richard Biener <rguenther@suse.de>
17025 PR tree-optimization/81410
17026 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17027 the gap in the ! slp_perm SLP case after each group.
17029 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17031 PR middle-end/81463
17032 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17035 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17037 PR middle-end/81462
17038 * predict.c (set_even_probabilities): Cleanup; do not affect
17039 probabilities that are already known.
17040 (combine_predictions_for_bb): Call even when count is set.
17042 2017-07-18 Nathan Sidwell <nathan@acm.org>
17044 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17047 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17050 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17051 optimization for loop niter analysis.
17053 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17056 * config/avr/avr.c (avr_optimize_casesi): Don't use
17057 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17059 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17061 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17062 body_cost_vec from _vect_peel_extended_info.
17063 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17064 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17067 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17069 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17071 2017-07-18 Richard Biener <rguenther@suse.de>
17073 PR tree-optimization/80620
17074 PR tree-optimization/81403
17075 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17076 info when re-using a VN table entry.
17078 2017-07-18 Richard Biener <rguenther@suse.de>
17080 PR tree-optimization/81418
17081 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17082 vectype_in. Verify that with lane-reducing reduction operations
17083 we have a single def-use cycle.
17085 2017-07-17 Carl Love <cel@us.ibm.com>
17087 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17089 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17090 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17091 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17092 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17093 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17094 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17095 VMULOSW): New enum "unspec" values.
17096 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17097 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17098 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17099 altivec_vmulosw): New patterns.
17100 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17101 VMULOSW): Add definitions.
17102 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17104 * config/alpha/alpha.c: Include predict.h.
17106 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17108 * tree-vrp.c (compare_assert_loc): Fix comparison function
17109 to return predictable results.
17111 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17113 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17115 (subdi3): Likewise.
17116 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17117 * doc/invoke.texi (mexpand-adddi): Update text.
17119 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17121 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17122 that also clobbers the CC register. The old expand code is moved
17124 (*arc_clzsi2): ... here.
17125 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17126 the CC register. The old expand code is moved to ...
17127 (arc_ctzsi2): ... here.
17129 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17131 * config/arc/arc.opt (mindexed-loads): Use initial value
17132 TARGET_INDEXED_LOADS_DEFAULT.
17133 (mauto-modify-reg): Use initial value
17134 TARGET_AUTO_MODIFY_REG_DEFAULT.
17135 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17136 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17137 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17138 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17140 2017-07-17 Martin Liska <mliska@suse.cz>
17143 * opts.c (finish_options): Do not allow -fgnu-tm
17144 w/ -fsanitize={kernel-,}address. Say sorry.
17146 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17149 * tree-loop-distribution.c (classify_partition): Only assert on
17150 numer of iterations.
17151 (merge_dep_scc_partitions): Delete prameter. Update function call.
17152 (distribute_loop): Remove code handling loop with unknown niters.
17153 (pass_loop_distribution::execute): Skip loop with unknown niters.
17155 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17158 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17159 function sort_partitions_by_post_order.
17161 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17163 PR tree-optimization/81374
17164 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17165 the max index of basic blocks, rather than number of basic blocks.
17167 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17169 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17171 (arc_legitimate_pic_operand_p): Likewise.
17172 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17174 (arc_needs_pcl_p): Likewise.
17175 (arc_legitimate_pc_offset_p): Likewise.
17176 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17177 function is also used in constrains.md.
17178 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17179 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17180 PLUS. Only return true/false in known cases, otherwise assert.
17181 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17182 is already called in arc_legitimate_constant_p.
17183 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17185 (LEGITIMATE_PIC_OPERAND_P): Use
17186 arc_raw_symbolic_reference_mentioned_p function.
17187 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17192 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17193 Andrew Burgess <andrew.burgess@embecosm.com>
17195 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17196 (arc_return_address_register): New function.
17197 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17198 (arc_handle_fndecl_attribute): Add naked attribute.
17199 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17200 (TARGET_WARN_FUNC_RETURN): Likewise.
17201 (arc_allocate_stack_slots_for_args): New function.
17202 (arc_warn_func_return): Likewise.
17203 (machine_function): Change type fn_type.
17204 (arc_compute_function_type): Consider new naked function type,
17205 change function return type.
17206 (arc_must_save_register): Adapt to handle new
17207 arc_compute_function_type's return type.
17208 (arc_expand_prologue): Likewise.
17209 (arc_expand_epilogue): Likewise.
17210 (arc_return_address_regs): Delete.
17211 (arc_return_address_register): New function.
17212 (arc_epilogue_uses): Use above function.
17213 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17214 (arc_function_type): Change encoding, add naked type.
17215 (ARC_INTERRUPT_P): Change to handle the new encoding.
17216 (ARC_FAST_INTERRUPT_P): Likewise.
17217 (ARC_NORMAL_P): Define.
17218 (ARC_NAKED_P): Likewise.
17219 (arc_compute_function_type): Delete prototype.
17220 * config/arc/arc.md (in_ret_delay_slot): Use
17221 arc_return_address_register function.
17222 (simple_return): Likewise.
17223 (p_return_i): Likewise.
17225 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17227 PR tree-optimization/81428
17228 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17229 can't be built for those types.
17231 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17233 Remove stuff dead since r239246.
17235 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17236 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17237 (avr_inform_devices): Remove dead stuff.
17239 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17241 * config/arm/arm_neon.h: Fix softp typo.
17243 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17245 PR tree-optimization/81365
17246 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17247 aggregate moves onto bb predecessor edges, make sure there are no
17248 loads that could alias the lhs in between the start of bb and the
17251 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17254 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17255 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17256 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17258 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17260 PR tree-optimization/81396
17261 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17262 (init_symbolic_number): Initialize it to 1.
17263 (perform_symbolic_merge): Add n_ops from both operands into the new
17265 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17266 without base_addr as useless if they need more than one operation.
17267 (bswap_replace): Handle !bswap case for NULL base_addr.
17269 2017-07-17 Tom de Vries <tom@codesourcery.com>
17272 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17275 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17277 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17278 conditional builtin define __FIX_LEON3FT_B2BST.
17280 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17282 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17283 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17286 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17288 PR rtl-optimization/81424
17289 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17290 to remove potential trapping from operands if -fnon-call-exceptions.
17292 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17294 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17295 profile_proability for scalling.
17296 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17298 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17300 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17302 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17304 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17305 fixpoint arithmetics.
17307 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17309 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17310 fixpoint arithmetics.
17312 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17314 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17315 fixpoint arithmetics.
17317 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17319 * profile-count.h (profile_probability::from_reg_br_prob_note,
17320 profile_probability::to_reg_br_prob_note): New functions.
17321 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17322 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17323 * predict.c (probability_reliable_p): Update.
17324 (edge_probability_reliable_p): Update.
17325 (br_prob_note_reliable_p): Update.
17326 (invert_br_probabilities): Update.
17327 (add_reg_br_prob_note): New function.
17328 (combine_predictions_for_insn): Update.
17329 * asan.c (asan_clear_shadow): Update.
17330 * cfgbuild.c (compute_outgoing_frequencies): Update.
17331 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17332 (update_br_prob_note): Update.
17333 (rtl_verify_edges): Update.
17334 (purge_dead_edges): Update.
17335 (fixup_reorder_chain): Update.
17336 * emit-rtl.c (try_split): Update.
17337 * ifcvt.c (cond_exec_process_insns): Update.
17338 (cond_exec_process_if_block): Update.
17339 (dead_or_predicable): Update.
17340 * internal-fn.c (expand_addsub_overflow): Update.
17341 (expand_neg_overflow): Update.
17342 (expand_mul_overflow): Update.
17343 * loop-doloop.c (doloop_modify): Update.
17344 * loop-unroll.c (compare_and_jump_seq): Update.
17345 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17346 * predict.h: Update.
17347 * reorg.c (mostly_true_jump): Update.
17349 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17350 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17351 * config/arc/arc.c: (emit_unlikely_jump): Update.
17352 * config/arm/arm.c: (emit_unlikely_jump): Update.
17353 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17354 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17355 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17356 (ix86_print_operand): Update.
17357 (ix86_split_fp_branch): Update.
17358 (predict_jump): Update.
17359 * config/ia64/ia64.c (ia64_print_operand): Update.
17360 * config/mmix/mmix.c (mmix_print_operand): Update.
17361 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17362 (rs6000_expand_split_stack_prologue): Update.
17363 * config/rs6000/rs6000.c: Update.
17364 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17365 (s390_expand_vec_movstr): Update.
17366 (s390_expand_cs_tdsi): Update.
17367 (s390_expand_split_stack_prologue): Update.
17368 * config/sh/sh.c (sh_print_operand): Update.
17369 (expand_cbranchsi4): Update.
17370 (expand_cbranchdi4): Update.
17371 * config/sparc/sparc.c (output_v9branch): Update.
17372 * config/spu/spu.c (get_branch_target): Update.
17373 (ea_load_store_inline): Update.
17374 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17375 * config/tilepro/tilepro.c: Update.
17377 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17379 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17380 (gimplify_save_expr): Likewise.
17382 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17384 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17386 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17388 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17390 * timevar.def (TV_IPA_FNSUMMARY): Define.
17392 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17394 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17395 to back store errata sensitive sequence from being generated.
17396 (sqrtdf2_fix): Likewise.
17398 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17400 * tree-ssa-threadupdate.c (compute_path_counts,
17401 update_joiner_offpath_counts): Use profile_probability.
17403 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17406 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17408 * config/arm/arm-c.c (arm_cpu_builtins): Define
17409 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17411 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17413 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17414 array entries to represent __ieee128 versions of the
17415 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17416 scalar_extract_sig, and scalar_insert_exp built-in functions.
17417 (altivec_resolve_overloaded_builtin): Add special case handling
17418 for the __builtin_scalar_insert_exp function, as represented by
17419 the P9V_BUILTIN_VEC_VSIEDP constant.
17420 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17421 exponent support for __ieee128 argument.
17422 (VSESQP): Add scalar extract signature support for __ieee128
17424 (VSTDCNQP): Add scalar test negative support for __ieee128
17426 (VSIEQP): Add scalar insert exponent support for __int128 argument
17427 with __ieee128 result.
17428 (VSIEQPF): Add scalar insert exponent support for __ieee128
17429 argument with __ieee128 result.
17430 (VSTDCQP): Add scalar test data class support for __ieee128
17432 (VSTDCNQP): Add overload support for scalar test negative with
17433 __ieee128 argument.
17434 (VSTDCQP): Add overload support for scalar test data class
17435 __ieee128 argument.
17436 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17437 UNSPEC_VSX_SXSIGDP.
17438 (UNSPEC_VSX_SIEXPQP): New constant.
17439 (xsxexpqp): New insn for VSX scalar extract exponent quad
17441 (xsxsigqp): New insn for VSX scalar extract significand quad
17443 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17444 precision with floating point argument.
17445 (xststdcqp): New expand for VSX scalar test data class quad
17447 (xststdcnegqp): New expand for VSX scalar test negative quad
17449 (xststdcqp): New insn to match expansions for VSX scalar test data
17450 class quad precision and VSX scalar test negative quad precision.
17451 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17452 special case operand checking to enforce that second operand of
17453 VSX scalar test data class with quad precision argument is a 7-bit
17455 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17456 prototypes and descriptions of __ieee128 versions of
17457 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17458 scalar_test_data_class, and scalar_test_neg built-in functions.
17460 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17462 PR tree-optimization/81162
17463 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17464 replace a negate with an add.
17466 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17468 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17470 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17472 * config/arm/arm-c.c (arm_cpu_builtins): Define
17473 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17475 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17477 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17478 (armv8-r): Set ARM Cortex-R52 as default CPU.
17479 * config/arm/arm-tables.opt: Regenerate.
17480 * config/arm/arm-tune.md: Regenerate.
17481 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17483 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17484 extension for -mcpu=cortex-r52.
17486 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17488 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17489 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17490 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17491 (fp-armv8): Define it as FP_ARMv8 only.
17492 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17493 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17495 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17496 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17497 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17498 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17499 than TARGET_FPU_ARMV8.
17500 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17501 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17502 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17504 * config/arm/neon.md (neon_vrint): Likewise.
17505 (neon_vcvt): Likewise.
17506 (neon_<fmaxmin_op><mode>): Likewise.
17507 (<fmaxmin><mode>3): Likewise.
17508 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17509 * config/arm/predicates.md (arm_cond_move_operator): Check against
17510 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17512 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17514 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17515 to top of function.
17517 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17519 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17520 loop in comment with memset.
17522 2017-07-14 Martin Liska <mliska@suse.cz>
17524 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17525 * dwarf2out.c (is_java): Remove the function.
17526 (output_pubname): Remove usage of the function.
17527 (lower_bound_default): Remove usage of DW_LANG_Java.
17528 (gen_compile_unit_die): Likewise.
17529 * gcc.c: Remove compiler defaults for .java and .zip files.
17530 * gimple-expr.c (remove_suffix): Change as there's no longer
17531 extension than 4-letter one.
17532 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17533 (gimplify_save_expr): Likewise.
17534 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17535 as it's possible even for other languages than Java.
17536 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17537 * lto-opts.c (lto_write_options): Remove reference to Java.
17538 * opts.c (strip_off_ending): Update file extension handling.
17539 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17540 * tree-eh.c (lower_resx): Likewise.
17541 * tree.c (free_lang_data_in_type): Remove dead code.
17542 (find_decls_types_r): Likewise.
17543 (build_common_builtin_nodes): Remove Java from a comment.
17544 (verify_type): Remove dead code.
17545 * varasm.c (assemble_external): Remove Java from a comment.
17547 2017-07-14 Martin Liska <mliska@suse.cz>
17549 * opts.c (finish_options): Add quotes.
17550 (common_handle_option): Likewise.
17552 2017-07-14 Martin Liska <mliska@suse.cz>
17554 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17555 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17556 Remove N_SO_PASCAL.
17557 * dwarf2out.c (lower_bound_default): Do not handle
17559 (gen_compile_unit_die): Likewise.
17560 * gcc.c: Remove default extension binding for GNU Pascal.
17561 * stmt.c: Remove Pascal language from a comment.
17562 * xcoffout.c: Likewise.
17564 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17567 * diagnostic-show-locus.c (fixit_cmp): New function.
17568 (layout::layout): Sort m_fixit_hints.
17569 (column_range::column_range): Assert that the values are valid.
17570 (struct char_span): New struct.
17571 (correction::overwrite): New method.
17572 (struct source_line): New struct.
17573 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17574 calls in terms of classes source_line and char_span, and
17575 correction::overwrite.
17576 (selftest::test_overlapped_fixit_printing_2): New function.
17577 (selftest::diagnostic_show_locus_c_tests): Call it.
17579 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17581 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17582 early if there is no lhs.
17584 2017-07-13 Martin Liska <mliska@suse.cz>
17586 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17587 (gen_reference_type_die): Likewise.
17588 * stor-layout.c: Remove Pascal-related comment.
17590 2017-07-13 Martin Liska <mliska@suse.cz>
17592 * opts.c (finish_options): Add quotes to error messages.
17593 (parse_sanitizer_options): Likewise.
17595 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17597 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17599 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17601 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17603 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17605 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17607 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17609 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17612 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17613 provides the hardware capability bits, define the macro
17614 __BUILTIN_CPU_SUPPORTS__.
17615 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17616 if GLIBC does not provide the hardware capability bits. Add a
17617 gcc_unreachable call if the built-in cpu function is neither
17618 __builtin_cpu_is nor __builtin_cpu_supports.
17619 (rs6000_get_function_versions_dispatcher): Change the warning
17620 that an old GLIBC is used which does not export the capability
17621 bits to be an error.
17622 * doc/extend.texi (target_clones attribute): Document the
17623 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17624 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17625 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17626 the macros defined by GCC if the newer GLIBC is available.
17628 2017-07-12 Jeff Law <law@redhat.com>
17630 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17631 remaining includes slightly.
17632 * config/riscv/riscv-builtins.c: Include profile-count.h.
17634 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17637 * config/avr/avr.c (avr_set_current_function): In diagnostic
17638 messages: Quote keywords and (parts of) identifiers.
17639 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17642 2017-07-12 Carl Love <cel@us.ibm.com>
17644 * config/rs6000/rs6000-c.c: Add support for built-in functions
17645 vector bool char vec_revb (vector bool char);
17646 vector bool short vec_revb (vector short char);
17647 vector bool int vec_revb (vector bool int);
17648 vector bool long long vec_revb (vector bool long long);
17649 * doc/extend.texi: Update the built-in documentation file for the
17650 new built-in functions.
17652 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17654 * config/s390/s390.md: Remove movcc splitter.
17656 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17658 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17659 load/store on condition.
17661 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17664 * config/avr/avr.c (avr_encode_section_info)
17665 [progmem && !TREE_READONLY]: Error if progmem object needs
17668 2017-07-11 Michael Collison <michael.collison@arm.com>
17670 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17673 2017-07-11 Carl Love <cel@us.ibm.com>
17675 * config/rs6000/rs6000-c.c: Add support for builtins
17676 vector unsigned int vec_parity_lsbb (vector signed int);
17677 vector unsigned int vec_parity_lsbb (vector unsigned int);
17678 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17679 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17680 vector unsigned long long vec_parity_lsbb (vector signed long long);
17681 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17682 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17683 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17684 * doc/extend.texi: Update the built-in documentation file for the
17685 new built-in functions.
17687 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17689 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17690 (layout::m_primary_loc): New field.
17691 (layout::layout): Initialize new field. Move location filtering
17692 logic from here to...
17693 (layout::maybe_add_location_range): ...this new method. Add
17694 support for filtering to just the lines already specified by other
17696 (layout::will_show_line_p): New method.
17697 (gcc_rich_location::add_location_if_nearby): New method.
17698 (selftest::test_add_location_if_nearby): New test function.
17699 (selftest::diagnostic_show_locus_c_tests): Call it.
17700 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17703 2017-07-11 Tom de Vries <tom@codesourcery.com>
17705 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17706 (bb_first_real_insn): New function.
17707 (nvptx_single): Add extra initialization of broadcasted condition
17710 2017-07-11 Nathan Sidwell <nathan@acm.org>
17712 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17714 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17716 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17717 Binutils doc as TEXI will mess them up.
17718 * doc/invoke.texi (AVR Options): Same here.
17720 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17722 * config/sparc/sparc.opt (mfix-ut700): New option.
17723 (mfix-gr712rc): Likewise.
17724 (sparc_fix_b2bst): New variable.
17725 * doc/invoke.texi (SPARC options): Document them.
17726 (ARM options): Fix warnings.
17727 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17728 instructions to prevent sequences that can trigger the store-store
17729 errata for certain LEON3FT processors.
17730 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17731 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17732 * config/sparc/sparc.md (fix_b2bst): New attribute.
17733 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17735 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17738 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17740 (*rsqrtsf2_sse): Ditto.
17742 (div<mode>3): Macroize insn from divdf3 and divsf3
17743 using MODEF mode iterator.
17745 2017-07-10 Martin Sebor <msebor@redhat.com>
17747 PR tree-optimization/80397
17748 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17749 instead of testing for equality to INTEGER_TYPE.
17751 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17753 * config.gcc: Remove uclibc from arc target spec.
17755 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17757 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17759 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17762 * lto-wrapper.c (remove_option): New function.
17763 (merge_and_complain): Merge PIC/PIE options more realistically.
17765 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17767 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17771 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17772 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17773 * config.in: Regenerate.
17774 * configure: Regenerate.
17775 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17776 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17777 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17778 (TARGET_GASISR_PROLOGUES): ...target mask.
17779 * common/config/avr/avr-common.c
17780 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17781 Set -mgas-isr-prologues.
17782 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17783 INSERT_PASS_BEFORE for it.
17784 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17785 * config/avr/avr.c (avr_option_override)
17786 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17787 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17788 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17789 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17790 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17792 (make_avr_pass_pre_proep): New function.
17793 (emit_push_sfr) <treg>: Add argument to function and use it
17794 instead of TMP_REG.
17795 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17796 and set machine->gasisr.yes.
17797 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17798 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17799 __gcc_isr.n_pushed to .L__stack_usage.
17800 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17801 (avr_asm_final_postscan_insn): ...this new static function.
17802 * config/avr/avr.h (machine_function)
17803 <is_no_gccisr, use_L__stack_usage>: New fields.
17804 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17805 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17806 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17807 (gasisr, *gasisr): New expander and insn.
17808 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17809 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17810 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17812 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17814 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17817 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17819 Move jump-tables out of .text again.
17822 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17823 (ASM_OUTPUT_ADDR_VEC): New function.
17824 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17825 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17826 INSN_ADDRESSes as asm comment.
17827 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17828 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17829 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17830 * config/avr/avr.md (*tablejump): Adjust comment.
17831 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17832 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17834 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17835 (avr_output_addr_vec): New proto.
17836 (avr_log_t) <insn_addresses>: New field.
17838 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17841 * config/i386/i386.c (ix86_function_arg_advance): Set
17842 outgoing_args_on_stack to true if there are outgoing arguments
17844 (ix86_function_arg): Likewise.
17845 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17846 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17847 * config/i386/i386.h (machine_function): Add
17848 outgoing_args_on_stack.
17850 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17852 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17853 supporting pthreds.
17854 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17856 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17858 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17859 (REAL_H): Remove $(MACHMODE_H).
17860 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17862 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17863 $(MACHMODE_H) and double-int.h.
17864 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17866 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17869 2017-07-07 Andrew Pinski <apinski@cavium.com>
17871 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17872 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17874 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17876 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17877 Add warning if GCC was not configured to link against a GLIBC that
17878 exports the hardware capability bits.
17879 (make_resolver_func): Make resolver function private and not a
17880 COMDAT function. Create the name with clone_function_name instead
17881 of make_unique_name.
17884 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17885 correct operand in doing the split.
17887 2017-07-07 Carl Love <cel@us.ibm.com>
17889 * config/rs6000/rs6000-c: Add support for built-in function
17890 vector unsigned short vec_pack_to_short_fp32 (vector float,
17892 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17893 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17894 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17895 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17896 (convert_4f32_8i16): Add define_expand.
17897 * doc/extend.texi: Update the built-in documentation file for the
17898 new built-in function.
17900 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17902 * config/sparc/m8.md: New file.
17903 * config/sparc/sparc.md: Include m8.md.
17905 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17907 * config/sparc/sparc.opt: New option -mvis4b.
17908 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17909 (sparc_option_override): Handle VIS4B.
17910 (enum sparc_builtins): Define
17911 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17912 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17913 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17914 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17915 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17916 (check_constant_argument): New function.
17917 (sparc_vis_init_builtins): Define builtins
17918 __builtin_vis_dictunpack{8,16,32},
17919 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17920 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17921 __builtin_vis_fpcmpde{8,16,32}shl and
17922 __builtin_vis_fpcmpur{8,16,32}shl.
17923 (sparc_expand_builtin): Check that the constant operands to
17924 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17925 constant and in range.
17926 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17928 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17929 (SPARC_IMM5_P): Likewise.
17930 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17931 (enabled): Handle vis4b.
17932 (UNSPEC_DICTUNPACK): New unspec.
17933 (UNSPEC_FPCMPSHL): Likewise.
17934 (UNSPEC_FPUCMPSHL): Likewise.
17935 (UNSPEC_FPCMPDESHL): Likewise.
17936 (UNSPEC_FPCMPURSHL): Likewise.
17937 (cpu_feature): New CPU feature `vis4b'.
17938 (dictunpack{8,16,32}): New insns.
17939 (FPCSMODE): New mode iterator.
17940 (fpcscond): New code iterator.
17941 (fpcsucond): Likewise.
17942 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17943 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17944 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17945 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17946 * config/sparc/constraints.md: Define constraints `q' for unsigned
17947 2-bit integer constants and `t' for unsigned 5-bit integer
17949 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17951 (imm5_operand_dictunpack16): Likewise.
17952 (imm5_operand_dictunpack32): Likewise.
17953 (imm2_operand): Likewise.
17954 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17955 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17956 ditunpack* and fpcmp*shl builtins.
17958 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17960 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17961 * config.in: Add HAVE_AS_SPARC6 define.
17962 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17964 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17966 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17967 (CPP_CPU_SPEC): Handle m8.
17968 (ASM_CPU_SPEC): Likewise.
17969 * config/sparc/sparc-opts.h (enum processor_type): Add
17971 * config/sparc/sparc.c (m8_costs): New struct.
17972 (sparc_option_override): Handle TARGET_CPU_m8.
17973 (sparc32_initialize_trampoline): Likewise.
17974 (sparc64_initialize_trampoline): Likewise.
17975 (sparc_issue_rate): Likewise.
17976 (sparc_register_move_cost): Likewise.
17977 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17978 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17979 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17980 (CPP_CPU_SPEC): Handle M8.
17981 (ASM_CPU_SPEC): Likewise.
17982 (AS_M8_FLAG): Define.
17983 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17984 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17985 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17987 * configure: Regenerate.
17988 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17991 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17993 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17995 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17996 ("*movdi_insn_sp32"): Do not set v3pipe.
17997 ("*movsi_insn"): Likewise.
17998 ("*movdi_insn_sp64"): Likewise.
17999 ("*movsf_insn"): Likewise.
18000 ("*movdf_insn_sp32"): Likewise.
18001 ("*movdf_insn_sp64"): Likewise.
18002 ("*zero_extendsidi2_insn_sp64"): Likewise.
18003 ("*sign_extendsidi2_insn"): Likewise.
18004 ("*mov<VM32:mode>_insn"): Likewise.
18005 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18006 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18007 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18008 ("<vlop:code><VL:mode>3"): Likewise.
18009 ("*not_<vlop:code><VL:mode>3"): Likewise.
18010 ("*nand<VL:mode>_vis"): Likewise.
18011 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18012 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18013 ("one_cmpl<VL:mode>2"): Likewise.
18014 ("faligndata<VM64:mode>_vis"): Likewise.
18015 ("alignaddrsi_vis"): Likewise.
18016 ("alignaddrdi_vis"): Likweise.
18017 ("alignaddrlsi_vis"): Likewise.
18018 ("alignaddrldi_vis"): Likewise.
18019 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18020 ("bmaskdi_vis"): Likewise.
18021 ("bmasksi_vis"): Likewise.
18022 ("bshuffle<VM64:mode>_vis"): Likewise.
18023 ("cmask8<P:mode>_vis"): Likewise.
18024 ("cmask16<P:mode>_vis"): Likewise.
18025 ("cmask32<P:mode>_vis"): Likewise.
18026 ("pdistn<P:mode>_vis"): Likewise.
18027 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18029 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18031 * config/sparc/sparc.md ("subtype"): New insn attribute.
18032 ("*wrgsr_sp64"): Set insn subtype.
18033 ("*rdgsr_sp64"): Likewise.
18034 ("alignaddrsi_vis"): Likewise.
18035 ("alignaddrdi_vis"): Likewise.
18036 ("alignaddrlsi_vis"): Likewise.
18037 ("alignaddrldi_vis"): Likewise.
18038 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18039 ("fexpand_vis"): Likewise.
18040 ("fpmerge_vis"): Likewise.
18041 ("faligndata<VM64:mode>_vis"): Likewise.
18042 ("bshuffle<VM64:mode>_vis"): Likewise.
18043 ("cmask8<P:mode>_vis"): Likewise.
18044 ("cmask16<P:mode>_vis"): Likewise.
18045 ("cmask32<P:mode>_vis"): Likewise.
18046 ("fchksm16_vis"): Likewise.
18047 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18048 ("fmean16_vis"): Likewise.
18049 ("fp<plusminus_insn>64_vis"): Likewise.
18050 ("<plusminus_insn>v8qi3"): Likewise.
18051 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18052 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18053 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18054 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18055 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18056 ("*movqi_insn"): Likewise.
18057 ("*movhi_insn"): Likewise.
18058 ("*movsi_insn"): Likewise.
18059 ("movsi_pic_gotdata_op"): Likewise.
18060 ("*movdi_insn_sp32"): Likewise.
18061 ("*movdi_insn_sp64"): Likewise.
18062 ("movdi_pic_gotdata_op"): Likewise.
18063 ("*movsf_insn"): Likewise.
18064 ("*movdf_insn_sp32"): Likewise.
18065 ("*movdf_insn_sp64"): Likewise.
18066 ("*zero_extendhisi2_insn"): Likewise.
18067 ("*zero_extendqihi2_insn"): Likewise.
18068 ("*zero_extendqisi2_insn"): Likewise.
18069 ("*zero_extendqidi2_insn"): Likewise.
18070 ("*zero_extendhidi2_insn"): Likewise.
18071 ("*zero_extendsidi2_insn_sp64"): Likewise.
18072 ("ldfsr"): Likewise.
18073 ("prefetch_64"): Likewise.
18074 ("prefetch_32"): Likewise.
18075 ("tie_ld32"): Likewise.
18076 ("tie_ld64"): Likewise.
18077 ("*tldo_ldub_sp32"): Likewise.
18078 ("*tldo_ldub1_sp32"): Likewise.
18079 ("*tldo_ldub2_sp32"): Likewise.
18080 ("*tldo_ldub_sp64"): Likewise.
18081 ("*tldo_ldub1_sp64"): Likewise.
18082 ("*tldo_ldub2_sp64"): Likewise.
18083 ("*tldo_ldub3_sp64"): Likewise.
18084 ("*tldo_lduh_sp32"): Likewise.
18085 ("*tldo_lduh1_sp32"): Likewise.
18086 ("*tldo_lduh_sp64"): Likewise.
18087 ("*tldo_lduh1_sp64"): Likewise.
18088 ("*tldo_lduh2_sp64"): Likewise.
18089 ("*tldo_lduw_sp32"): Likewise.
18090 ("*tldo_lduw_sp64"): Likewise.
18091 ("*tldo_lduw1_sp64"): Likewise.
18092 ("*tldo_ldx_sp64"): Likewise.
18093 ("*mov<VM32:mode>_insn"): Likewise.
18094 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18095 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18097 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18099 * config/sparc/sparc.md ("type"): New insn type viscmp.
18100 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18102 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18103 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18104 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18105 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18107 ("n7_vis_logical_11cycle"): Likewise.
18108 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18109 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18110 * config/sparc/niagara.md ("niag_vis"): Likewise.
18111 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18112 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18114 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18116 * config/sparc/sparc.md: New instruction type `bmask'.
18117 (bmaskdi_vis): Use the `bmask' type.
18118 (bmasksi_vis): Likewise.
18119 * config/sparc/ultra3.md (us3_array): Likewise.
18120 * config/sparc/niagara7.md (n7_array): Likewise.
18121 * config/sparc/niagara4.md (n4_array): Likewise.
18122 * config/sparc/niagara2.md (niag2_vis): Likewise.
18123 (niag3_vis): Likewise.
18124 * config/sparc/niagara.md (niag_vis): Likewise.
18126 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18128 * ipa-comdats.c: Remove optimize check from gate.
18129 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18130 for functions not optimized.
18131 (ipa_fn_summary_read): Skip optimize check.
18132 (ipa_fn_summary_write): Likewise.
18133 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18135 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18137 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18138 (check_callers): Give up on caller which is not optimized.
18139 (inline_small_functions): Likewise.
18140 (ipa_inline): Do not give up when not optimizing.
18141 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18142 away unoptimizes cdtors.
18143 (whole_program_function_and_variable_visibility): Do
18144 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18145 * ipa.c (process_references): Do not check optimize.
18146 (symbol_table::remove_unreachable_nodes): Update optimize check.
18147 (set_writeonly_bit): Update optimize check.
18148 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18149 (pass_ipa_single_use::gate): Remove.
18151 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18153 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18154 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18155 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18156 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18157 permute_load, permute_store, adjust_extract, adjust_splat,
18158 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18159 replace_swap_with_copy, dump_swap_insn_table,
18160 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18161 recombine_lvx_pattern, recombine_stvx_pattern,
18162 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18163 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18164 to file rs6000-p8swap.c.
18165 * config/rs6000/rs6000-p8swap.c: New file.
18166 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18167 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18168 and rs6000*-*-* targets.
18170 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18172 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18174 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18176 * lto-wrapper.c (merge_and_complain): Do not merge
18177 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18178 fsigned_zeros, ftrapping_math, fwrapv.
18179 (append_compiler_options): Do not track these options.
18180 (append_linker_options): Likewie
18182 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18184 * cgraphunit.c (cgraph_node::finalize_function): When
18185 !flag_toplevel_reorde set no_reorder flag.
18186 (varpool_node::finalize_decl): Likewise.
18187 (symbol_table::compile): Drop no toplevel reorder path.
18189 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18191 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18192 edges; zero probability is not better than uninitialized.
18194 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18196 * asan.h (asan_sanitize_allocas_p): Declare.
18197 * asan.c (asan_sanitize_allocas_p): New function.
18198 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18199 (handle_builtin_alloca): Likewise.
18200 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18201 if !asan_sanitize_allocas_p.
18202 * params.def (asan-instrument-allocas): Add new option.
18203 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18204 * opts.c (common_handle_option): Disable allocas sanitization for
18207 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18209 * asan.c: Include gimple-fold.h.
18210 (get_last_alloca_addr): New function.
18211 (handle_builtin_stackrestore): Likewise.
18212 (handle_builtin_alloca): Likewise.
18213 (asan_emit_allocas_unpoison): Likewise.
18214 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18215 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18216 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18217 (instrument_builtin_call): Pass gimple iterator to
18218 get_mem_refs_of_builtin_call.
18219 (last_alloca_addr): New global.
18220 * asan.h (asan_emit_allocas_unpoison): Declare.
18221 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18222 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18223 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18224 if function calls alloca.
18225 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18226 * gimple-fold.h (replace_call_with_value): Declare.
18227 * internal-fn.c: Include asan.h.
18228 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18229 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18231 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18233 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18234 (C_SELFTEST_FLAGS): New.
18235 (CPP_SELFTEST_FLAGS): New.
18236 (SELFTEST_DEPS): New, from deps of s-selftest.
18237 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18238 (CPP_SELFTEST_DEPS): New.
18239 (selftest): Add dependency on s-selftest-c++.
18240 (s-selftest): Rename to...
18241 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18242 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18243 than SELFTEST_FLAGS.
18244 (selftest-gdb): Rename to...
18245 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18247 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18248 (selftest-valgrind): Rename to...
18249 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18251 (selftest-valgrind): Reintroduce as an alias for
18252 selftest-c-valgrind.
18253 (s-selftest-c++): New.
18254 (selftest-c++-gdb): New.
18255 (selftest-c++-valgrind): New.
18257 2017-07-06 Olivier Hainque <hainque@adacore.com>
18259 * gcc.c (process_command): When deciding if undefined variables
18260 should be ignored when processing specs, accept "gcc -v" as well.
18262 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18264 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18265 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18267 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18269 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18270 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18271 * config/arm/arm-tables.opt: Regenerate.
18272 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18274 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18276 2017-07-06 Carl Love <cel@us.ibm.com>
18278 * ChangeLog: Clean up from mid air collision
18280 2017-07-06 Carl Love <cel@us.ibm.com>
18282 * config/rs6000/rs6000-c.c: Add support for built-in functions
18283 vector signed int vec_subc (vector signed int, vector signed int);
18284 vector signed __int128 vec_subc (vector signed __int128,
18285 vector signed __int128);
18286 vector unsigned __int128 vec_subc (vector unsigned __int128,
18287 vector unsigned __int128);
18288 vector signed int vec_sube (vector signed int, vector signed int,
18289 vector signed int);
18290 vector unsigned int vec_sube (vector unsigned int,
18291 vector unsigned int,
18292 vector unsigned int);
18293 vector signed __int128 vec_sube (vector signed __int128,
18294 vector signed __int128,
18295 vector signed__int128);
18296 vector unsigned __int128 vec_sube (vector unsigned __int128,
18297 vector unsigned __int128,
18298 vector unsigned __int128);
18299 vector signed int vec_subec (vector signed int, vector signed int,
18300 vector signed int);
18301 vector unsigned int vec_subec (vector unsigned int,
18302 vector unsigned int,
18303 vector unsigned int);
18304 vector signed __int128 vec_subec (vector signed __int128,
18305 vector signed __int128,
18306 vector signed__int128);
18307 vector unsigned __int128 vec_subec (vector unsigned __int128,
18308 vector unsigned __int128,
18309 vector unsigned __int128);
18310 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18311 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18312 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18313 BU_ALTIVEC_OVERLOAD_X definitions.
18314 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18315 * doc/extend.texi: Update the built-in documentation file for the new
18316 built-in functions.
18318 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18321 * diagnostic-show-locus.c (layout::layout): Use start and finish
18322 spelling location for the start and finish of each range.
18323 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18324 Add unused aspect param.
18325 * input.c (expand_location_1): Add "aspect" param, and use it
18326 to access the correct part of the location.
18327 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18329 (expand_location_to_spelling_point): Likewise.
18330 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18331 param, and pass it to expand_location_1.
18333 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18335 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18336 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18337 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18338 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18339 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18340 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18341 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18342 _mm_maskz_getmant_ss): New intrinsics.
18343 (__builtin_ia32_getexpss128_mask): Changed to ...
18344 __builtin_ia32_getexpss128_round ... this.
18345 (__builtin_ia32_getexpsd128_mask): Changed to ...
18346 __builtin_ia32_getexpsd128_round ... this.
18347 * config/i386/i386-builtin-types.def
18348 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18349 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18350 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18351 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18352 __builtin_ia32_getmantss_mask_round): New builtins.
18353 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18354 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18355 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18356 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18357 * config/i386/sse.md
18358 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18359 avx512f_sgetexp<mode><mask_scalar_name>
18360 <round_saeonly_scalar_name> ... this.
18361 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18362 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18363 vgetexp<ssescalarmodesuffix>
18364 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18365 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18366 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18367 avx512f_vgetmant<mode><mask_scalar_name>
18368 <round_saeonly_scalar_name> ... this.
18369 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18370 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18371 vgetmant<ssescalarmodesuffix>
18372 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18373 %0<mask_scalar_operand4>, %1, %2
18374 <round_saeonly_scalar_mask_op4>, %3} ... this.
18375 * config/i386/subst.md (mask_scalar_operand4,
18376 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18377 round_saeonly_scalar_nimm_predicate): New subst attributes.
18379 2017-07-06 Julia Koval <julia.koval@intel.com>
18381 * config/i386/i386.c (ix86_erase_embedded_rounding):
18382 Remove code for old rounding pattern.
18384 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18386 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18388 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18390 * doc/sourcebuild.texi (Test Directives, Variants of
18391 dg-require-support): Add documentation for dg-require-stack-check.
18393 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18395 * config/i386/subst.md (mask_scalar, round_scalar,
18396 round_saeonly_scalar): New meta-templates.
18397 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18398 round_scalar_mask_operand3, round_scalar_mask_op3,
18399 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18400 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18401 round_saeonly_scalar_constraint,
18402 round_saeonly_scalar_prefix): New subst attribute.
18403 * config/i386/sse.md
18404 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18405 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18406 <round_scalar_name> ... this.
18407 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18408 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18409 <round_scalar_name> ... this.
18410 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18411 <sse>_vm<code><mode>3<mask_scalar_name>
18412 <round_saeonly_scalar_name> ... this.
18413 (v<plusminus_mnemonic><ssescalarmodesuffix>
18414 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18415 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18416 v<plusminus_mnemonic><ssescalarmodesuffix>
18417 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18418 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18419 (v<multdiv_mnemonic><ssescalarmodesuffix>
18420 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18421 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18422 v<multdiv_mnemonic><ssescalarmodesuffix>
18423 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18424 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18425 (v<maxmin_float><ssescalarmodesuffix>
18426 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18427 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18428 v<maxmin_float><ssescalarmodesuffix>
18429 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18430 %0<mask_scalar_operand3>, %1, %<iptr>2
18431 <round_saeonly_scalar_mask_op3>} ... this.
18433 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18435 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18436 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18438 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18439 Alan Hayward <alan.hayward@arm.com>
18440 David Sherwood <david.sherwood@arm.com>
18442 * combine.c (simplify_if_then_else): Remove "enum" before
18444 * compare-elim.c (can_eliminate_compare): Likewise.
18445 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18447 (aarch64_lookup_simd_builtin_type): Likewise.
18448 (aarch64_simd_builtin_type): Likewise.
18449 (aarch64_init_simd_builtin_types): Likewise.
18450 (aarch64_simd_expand_args): Likewise.
18451 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18453 (aarch64_reverse_mask): Likewise.
18454 (aarch64_simd_emit_reg_reg_move): Likewise.
18455 (aarch64_gen_adjusted_ldpstp): Likewise.
18456 (aarch64_ccmp_mode_to_code): Likewise.
18457 (aarch64_operands_ok_for_ldpstp): Likewise.
18458 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18459 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18461 (aarch64_min_divisions_for_recip_mul): Likewise.
18462 (aarch64_reassociation_width): Likewise.
18463 (aarch64_get_condition_code_1): Likewise.
18464 (aarch64_simd_emit_reg_reg_move): Likewise.
18465 (aarch64_simd_attr_length_rglist): Likewise.
18466 (aarch64_reverse_mask): Likewise.
18467 (aarch64_operands_ok_for_ldpstp): Likewise.
18468 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18469 (aarch64_gen_adjusted_ldpstp): Likewise.
18470 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18472 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18473 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18474 (arm_lookup_simd_builtin_type): Likewise.
18475 (arm_simd_builtin_type): Likewise.
18476 (arm_init_simd_builtin_types): Likewise.
18477 (arm_expand_builtin_args): Likewise.
18478 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18479 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18480 (ft32_setup_incoming_varargs): Likewise.
18481 (ft32_function_arg): Likewise.
18482 (ft32_function_arg_advance): Likewise.
18483 (ft32_pass_by_reference): Likewise.
18484 (ft32_arg_partial_bytes): Likewise.
18485 (ft32_valid_pointer_mode): Likewise.
18486 (ft32_addr_space_pointer_mode): Likewise.
18487 (ft32_addr_space_legitimate_address_p): Likewise.
18488 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18490 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18491 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18492 (iamcu_alignment): Likewise.
18493 (canonicalize_vector_int_perm): Likewise.
18494 (ix86_noce_conversion_profitable_p): Likewise.
18495 (ix86_mpx_bound_mode): Likewise.
18496 (ix86_operands_ok_for_move_multiple): Likewise.
18497 * config/microblaze/microblaze-protos.h
18498 (microblaze_expand_conditional_branch_reg): Likewise.
18499 * config/microblaze/microblaze.c
18500 (microblaze_expand_conditional_branch_reg): Likewise.
18501 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18503 (rs6000_reassociation_width): Likewise.
18504 (rs6000_invalid_binary_op): Likewise.
18505 (fusion_p9_p): Likewise.
18506 (emit_fusion_p9_load): Likewise.
18507 (emit_fusion_p9_store): Likewise.
18508 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18510 (riscv_hard_regno_mode_ok_p): Likewise.
18511 (riscv_address_insns): Likewise.
18512 (riscv_split_symbol): Likewise.
18513 (riscv_legitimize_move): Likewise.
18514 (riscv_function_value): Likewise.
18515 (riscv_hard_regno_nregs): Likewise.
18516 (riscv_expand_builtin): Likewise.
18517 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18518 (riscv_build_integer): Likewise.
18519 (riscv_split_integer): Likewise.
18520 (riscv_legitimate_constant_p): Likewise.
18521 (riscv_cannot_force_const_mem): Likewise.
18522 (riscv_regno_mode_ok_for_base_p): Likewise.
18523 (riscv_valid_base_register_p): Likewise.
18524 (riscv_valid_offset_p): Likewise.
18525 (riscv_valid_lo_sum_p): Likewise.
18526 (riscv_classify_address): Likewise.
18527 (riscv_legitimate_address_p): Likewise.
18528 (riscv_address_insns): Likewise.
18529 (riscv_load_store_insns): Likewise.
18530 (riscv_force_binary): Likewise.
18531 (riscv_split_symbol): Likewise.
18532 (riscv_force_address): Likewise.
18533 (riscv_legitimize_address): Likewise.
18534 (riscv_move_integer): Likewise.
18535 (riscv_legitimize_const_move): Likewise.
18536 (riscv_legitimize_move): Likewise.
18537 (riscv_address_cost): Likewise.
18538 (riscv_subword): Likewise.
18539 (riscv_output_move): Likewise.
18540 (riscv_canonicalize_int_order_test): Likewise.
18541 (riscv_emit_int_order_test): Likewise.
18542 (riscv_function_arg_boundary): Likewise.
18543 (riscv_pass_mode_in_fpr_p): Likewise.
18544 (riscv_pass_fpr_single): Likewise.
18545 (riscv_pass_fpr_pair): Likewise.
18546 (riscv_get_arg_info): Likewise.
18547 (riscv_function_arg): Likewise.
18548 (riscv_function_arg_advance): Likewise.
18549 (riscv_arg_partial_bytes): Likewise.
18550 (riscv_function_value): Likewise.
18551 (riscv_pass_by_reference): Likewise.
18552 (riscv_setup_incoming_varargs): Likewise.
18553 (riscv_print_operand): Likewise.
18554 (riscv_elf_select_rtx_section): Likewise.
18555 (riscv_save_restore_reg): Likewise.
18556 (riscv_for_each_saved_reg): Likewise.
18557 (riscv_register_move_cost): Likewise.
18558 (riscv_hard_regno_mode_ok_p): Likewise.
18559 (riscv_hard_regno_nregs): Likewise.
18560 (riscv_class_max_nregs): Likewise.
18561 (riscv_memory_move_cost): Likewise.
18562 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18563 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18564 (rl78_addr_space_address_mode): Likewise.
18565 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18567 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18568 (rs6000_reassociation_width): Likewise.
18569 (rs6000_invalid_binary_op): Likewise.
18570 (fusion_p9_p): Likewise.
18571 (emit_fusion_p9_load): Likewise.
18572 (emit_fusion_p9_store): Likewise.
18573 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18574 (ok_for_simple_move_operands): Likewise.
18575 (ok_for_simple_move_strict_operands): Likewise.
18576 (ok_for_simple_arith_logic_operands): Likewise.
18577 (visium_legitimize_reload_address): Likewise.
18578 (visium_select_cc_mode): Likewise.
18579 (output_cbranch): Likewise.
18580 (visium_split_double_move): Likewise.
18581 (visium_expand_copysign): Likewise.
18582 (visium_expand_int_cstore): Likewise.
18583 (visium_expand_fp_cstore): Likewise.
18584 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18585 (visium_function_arg): Likewise.
18586 (visium_function_arg_advance): Likewise.
18587 (visium_libcall_value): Likewise.
18588 (visium_setup_incoming_varargs): Likewise.
18589 (visium_legitimate_constant_p): Likewise.
18590 (visium_legitimate_address_p): Likewise.
18591 (visium_legitimize_address): Likewise.
18592 (visium_secondary_reload): Likewise.
18593 (visium_register_move_cost): Likewise.
18594 (visium_memory_move_cost): Likewise.
18595 (prepare_move_operands): Likewise.
18596 (ok_for_simple_move_operands): Likewise.
18597 (ok_for_simple_move_strict_operands): Likewise.
18598 (ok_for_simple_arith_logic_operands): Likewise.
18599 (visium_function_value_1): Likewise.
18600 (rtx_ok_for_offset_p): Likewise.
18601 (visium_legitimize_reload_address): Likewise.
18602 (visium_split_double_move): Likewise.
18603 (visium_expand_copysign): Likewise.
18604 (visium_expand_int_cstore): Likewise.
18605 (visium_expand_fp_cstore): Likewise.
18606 (visium_split_cstore): Likewise.
18607 (visium_select_cc_mode): Likewise.
18608 (visium_split_cbranch): Likewise.
18609 (output_cbranch): Likewise.
18610 (visium_print_operand_address): Likewise.
18611 * expmed.c (flip_storage_order): Likewise.
18612 * expmed.h (emit_cstore): Likewise.
18613 (flip_storage_order): Likewise.
18614 * genrecog.c (validate_pattern): Likewise.
18615 * hsa-gen.c (gen_hsa_addr): Likewise.
18616 * internal-fn.c (expand_arith_overflow): Likewise.
18617 * ira-color.c (allocno_copy_cost_saving): Likewise.
18618 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18619 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18620 (process_invariant_for_inheritance): Likewise.
18621 * lra-eliminations.c (move_plus_up): Likewise.
18622 * omp-low.c (lower_oacc_reductions): Likewise.
18623 * simplify-rtx.c (simplify_subreg): Likewise.
18624 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18625 (TARGET_CHKP_BOUND_MODE): Likewise..
18626 * targhooks.c (default_chkp_bound_mode): Likewise.
18627 (default_setup_incoming_vararg_bounds): Likewise.
18628 * targhooks.h (default_chkp_bound_mode): Likewise.
18629 (default_setup_incoming_vararg_bounds): Likewise.
18630 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18631 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18632 (have_whole_vector_shift): Likewise.
18633 * tree-vect-stmts.c (vectorizable_load): Likewise.
18634 * doc/tm.texi: Regenerate.
18636 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18638 Graceful degrade if Binutils PR21472 is not available.
18641 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18642 .rodata in flash test fails.
18643 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18644 * confgure: Regenerate.
18645 * config.in: Regenerate.
18646 * config/avr/avr.c (avr_asm_named_section)
18647 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18648 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18649 (avr_asm_init_sections): Same.
18651 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18653 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18654 (fma<VH:mode>4_intrinsic): Likewise.
18655 (*fmsub<VCVTF:mode>4): Likewise.
18656 (*fmsub<VH:mode>4_intrinsic): Likewise.
18658 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18661 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18662 Don't depend on "optimize > 0".
18663 (out_movhi_r_mr, out_movqi_mr_r): Same.
18664 (out_movhi_mr_r, out_movqi_r_mr): Same.
18665 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18666 io_address_operand on "optimize > 0".
18668 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18670 * tree-loop-distribution.c: Add general explanantion on the pass.
18671 (generate_loops_for_partition): Mark distributed loop.
18672 (pg_add_dependence_edges): New parameter. Handle alias data
18673 dependence specially and record it in the parameter if asked.
18674 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18675 (init_partition_graph_vertices, add_partition_graph_edge): New.
18676 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18677 (free_partition_graph_vdata, build_partition_graph): New.
18678 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18679 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18680 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18681 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18682 (version_for_distribution_p, finalize_partitions): New.
18683 (distribute_loop): Handle alias data dependence specially. Factor
18684 out loop fusion code as functions and call these functions.
18686 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18688 * tree-loop-distribution.c (classify_partition): New parameter and
18689 better handle reduction statement.
18690 (rdg_build_partitions): Revise comment.
18691 (distribute_loop): Compute statements in all partitions and pass it
18692 to classify_partition.
18694 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18696 * tree-loop-distribution.c (enum partition_type): New.
18697 (struct partition): New field type.
18698 (partition_merge_into): Add parameter. Update partition type.
18699 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18700 (build_rdg_partition_for_vertex): Compute partition type.
18701 (rdg_build_partitions): Dump partition type.
18702 (distribute_loop): Update calls to partition_merge_into.
18704 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18706 * tree-loop-distribution.c (struct ddr_hasher): New.
18707 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18709 (classify_partition): Call get_data_dependence.
18710 (pg_add_dependence_edges): Ditto.
18711 (distribute_loop): Release data dependence hash table.
18713 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18715 * tree-loop-distribution.c (ref_base_address): Delete.
18716 (similar_memory_accesses): Rename ...
18717 (share_memory_accesses): ... to this. Check if partitions access
18718 the same memory reference.
18719 (distribute_loop): Call share_memory_accesses.
18721 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18723 * tree-loop-distribution.c (struct partition): New field recording
18724 its data reference.
18725 (partition_alloc, partition_free): Init and release data refs.
18726 (partition_merge_into): Merge data refs.
18727 (build_rdg_partition_for_vertex): Collect data refs for partition.
18728 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18730 (distribute_loop): Remve data refs from vertice data of partition
18733 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18735 * tree-loop-distribution.c (params.h): Include header file.
18736 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18737 (datarefs_vec): New global var.
18738 (create_rdg_vertices): Use datarefs_vec directly.
18739 (free_rdg): Don't free data references.
18740 (build_rdg): Update use. Don't free data references.
18741 (distribute_loop): Compute global variable for data references.
18742 Bail out if there are too many data references.
18744 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18746 * tree-loop-distribution.c (loop_nest): New global var.
18747 (build_rdg): Use loop directly, rather than loop nest.
18748 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18750 (distribute_loop): Compute global variable loop nest. Update use.
18752 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18754 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18755 (partition_merge_into): New parameter. Dump reason for fusion.
18756 (distribute_loop): Update use of partition_merge_into.
18758 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18760 * tree-loop-distribution.c (bb_top_order_index): New.
18761 (bb_top_order_index_size, bb_top_order_cmp): New.
18762 (stmts_from_loop): Use topological order.
18763 (pass_loop_distribution::execute): Compute and release topological
18764 order for basic blocks.
18766 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18768 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18771 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18773 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18774 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18775 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18776 * internal-fn.def (LOOP_DIST_ALIAS): New.
18777 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18778 (fold_loop_internal_call): ... this.
18779 (vect_loop_dist_alias_call): New function.
18780 (set_uid_loop_bbs): Call fold_loop_internal_call.
18781 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18784 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18787 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18788 Require dead FLAGS_REG at the beginning of a peephole.
18790 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18793 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18794 arguments in the call to __builtin_ia32_sbb_u32.
18795 (_subborrow_u64): Swap _X and _Y arguments in the call to
18796 __builtin_ia32_sbb_u64.
18798 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18801 * tree-vrp.c (compare_assert_loc): Turn into a function template
18802 with stable template parameter. Only test if a->e is NULL,
18803 !a->e == !b->e has been verified already. Use e == NULL or
18804 e != NULL instead of e or ! e tests. If stable is true, don't use
18805 iterative_hash_expr, on the other side allow a or b or both NULL
18806 and sort the NULLs last.
18807 (process_assert_insertions): Sort using compare_assert_loc<false>
18808 instead of compare_assert_loc, later sort using
18809 compare_assert_loc<true> before calling process_assert_insertions_for
18810 in a loop. Use break instead of continue once seen NULL pointer.
18812 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18814 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18815 Cortex-R7 and Cortex-R8 processors.
18817 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18819 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18820 uninitialized while src is not.
18822 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18824 * common/config/arm/arm-common.c: Adjust include path for
18826 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18827 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18828 (arm-cpu-data.h): Likewise.
18829 (arm-cpu-cdata.h): Likewise.
18830 * config/arm/arm-cpu.h: Delete.
18831 * config/arm/arm-cpu-cdata.h: Delete.
18832 * config/arm/arm-cpu-data.h: Delete.
18834 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18836 * config/arm/arm-cpus.in (cortex-a55): New.
18837 (cortex-a75): Likewise.
18838 (cortex-a75.cortex-a55): Likewise.
18839 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18841 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18842 * config/arm/arm-cpu-cdata.h: Regenerate.
18843 * config/arm/arm-cpu-data.h: Regenerate.
18844 * config/arm/arm-cpu.h: Regenerate.
18845 * config/arm/arm-tables.opt: Regenerate.
18846 * config/arm/arm-tune.md: Regenerate.
18848 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18850 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18852 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18854 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18857 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18859 PR tree-optimization/81292
18860 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18861 full_string_p, also call adjust_related_strinfos if the adjustment
18862 is simple, otherwise invalidate related strinfos.
18864 2017-07-04 Martin Liska <mliska@suse.cz>
18867 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18868 newly created variable as DECL_IGNORED_P.
18870 2017-07-04 Martin Liska <mliska@suse.cz>
18873 * ipa-inline.c (inline_small_functions):
18874 Use xstrdup_for_dump.
18876 2017-07-04 Tom de Vries <tom@codesourcery.com>
18878 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18880 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18883 * config/darwin.c (darwin_function_switched_text_sections):
18886 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18888 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18890 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18892 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18894 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18896 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18897 min_profitable_iters, and th as inclusive lower bounds.
18898 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18899 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18900 for min_profitable_iters and min_profitable_estimate.
18901 (vect_transform_loop): Treat th as an inclusive lower bound.
18902 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18904 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18907 * config/darwin.c (darwin_function_switched_text_sections):
18908 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18909 in two pieces, and suppress the use of buf.
18911 2017-07-03 Nathan Sidwell <nathan@acm.org>
18913 * hash-table.h (hash_table_mod1): Fix indentation.
18915 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18917 PR middle-end/81290
18918 * predict.c (force_edge_cold): Be more careful about propagation
18920 * profile-count.h (profile_probability::guessed,
18921 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18923 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18925 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18927 * doc/invoke.texi (rcpc architecture extension): Document it.
18929 2017-07-03 Richard Biener <rguenther@suse.de>
18931 PR tree-optimization/60510
18932 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18933 the scalar reduction PHI and use it.
18934 (vectorizable_reduction): Properly guard the single_defuse_cycle
18935 path for non-SLP reduction chains where we cannot use it.
18936 Rework reduc_def/index and vector type deduction. Rework
18937 vector operand gathering during reduction op code-gen.
18938 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18939 chains dissolve the chain and leave it to non-SLP reduction
18942 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18944 * tree-data-ref.h (dr_alignment): Declare.
18945 * tree-data-ref.c (dr_alignment): New function.
18946 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18947 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18949 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18951 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18953 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18954 and base_misalignment fields.
18955 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18956 * tree-data-ref.c: Include builtins.h.
18957 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18958 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18959 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18960 * tree-vect-data-refs.c: Include tree-cfg.h.
18961 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18962 fields instead of calculating an alignment here.
18963 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18964 innermost_loop_behavior fields.
18966 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18968 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18970 (DR_STEP_ALIGNMENT): New macro.
18971 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18972 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18973 (create_data_ref): Print it.
18974 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18975 to tell whether the step preserves vector (mis)alignment.
18976 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18977 Move the check for an integer step and generalise to all INTEGER_CST.
18978 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18979 Print the outer step alignment.
18981 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18983 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18984 with offset_alignment.
18985 (DR_ALIGNED_TO): Delete.
18986 (DR_OFFSET_ALIGNMENT): New macro.
18987 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18988 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18989 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18990 (create_data_ref): Likewise.
18991 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18992 (vect_analyze_data_refs): Likewise.
18993 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18994 creating dummy innermost behavior.
18996 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18998 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18999 with a "innermost_loop_behavior *" and refeence tree.
19000 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19001 (create_data_ref): Update call accordingly.
19002 * tree-predcom.c (find_looparound_phi): Likewise.
19004 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19006 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19007 fields with dr_wrt_vec_loop.
19008 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19009 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19010 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19011 (vect_dr_behavior): New function.
19012 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19013 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19014 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19015 track whether the step preserves the misalignment.
19016 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19017 Use vect_dr_behavior.
19018 (vect_setup_realignment): Update call accordingly.
19019 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19020 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19021 call to vect_create_addr_base_for_vector_ref.
19022 (vect_create_cond_for_align_checks): Likewise.
19023 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19024 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19025 (vect_recog_mask_conversion_pattern): Likewise.
19026 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19027 (new_stmt_vec_info): Remove redundant zeroing.
19029 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19031 * common/config/arm/arm-common.c (arm_be8_option): New function.
19032 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19033 (ISA_ARMv6): Add isa_bit_be8.
19034 * config/arm/arm.h (arm_be8_option): Add prototype.
19035 (BE8_SPEC_FUNCTION): New define.
19036 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19037 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19038 (mlittle-endian): Similarly.
19039 (mbe8, mbe32): New options.
19040 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19041 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19043 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19045 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19047 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19049 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19050 (cleanup_tree_cfg_bb): Use it.
19051 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19053 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19055 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19058 * loop-doloop.c (add_test): Update profile.
19060 2017-07-03 Martin Liska <mliska@suse.cz>
19063 * sanopt.c (rewrite_usage_of_param): New function.
19064 (sanitize_rewrite_addressable_params): Likewise.
19065 (pass_sanopt::execute): Call rewrite_usage_of_param.
19067 2017-07-03 Richard Biener <rguenther@suse.de>
19069 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19070 back to using VIEW_CONVERT_EXPR.
19072 2017-07-03 Martin Liska <mliska@suse.cz>
19075 * doc/extend.texi: Document when a resolver function is
19076 generated for target_clones.
19078 2017-07-03 Martin Liska <mliska@suse.cz>
19080 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19081 and shadow memory of auto variables which are subject of
19082 use-after-scope sanitization.
19083 (asan_expand_mark_ifn): Add do set only when is_poison.
19085 2016-07-03 Richard Biener <rguenther@suse.de>
19087 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19089 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19090 (vectorizable_reduction): Perform reduction PHI creation when
19091 visiting a reduction PHI and adjust and simplify code generation
19092 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19093 (vect_transform_loop): Visit reduction PHIs.
19094 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19095 defs into the SLP tree.
19096 (vect_build_slp_tree): Reduction defs terminate the recursion.
19097 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19099 (vect_get_vec_defs_for_stmt_copy): Export.
19100 (vect_get_vec_defs): Likewise.
19101 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19103 (vect_get_vec_defs_for_stmt_copy): Declare.
19104 (vect_get_vec_defs): Likewise.
19106 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19108 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19109 parameter with a "loop" parameter and use it instead of the
19110 loop containing DR_STMT. Don't check simple_iv when doing
19111 BB analysis. Describe the two analysis modes in the comment.
19113 2017-07-03 Tom de Vries <tom@codesourcery.com>
19115 PR tree-optimization/69468
19116 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19117 (find_same_succ_bb): Handle ignore_edge_flags.
19119 2017-07-03 Tom de Vries <tom@codesourcery.com>
19121 PR tree-optimization/81192
19122 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19124 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19126 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19128 2017-07-03 Tom de Vries <tom@codesourcery.com>
19130 PR tree-optimization/81192
19131 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19132 BB_SAME_SUCC (bb) == NULL.
19134 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19136 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19139 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19141 * dumpfile.c: Include profile-count.h
19142 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19144 (insert_cond_bb): Update profile.
19145 * tree-cfg.h (insert_cond_bb): Update prototype.
19146 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19147 * tree-dump.c: Do not include tree-cfg.
19149 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19151 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19153 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19155 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19158 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19160 * tree-complex.c (expand_complex_div_wide): update profile.
19162 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19163 Alan Hayward <alan.hayward@arm.com>
19164 David Sherwood <david.sherwood@arm.com>
19166 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19167 (CORETYPES_H): New define.
19168 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19169 (insn-modes-inline.h, s-modes-inline-h): New rules.
19170 (generated_files): Add insn-modes-inline.h.
19171 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19172 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19173 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19174 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19175 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19176 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19177 (build/gencodes.o, build/genconditions.o): Likewise.
19178 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19179 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19180 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19181 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19182 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19183 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19184 * coretypes.h: Include everything up to real.h for generators.
19185 Include insn-modes.h first. Include wide-int-print.h after
19186 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19187 * machmode.h: Don't include insn-modes.h here.
19188 * function-tests.c: Remove includes of signop.h, machmode.h,
19189 double-int.h and wide-int.h.
19191 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19193 * optc-save-gen.awk: Likewise.
19194 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19195 * godump.c: Remove include of wide-int-print.h.
19196 * pretty-print.h: Likewise.
19197 * wide-int-print.cc: Likewise.
19198 * wide-int.cc: Likewise.
19199 * hash-map-tests.c: Remove include of signop.h.
19200 * hash-set-tests.c: Likewise.
19201 * rtl-tests.c: Likewise.
19202 * mkconfig.sh: Remove include of machmode.h.
19203 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19205 (emit_insn_modes_inline_h): ...this new function. Emit the code
19206 into an insn-modes-inline.h header file, adding appropriate
19207 include guards and end comments.
19208 (emit_insn_modes_c_header): Remove include of machmode.h.
19209 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19211 (main): Handle -i flag and call emit_insn_modes_inline_h when
19214 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19216 * tree-ssa-strlen.c (strinfo): Rename the length field to
19217 nonzero_chars. Add a full_string_p field.
19218 (compare_nonzero_chars, zero_length_string_p): New functions.
19219 (get_addr_stridx): Add an offset_out parameter.
19220 Use compare_nonzero_chars.
19221 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19222 (new_strinfo): Update after above changes to strinfo.
19223 (set_endptr_and_length): Set full_string_p.
19224 (get_string_length): Update after above changes to strinfo.
19225 (unshare_strinfo): Update call to new_strinfo.
19226 (maybe_invalidate): Likewise.
19227 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19228 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19229 as a uhwi instead of an shwi. Update after above changes to
19230 strinfo and new_strinfo.
19231 (zero_length_string): Assert that chainsi contains full strings.
19232 Use zero_length_string_p. Update call to new_strinfo.
19233 (adjust_related_strinfos): Update after above changes to strinfo.
19234 Copy full_string_p from origsi.
19235 (adjust_last_stmt): Use zero_length_string_p.
19236 (handle_builtin_strlen): Update after above changes to strinfo and
19237 new_strinfo. Install the lhs as the string length if the previous
19238 entry didn't describe a full string.
19239 (handle_builtin_strchr): Update after above changes to strinfo
19241 (handle_builtin_strcpy): Likewise.
19242 (handle_builtin_strcat): Likewise.
19243 (handle_builtin_malloc): Likewise.
19244 (handle_pointer_plus): Likewise.
19245 (handle_builtin_memcpy): Likewise. Track nonzero characters
19246 that aren't necessarily followed by a nul terminator.
19247 (handle_char_store): Likewise.
19249 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19251 PR tree-optimization/80769
19252 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19253 for malloc and calloc. Document the new invariant that all related
19254 strinfos have delayed lengths or none do.
19255 (verify_related_strinfos): Move earlier in file.
19256 (set_endptr_and_length): New function, split out from...
19257 (get_string_length): ...here. Also set the lengths of related
19259 (zero_length_string): Assert that chainsi has known (rather than
19261 (adjust_related_strinfos): Likewise.
19263 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19265 PR tree-optimization/81136
19266 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19267 assert that two references with the same misalignment have the same
19268 compile-time misalignment if those compile-time misalignments
19271 2017-07-01 Andi Kleen <ak@linux.intel.com>
19273 * print-tree.c (print_node): Print all attributes.
19275 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19277 * cfg.c (scale_bbs_frequencies): New function.
19278 * cfg.h (scale_bbs_frequencies): Declare it.
19279 * cfgloopanal.c (single_likely_exit): Cleanup.
19280 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19282 (scale_loop_profile): Likewise.
19283 (loop_version): Likewise.
19284 (create_empty_loop_on_edge): Update.
19285 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19286 scale_loop_frequencies, scale_loop_profile, loopify,
19287 loop_version): Update prototypes.
19288 * modulo-sched.c (sms_schedule): Update.
19289 * predict.c (unlikely_executed_edge_p): Also check probability.
19290 (probably_never_executed_edge_p): Fix typo.
19291 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19292 * tree-parloops.c (gen_parallel_loop): Update.
19293 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19294 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19295 * tree-ssa-loop-split.c (split_loop): Update.
19296 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19297 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19298 (vect_loop_versioning): Update.
19299 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19301 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19303 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19305 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19307 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19308 to keep profile consistent.
19310 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19312 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19313 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19314 * profile-count.h (max_safe_multiplier): Make unsigned.
19315 (profile_count::guessed_zero): New.
19317 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19319 * bb-reorder.c (fix_up_crossing_landing_pad,
19320 fix_crossing_conditional_branches): Use make_single_succ_edge
19321 to keep profile consistent.
19323 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19325 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19328 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19331 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19332 the right scopes, make sure cond_jump isn't preserved between multiple
19333 iterations. Search for fallthru edge whenever there are 3+ edges and
19334 use find_fallthru_edge for it.
19336 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19338 Patch by Alexander Monakov <amonakov@ispras.ru>
19339 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19340 probabilities consistently.
19342 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19344 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19345 emit_cmp_and_jump_insns.
19347 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19350 * tree-inline.c (expand_call_inline): Combine profile statuses.
19352 2017-06-30 Andrew Pinski <apinski@cavium.com>
19354 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19355 fold_stmt returned true.
19357 2017-06-30 Nathan Sidwell <nathan@acm.org>
19359 * ggc.h (empty_string): Delete.
19360 * cfgexpand.c (expand_asm_stmt): Use plain "".
19361 * optabs.c (expand_asm_memory_barrier): Likewise.
19362 * stringpool.c (empty_string): Delete.
19363 (digit_vector, digit_string): Delete.
19364 (ggc_alloc_string): Use plain "", don't optimize single digit
19365 strings. Use ggc_alloc_atomic.
19367 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19369 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19370 comparison set and one other set, use the cost of the non-comparison
19373 2017-06-30 Nathan Sidwell <nathan@acm.org>
19375 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19378 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19380 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19381 loops. Remove now unneeded calls to gimple_switch_set_label() that
19382 just set removed labels to NULL_TREE.
19384 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19386 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19387 (set_range_info): ...here. Only call set_range_info_raw if domain
19389 (set_nonzero_bits): Call set_range_info_raw.
19390 * tree-ssanames.h (set_range_info_raw): New.
19392 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19395 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19396 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19397 of nonimmediate_operand and <store_mask_constraint> instead of m
19398 for the input operand. For V8FI iterator, always split if input
19399 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19400 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19401 <store_mask_predicate> instead of register_operand and
19402 <store_mask_constraint> instead of v for the input operand. Make
19403 sure both operands aren't MEMs for if not <mask_applied>.
19405 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19407 * lto-wrapper.c (copy_file) Close both file descriptors before
19410 2017-06-30 Martin Liska <mliska@suse.cz>
19413 * multiple_target.c (create_dispatcher_calls): Make ifunc
19414 also for function that don't have calls or are not referenced.
19416 2017-06-30 Richard Biener <rguenther@suse.de>
19418 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19419 analyze the first scalar stmt. Move vector type computation
19420 for the BB case here from ...
19421 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19422 live operation processing in the SLP case properly.
19424 2017-06-30 Richard Biener <rguenther@suse.de>
19426 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19428 2017-06-30 Martin Liska <mliska@suse.cz>
19431 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19432 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19434 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19436 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19439 2017-06-30 Richard Biener <rguenther@suse.de>
19441 PR tree-optimization/81249
19442 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19443 condition reduction result to original scalar type.
19445 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19447 * profile-count.h (enum profile_quality): Fix typos and whitespace
19450 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19452 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19453 type for branch probabilities.
19455 2017-06-29 Julian Brown <julian@codesourcery.com>
19456 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19458 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19459 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19460 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19461 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19463 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19465 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19466 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19467 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19468 CC usage from generic code to here.
19469 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19470 CC usage into the target macros.
19472 2017-06-29 Maya Rashish <coypu@sdf.org>
19474 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19477 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19479 * arm/arm-builtins.c: Include profile-count.h
19480 * except.c (sjlj_emit_function_enter): Use
19481 profile_probability::unlikely.
19483 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19485 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19486 and tocrel_offset be pointer args rather than implicitly using
19488 (legitimate_constant_pool_address_p, rs6000_emit_move,
19489 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19490 tocrel_offset and use in toc_relative_expr_p call.
19491 (print_operand, print_operand_address): Use static tocrel_base_oac
19492 and tocrel_offset_oac.
19493 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19496 2017-06-29 Maya Rashish <coypu@sdf.org>
19498 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19500 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19502 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19503 objects, take into account only the alignment of 'op0' and 'mode1' if
19506 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19508 * ccmp.c (ccmp_tree_comparison_p): New function.
19509 (ccmp_candidate_p): Update to use above function.
19510 (get_compare_parts): New function.
19511 (expand_ccmp_next): Update to use new functions.
19512 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19514 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19515 take mode as argument.
19516 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19517 * expr.c (expand_expr_real_1): Pass mode as argument.
19519 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19521 * combine.c (combine_instructions): Print insns to dump_file, together
19524 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19526 * asan.c (asan_emit_stack_protection): Update.
19527 (create_cond_insert_point): Update.
19528 * auto-profile.c (afdo_propagate_circuit): Update.
19529 * basic-block.h (struct edge_def): Turn probability to
19530 profile_probability.
19531 (EDGE_FREQUENCY): Update.
19532 * bb-reorder.c (find_traces_1_round): Update.
19533 (better_edge_p): Update.
19534 (sanitize_hot_paths): Update.
19535 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19536 (make_single_succ_edge): Update.
19537 (check_bb_profile): Update.
19538 (dump_edge_info): Update.
19539 (update_bb_profile_for_threading): Update.
19540 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19541 probabilitycount to 0.
19542 * cfgbuild.c (compute_outgoing_frequencies): Update.
19543 * cfgcleanup.c (try_forward_edges): Update.
19544 (outgoing_edges_match): Update.
19545 (try_crossjump_to_edge): Update.
19546 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19547 (expand_gimple_tailcall): Update.
19548 (construct_init_block): Use make_single_succ_edge.
19549 (construct_exit_block): Use make_single_succ_edge.
19550 * cfghooks.c (verify_flow_info): Update.
19551 (redirect_edge_succ_nodup): Update.
19552 (split_edge): Update.
19553 (account_profile_record): Update.
19554 * cfgloopanal.c (single_likely_exit): Update.
19555 * cfgloopmanip.c (scale_loop_profile): Update.
19556 (set_zero_probability): Remove.
19557 (duplicate_loop_to_header_edge): Update.
19558 * cfgloopmanip.h (loop_version): Update prototype.
19559 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19560 (force_nonfallthru_and_redirect): Update.
19561 (update_br_prob_note): Update.
19562 (rtl_verify_edges): Update.
19563 (purge_dead_edges): Update.
19564 (rtl_lv_add_condition_to_bb): Update.
19565 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19566 * cgraphunit.c (init_lowered_empty_function): Update.
19567 (cgraph_node::expand_thunk): Update.
19568 * cilk-common.c: Include profile-count.h
19569 * dojump.c (inv): Remove.
19570 (jumpifnot): Update.
19571 (jumpifnot_1): Update.
19572 (do_jump_1): Update.
19574 (do_jump_by_parts_greater_rtx): Update.
19575 (do_compare_rtx_and_jump): Update.
19576 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19577 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19578 * dwarf2cfi.c: Include profile-count.h
19579 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19580 (sjlj_emit_dispatch_table): Likewise.
19581 * explow.c: Include profile-count.h
19582 * expmed.c (emit_store_flag_force): Update.
19583 (do_cmp_and_jump): Update.
19584 * expr.c (compare_by_pieces_d::generate): Update.
19585 (compare_by_pieces_d::finish_mode): Update.
19586 (emit_block_move_via_loop): Update.
19587 (store_expr_with_bounds): Update.
19588 (store_constructor): Update.
19589 (expand_expr_real_2): Update.
19590 (expand_expr_real_1): Update.
19591 * expr.h (try_casesi, try_tablejump): Update prototypes.
19592 * gimple-pretty-print.c (dump_probability): Update.
19593 (dump_profile): New.
19594 (dump_gimple_label): Update.
19595 (dump_gimple_bb_header): Update.
19596 * graph.c (draw_cfg_node_succ_edges): Update.
19597 * hsa-gen.c (convert_switch_statements): Update.
19598 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19599 (find_if_case_1): Update.
19600 (find_if_case_2): Update.
19601 * internal-fn.c (expand_arith_overflow_result_store): Update.
19602 (expand_addsub_overflow): Update.
19603 (expand_neg_overflow): Update.
19604 (expand_mul_overflow): Update.
19605 (expand_vector_ubsan_overflow): Update.
19606 * ipa-cp.c (good_cloning_opportunity_p): Update.
19607 * ipa-split.c (split_function): Use make_single_succ_edge.
19608 * ipa-utils.c (ipa_merge_profiles): Update.
19609 * loop-doloop.c (add_test): Update.
19610 (doloop_modify): Update.
19611 * loop-unroll.c (compare_and_jump_seq): Update.
19612 (unroll_loop_runtime_iterations): Update.
19613 * lra-constraints.c (lra_inheritance): Update.
19614 * lto-streamer-in.c (input_cfg): Update.
19615 * lto-streamer-out.c (output_cfg): Update.
19616 * mcf.c (adjust_cfg_counts): Update.
19617 * modulo-sched.c (sms_schedule): Update.
19618 * omp-expand.c (expand_omp_for_init_counts): Update.
19619 (extract_omp_for_update_vars): Update.
19620 (expand_omp_ordered_sink): Update.
19621 (expand_omp_for_ordered_loops): Update.
19622 (expand_omp_for_generic): Update.
19623 (expand_omp_for_static_nochunk): Update.
19624 (expand_omp_for_static_chunk): Update.
19625 (expand_cilk_for): Update.
19626 (expand_omp_simd): Update.
19627 (expand_omp_taskloop_for_outer): Update.
19628 (expand_omp_taskloop_for_inner): Update.
19629 * omp-simd-clone.c (simd_clone_adjust): Update.
19630 * optabs.c (expand_doubleword_shift): Update.
19631 (expand_abs): Update.
19632 (emit_cmp_and_jump_insn_1): Update.
19633 (expand_compare_and_swap_loop): Update.
19634 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19635 * predict.c (predictable_edge_p): Update.
19636 (edge_probability_reliable_p): Update.
19637 (set_even_probabilities): Update.
19638 (combine_predictions_for_insn): Update.
19639 (combine_predictions_for_bb): Update.
19640 (propagate_freq): Update.
19641 (estimate_bb_frequencies): Update.
19642 (force_edge_cold): Update.
19643 * profile-count.c (profile_count::dump): Add missing space into dump.
19644 (profile_count::debug): Add newline.
19645 (profile_count::differs_from_p): Explicitly convert to unsigned.
19646 (profile_count::stream_in): Update.
19647 (profile_probability::dump): New member function.
19648 (profile_probability::debug): New member function.
19649 (profile_probability::differs_from_p): New member function.
19650 (profile_probability::differs_lot_from_p): New member function.
19651 (profile_probability::stream_in): New member function.
19652 (profile_probability::stream_out): New member function.
19653 * profile-count.h (profile_count_quality): Rename to ...
19654 (profile_quality): ... this one.
19655 (profile_probability): New.
19656 (profile_count): Update.
19657 * profile.c (compute_branch_probabilities): Update.
19658 * recog.c (peep2_attempt): Update.
19659 * sched-ebb.c (schedule_ebbs): Update.
19660 * sched-rgn.c (find_single_block_region): Update.
19661 (compute_dom_prob_ps): Update.
19662 (schedule_region): Update.
19663 * sel-sched-ir.c (compute_succs_info): Update.
19664 * stmt.c (struct case_node): Update.
19665 (do_jump_if_equal): Update.
19666 (get_outgoing_edge_probs): Update.
19667 (conditional_probability): Update.
19668 (emit_case_dispatch_table): Update.
19669 (expand_case): Update.
19670 (expand_sjlj_dispatch_table): Update.
19671 (emit_case_nodes): Update.
19672 * targhooks.c: Update.
19673 * tracer.c (better_p): Update.
19674 (find_best_successor): Update.
19675 * trans-mem.c (expand_transaction): Update.
19676 * tree-call-cdce.c: Update.
19677 * tree-cfg.c (gimple_split_edge): Upate.
19678 (move_sese_region_to_fn): Upate.
19679 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19680 * tree-eh.c (lower_resx): Upate.
19681 (cleanup_empty_eh_move_lp): Upate.
19682 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19683 * tree-inline.c (copy_edges_for_bb): Update.
19684 (copy_cfg_body): Update.
19685 * tree-parloops.c (gen_parallel_loop): Update.
19686 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19687 (gimple_gen_time_profiler): Update.
19688 * tree-ssa-dce.c (remove_dead_stmt): Update.
19689 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19690 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19691 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19692 (unloop_loops): Update.
19693 (try_peel_loop): Update.
19694 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19695 * tree-ssa-loop-split.c (connect_loops): Update.
19696 (split_loop): Update.
19697 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19698 (hoist_guard): Update.
19699 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19700 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19701 (value_replacement): Update.
19702 * tree-ssa-reassoc.c (branch_fixup): Update.
19703 * tree-ssa-tail-merge.c (replace_block_by): Update.
19704 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19705 (create_edge_and_update_destination_phis): Update.
19706 (compute_path_counts): Update.
19707 (recompute_probabilities): Update.
19708 (update_joiner_offpath_counts): Update.
19709 (freqs_to_counts_path): Update.
19710 (duplicate_thread_path): Update.
19711 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19712 (struct switch_conv_info): Update.
19713 (gen_inbound_check): Update.
19714 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19715 (vect_do_peeling): Update.
19716 (vect_loop_versioning): Update.
19717 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19718 (optimize_mask_stores): Update.
19719 * ubsan.c (ubsan_expand_null_ifn): Update.
19720 * value-prof.c (gimple_divmod_fixed_value): Update.
19721 (gimple_divmod_fixed_value_transform): Update.
19722 (gimple_mod_pow2): Update.
19723 (gimple_mod_pow2_value_transform): Update.
19724 (gimple_mod_subtract): Update.
19725 (gimple_mod_subtract_transform): Update.
19726 (gimple_ic): Update.
19727 (gimple_stringop_fixed_value): Update.
19728 (gimple_stringops_transform): Update.
19729 * value-prof.h: Update.
19731 2017-06-29 Carl Love <cel@us.ibm.com>
19733 * config/rs6000/rs6000-c.c: Add support for built-in functions
19734 vector signed int vec_signed (vector float);
19735 vector signed long long vec_signed (vector double);
19736 vector signed int vec_signed2 (vector double, vector double);
19737 vector signed int vec_signede (vector double);
19738 vector signed int vec_signedo (vector double);
19739 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19740 instruction generator.
19741 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19742 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19743 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19745 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19746 vunsignede_v2df): Add define_expands.
19747 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19748 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19749 VEC_UNSIGNEDO): Add definitions.
19750 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19751 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19752 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19753 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19754 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19755 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19756 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19757 vec_unsignede, vec_unsignedo): Add builtin defines.
19758 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19760 * doc/extend.texi: Update the built-in documentation file for the
19761 new built-in functions.
19763 2017-06-29 Richard Biener <rguenther@suse.de>
19765 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19766 reduction chains to LOOP_VINFO_REDUCTIONS.
19767 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19768 SLP reductions after processing reduction chains.
19770 2017-06-29 Nathan Sidwell <nathan@acm.org>
19772 * builtins.c (fold_builtin_FUNCTION): Use
19773 lang_hooks.decl_printable_name.
19775 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19777 PR middle-end/81194
19778 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19779 with only one label.
19780 * stmt.c (expand_case): Assert NCASES is greater than one.
19782 2017-06-29 Richard Biener <rguenther@suse.de>
19784 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19786 (group_case_labels): Likewise.
19787 (find_taken_edge): Push sanity checking on val to workers...
19788 (find_taken_edge_cond_expr): ... here
19789 (find_taken_edge_switch_expr): ... and here, handle cases
19790 with just a default label.
19791 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19792 (group_case_labels): Likewise.
19793 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19794 group_case_labels does anything cleanup the CFG again.
19796 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19798 PR tree-optimization/81196
19799 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19800 exit condition comparing two IVs.
19802 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19804 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19805 profile to the dummy entry at the end of the list of architectures.
19806 * config/arm/arm-cpu-cdata.h: Regenerated.
19808 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19809 Michael Collison <michael.collison@arm.com>
19812 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19814 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19815 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19816 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19817 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19818 of shift when the shift amount is masked with constant equal to
19819 the size of the mode.
19820 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19823 2017-06-29 Martin Liska <mliska@suse.cz>
19825 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19827 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19830 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19833 (edge_growth_cache_entry::edge_growth_cache_entry): New
19835 (reset_edge_growth_cache): Update to use constructor.
19837 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19839 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19840 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19841 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19843 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19845 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19846 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19848 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19850 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19851 (*-linux-uclibc*): Add t-uclibc tmake_file.
19852 * config/t-musl: New.
19853 * config/t-uclibc: New.
19855 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19857 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19859 (gen_comm_data): Emit architectural setting of arch_prof.
19860 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19862 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19863 (armv8-m.base, armv8-m.main): Likewise.
19864 * arm-protos.h (arm_build_target): Add profile field.
19865 (arch_option): Likewise.
19866 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19868 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19869 arm_active_target.profile.
19871 2017-06-28 Richard Biener <rguenther@suse.de>
19873 PR middle-end/81227
19874 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19875 TYPE_OVERFLOW_WRAPS.
19876 * match.pd (negate_expr_p): Likewise.
19877 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19878 fold_build2, not fold_binary.
19880 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19882 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19883 Convert memory address to Pmode.
19884 (aarch64_print_operand): Assert MEM operands are always Pmode.
19886 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19889 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19890 Remove redundant if.
19891 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19892 * config/arm/aarch-common-protos.h
19893 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19894 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19896 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19899 * multiple_target.c (create_dispatcher_calls): Set the default
19900 clone to be static, not public.
19902 2017-06-28 Richard Biener <rguenther@suse.de>
19904 * tree-vect-loop.c (vectorizable_reduction): Move special
19905 cond reduction IV var creation ...
19906 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19907 parameter. Use STMT_VINFO_VECTYPE.
19908 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19911 2017-06-28 Martin Liska <mliska@suse.cz>
19914 * ipa-visibility.c (non_local_p): Handle visibility.
19916 2017-06-28 Martin Liska <mliska@suse.cz>
19919 * common.opt: Add IntegerRange to various options.
19920 * opt-functions.awk (integer_range_info): New function.
19921 * optc-gen.awk: Add integer_range_info to cl_options struct.
19922 * opts-common.c (decode_cmdline_option): Handle
19923 CL_ERR_INT_RANGE_ARG.
19924 (cmdline_handle_error): Likewise.
19925 * opts.c (print_filtered_help): Show valid interval in
19926 when --help is provided.
19927 * opts.h (struct cl_option): Add range_min and range_max fields.
19928 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19930 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19932 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19933 (x * C EQ/NE y * C): New transformation.
19935 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19937 * genmultilib (combination_space): Accept '+' in option names.
19939 2017-06-28 Martin Liska <mliska@suse.cz>
19942 * asan.c (instrument_derefs): Bail out inner references
19943 that are hard register variables.
19945 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19948 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19949 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19951 2017-06-28 Richard Biener <rguenther@suse.de>
19953 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19954 (vect_get_slp_defs): Adjust.
19955 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19957 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19959 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19960 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19961 (vectorizable_reduction): Adjust.
19962 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19964 (vect_get_slp_defs): Likewise.
19965 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19966 (vectorizable_bswap): Adjust.
19967 (vectorizable_call): Likewise.
19968 (vectorizable_conversion): Likewise.
19969 (vectorizable_assignment): Likewise.
19970 (vectorizable_shift): Likewise.
19971 (vectorizable_operation): Likewise.
19972 (vectorizable_store): Likewise.
19973 (vectorizable_condition): Likewise.
19974 (vectorizable_comparison): Likewise.
19976 2017-06-28 Michael Collison <michael.collison@arm.com>
19979 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19981 (arm_gen_movmemqi): Removed unused variable 'i'.
19982 Convert 'for' loop into 'while' loop.
19983 (arm_expand_prologue): Remove last unnecessary set of insn.
19984 (thumb_pop): Remove unused variable 'pushed_words'.
19985 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19987 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19989 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19990 * config/s390/s390-protos.h: Add prototype of
19991 s390_rel_address_ok_p.
19992 * config/s390/s390.c (s390_got_symbol): New function.
19993 (s390_rel_address_ok_p): New function.
19994 (legitimize_pic_address): Use s390_rel_address_ok_p.
19995 (s390_load_got): Use s390_got_symbol.
19996 (s390_option_override): Issue error if
19997 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19998 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20000 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20002 2017-06-27 Andrew Pinski <apinski@cavium.com>
20004 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20005 (X * copysign (1.0, X)): New pattern.
20006 (X * copysign (1.0, -X)): New pattern.
20007 (copysign (-1.0, CST)): New pattern.
20009 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20011 * genmultilib (combination_space): Remove variable.
20012 Validate reuse rules against regular expression for any sequence
20013 of multilib options in any order.
20015 2017-06-27 Michael Collison <michael.collison@arm.com>
20017 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20018 call aarch64_split_simd_combine.
20019 * (aarch64_combine_internal<mode>): Delete pattern.
20020 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20021 Allow register and subreg operands.
20023 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20025 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20026 specific need, just fallback on defaults.
20027 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20029 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20030 Olivier Hainque <hainque@adacore.com>
20032 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20034 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20035 targets. Pick a default if no particular attempt applied.
20036 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20039 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20041 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20042 (x86_64-wrs-vxworks7): Likewise.
20044 2017-06-27 Marek Polacek <polacek@redhat.com>
20047 * ubsan.c (instrument_null): Check get_base_address's result for null.
20049 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20051 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20053 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20055 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20056 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20057 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20058 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20059 New function types.
20060 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20061 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20062 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20063 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20064 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20065 BUILT_IN_FEUPDATEENV): New builtins.
20066 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20067 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20068 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20069 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20071 (builtin_structptr_types): Adjust size.
20072 * tree.c (builtin_structptr_types): Add four entries.
20074 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20075 Olivier Hainque <hainque@adacore.com>
20077 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20078 (TLS_SYM): New local macro, forcing reference to __tls__ on
20079 link command lines for VxWorks 7 RTPs, triggering initialization
20081 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20082 OS features TLS support, true for RTPs on VxWorks 7.
20083 * config/vxworks.c (vxworks_override_options): Setup emutls
20086 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20088 * predict.c (test_prediction_value_range): Use -1U instead of -1
20089 to avoid narrowing conversion warning.
20090 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20091 to avoid narrowing conversion warning.
20092 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20094 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20096 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20098 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20099 64bit configurations.
20100 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20101 (SIZE_TYPE): Likewise.
20102 * config/vxworks.c (vxworks_emutls_var_fields): Use
20103 long_unsigned_type_node instead of unsigned_type_node as the offset
20104 field type, which is "pointer" mode in emutls.c.
20106 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20109 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20111 PR middle-end/81207
20112 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20113 gimple_vuse copying separately from gimple_vdef copying.
20115 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20117 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20120 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20121 Olivier Hainque <hainque@adacore.com>
20123 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20124 for all vxworks7 targets.
20125 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20126 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20127 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20128 variations for VX6/VX7 and 32/64bits later on in ...
20129 (VXWORKS_LIB_SPEC): Leverage new macros.
20130 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20131 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20133 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20135 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20136 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20138 2017-06-26 Carl Love <cel@us.ibm.com>
20140 * config/rs6000/rs6000-c.c: Add support for built-in functions
20141 vector bool char vec_reve (vector bool char);
20142 vector signed char vec_reve (vector signed char);
20143 vector unsigned char vec_reve (vector unsigned char);
20144 vector bool int vec_reve (vector bool int);
20145 vector signed int vec_reve (vector signed int);
20146 vector unsigned int vec_reve (vector unsigned int);
20147 vector bool long long vec_reve (vector bool long long);
20148 vector signed long long vec_reve (vector signed long long);
20149 vector unsigned long long vec_reve (vector unsigned long long);
20150 vector bool short vec_reve (vector bool short);
20151 vector signed short vec_reve (vector signed short);
20152 vector double vec_reve (vector double);
20153 vector float vec_reve (vector float);
20154 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20155 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20156 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20157 (altivec_vreve): New pattern.
20158 * config/rs6000/altivec.h (vec_reve): New define.
20159 * doc/extend.texi (vec_rev): Update the built-in documentation file
20160 for the new built-in functions.
20162 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20164 PR tree-optimization/71815
20165 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20167 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20169 (slsr_process_phi): Likewise.
20170 (replace_uncond_cands_and_profitable_phis): Don't replace a
20171 multiply candidate with a stride of 1 (copy or cast).
20172 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20174 (lowest_cost_path): Likewise.
20175 (total_savings): Likewise.
20177 2017-06-26 Richard Biener <rguenther@suse.de>
20180 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20181 Use def_builtin_pure for all gather builtins.
20183 2017-06-26 Richard Biener <rguenther@suse.de>
20185 PR tree-optimization/81203
20186 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20187 non-dominating BBs.
20189 2017-06-26 Marek Polacek <polacek@redhat.com>
20192 * doc/invoke.texi: Document -Wmultistatement-macros.
20194 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20196 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20197 arm_neon_ok_no_float_abi effective target.
20199 2017-06-26 Richard Biener <rguenther@suse.de>
20201 PR tree-optimization/80928
20202 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20203 (copy_bbs): Set BB_DUPLICATED flag early.
20204 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20206 (execute_on_shrinking_pred): Likewise.
20207 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20208 BB_DUPLICATED blocks.
20209 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20210 iterate over all PHIs considering removal of *gsi.
20212 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20214 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20217 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20219 * config/rs6000/rs6000-string.c: (expand_block_clear,
20220 do_load_for_compare, select_block_compare_mode,
20221 compute_current_alignment, expand_block_compare,
20222 expand_strncmp_align_check, expand_strn_compare,
20223 expand_block_move, rs6000_output_load_multiple)
20224 Move functions related to string/block move/compare
20225 to a separate file.
20226 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20227 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20228 for this function which is now used in two files.
20229 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20230 * config.gcc: Add rs6000-string.o to extra_objs for
20231 targets powerpc*-*-* and rs6000*-*-*.
20233 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20236 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20237 32-bit, since indexed is not valid for DImode.
20238 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20239 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20240 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20241 (define_peephole2 for Altivec d-form store): Likewise.
20244 * multiple_target.c (create_dispatcher_calls): Only create the
20245 dispatcher call if the function is the default clone of a
20246 versioned function.
20248 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20250 PR middle-end/80902
20251 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20252 a call, force the call to not be a tail call.
20254 2017-06-23 Jeff Law <law@redhat.com>
20256 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20259 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20261 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20262 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20263 handling for early expansion of vector shifts (sl,sr,sra,rl).
20264 (builtin_function_type): Add vector shift right instructions
20265 to the unsigned argument list.
20267 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20269 rtl-optimizatoin/79286
20270 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20271 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20272 trap. PIC register plus a const unspec without offset can never trap.
20274 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20276 * tree.h (builtin_structptr_type): New type.
20277 (builtin_structptr_types): Declare new array.
20278 * tree.c (builtin_structptr_types): New array.
20279 (free_lang_data, build_common_tree_nodes): Use it.
20281 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20284 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20287 2017-06-22 Matt Turner <mattst88@gmail.com>
20289 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20290 Lake models to skylake case. Assume skylake for unknown
20291 models with clflushopt.
20293 2017-06-22 Jeff Law <law@redhat.com>
20295 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20296 frame sizes that do not satisfy aarch64_uimm12_shift.
20298 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20300 * profile-count.h (apply_probability,
20301 apply_scale, probability_in): Fix checks for zero.
20303 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20305 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20306 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20308 2016-06-22 Richard Biener <rguenther@suse.de>
20310 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20311 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20312 REDUC_MAX_EXPR support.
20313 (vectorizable_reduction): Likewise.
20314 (vect_create_epilog_for_reduction): Likewise.
20316 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20318 * match.pd (A / (1 << B) -> A >> B): New.
20319 * generic-match-head.c: Include optabs-tree.h.
20320 * gimple-match-head.c: Likewise.
20321 * optabs-tree.h (target_supports_op_p): New.
20322 * optabs-tree.c (target_supports_op_p): New.
20324 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20326 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20327 $gcc_cv_ld --help output.
20328 (gcc_cv_ld_demangle): Likewise.
20329 (gcc_cv_ld_eh_frame_hdr): Likewise.
20330 (gcc_cv_ld_pie): Likewise.
20331 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20332 (gcc_cv_ld_buildid): Likewise.
20333 (gcc_cv_ld_sysroot): Likewise.
20334 (ld_bndplt_support): Likewise.
20335 (ld_pushpopstate_support): Likewise.
20336 * configure: Regenerate.
20337 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20339 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20342 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20343 operands indexes to avoid gap between operands and match_dups.
20345 2017-06-21 Andrew Pinski <apinski@cavium.com>
20347 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20348 Increment Arith_shift and Arith_shift_reg by 1.
20349 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20351 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20352 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20353 (aarch64_strip_extend): Add new argument and test for it.
20354 (aarch64_cheap_mult_shift_p): New function.
20355 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20356 add a cost if it is true.
20357 Update calls to aarch64_strip_extend.
20358 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20360 2017-06-21 Andrew Pinski <apinski@cavium.com>
20362 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20364 (thunderxt88): Likewise.
20365 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20366 (thunderx_prefetch_tune): New variable.
20367 (thunderx2t99_prefetch_tune): Update for the correct values.
20368 (thunderxt88_tunings): New variable.
20369 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20370 generic_prefetch_tune.
20371 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20373 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20375 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20376 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20377 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20378 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20379 (aarch64_atomic_cas<mode>, GPI): Likewise.
20381 2017-06-21 Martin Liska <mliska@suse.cz>
20383 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20384 statements on cold and hot labels.
20385 * predict.c (tree_estimate_probability_bb): Remove the
20386 prediction from this place.
20388 2017-06-21 Martin Liska <mliska@suse.cz>
20390 PR tree-optimization/79489
20391 * gimplify.c (maybe_add_early_return_predict_stmt): New
20393 (gimplify_return_expr): Call the function.
20394 * predict.c (tree_estimate_probability_bb): Remove handling
20396 * predict.def: Update comment about early return predictor.
20397 * gimple-predict.h (is_gimple_predict): New function.
20398 * predict.def: Change default value of early return to 66.
20399 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20401 * passes.def: Put pass_strip_predict_hints to the beginning of
20404 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20406 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20407 FUNCTION_DECL declarations.
20408 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20410 (dwaf2out_decl): Likewise.
20411 * godump.c (go_early_global_decl): Skip call to the real debug hook
20412 for FUNCTION_DECL declarations.
20413 * passes.c (rest_of_decl_compilation): Skip call to the
20414 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20415 -fdump-go-spec is passed.
20417 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20419 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20420 Reorder for compactness.
20421 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20422 (def_builtin_pure, def_builtin_pure2): New functions.
20423 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20425 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20427 * match.pd (nop_convert): New predicate.
20428 ((A +- CST1) +- CST2): Allow some NOP conversions.
20430 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20433 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20434 with ctors/dtors if GOVD_SHARED is set.
20436 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20438 * config/aarch64/aarch64.md (movti_aarch64):
20439 Emit mov rather than orr.
20440 (movtf_aarch64): Likewise.
20441 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20442 Emit mov rather than orr.
20444 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20446 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20447 Swap alternatives, make integer dup more expensive.
20449 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20451 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20452 Return true for non-tls symbols.
20454 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20456 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20457 (cortex-a75): Likewise.
20458 (cortex-a75.cortex-a55): Likewise.
20459 * config/aarch64/aarch64-tune.md: Regenerate.
20460 * doc/invoke.texi (-mtune): Document new values for -mtune.
20462 2017-06-21 Tom de Vries <tom@codesourcery.com>
20464 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20465 stack_size feature.
20466 (Effective-Target Keywords, Other attributes): Suggest using
20467 dg-add-options stack_size feature to get stack limit in stack_size
20468 effective target documentation.
20470 2017-06-21 Julian Brown <julian@codesourcery.com>
20471 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20473 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20474 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20475 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20477 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20478 attribute type list for neon_multiply.
20479 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20480 attribute type list for neon_multiply.
20481 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20482 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20483 attribute type list for neon_multiply.
20484 * config/arm/types.md (crypto_pmull): Add.
20485 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20486 attribute type list.
20488 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20490 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20493 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20495 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20496 to make sure not to dereference a NULL cost_classes_ptr pointer.
20498 2017-06-20 Carl Love <cel@us.ibm.com>
20500 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20501 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20502 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20503 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20504 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20505 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20506 VMULOSW): New enum "unspec" values.
20507 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20508 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20509 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20510 altivec_vmulosw): New patterns.
20511 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20512 VMULOSW): Add definitions.
20514 2017-06-20 Julia Koval <julia.koval@intel.com>
20516 * config/i386/i386.c: Fix rounding expand for new pattern.
20517 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20519 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20521 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20522 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20524 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20526 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20529 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20531 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20532 architecture, then by implementer ID.
20533 * config/aarch64/aarch64-tune.md: Regenerate.
20535 2017-06-20 Richard Biener <rguenther@suse.de>
20537 PR middle-end/81097
20538 * fold-const.c (split_tree): Fold to type before negating.
20540 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20542 * diagnostic-show-locus.c
20543 (selftest::test_fixit_deletion_affecting_newline): New function.
20544 (selftest::diagnostic_show_locus_c_tests): Call it.
20546 2017-06-20 Andreas Schwab <schwab@suse.de>
20549 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20552 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20554 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20555 __ARM_FEATURE_COPROC according to support.
20557 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20559 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20560 Rewritten to avoid overflow for > 32-bit pointers.
20563 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20564 by removing enum keyword.
20565 (ubsan_type_descriptor): Likewise. Formatting fix.
20568 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20569 splitter): Require TARGET_SSE2 in the condition.
20571 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20574 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20575 for doing vector set of SFmode on ISA 3.0.
20576 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20577 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20579 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20580 SFmode value into a V4SF variable that was extracted from another
20581 V4SF variable without converting the element to double precision
20582 and back to single precision vector format.
20583 (vsx_insert_extract_v4sf_p9_2): Likewise.
20585 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20587 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20588 in UWHI to avoid undefined overflow.
20591 * ubsan.h (enum ubsan_encode_value_phase): New.
20592 (ubsan_encode_value): Change second argument to
20593 enum ubsan_encode_value_phase with default value of
20594 UBSAN_ENCODE_VALUE_GENERIC.
20595 * ubsan.c (ubsan_encode_value): Change second argument to
20596 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20597 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20598 create_tmp_var_raw instead of create_tmp_var and use a
20600 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20601 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20602 ubsan_encode_value callers.
20605 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20606 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20607 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20609 2017-06-19 Richard Biener <rguenther@suse.de>
20611 PR middle-end/81118
20612 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20613 estimates if we changed anything.
20615 2017-06-19 Richard Biener <rguenther@suse.de>
20617 PR tree-optimization/80887
20618 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20619 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20620 simplified lookups, then reset mprts_hook.
20621 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20623 (try_to_simplify): Likewise.
20625 2017-06-19 Martin Liska <mliska@suse.cz>
20628 * gimplify.c (gimplify_switch_expr):
20629 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20631 2017-06-19 Martin Liska <mliska@suse.cz>
20633 * doc/install.texi: Document that PGO runs in 4 stages.
20635 2017-06-19 Martin Liska <mliska@suse.cz>
20638 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20639 to dispatcher function name.
20640 * multiple_target.c (replace_function_decl): New function.
20641 (create_dispatcher_calls): Redirect both edges and references.
20643 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20645 * profile-count.c (profile_count::dump): Dump quality.
20646 (profile_count::differs_from_p): Update for unsigned val.
20647 * profile-count.h (profile_count_quality): New enum.
20648 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20650 2017-06-19 Richard Biener <rguenther@suse.de>
20652 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20653 struct function as arg.
20654 (estimate_numbers_of_iterations): Export overload with loop arg.
20655 (free_numbers_of_iterations_estimates_loop): Use an overload of
20656 free_numbers_of_iterations_estimates instead.
20657 * tree-cfg.c (remove_bb): Adjust.
20658 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20659 * tree-parloops.c (gen_parallel_loop): Likewise.
20660 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20662 (tree_unroll_loops_completely): Likewise.
20663 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20664 Use an overload instead and export.
20665 (estimated_loop_iterations): Adjust.
20666 (max_loop_iterations): Likewise.
20667 (likely_max_loop_iterations): Likewise.
20668 (estimate_numbers_of_iterations): Take struct function as arg
20670 (loop_exits_before_overflow): Adjust.
20671 (free_numbers_of_iterations_estimates_loop): Use an overload.
20672 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20673 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20675 2017-06-19 Richard Biener <rguenther@suse.de>
20678 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20679 RANGE_EXPR conservatively.
20681 2017-06-16 Carl Love <cel@us.ibm.com>
20683 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20684 definitions for vec_float, vec_float2, vec_floato,
20685 vec_floate built-ins.
20686 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20687 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20689 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20690 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20691 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20692 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20693 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20694 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20695 vec_floato): Add builtin defines.
20696 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20697 Update the built-in documentation file for the new built-in
20700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20702 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20703 (mthumb): Mark as the negative of -marm.
20705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20707 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20709 (ARM Options, -mtune): Document that this accepts the same
20710 extension options as -mcpu.
20711 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20713 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20715 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20716 permitted extensions.
20718 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20720 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20721 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20722 (armv8-m.main): Add option +nodsp.
20723 * config/arm/arm-cpu-cdata.h: Regenerated.
20725 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20727 * config/arm/t-fuchsia: New file.
20728 * config.gcc (arm*-*-fuchsia*): Use it.
20730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20732 * config/arm/t-symbian: Rewrite for new option infrastructure.
20734 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20736 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20737 (MULTILIB_REQUIRED): Likewise.
20739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20741 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20742 (MULTILIB_RESUE): Likewise.
20743 (MULTILIB_MATCHES): Likewise.
20744 (MULTLIB_REQUIRED): Likewise.
20746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20748 * config/arm/t-rtems: Rewrite for new option framework.
20750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20752 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20753 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20754 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20755 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20756 * config/arm/t-multilib: ... here.
20757 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20758 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20759 armv7-a and armv8*-a when A-profile libraries have not been built.
20760 * config/arm/t-rmprofile: Rewrite.
20762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20764 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20765 with a backslash. Remove the backslash after substituting unescaped
20767 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20769 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20771 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20772 the driver to pass through the default setting of -mfloat-abi.
20773 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20775 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20776 (all_feat_combs): New rule.
20777 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20779 * config/arm/t-aprofile: Rewrite.
20781 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20783 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20784 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20785 fpu is not specified by the user/command-line.
20786 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20787 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20788 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20789 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20790 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20791 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20793 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20795 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20796 * config/arm/t-arm-elf: Rewritten.
20798 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20800 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20801 have some floating-point instructions.
20802 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20803 (TARGET_MAYBE_HARD_FLOAT): New macro.
20804 * config/arm/arm-builtins.c (arm_init_builtins): Use
20805 TARGET_MAYBE_HARD_FLOAT.
20806 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20808 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20810 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20811 (configargs.h): Include it.
20812 (arm_print_hint_for_fpu_option): New function.
20813 (arm_parse_fpu_option): New function.
20814 (candidate_extension): New class.
20815 (arm_canon_for_multilib): New function.
20816 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20817 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20818 (ARCH_CANONICAL_SPECS): New macro.
20819 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20821 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20823 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20824 are set after handling multilib fragments. Set target_cpu_default2
20827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20829 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20831 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20832 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20834 (check_arch): Likewise.
20835 * config/arm/arm.c (arm_configure_build_target): Handle
20836 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20837 options in the default.
20839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20841 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20842 when an option is an alias of another.
20843 * config/arm/parsecpu.awk (optalias): New parser token.
20844 (gen_comm_data): Mark non-alias options as such. Emit entries
20845 for extension aliases.
20846 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20847 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20848 (armv6kz, armv6zk, armv6t2): Likewise.
20849 (armv7): Make vfpv3-d16 an alias.
20850 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20852 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20853 Sort in canonical order.
20854 (armv8-a): Sort in canonical order.
20855 (armv8.1-a, armv8.2-a): Likewise.
20856 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20858 (cortex-a9): Sort in canonical order.
20859 * config/arm/arm.c (selftests.h): Include it.
20860 (arm_test_cpu_arch_data): New function.
20861 (arm_run_self_tests): New function.
20862 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20863 (targetm): Move declaration to the end of the file.
20864 * arm-cpu-cdata.h: Regenerated.
20866 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20868 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20869 call to target_mode_check describing the type of option passed.
20870 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20871 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20872 arm_parse_cpu_option_name to match parameters against list of
20874 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20875 arm_arch_core_flags data structure.
20876 * config/arm/arm-cpu_cdata.h: Regenerated.
20878 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20880 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20882 (arm_print_hint_for_cpu_option): Likewise.
20883 (arm_print_hint_for_arch_option): Likewise.
20884 (arm_parse_cpu_option_name): Likewise.
20885 (arm_parse_arch_option_name): Likewise.
20886 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20887 of entries in the all_fpus list.
20888 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20889 (arm_parse_cpu_option_name): Declare.
20890 (arm_parse_arch_option_name): Declare.
20891 (arm_parse_option_features): Declare.
20892 (arm_intialize_isa): Declare.
20893 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20895 (gen_comm_data): ... here. Make definitions non-static.
20896 * config/arm/arm-cpu-data.h: Regenerated.
20897 * config/arm/arm-cpu-cdata.h: Regenerated.
20899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20901 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20902 (cpu_arch_extension): New structure.
20903 (cpu_arch_option, arch_option, cpu_option): New structures.
20904 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20905 architecture types.
20906 (gen_data): Generate new format data tables.
20907 * config/arm/arm.c (cpu_tune): New structure.
20908 (cpu_option, processors): Delete.
20909 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20910 (arm_print_hint_for_cpu_option): ... this and ...
20911 (arm_print_hint_for_arch_option): ... this.
20912 (arm_parse_arch_cpu_name): Delete. Replace with ...
20913 (arm_parse_cpu_option_name): ... this and ...
20914 (arm_parse_arch_option_name): ... this.
20915 (arm_unrecognized_feature): Change type of target parameter to
20917 (arm_parse_arch_cpu_features): Delete. Replace with ...
20918 (arm_parse_option_features): ... this.
20919 (arm_configure_build_target): Rework to use new configuration data
20921 (arm_print_tune_info): Rework for new configuration data tables.
20922 * config/arm/arm-cpu-data.h: Regenerated.
20923 * config/arm/arm-cpu.h: Regenerated.
20925 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20927 * Makefile.in (OBJS): Move sbitmap.o from here ...
20928 (OBJS-libcommon): ... to here.
20930 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20932 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20933 (ISA_ALL_CRYPTO): New macro.
20934 (ISA_ALL_SIMD): New macro
20935 (ISA_ALL_FP): New macro.
20936 * config/arm/arm.c (fpu_bitlist): Update initializer.
20937 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20939 (arm9e): Add fpu. Add option for nofp
20940 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20941 (arm926ej-s, arm1026ej-s): Likewise.
20942 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20943 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20944 neon-fp16, neon-vfpv4, nofp and nosimd.
20945 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20946 (cortex-a8): Add fpu. Add option for nofp.
20947 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20948 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20949 (cortex-r4f): Add fpu.
20950 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20951 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20953 (cortex-r8): Likewise.
20954 (cortex-m4): Add fpu. Add option for nofp.
20955 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20956 (cortex-a17.cortex-a7): Likewise.
20957 (cortex-a32): Add fpu. Add options for crypto and nofp.
20958 (cortex-a35, cortex-a53): Likewise.
20959 (cortex-a57): Add fpu. Add option for crypto.
20960 (cortex-a72, cortex-a73): Likewise.
20961 (exynos-m1): Likewise.
20962 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20963 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20964 (cortex-m33): Add fpu. Add option for nofp.
20965 * config/arm/arm-cpu-cdata.h: Regenerated
20966 * config/arm/arm-cpu-data.h: Regenerated.
20968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20970 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20971 (armv5te, armv5tej): Likewise.
20972 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20973 (armv7): Add options fp and vfpv3-d16.
20974 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20975 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20977 (armv7ve): Likewise.
20978 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20979 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20980 (armv8-a): Add nocrypto option.
20981 (armv8.1-a, armv8.2-a): Likewise.
20982 (armv8-m.main): add options fp, fp.dp and nofp.
20984 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20986 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20988 (armv8-a+crc): Delete.
20989 (armv8.1-a): Add options simd, crypto and nofp.
20990 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20991 (armv8.2-a+fp16): Delete.
20992 (armv8-m.main): Add option dsp.
20993 (armv8-m.main+dsp): Delete.
20994 (cortex-a8): Add fpu. Add nofp option.
20995 (cortex-a9): Add fpu. Add nofp and nosimd options.
20996 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20997 link to main cpu and architecture data structures.
20998 (gen_comm_data): Only put isa attributes from the main architecture
21000 (option): New statement for architecture and CPU entries.
21001 * arm.c (struct cpu_option): New structure.
21002 (struct processors): Add entry for options.
21003 (arm_unrecognized_feature): New function.
21004 (arm_parse_arch_cpu_name): Ignore any characters after the first
21006 (arm_parse_arch_cpu_feature): New function.
21007 (arm_configure_build_target): Separate out any CPU and architecture
21008 features and parse separately. Don't error out if -mfpu=auto is
21009 used with only an architecture string.
21010 (arm_print_asm_arch_directives): New function.
21011 (arm_file_start): Call it.
21012 * config/arm/arm-cpu-cdata.h: Regenerated.
21013 * config/arm/arm-cpu-data.h: Likewise.
21014 * config/arm/arm-tables.opt: Likewise.
21016 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21018 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21019 assembler when it is not -mfpu=auto.
21021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21023 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21024 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21025 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21026 (ASM_CPU_SPEC): Rewrite.
21027 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21028 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21029 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21030 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21031 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21032 copied string is NUL-terminated. Also strip any characters prefixed
21034 (arm_rewrite_selected_arch): New function.
21035 (arm_rewrite_march): New function.
21037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21039 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21040 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21041 (march, mcpu, mtune): Convert to string-based options.
21042 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21043 (arm_parse_arch_cpu_name): New function.
21044 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21045 identify selected architecture or CPU.
21046 (arm_option_save): New function.
21047 (TARGET_OPTION_SAVE): Redefine.
21048 (arm_option_restore): Restore string options.
21049 (arm_option_print): Print string options.
21051 2017-06-16 Martin Sebor <msebor@redhat.com>
21053 PR tree-optimization/80933
21054 PR tree-optimization/80934
21055 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21056 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21057 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21058 (gimple_fold_builtin): Call them.
21060 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21062 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21063 as unlikely; update profile.
21065 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21067 * predict.c (force_edge_cold): Handle declaring edges impossible
21070 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21072 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21074 (try_unroll_loop_completely): Fix reporting.
21076 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21078 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21080 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21083 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21084 if given a non-constant argument for an intrinsic which requires a
21087 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21089 * profile.c (compare_freqs): New function.
21090 (branch_prob): Sort edge list.
21091 (find_spanning_tree): Assume that the list is priority sorted.
21093 2017-06-16 Richard Biener <rguenther@suse.de>
21095 PR tree-optimization/81090
21096 * passes.def (pass_record_bounds): Remove.
21097 * tree-pass.h (make_pass_record_bounds): Likewise.
21098 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21099 make_pass_record_bounds): Likewise.
21100 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21101 not free niter estimates at the beginning but at the end.
21102 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21104 2017-06-16 Richard Biener <rguenther@suse.de>
21106 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21107 initializer to workaround ICE in host GCC 4.8.
21109 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21111 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21113 (clone_inlined_nodes): Update.
21115 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21117 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21118 prefetch settings, and enable prefetching by default at -O3.
21120 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21122 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21123 Set flag_prefetch_loop_arrays according to tuning data.
21125 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21127 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21128 New tune structure.
21129 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21130 [Unrelated to main purpose of the patch] Place the pointer field last
21131 to enable type checking errors when tune structure are wrongly merged.
21132 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21133 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21134 (thunderx2t99_prefetch_tune): New tune constants.
21135 (tune_params *_tunings): Update all tunings (no functional change).
21136 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21137 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21138 from tunings structures.
21140 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21143 * ubsan.c (instrument_null): Add T argument, use it instead
21144 of computing it based on IS_LHS.
21145 (instrument_object_size): Likewise.
21146 (pass_ubsan::execute): Adjust instrument_null and
21147 instrument_object_size callers to pass gimple_get_lhs or
21148 gimple_assign_rhs1 result to it. Use instrument_null instead of
21149 calling get_base_address and instrument_mem_ref. Handle
21150 aggregate call arguments for object-size sanitization.
21152 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21154 PR tree-optimization/81089
21155 * tree-vrp.c (is_masked_range_test): Validate operands of
21158 2017-06-15 Martin Sebor <msebor@redhat.com>
21161 * dumpfile.c (dump_register): Avoid calling memset to initialize
21162 a class with a default ctor.
21163 * gcc.c (struct compiler): Remove const qualification.
21164 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21165 * hash-table.h: Ditto.
21166 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21168 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21169 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21171 * params.h (struct param_info): Make struct members non-const.
21172 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21173 with default initialization.
21174 * vec.h (vec_copy_construct, vec_default_construct): New helper
21176 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21177 with vec_copy_construct.
21178 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21179 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21180 * doc/invoke.texi (-Wclass-memaccess): Document.
21182 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21184 * emit-rtl.h (is_leaf): Update comment about local
21185 register allocator.
21187 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21190 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21191 for a variable to have a section before checking if the section has a
21193 Set section to.persistent if persistent attribute is set.
21194 Warn if .persistent attribute is used on an automatic variable.
21196 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21198 PR rtl-optimization/80474
21199 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21201 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21203 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21206 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21208 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21209 (casesi): Emit bounds checking as RTL.
21210 (casesi_internal_mips16_<mode>): Remove bounds checking.
21212 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21214 * config/xtensa/xtensa.c (xtensa_option_override): Append
21215 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21216 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21217 xtensa_doloop_hooks): Define unconditionally.
21218 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21220 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21221 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21222 for it in xtensa_option_override.
21223 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21224 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21226 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21228 * doc/cppopts.texi: Document '-' special value to -MF.
21230 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21232 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21233 (cortex_a53_fconst): Likewise.
21234 (cortex_a53_fpmul): Likewise.
21235 (cortex_a53_f_load_64): Likewise.
21236 (cortex_a53_f_load_many): Likewise.
21237 (cortex_a53_advsimd_alu): Likewise.
21238 (cortex_a53_advsimd_alu_q): Likewise.
21239 (cortex_a53_advsimd_mul): Likewise.
21240 (cortex_a53_advsimd_mul_q): Likewise.
21241 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21242 Add missing fmul, r2f_cvt and fconst cases.
21244 2017-06-14 Richard Biener <rguenther@suse.de>
21246 PR middle-end/81088
21247 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21249 (fold_binary_loc): When associating do not treat pre-existing
21250 TREE_OVERFLOW on literal constants as a reason to allow
21251 TREE_OVERFLOW on associated literal constants.
21253 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21255 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21256 (MASK_FEATURES): New macro.
21257 * config/sparc/sparc.c (sparc_option_override): Remove the special
21258 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21260 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21262 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21263 a division of 0 if non-call exceptions are enabled.
21265 2017-06-14 Andrew Pinski <apinski@cavium.com>
21266 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21269 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21270 Improve vector initialization code gen for only variable case.
21272 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21274 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21276 2017-06-14 Richard Biener <rguenther@suse.de>
21278 PR tree-optimization/81083
21279 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21282 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21284 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21285 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21286 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21287 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21288 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21289 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21291 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21293 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21294 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21296 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21298 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21300 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21302 * config/rs6000/t-rtems: Don't handle SPE.
21304 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21306 * config/rs6000/t-linux: Don't handle SPE.
21308 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21310 * config/rs6000/eabispe.h: Delete file.
21312 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21314 * config/rs6000/t-spe: Delete file.
21316 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21318 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21319 (rs6000_legitimate_offset_address_p): Return false for anything in
21320 V2SImode or V2SFmode.
21322 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21324 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21325 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21326 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21328 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21329 (rs6000_legitimate_offset_address_p): Ditto.
21330 (rs6000_emit_move): Ditto.
21331 (rs6000_init_builtins): Remove V4HI_type_node.
21333 2017-06-13 Martin Liska <mliska@suse.cz>
21336 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21337 (gate_asan): Likewise.
21338 * asan.h (asan_no_sanitize_address_p): Remove the function.
21339 (sanitize_flags_p): New function.
21340 * builtins.def: Fix coding style.
21341 * common.opt: Use renamed enum value.
21342 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21343 * doc/extend.texi: Document no_sanitize attribute.
21344 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21345 to SANITIZE_UNDEFINED_NONDEFAULT.
21346 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21347 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21348 Use sanitize_flags_p.
21349 * gimplify.c (gimplify_function_tree): Likewise.
21350 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21351 * opts.c (parse_no_sanitize_attribute): New function.
21352 (common_handle_option): Use renamed enum value.
21353 * opts.h (parse_no_sanitize_attribute): Declare.
21354 * tree.c (sanitize_flags_p): New function.
21355 * tree.h: Declared here.
21356 * tsan.c: Use sanitize_flags_p.
21357 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21358 (instrument_mem_ref): Likewise.
21359 (instrument_bool_enum_load): Likewise.
21360 (do_ubsan_in_current_function): Remove the function.
21361 (pass_ubsan::execute): Use sanitize_flags_p.
21362 * ubsan.h: Remove do_ubsan_in_current_function
21363 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21364 (dump_function_to_file): Use it here.
21366 2017-06-13 Martin Jambor <mjambor@suse.cz>
21368 PR tree-optimization/80803
21369 PR tree-optimization/81063
21370 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21371 (propagate_subaccesses_across_link): Enqueue subtree whenever
21372 necessary instead of relying on the caller.
21374 2017-06-13 Martin Jambor <mjambor@suse.cz>
21376 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21377 that have a first_link.
21378 (sort_and_splice_var_accesses): Do not check first_link before
21380 (subtree_mark_written_and_enqueue): Likewise.
21381 (propagate_all_subaccesses): Likewise and do not stop at first
21382 parent with a first_link.
21384 2017-06-13 Martin Jambor <mjambor@suse.cz>
21386 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21389 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21391 * match.pd: New pattern.
21393 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21395 * tree-vrp.c (is_masked_range_test): New function.
21396 (register_edge_assert_for): Determine ranges for
21399 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21401 PR tree-optimization/67328
21402 * fold-const.c (maskable_range_p): New function.
21403 (build_range_check): Generate bittests if possible.
21405 2017-06-13 Martin Liska <mliska@suse.cz>
21407 * gimple-pretty-print.c (dump_probability): Add new argument.
21408 (dump_edge_probability): Dump both probability and count.
21409 (dump_gimple_label): Likewise.
21410 (dump_gimple_bb_header): Likewise.
21412 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21415 * config/avr/avr-devices.c: Fix indentation.
21416 * config/avr/gen-avr-mmcu-specs.c: Dito.
21418 2017-06-13 Richard Biener <rguenther@suse.de>
21420 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21421 instead get vector type from stmt_info.
21422 (vectorizable_reduction): Adjust. Remove dead code.
21424 2017-06-13 Richard Biener <rguenther@suse.de>
21426 PR middle-end/81065
21427 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21428 case of C * (x * C2 + C3).
21429 (fold_addr_of_array_ref_difference): Properly fold index difference.
21431 2017-06-12 David S. Miller <davem@davemloft.net>
21434 * config/sparc/sparc.md (return expander): Emit frame blockage if
21435 function uses alloca.
21437 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21439 * combine.c (make_field_assignment): Check len rather than the mode
21440 precision when calling force_to_mode.
21442 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21444 Support multilibs and devices that see flash in RAM address range.
21447 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21448 (avr_mcu_t) <flash_pm_offset>: New field.
21449 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21450 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21451 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21452 (AVR_TINY_PM_OFFSET): Remove macro.
21453 * config/avr/avr.opt (-mshort-calls): New option.
21454 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21455 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21456 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21457 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21458 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21459 instead of avr_arch->have_jmp_call.
21460 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21461 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21462 avr_arch->flash_pm_offset to define.
21463 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21464 new field flash_pm_offset. Add entry for avrxmega3.
21465 (avr_texinfo): Add entry for avrxmega3.
21466 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21467 attiny212, attiny214,
21468 attiny412, attiny414, attiny416, attiny417,
21469 attiny814, attiny816, attiny817,
21470 attiny1614, attiny1616, attiny1617,
21471 attiny3214, attiny3216, attiny3217.
21472 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21473 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21474 (avr_print_operand_address) [AVR_TINY]: Same.
21475 (avr_asm_init_sections) <readonly_data_section>: Only patch
21476 callback if avr_arch->flash_pm_offset = 0.
21477 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21478 for rodata if avr_arch->flash_pm_offset != 0.
21479 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21480 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21481 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21482 (m_options): Append opt_rcall.
21483 (m_dirnames): Append dir_rcall.
21484 * config/avr/t-multilib: Regenerate.
21486 * configure.ac [target=avr]: Check whether avrxmega3 default
21487 linker description file works as needed.
21488 * configure: Regenerate.
21489 * doc/avr-mmcu.texi: Regenerate.
21490 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21491 <__AVR_ARCH__>: Document avrxmega3 and 103.
21492 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21493 <__AVR_SHORT_CALLS__>: Document it.
21494 <__AVR_PM_BASE_ADDRESS__>: Document it.
21495 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21496 (AVR Variable Attributes) <progmem>: Document this is
21497 not needed for avrxmega3.
21498 (AVR Named Address Spaces) <__flash>: Dito.
21500 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21502 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21504 2017-06-12 Doug Rupp <rupp@adacore.com>
21506 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21507 Append vxworks-stdint.h to the tm_file list.
21508 * config/vxworks-stdint.h: New file.
21510 2017-06-12 Martin Liska <mliska@suse.cz>
21512 PR tree-optimization/81041
21513 * tree-profile.c (gimple_gen_ic_func_profiler):
21514 Create an extra BB in profile-generate
21515 (gimple_gen_time_profiler): Likewise.
21517 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21519 PR tree-optimization/81003
21520 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21521 (update_range_test): Use it instead of force_gimple_operand_gsi.
21523 2017-06-12 Richard Biener <rguenther@suse.de>
21525 PR tree-optimization/81053
21526 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21527 with backedge value not defined in loop. Simplify def stmt
21530 2017-06-11 Tom de Vries <tom@codesourcery.com>
21533 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21535 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21536 nvptx_cannot_force_const_mem.
21538 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21540 * opts.c (finish_options): Move test for flag_split_stack after
21541 it has been initialized.
21543 2017-06-11 Jason Merrill <jason@redhat.com>
21545 * tree.h (id_equal): New.
21546 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21547 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21548 instead of strcmp of IDENTIFIER_POINTER.
21550 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21552 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21553 (mark_all_inlined_calls_cdtor): Fix formating.
21554 (inline_transform): Rescale profile before inlining.
21556 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21558 * cgraph.h (cgraph_edge::clone): Update prototype.
21559 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21560 (cgraph_node::create_clone): Update.
21561 (cgraph_node::create_version_clone): Update.
21562 * tree-inline.c (copy_bb): Update.
21563 (expand_call_inline): Update.
21565 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21567 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21568 factored out from ...
21569 (rs6000_emit_prologue): ... here.
21571 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21573 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21574 factored out from ...
21575 (rs6000_emit_prologue): ... here.
21577 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21579 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21581 (handle_missing_profiles): Fix computation of tp_first_run.
21582 (counts_to_freqs): Do not touch freqs when count is 0.
21584 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21586 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21589 2017-06-10 Tom de Vries <tom@codesourcery.com>
21591 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21592 attributes): Document signal effective target.
21594 2017-06-10 Tom de Vries <tom@codesourcery.com>
21596 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21597 Document effective target stack_size.
21599 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21601 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21602 to the edit_context if they can be auto-applied.
21604 2017-06-9 Ian Lance Taylor <iant@golang.org>
21606 * opts.c (finish_options): If -fsplit-stack, disable implicit
21607 -forder-blocks-and-partition.
21608 * doc/invoke.texi (Optimize Options): Document that when using
21609 -fsplit-stack -forder-blocks-and-partition is not implicitly
21612 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21614 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21615 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21616 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21617 * builtins.def (abort, trap, unreachable): Declare cold.
21618 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21619 * tree-core.h (ECF_COLD): New.
21620 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21621 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21623 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21625 * predict.c (unlikely_executed_stmt_p): Cleanup.
21627 2017-06-09 Richard Biener <rguenther@suse.de>
21629 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21630 model if the ref is always written to.
21632 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21634 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21636 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21638 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21641 2017-06-09 Tom de Vries <tom@codesourcery.com>
21644 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21645 "target cannot support label values" when encountering LABEL_REF.
21647 2017-06-09 Martin Liska <mliska@suse.cz>
21649 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21650 (gimple_gen_ic_func_profiler): Emit direct comparison
21651 of __gcov_indirect_call_callee with NULL.
21652 (gimple_gen_time_profiler): Change probability from
21653 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21655 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21657 * profile.c (edge_gcov_counts): Turn to pointer.
21658 (compute_branch_probabilities, compute_branch_probabilities): Update.
21659 (branch_prob): Do not clear edge_gcov_count.
21660 * profile.h (edge_gcov_counts): Turn to pointer.
21661 (edge_gcov_count): Update.
21663 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21665 * gimple.h (gimple_check_failed): Mark cold.
21667 2017-06-09 Richard Biener <rguenther@suse.de>
21669 PR tree-optimization/66623
21670 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21671 refactor check_reduction into two parts, properly computing
21672 whether we have to check reduction validity for outer loop
21675 2017-06-09 Richard Biener <rguenther@suse.de>
21677 PR tree-optimization/79483
21678 * graphite-scop-detection.c (order): New global.
21679 (get_order): Compute bb to order mapping that satisfies code
21680 generation constraints.
21681 (cmp_pbbs): New helper.
21682 (build_scops): Start domwalk at entry block, sort generated
21685 2017-06-09 Richard Biener <rguenther@suse.de>
21687 PR middle-end/81007
21688 * ipa-polymorphic-call.c
21689 (ipa_polymorphic_call_context::restrict_to_inner_class):
21690 Skip FIELD_DECLs with error_mark_node type.
21691 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21694 2017-06-09 Martin Liska <mliska@suse.cz>
21696 * predict.c (struct branch_predictor): New struct.
21697 (test_prediction_value_range): New test.
21698 (predict_c_tests): New function.
21699 * selftest-run-tests.c (selftest::run_tests): Run the function.
21700 * selftest.h: Declare new tests.
21702 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21705 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21706 gen_add3_insn did not fail.
21707 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21708 r0, construct that number in a temporary reg and add that reg to r0.
21709 If asked to put the result in r0 as well, fail.
21711 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21713 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21714 for early expansion of vec_eqv.
21716 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21718 PR middle-end/81005
21719 * ubsan.c (instrument_null): Avoid pointless code temporary.
21720 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21722 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21726 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21729 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21732 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21734 * predict.c (unlikely_executed_edge_p): Move ahead.
21735 (probably_never_executed_edge_p): Use it.
21737 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21739 PR middle-end/79988
21740 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21741 gimple_call_builtin_p call.
21743 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21745 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21746 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21747 rtl_check_failed_type2, rtl_check_failed_code1,
21748 rtl_check_failed_code2, rtl_check_failed_code_mode,
21749 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21750 rtvec_check_failed_bounds, rtl_check_failed_flag,
21751 _fatal_insn_not_found, _fatal_insn): Likewise.
21752 * tree.h (tree_contains_struct_check_failed,
21753 tree_check_failed, tree_not_check_failed,
21754 tree_class_check_failed, tree_range_check_failed,
21755 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21756 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21757 tree_operand_check_failed, omp_clause_check_failed,
21758 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21761 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21763 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21764 flag_branch_probabilities.
21765 * ipa-inline.c (edge_badness): Likewise.
21766 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21767 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21768 * predict.c (maybe_hot_frequency_p): Likewise.
21769 (probably_never_executed): Likewise.
21770 * sched-ebb.c (schedule_ebbs): Likewise.
21771 * sched-rgn.c (find_single_block_region): Likewise.
21772 * tracer.c (tail_duplicate): Likewise.
21774 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21776 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21777 longer requires x_flag_profile_use.
21779 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21781 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21782 instead of flag_reorder_blocks_and_partition.
21783 * dbxout.c (dbxout_function_end): Likewise.
21784 * dwarf2out.c (gen_subprogram_die): Likewise.
21785 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21786 * hw-doloop.c (reorg_loops): Likewise.
21787 * varasm.c (assemble_start_function,
21788 assemble_end_function): Likewise.
21789 (decide_function_section): Do not check for
21790 flag_reorder_blocks_and_partition.
21792 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21794 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21796 (chkp_get_hard_register_fake_addr_expr): Ditto.
21797 (chkp_build_addr_expr): Add check for hard reg case.
21798 (chkp_parse_array_and_component_ref): Ditto.
21799 (chkp_find_bounds_1): Ditto.
21800 (chkp_process_stmt): Don't generate bounds store for
21803 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21805 * predict.c (maybe_hot_bb_p): Do not check profile status.
21806 (maybe_hot_edge_p): Likewise.
21807 (probably_never_executed): Check for zero counts even if profile
21809 (unlikely_executed_edge_p): New function.
21810 (unlikely_executed_stmt_p): New function.
21811 (unlikely_executed_bb_p): New function.
21812 (set_even_probabilities): Use unlikely predicates.
21813 (combine_predictions_for_bb): Likewise.
21814 (predict_paths_for_bb): Likewise.
21815 (predict_paths_leading_to_edge): Likewise.
21816 (determine_unlikely_bbs): New function.
21817 (estimate_bb_frequencies): Use it.
21818 (compute_function_frequency): Use zero counts even if profile is
21820 * profile-count.h: Fix typo.
21822 2017-08-08 Julia Koval <julia.koval@intel.com>
21824 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21825 _mm512_mask_cvtsepi16_storeu_epi8,
21826 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21827 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21828 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21829 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21830 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21831 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21832 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21833 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21834 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21835 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21836 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21837 __builtin_ia32_pmovuswb256mem_mask,
21838 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21839 __builtin_ia32_pmovwb512mem_mask): New builtins.
21841 2017-08-08 Julia Koval <julia.koval@intel.com>
21843 PR target/73350,80862
21844 * config/i386/subst.md (round): Fix round pattern.
21845 * config/i386/i386.c (ix86_erase_embedded_rounding):
21846 Fix erasing rounding for the fixed pattern.
21848 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21850 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21852 2017-06-08 Martin Liska <mliska@suse.cz>
21854 PR gcov-profile/80911
21855 * gcov.c (block_info::block_info): New constructor.
21857 2017-06-07 Carl Love <cel@us.ibm.com>
21859 * config/rs6000/rs6000-c: The return type of the following
21860 built-in functions was implemented as int not long long. Fix sign
21861 of return value for the unsigned version of vec_mulo and vec_mule.
21862 vector unsigned long long vec_bperm (vector unsigned long long,
21863 vector unsigned char)
21864 vector signed long long vec_mule (vector signed int,
21866 vector unsigned long long vec_mule (vector unsigned int,
21867 vector unsigned int)
21868 vector signed long long vec_mulo (vector signed int,
21870 vector unsigned long long vec_mulo (vector unsigned int,
21871 vector unsigned int)
21872 * doc/extend.texi: Fix the documentation for the built-in
21875 2017-06-07 Carl Love <cel@us.ibm.com>
21878 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21881 2017-06-07 Carl Love <cel@us.ibm.com>
21883 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21884 support, Generate doublehv for signed int/float for BE case only.
21886 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21888 * doc/invoke.texi (mcx16): Rewrite.
21890 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21892 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21893 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21894 *mov<mode>_softfloat, and an anonymous splitter): Use
21895 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21897 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21899 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21901 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21902 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21903 (rs6000_debug_reg_global): Adjust.
21904 (rs6000_init_hard_regno_mode_ok): Adjust.
21905 (rs6000_dbx_register_number): Adjust.
21906 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21907 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21908 Remove SPE_ACC and SPEFSCR.
21909 (REG_ALLOC_ORDER): Ditto.
21910 (FRAME_POINTER_REGNUM): Change to 111.
21911 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21912 (REG_CLASS_NAMES): Ditto.
21913 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21914 (REGISTER_NAMES): Ditto.
21915 (ADDITIONAL_REG_NAMES): Ditto.
21916 (rs6000_reg_names): Ditto.
21917 * config/rs6000/rs6000.md: Renumber some register number
21920 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21922 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21924 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21925 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21927 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21928 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21929 Delete the SPE high registers.
21930 (REG_ALLOC_ORDER): Ditto.
21931 (enum reg_class): Remove SPE_HIGH_REGS.
21932 (REG_CLASS_NAMES): Ditto.
21933 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21934 (REGISTER_NAMES): Ditto.
21935 (rs6000_reg_names): Ditto.
21936 * doc/tm.texi.in: Remove SPE as example.
21937 * doc/tm.texi: Regenerate.
21939 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21941 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21942 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21943 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21944 * config/rs6000/rs6000.md (type): Remove "brinc".
21946 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21948 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21949 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21950 * config/rs6000/linuxspe.h: Delete file.
21951 * config/rs6000/rs6000.md: Don't include spe.md.
21952 * config/rs6000/spe.h: Delete file.
21953 * config/rs6000/spe.md: Delete file.
21954 * config/rs6000/t-rs6000: Remove spe.md.
21956 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21958 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21959 (reg_or_none500mem_operand): Delete.
21960 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21961 instead of reg_or_none500mem_operand.
21963 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21965 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21966 handling of SPE flags.
21967 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21969 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21971 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21973 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21974 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21975 paired_divv2sf3): Similar.
21976 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21977 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21978 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21980 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21981 Rename the paired_* instruction patterns.
21982 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21984 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21985 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21986 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21987 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21988 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21989 PAIRED_VECTOR_MODE.
21990 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21991 (spe_func_has_64bit_regs_p): Delete.
21992 (spe_expand_predicate_builtin): Delete.
21993 (spe_expand_evsel_builtin): Delete.
21994 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21995 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21996 (invalid_e500_subreg): Delete.
21997 (rs6000_legitimize_address): Always force_reg op2 as well, for
21998 paired single memory accesses.
21999 (rs6000_member_type_forces_blk): Delete.
22000 (rs6000_spe_function_arg): Delete.
22001 (rs6000_expand_unop_builtin): Delete SPE handling.
22002 (rs6000_expand_binop_builtin): Ditto.
22003 (spe_expand_stv_builtin): Delete.
22004 (bdesc_2arg_spe): Delete.
22005 (spe_expand_builtin): Delete.
22006 (spe_expand_predicate_builtin): Delete.
22007 (spe_expand_evsel_builtin): Delete.
22008 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22009 (spe_init_builtins): Delete.
22010 (spe_func_has_64bit_regs_p): Delete.
22011 (savres_routine_name): Delete "info" parameter. Adjust callers.
22012 (rs6000_emit_stack_reset): Ditto.
22013 (rs6000_dwarf_register_span): Delete.
22014 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22015 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22016 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22018 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22020 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22021 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22022 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22023 mulv2sf3, divv2sf3): Delete expanders.
22025 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22027 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22029 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22031 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22032 * config/rs6000/rs6000.c: Ditto.
22034 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22036 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22037 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22038 comparison_operator.
22040 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22042 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22043 * config/rs6000/rs6000.opt: Ditto.
22044 * config/rs6000/t-rtems: Ditto.
22046 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22048 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22049 TARGET_E500_SINGLE by 0, simplify.
22050 * config/rs6000/rs6000.c: Ditto.
22051 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22052 (spe_build_register_parallel): Delete.
22053 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22054 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22055 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22056 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22057 (E500_CONVERT): Delete.
22058 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22060 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22062 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22063 * config/rs6000/dfp.md: Ditto.
22064 (negdd2, *negdd2_fpr): Merge.
22065 (absdd2, *absdd2_fpr): Merge.
22066 (negtd2, *negtd2_fpr): Merge.
22067 (abstd2, *abstd2_fpr): Merge.
22068 * config/rs6000/e500.h: Delete file.
22069 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22070 TARGET_FPRS by 1 and simplify.
22071 * config/rs6000/rs6000-c.c: Ditto.
22072 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22073 TARGET_DF_SPE by 0.
22074 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22076 * config/rs6000/rs6000.md: Ditto.
22077 (floatdidf2, *floatdidf2_fpr): Merge.
22078 (move_from_CR_gt_bit): Delete.
22079 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22080 (E500_CR_IOR_COMPARE): Delete.
22081 (All patterns that require !TARGET_FPRS): Delete.
22082 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22084 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22086 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22088 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22090 * graphds.c (add_edge): Intitialize edge's attached data.
22091 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22092 pointer parameter. Call pointed function on each edge during
22093 graph traversing. Skip traversing the edge when the function
22095 (graphds_dfs, graphds_scc): Ditto.
22096 (for_each_edge): New parameter. Pass the new parameter to callback
22098 * graphds.h (skip_edge_callback): New function pointer type.
22099 (graphds_dfs, graphds_scc): New function pointer parameter.
22100 (graphds_edge_callback, for_each_edge): New parameter.
22102 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22104 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22105 out code checking if runtime alias check is possible to below ...
22106 Call the new function.
22107 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22108 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22110 2017-06-07 Marek Polacek <polacek@redhat.com>
22113 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22114 TYPE_OVERFLOW_WRAPS check.
22116 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22118 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22119 if versioning is required.
22120 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22121 peeling with the check for versioning.
22123 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22125 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22126 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22127 Set true to new parameter if new ssa variable is defined.
22128 (vect_gen_vector_loop_niters): Refactor. Set range information
22129 for the new vector loop bound variable.
22130 (vect_do_peeling): Ditto.
22132 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22134 * tree-affine.c (ssa.h): Include header file.
22135 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22136 has wrapping overflow behavior.
22138 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22140 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22142 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22144 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22145 (tree_to_aff_combination): ... here.
22147 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22149 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22150 reg_pressure model function.
22151 (ivopts_global_cost_for_size): Delete.
22152 (determine_set_costs, iv_ca_recount_cost): Call new model function
22153 ivopts_estimate_reg_pressure.
22155 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22157 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22158 expensive than udiv. Remove floating point cases from mod.
22160 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22162 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22163 Increase idiv cost.
22165 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22167 * config/aarch64/aarch64.md
22168 (copysignsf3): Fix mask generation.
22170 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22172 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22174 (class dump_manager): Add register_dumps method.
22175 * dumpfile.c: Include langhooks.h.
22176 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22177 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22178 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22179 (dump_manager::dump_register): Start with 512 entries instead of 32.
22180 (dump_manager::register_dumps): New method.
22181 * toplev.c (general_init): Instead of invoking register_dumps
22182 langhook, invoke register_dumps method on the dump manager.
22183 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22186 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22188 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22189 Say that # requires an associated define_split to exist, and that
22190 the define_split must be suitable for use after register allocation.
22192 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22194 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22195 (compute_outgoing_frequencies): Also initialize zero counts.
22196 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22197 around loops; preserve more of profile when nothing changes.
22199 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22201 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22203 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22204 * config/arm/arm-cpu-cdata.h: Regenerate.
22205 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22206 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22207 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22208 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22209 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22211 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22213 * config/arm/t-rmprofile: Likewise.
22214 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22216 2017-06-06 David S. Miller <davem@davemloft.net>
22219 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22220 blockage if function uses alloca.
22222 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22224 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22225 New "uid" fields to hold pretty-print IDs of group and ref.
22226 Memory references are now identified as <group_id>:<ref_id>
22227 instead of using [random] addresses.
22228 (dump_mem_details): Simplify, no functional change.
22229 (dump_mem_ref): Simplify and make output more concise.
22230 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22231 (find_or_create_group): Initialize group uid.
22232 (record_ref): Initialize ref uid. Improve debug output.
22233 (prune_group_by_reuse, should_issue_prefetch_p,)
22234 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22235 (mark_nontemporal_store, determine_loop_nest_reuse):
22236 Improve debug output.
22238 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22240 * dbgcnt.def (prefetch): New debug counter.
22241 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22242 (schedule_prefetches): Stop issueing prefetches if debug counter
22245 2017-06-06 Tom de Vries <tom@codesourcery.com>
22247 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22248 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22250 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22252 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22253 Use aarch64_reg_or_zero predicate for operand 4.
22254 (aarch64_compare_and_swap<mode> define_insn_and_split):
22255 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22256 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22258 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22260 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22261 (arm_compute_save_core_reg_mask): This.
22262 (thumb1_compute_save_reg_mask): Rename into ...
22263 (thumb1_compute_save_core_reg_mask): This.
22264 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22265 (arm_compute_frame_layout): Likewise.
22267 2017-06-06 Richard Biener <rguenther@suse.de>
22269 PR tree-optimization/80974
22270 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22271 keep or clear leaders SSA info.
22273 2017-06-06 Tom de Vries <tom@codesourcery.com>
22275 * config/nvptx/nvptx.c (split_mode_p): New function.
22276 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22278 2017-06-06 Tom de Vries <tom@codesourcery.com>
22280 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22282 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22285 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22288 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22290 * shrink-wrap.c (handle_simple_exit): Update profile.
22291 (try_shrink_wrapping): Upate profile.
22293 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22295 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22296 (tree_guess_outgoing_edge_probabilities): New.
22297 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22298 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22300 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22302 * ipa-split.c (split_function): Initialize return bb profile.
22304 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22306 * profile.c (compute_branch_probabilities): Also initialize
22307 EXIT_BLOCK profile.
22309 2017-06-06 Richard Biener <rguenther@suse.de>
22311 PR tree-optimization/80928
22312 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22313 (vect_analyze_loop_operations): Properly guard analysis for
22315 (vect_transform_loop): Likewise.
22316 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22317 (vect_model_induction_cost): Do not cost for pure SLP.
22318 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22319 of induction in inner loop vectorization.
22320 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22321 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22322 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22324 (vect_analyze_slp_cost_1): Cost induction.
22325 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22326 (vect_get_slp_vect_defs): Likewise.
22327 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22328 (vect_transform_stmt): Handle SLP reductions.
22329 * tree-vectorizer.h (vectorizable_induction): Adjust.
22331 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22333 * config/rs6000/rs6000.c (make_resolver_func): Update
22334 init_lowered_empty_function call.
22336 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22338 * doc/invoke.texi: Document the -fprofile-abs-path option.
22339 * common.opt (fprofile-abs-path): New option.
22340 * gcov-io.h (gcov_write_filename): Declare.
22341 * gcov-io.c (gcov_write_filename): New function.
22342 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22343 * profile.c (output_location): Likewise.
22345 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22347 * shring-wrap.c: Revert accidental commit.
22349 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22351 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22353 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22355 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22357 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22358 profile in callgraph edge.
22359 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22360 (apply_scale): Likewise.
22361 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22362 Also scale profile when inlining function with zero profile.
22363 (initialize_cfun): Update exit block profile even when it is zero.
22364 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22365 when profile is read.
22367 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22369 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22370 (CLONE_*): New constants to define the processors we can generate
22371 code for with the target_clone attribute.
22372 (rs6000_clone_map): New array to identify which clone processors
22373 the current program is running on.
22374 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22375 target_clone attribute.
22376 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22377 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22378 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22379 (cpu_expand_builtin): Add support for target_clone attribute.
22380 (rs6000_valid_attribute_p): Allow "default" attribute.
22381 (get_decl_name): New debug function to simplify printing the
22382 current function name in debugging statements.
22383 (rs6000_clone_priority): New functions to support the target_clone
22384 attribute, and be able to generate code to switch between ISA 2.05
22385 through ISA 3.0 (power6 through power9).
22386 (rs6000_compare_version_priority): Likewise.
22387 (rs6000_get_function_versions_dispatcher): Likewise.
22388 (make_resolver_func): Likewise.
22389 (add_condition_to_bb): Likewise.
22390 (dispatch_function_versions): Likewise.
22391 (rs6000_generate_version_dispatcher_body): Likewise.
22392 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22393 (fusion_gpr_load_p): Fix a spacing issue.
22394 * doc/extend.texi (Common Function Attributes): Document that the
22395 PowerPC supports the target_clone attribute.
22397 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22399 * config/arm/arm.h: explain F symbol found in description of ARM
22400 register allocation in its legend.
22402 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22404 * config/mips/frame-header-opt.c: Include profile-count.h.
22405 * config/riscv/riscv.c: Include profile-count.h
22407 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22409 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22411 (sm_set_flag_if_changed): Add bbs field.
22412 (execute_sm_if_changed_flag_set): Pass BBS.
22413 (execute_sm): Update.
22415 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22417 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22420 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22422 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22423 (peephole2): New peephole2 to emit the above.
22424 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22426 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22428 * config/aarch64/aarch64.c (define_peephole2 above
22429 *sub_<shift>_<mode>): New peephole.
22431 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22433 * config/i386/i386.c (make_resolver_func): Update.
22434 * Makefile.in: Add profile-count.h and profile-count.o
22435 * auto-profile.c (afdo_indirect_call): Update to new API.
22436 (afdo_set_bb_count): Update.
22437 (afdo_propagate_edge): Update.
22438 (afdo_propagate_circuit): Update.
22439 (afdo_calculate_branch_prob): Update.
22440 (afdo_annotate_cfg): Update.
22441 * basic-block.h: Include profile-count.h
22442 (struct edge_def): Turn count to profile_count.
22443 (struct basic_block_def): Likewie.
22444 (REG_BR_PROB_BASE): Move to profile-count.h
22445 (RDIV): Move to profile-count.h
22446 * bb-reorder.c (max_entry_count): Turn to profile_count.
22447 (find_traces): Update.
22448 (rotate_loop):Update.
22449 (connect_traces):Update.
22450 (sanitize_hot_paths):Update.
22451 * bt-load.c (migrate_btr_defs): Update.
22452 * cfg.c (RDIV): Remove.
22453 (init_flow): Use alloc_block.
22454 (alloc_block): Uninitialize count.
22455 (unchecked_make_edge): Uninitialize count.
22456 (check_bb_profile): Update.
22457 (dump_edge_info): Update.
22458 (dump_bb_info): Update.
22459 (update_bb_profile_for_threading): Update.
22460 (scale_bbs_frequencies_int): Update.
22461 (scale_bbs_frequencies_gcov_type): Update.
22462 (scale_bbs_frequencies_profile_count): New.
22463 * cfg.h (update_bb_profile_for_threading): Update.
22464 (scale_bbs_frequencies_profile_count): Declare.
22465 * cfgbuild.c (compute_outgoing_frequencies): Update.
22466 (find_many_sub_basic_blocks): Update.
22467 * cfgcleanup.c (try_forward_edges): Update.
22468 (try_crossjump_to_edge): Update.
22469 * cfgexpand.c (expand_gimple_tailcall): Update.
22470 (construct_exit_block): Update.
22471 * cfghooks.c (verify_flow_info): Update.
22472 (dump_bb_for_graph): Update.
22473 (split_edge): Update.
22474 (make_forwarder_block): Update.
22475 (duplicate_block): Update.
22476 (account_profile_record): Update.
22477 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22478 (get_estimated_loop_iterations): Update.
22479 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22480 (single_likely_exit): Update.
22481 * cfgloopmanip.c (scale_loop_profile): Update.
22483 (set_zero_probability): Update.
22484 (lv_adjust_loop_entry_edge): Update.
22485 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22486 (purge_dead_edges): Update.
22487 (rtl_account_profile_record): Update.
22488 * cgraph.c (cgraph_node::create): Uninitialize count.
22489 (symbol_table::create_edge): Uninitialize count.
22490 (cgraph_update_edges_for_call_stmt_node): Update.
22491 (cgraph_edge::dump_edge_flags): Update.
22492 (cgraph_node::dump): Update.
22493 (cgraph_edge::maybe_hot_p): Update.
22494 * cgraph.h: Include profile-count.h
22495 (create_clone), create_edge, create_indirect_edge): Update.
22496 (cgraph_node): Turn count to profile_count.
22497 (cgraph_edge0: Likewise.
22498 (make_speculative, clone): Update.
22499 (create_edge): Update.
22500 (init_lowered_empty_function): Update.
22501 * cgraphclones.c (cgraph_edge::clone): Update.
22502 (duplicate_thunk_for_node): Update.
22503 (cgraph_node::create_clone): Update.
22504 * cgraphunit.c (cgraph_node::analyze): Update.
22505 (cgraph_node::expand_thunk): Update.
22506 * final.c (dump_basic_block_info): Update.
22507 * gimple-streamer-in.c (input_bb): Update.
22508 * gimple-streamer-out.c (output_bb): Update.
22509 * graphite.c (print_global_statistics): Update.
22510 (print_graphite_scop_statistics): Update.
22511 * hsa-brig.c: Include basic-block.h.
22512 * hsa-dump.c: Include basic-block.h.
22513 * hsa-gen.c (T sum_slice): Update.
22514 (convert_switch_statements):Update.
22515 * hsa-regalloc.c: Include basic-block.h.
22516 * ipa-chkp.c (chkp_produce_thunks): Update.
22517 * ipa-cp.c (struct caller_statistics): Update.
22518 (init_caller_stats): Update.
22519 (gather_caller_stats): Update.
22520 (ipcp_cloning_candidate_p): Update.
22521 (good_cloning_opportunity_p): Update.
22522 (get_info_about_necessary_edges): Update.
22523 (dump_profile_updates): Update.
22524 (update_profiling_info): Update.
22525 (update_specialized_profile): Update.
22526 (perhaps_add_new_callers): Update.
22527 (decide_about_value): Update.
22528 (ipa_cp_c_finalize): Update.
22529 * ipa-devirt.c (struct odr_type_warn_count): Update.
22530 (struct decl_warn_count): Update.
22531 (struct final_warning_record): Update.
22532 (possible_polymorphic_call_targets): Update.
22533 (ipa_devirt): Update.
22534 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22535 * ipa-icf.c (sem_function::merge): Update.
22536 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22537 * ipa-inline.c (compute_uninlined_call_time): Update.
22538 (compute_inlined_call_time): Update.
22539 (want_inline_small_function_p): Update.
22540 (want_inline_self_recursive_call_p): Update.
22541 (edge_badness): Update.
22542 (lookup_recursive_calls): Update.
22543 (recursive_inlining): Update.
22544 (inline_small_functions): Update.
22545 (dump_overall_stats): Update.
22546 (dump_inline_stats): Update.
22547 * ipa-profile.c (ipa_profile_generate_summary): Update.
22548 (ipa_propagate_frequency): Update.
22549 (ipa_profile): Update.
22550 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22551 * ipa-utils.c (ipa_merge_profiles): Update.
22552 * loop-doloop.c (doloop_modify): Update.
22553 * loop-unroll.c (report_unroll): Update.
22554 (unroll_loop_runtime_iterations): Update.
22555 * lto-cgraph.c (lto_output_edge): Update.
22556 (lto_output_node): Update.
22557 (input_node): Update.
22558 (input_edge): Update.
22559 (merge_profile_summaries): Update.
22560 * lto-streamer-in.c (input_cfg): Update.
22561 * lto-streamer-out.c (output_cfg): Update.
22562 * mcf.c (create_fixup_graph): Update.
22563 (adjust_cfg_counts): Update.
22564 (sum_edge_counts): Update.
22565 * modulo-sched.c (sms_schedule): Update.
22566 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22567 * predict.c (maybe_hot_count_p): Update.
22568 (probably_never_executed): Update.
22569 (dump_prediction): Update.
22570 (combine_predictions_for_bb): Update.
22571 (propagate_freq): Update.
22572 (handle_missing_profiles): Update.
22573 (counts_to_freqs): Update.
22574 (rebuild_frequencies): Update.
22575 (force_edge_cold): Update.
22576 * predict.h: Include profile-count.h
22577 (maybe_hot_count_p, counts_to_freqs): UPdate.
22578 * print-rtl-function.c: Do not include cfg.h
22579 * print-rtl.c: Include basic-block.h
22580 * profile-count.c: New file.
22581 * profile-count.h: New file.
22582 * profile.c (is_edge_inconsistent): Update.
22583 (correct_negative_edge_counts): Update.
22584 (is_inconsistent): Update.
22585 (set_bb_counts): Update.
22586 (read_profile_edge_counts): Update.
22587 (compute_frequency_overlap): Update.
22588 (compute_branch_probabilities): Update; Initialize and deinitialize
22590 (branch_prob): Update.
22591 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22592 (edge_gcov_count): New.
22593 (bb_gcov_count): New.
22594 * shrink-wrap.c (try_shrink_wrapping): Update.
22595 * tracer.c (better_p): Update.
22596 * trans-mem.c (expand_transaction): Update.
22597 (ipa_tm_insert_irr_call): Update.
22598 (ipa_tm_insert_gettmclone_call): Update.
22599 * tree-call-cdce.c: Update.
22600 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22601 (gimple_duplicate_sese_tail): Update.
22602 (gimple_account_profile_record): Update.
22603 (execute_fixup_cfg): Update.
22604 * tree-inline.c (copy_bb): Update.
22605 (copy_edges_for_bb): Update.
22606 (initialize_cfun): Update.
22607 (freqs_to_counts): Update.
22608 (copy_cfg_body): Update.
22609 (expand_call_inline): Update.
22610 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22611 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22612 (try_unroll_loop_completely): Update.
22613 (try_peel_loop): Update.
22614 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22615 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22616 * tree-ssa-loop-split.c (connect_loops): Update.
22617 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22618 * tree-ssa-reassoc.c (branch_fixup): Update.
22619 * tree-ssa-tail-merge.c (replace_block_by): Update.
22620 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22621 (compute_path_counts): Update.
22622 (update_profile): Update.
22623 (recompute_probabilities): Update.
22624 (update_joiner_offpath_counts): Update.
22625 (estimated_freqs_path): Update.
22626 (freqs_to_counts_path): Update.
22627 (clear_counts_path): Update.
22628 (ssa_fix_duplicate_block_edges): Update.
22629 (duplicate_thread_path): Update.
22630 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22631 (struct switch_conv_info): Update.
22632 * tree-tailcall.c (decrease_profile): Update.
22633 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22634 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22635 * value-prof.c (check_counter): Update.
22636 (gimple_divmod_fixed_value): Update.
22637 (gimple_mod_pow2): Update.
22638 (gimple_mod_subtract): Update.
22639 (gimple_ic_transform): Update.
22640 (gimple_stringop_fixed_value): Update.
22641 * value-prof.h (gimple_ic): Update.
22643 2017-06-02 Carl Love <cel@us.ibm.com>
22645 * config/rs6000/rs6000-c: Add support for built-in functions
22646 vector double vec_doublee (vector signed int);
22647 vector double vec_doublee (vector unsigned int);
22648 vector double vec_doublee (vector float);
22649 vector double vec_doubleh (vector signed int);
22650 vector double vec_doubleh (vector unsigned int);
22651 vector double vec_doubleh (vector float);
22652 vector double vec_doublel (vector signed int);
22653 vector double vec_doublel (vector unsigned int);
22654 vector double vec_doublel (vector float);
22655 vector double vec_doubleo (vector signed int);
22656 vector double vec_doubleo (vector unsigned int);
22657 vector double vec_doubleo (vector float);.
22658 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22659 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22661 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22662 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22663 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22665 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22666 vec_doublel, vec_doubleh.
22667 * doc/extend.texi: Update the built-in documentation file for the
22668 new built-in functions.
22670 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22673 * ipa-inline-analysis.c (free_growth_caches): Set
22674 edge_removal_hook_holder to NULL after removing it.
22676 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22678 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22679 comparision with zero.
22681 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22682 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22683 for early expansion of vec_min and vec_max builtins.
22684 (builtin_function_type): Add min/max unsigned variants to those
22685 identified as having unsigned arguments.
22687 2017-06-02 Olivier Hainque <hainque@adacore.com>
22689 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22691 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22693 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22694 Use VALL_F16 iterator rather than VALL.
22696 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22698 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22699 Emit CBNZ inside loop when doing a strong exchange and comparing
22700 against zero. Generate the CC flags after the loop.
22702 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22704 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22705 (dl_section_ref): New.
22706 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22707 On AIX, append an expression to subtract the size of the
22708 section length to dl_section_ref.
22710 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22712 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22713 for early expansion of vector absolute builtins.
22715 2017-06-02 Richard Biener <rguenther@suse.de>
22717 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22718 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22720 2017-06-02 Richard Biener <rguenther@suse.de>
22722 PR tree-optimization/80948
22723 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22724 stmt order as well.
22726 2017-06-02 Richard Biener <rguenther@suse.de>
22728 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22730 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22731 for inductions as relevant.
22733 2017-06-02 Richard Biener <rguenther@suse.de>
22735 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22736 (vectorizable_induction): ... this. Remove dead code.
22738 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22740 * builtins. (expand_builtin_alloca): Remove second parameter and
22741 infer its value from the first parameter instead.
22742 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22744 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22746 PR rtl-optimization/80903
22747 * loop-doloop.c (add_test): Unshare sequence.
22749 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22751 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22753 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22755 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22757 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22758 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22759 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22760 (xlogue_layout::get_stub_rtx): Make static.
22761 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22762 (xlogue_layout::compute_stub_managed_regs): Rename to...
22763 (xlogue_layout::count_stub_managed_regs): ...this.
22764 (xlogue_layout::is_stub_managed_reg): New function.
22765 (xlogue_layout::m_stub_names): Rename to...
22766 (xlogue_layout::s_stub_names): ...this, make static.
22767 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22768 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22769 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22770 xlogue_layout::s_stub_names): Instantiate statics.
22771 (stub_managed_regs): Remove.
22772 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22773 (disable_call_ms2sysv_xlogues): Rename to...
22774 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22775 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22777 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22778 change after reload_completed.
22779 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22781 (ix86_expand_prologue): Likewise.
22782 (ix86_expand_epilogue): Likewise.
22783 (ix86_expand_split_stack_prologue): Likewise.
22784 (ix86_compute_frame_layout): Remove frame parameter ...
22785 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22786 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22788 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22789 (ix86_frame): Move from here ...
22790 * config/i386/i386.h (ix86_frame): ... to here.
22791 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22792 complete ix86_frame data structure instead. Remove some_ld_name.
22794 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22796 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22797 symbols that hold a DECL_VALUE_EXPR.
22799 2017-06-01 Martin Jambor <mjambor@suse.cz>
22801 PR tree-optimization/80898
22802 * tree-sra.c (process_subtree_disqualification): Removed.
22803 (disqualify_candidate): Do not acll
22804 process_subtree_disqualification.
22805 (subtree_mark_written_and_enqueue): New function.
22806 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22807 RHS has been disqualified and re-queue LHS if necessary. Apart
22808 from that, ignore disqualified RHS.
22810 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22812 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22813 address fetch for z10 or later.
22815 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22817 * config/arc/arc.md (tst_movb): Add guard when splitting.
22819 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22821 * config/arc/arc.c (arc_can_eliminate): Test against
22822 arc_frame_pointer_needed.
22824 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22826 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22827 to prevent store reordering.
22828 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22829 (type): Add block type.
22830 (stack_tie): Define special instruction to be used in
22833 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22835 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22836 constraint. It is not valid for the pattern.
22837 (noncommutative_binary_comparison): Likewise.
22839 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22841 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22844 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22846 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22847 be used by the reg-alloc.
22849 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22851 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22852 reg-alloc when having mul64 or mul32x16 instructions.
22853 (mulsidi3): Likewise.
22854 (umulsidi3): Likewise.
22855 (mulsi32x16): New pattern.
22856 (mulsi64): Likewise.
22857 (mulsidi64): Likewise.
22858 (umulsidi64): Likewise.
22859 (MUL32x16_REG): Define.
22860 (mul64_600): Use MUL32x16_REG.
22861 (mac64_600): Likewise.
22862 (umul64_600): Likewise.
22863 (umac64_600): Likewise.
22865 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22867 * config/arc/arc.md (mulsi3_700): Make it commutative.
22869 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22871 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22873 (*sign_extendsidi2_insn): Likewise for movstosw.
22875 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22877 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22878 the type of the input discriminant value. Convert the
22879 discriminant value of signedness vary.
22881 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22883 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22884 Add to -Wall section.
22886 2017-06-01 Richard Biener <rguenther@suse.de>
22888 PR middle-end/66313
22889 * fold-const.c (fold_plusminus_mult_expr): If the factored
22890 factor may be zero use a wrapping type for the inner operation.
22891 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22892 and handle moved defs.
22893 (process_assignment): Properly guard the unary op case. Return a
22894 tri-state indicating that moving the stmt before the call may allow
22895 to continue. Pass through to_move.
22896 (find_tail_calls): Handle moving unrelated defs before
22899 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22902 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22903 splitter result in the canonical way.
22905 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22907 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22908 also for 32bit target. Update insn attributes.
22909 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22911 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22913 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22914 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22915 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22916 (_mm_maskz_min_ss): New intrinsics.
22918 2017-05-31 Martin Liska <mliska@suse.cz>
22920 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22921 Change comment style to one we normally use.
22922 (vectorizable_reduction): Likewise.
22923 (vectorizable_induction): Likewise.
22924 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22925 (vectorizable_call): Likewise.
22926 (vectorizable_simd_clone_call): Likewise.
22927 (vectorizable_conversion): Likewise.
22928 (vectorizable_assignment): Likewise.
22929 (vectorizable_shift): Likewise.
22930 (vectorizable_operation): Likewise.
22931 (vectorizable_store): Likewise.
22932 (vectorizable_load): Likewise.
22933 * tree-vectorizer.h: Likewise.
22935 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22937 * passes.c (emergency_dump_function): New.
22938 * tree-pass.h (emergency_dump_function): Declare.
22939 * plugin.c (plugins_internal_error_function): Remove.
22940 * plugin.h (plugins_internal_error_function): Remove declaration.
22941 * toplev.c (internal_error_function): New static function. Use it...
22942 (general_init): ...here.
22944 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22946 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22947 (arc_rtx_costs): Add costs for new patterns.
22948 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22949 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22951 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22953 * tree-ssa-strlen.c (get_next_strinfo): New function.
22954 (get_stridx_plus_constant): Use it.
22955 (zero_length_string): Likewise.
22956 (adjust_related_strinfos): Likewise.
22957 (adjust_last_stmt): Likewise.
22959 2017-05-31 Richard Biener <rguenther@suse.de>
22962 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22963 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22965 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22967 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22968 loop_vinfo argument and use of dependence distance vectors.
22969 Check instead whether the two references differ only in their
22970 initial value and assume that they have the same alignment if the
22971 difference is a multiple of the vector alignment.
22972 (vect_analyze_data_refs_alignment): Update call accordingly.
22974 2017-05-31 Martin Liska <mliska@suse.cz>
22977 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22979 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22981 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22982 (create_intersect_range_checks): Move from ...
22983 * tree-data-ref.c (create_intersect_range_checks_index)
22984 (create_intersect_range_checks): ... to here.
22985 (create_runtime_alias_checks): New function factored from ...
22986 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22987 here. Call above function.
22988 * tree-data-ref.h (create_runtime_alias_checks): New function.
22990 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22992 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22993 segment length for dr_b and compute it in wide_int.
22995 2017-05-31 Richard Biener <rguenther@suse.de>
22997 PR tree-optimization/80906
22998 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22999 and pass through iv_map.
23000 (copy_bb_and_scalar_dependences): Adjust.
23001 (translate_pending_phi_nodes): Likewise.
23002 (copy_loop_close_phi_args): Handle code-generating IVs instead
23005 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23007 * diagnostic-color.c (color_dict): Add "type-diff".
23008 (parse_gcc_colors): Update comment.
23009 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23010 -fdiagnostics-show-template-tree and -fno-elide-type.
23011 (GCC_COLORS): Add type-diff to example.
23013 (-fdiagnostics-show-template-tree): New.
23014 (-fno-elide-type): New.
23015 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23016 the pp_format_decoder callback. Call any m_format_postprocessor's
23018 (pretty_printer::pretty_printer): Initialize
23019 m_format_postprocessor.
23020 (pretty_printer::~pretty_printer): Delete any
23021 m_format_postprocessor.
23022 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23023 (class format_postprocessor): New class.
23024 (struct pretty_printer::format_decoder): Document the new parameters.
23025 (struct pretty_printer::m_format_postprocessor): New field.
23026 * tree-diagnostic.c (default_tree_printer): Update for new
23027 bool and const char ** params.
23028 * tree-diagnostic.h (default_tree_printer): Likewise.
23030 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23032 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23033 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23034 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23035 rs6000_gen_cell_microcode code.
23036 (rs6000_final_prescan_insn): Delete.
23037 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23038 "warn-cell-microcode" entries.
23039 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23040 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23041 throughout. Change cc_reg_not_micro_cr0_operand to
23042 cc_reg_not_cr0_operand throughout.
23043 (*extendhi<mode>2_noload): Delete.
23044 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23045 (mwarn-cell-microcode): Delete.
23046 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23047 -mgen-cell-microcode and -mwarn-cell-microcode.
23049 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23052 * config/i386/constraints.md (Yd): New constraint.
23054 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23055 and (?Yd, r) alternatives. Update insn attributes.
23056 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23057 and (?*Yd, r) alternatives. Update insn attributes.
23058 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23060 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23062 * gimplify.c (gimplify_modify_expr): Don't create a
23063 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23066 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23068 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23070 2017-05-30 Richard Biener <rguenther@suse.de>
23072 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23073 and reduc_def fields.
23074 (STMT_VINFO_REDUC_TYPE): New define.
23075 (STMT_VINFO_REDUC_DEF): Likewise.
23076 (vect_force_simple_reduction): Adjust prototype.
23077 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23078 (vect_is_simple_reduction): Remove check_reduction argument.
23079 (vect_force_simple_reduction): Adjust and set
23080 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23081 (vectorizable_reduction): Do not re-do reduction analysis
23082 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23083 * tree-parloops.c (gather_scalar_reductions): Adjust.
23085 2017-05-30 Richard Biener <rguenther@suse.de>
23087 PR middle-end/80901
23088 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23091 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23093 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23094 Introduce unknown_misalignment parameter and remove vf.
23095 (vect_peeling_hash_get_lowest_cost):
23096 Pass unknown_misalignment parameter.
23097 (vect_enhance_data_refs_alignment):
23098 Fix unsupportable data ref treatment.
23100 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23102 * tree-vect-data-refs.c (vect_get_data_access_cost):
23103 Workaround for SLP handling.
23104 (vect_enhance_data_refs_alignment):
23105 Compute costs for doing no peeling at all, compare to the best
23106 peeling costs so far and avoid peeling if cheaper.
23108 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23110 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23111 Return peeling info and set costs to zero for unlimited cost
23113 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23114 with unknown misalignment. Compute and costs for unknown
23115 misalignment, compare them to the costs for known misalignment
23116 and choose the cheapest for peeling.
23118 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23120 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23121 (vect_get_peeling_costs_all_drs): Create function.
23122 (vect_peeling_hash_get_lowest_cost):
23123 Use vect_get_peeling_costs_all_drs.
23124 (vect_peeling_supportable): Create function.
23125 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23127 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23129 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23130 DR_HAS_NEGATIVE_STEP.
23131 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23132 (vect_enhance_data_refs_alignment): Use.
23133 (vect_duplicate_ssa_name_ptr_info): Use.
23134 * tree-vectorizer.h (dr_misalignment): Use.
23135 (known_alignment_for_access_p): Use.
23137 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23140 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23142 (has_section_name): New function.
23144 2017-05-30 Martin Liska <mliska@suse.cz>
23147 * auto-profile.c (get_function_decl_from_block): Fix
23150 2017-05-30 Richard Biener <rguenther@suse.de>
23152 PR middle-end/80876
23153 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23155 2017-05-30 Martin Liska <mliska@suse.cz>
23157 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23158 * dumpfile.h (struct dump_file_info): Remove ctors.
23160 2017-05-30 Martin Liska <mliska@suse.cz>
23162 * predict.def: Fix GNU coding style.
23164 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23166 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23167 Mark 'to' argument with ATTRIBUTE_UNUSED.
23169 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23171 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23172 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23173 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23176 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23178 * doc/install.texi (Options specification): Restore entry of
23179 --enable-sjlj-exceptions.
23181 2017-05-27 Michael Eager <eager@eagercon.com>
23184 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23186 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23188 * config/microblaze/microblaze.h
23189 (FIXED_REGISTERS): Update in macro.
23190 (CALL_USED_REGISTERS): Update in macro.
23192 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23194 * doc/install.texi: Add links to macOS binary distributions.
23196 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23200 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23202 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23204 2017-05-26 Martin Liska <mliska@suse.cz>
23206 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23208 2017-05-26 Martin Liska <mliska@suse.cz>
23210 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23211 always leading ';; '.
23212 (dump_bb_info): Likewise.
23213 (brief_dump_cfg): Likewise.
23214 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23215 * dumpfile.c: Remove usage of TDF_VERBOSE.
23216 * dumpfile.h (enum dump_kind): Likewise.
23217 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23218 * print-tree.c (debug_verbose): Remove.
23219 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23220 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23221 * tree-diagnostic.c (default_tree_printer): Replace
23222 TDF_DIAGNOSTIC with TDF_SLIM.
23224 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23226 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23227 in parameter loop, rather than loop_vinfo.
23228 (create_intersect_range_checks): Ditto.
23229 (vect_create_cond_for_alias_checks): Update call to above functions.
23231 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23233 PR tree-optimization/80815
23234 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23235 for merging runtime alias checks. Handle negative DR_STEPs.
23237 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23239 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23241 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23242 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23243 out code pruning runtime alias checks.
23244 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23245 factored out from above.
23246 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23248 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23250 (prune_runtime_alias_test_list): New decalaration.
23252 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23254 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23255 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23256 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23257 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23258 (operator==, comp_dr_with_seg_len_pair): Ditto.
23259 (vect_prune_runtime_alias_test_list): Ditto.
23261 2017-05-26 Martin Liska <mliska@suse.cz>
23264 * params.def: Bound partial-inlining-entry-probability param.
23266 2017-05-26 Marek Polacek <polacek@redhat.com>
23269 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23272 2017-05-26 Richard Biener <rguenther@suse.de>
23274 PR tree-optimization/80842
23275 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23278 2017-05-26 Richard Biener <rguenther@suse.de>
23280 PR tree-optimization/80844
23281 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23283 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23285 * doc/md.texi (Machine Constraints): Update x86 family
23286 machine constraints section to match 'config/i386/constraints.md'.
23288 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23290 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23292 2017-05-25 Nathan Sidwell <nathan@acm.org>
23294 * doc/invoke.texi (--enable-languages): Update documentation.
23296 2017-05-25 Martin Liska <mliska@suse.cz>
23298 * dumpfile.c: Add TDF_FOLDING.
23299 * dumpfile.h (enum dump_kind): Likewise.
23300 * genmatch.c (dt_simplify::gen_1): Use it.
23302 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23304 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23306 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23308 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23309 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23311 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23313 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23314 * match.pd (X == C): Rewrite it here.
23315 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23316 with_certain_nonzero_bits2): New predicates.
23317 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23319 2017-05-24 Nathan Sidwell <nathan@acm.org>
23321 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23324 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23327 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23329 * config/powerpcspe: New port. Files are copied from the rs6000
23330 port, with "rs6000" in filenames replaced by "powerpcspe".
23332 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23334 PR rtl-optimization/80754
23335 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23337 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23339 * config/sparc/sparc.md (length): Return the correct value for -mflat
23340 sibcalls to match output_sibcall.
23342 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23346 * config/rs6000/rs6000.c (struct machine_function): Add new field
23348 (rs6000_get_separate_components): Init that field, use it.
23349 (rs6000_components_for_bb): Use the field.
23351 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23353 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23355 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23357 PR middle-end/80823
23358 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23360 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23363 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23364 address against address_operand predicate.
23365 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23367 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23369 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23370 parameters passed indirectly.
23372 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23374 * config/i386/i386.md (*movdi_internal): Remove SSE4
23375 alternative 18 (?r, *v). Update insn attributes.
23376 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23377 Update insn attributes.
23378 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23379 Update insn attributes.
23380 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23381 alternative 1 (r, v). Remove isa attribute.
23382 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23383 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23384 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23386 2017-05-23 Tom de Vries <tom@codesourcery.com>
23388 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23391 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23393 * cgraphunit.c (symbol_table::process_new_functions): Update.
23394 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23395 (inline_generate_summary): Rename to ...
23396 (ipa_fn_summary_generate): ... this one.
23397 (inline_read_summary): Rename to ...
23398 (ipa_fn_summary_read): ... this one.
23399 (inline_write_summary): Rename to ...
23400 (ipa_fn_summary_write): ... this one.
23401 (inline_free_summary): Rename to ...
23402 (ipa_free_fn_summary): ... this one.
23403 (pass_data_local_fn_summary, pass_local_fn_summary,
23404 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23405 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23406 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23407 make_pass_ipa_fn_summary): New.
23408 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23409 inline_write_summary, inline_free_summary): Remove.
23410 (ipa_free_fn_summary) : New.
23411 * ipa-inline.c (ipa_inline): Update.
23412 (pass_ipa_inline): Do not generate summaries.
23413 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23415 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23416 and add pass_ipa_fn_summary.
23417 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23419 (make_pass_inline_parameters): Remove.
23421 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23423 * omp-low.c (struct omp_context): Remove "default_kind" member.
23426 * omp-offload.c (execute_oacc_device_lower): Remove the
23427 parallelism dimensions function attributes for unparallelized
23428 OpenACC kernels constructs.
23430 2017-05-23 Martin Liska <mliska@suse.cz>
23432 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23434 (cgraph_edge::make_speculative): Likewise.
23435 (cgraph_edge::resolve_speculation): Likewise.
23436 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23437 (cgraph_node::dump): Likewise.
23438 * cgraph.h: Likewise.
23439 * cgraphunit.c (analyze_functions): Likewise.
23440 (symbol_table::compile): Likewise.
23441 * ipa-cp.c (print_all_lattices): Likewise.
23442 (determine_versionability): Likewise.
23443 (initialize_node_lattices): Likewise.
23444 (ipcp_verify_propagated_values): Likewise.
23445 (estimate_local_effects): Likewise.
23446 (update_profiling_info): Likewise.
23447 (create_specialized_node): Likewise.
23448 (perhaps_add_new_callers): Likewise.
23449 (decide_about_value): Likewise.
23450 (decide_whether_version_node): Likewise.
23451 (identify_dead_nodes): Likewise.
23452 (ipcp_store_bits_results): Likewise.
23453 * ipa-devirt.c (dump_targets): Likewise.
23454 (ipa_devirt): Likewise.
23455 * ipa-icf.c (sem_item::dump): Likewise.
23456 (sem_function::equals): Likewise.
23457 (sem_variable::equals): Likewise.
23458 (sem_item_optimizer::read_section): Likewise.
23459 (sem_item_optimizer::execute): Likewise.
23460 (congruence_class::dump): Likewise.
23461 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23462 (dump_inline_summary): Likewise.
23463 (estimate_node_size_and_time): Likewise.
23464 (inline_analyze_function): Likewise.
23465 * ipa-inline-transform.c (inline_call): Likewise.
23466 * ipa-inline.c (report_inline_failed_reason): Likewise.
23467 (want_early_inline_function_p): Likewise.
23468 (edge_badness): Likewise.
23469 (update_edge_key): Likewise.
23470 (inline_small_functions): Likewise.
23471 * ipa-profile.c (ipa_profile): Likewise.
23472 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23473 (ipa_make_edge_direct_to_target): Likewise.
23474 (remove_described_reference): Likewise.
23475 (ipa_impossible_devirt_target): Likewise.
23476 (propagate_controlled_uses): Likewise.
23477 (ipa_print_node_params): Likewise.
23478 (ipcp_transform_function): Likewise.
23479 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23480 (propagate_pure_const): Likewise.
23481 * ipa-reference.c (generate_summary): Likewise.
23482 (read_write_all_from_decl): Likewise.
23483 (propagate): Likewise.
23484 (ipa_reference_read_optimization_summary): Likewise.
23485 * ipa-utils.c (ipa_merge_profiles): Likewise.
23486 * ipa.c (walk_polymorphic_call_targets): Likewise.
23487 (symbol_table::remove_unreachable_nodes): Likewise.
23488 (ipa_single_use): Likewise.
23489 * passes.c (execute_todo): Likewise.
23490 * predict.c (drop_profile): Likewise.
23491 * symtab.c (symtab_node::get_dump_name): New function.
23492 (symtab_node::dump_name): Likewise.
23493 (symtab_node::dump_asm_name): Likewise.
23494 (symtab_node::dump_references): Likewise.
23495 (symtab_node::dump_referring): Likewise.
23496 (symtab_node::dump_base): Likewise.
23497 (symtab_node::debug_symtab): Likewise.
23498 * tree-sra.c (convert_callers_for_node): Likewise.
23499 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23500 * value-prof.c (init_node_map): Likewise.
23502 2017-05-23 Martin Liska <mliska@suse.cz>
23504 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23505 and symtab_node::debug_symtab to symbol_table::debug.
23506 * cgraphunit.c (analyze_functions): Use the renamed function.
23507 (symbol_table::compile): Likewise.
23508 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23509 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23510 * passes.c (execute_todo): Likewise.
23511 * symtab.c (symbol_table::dump): New function.
23512 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23514 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23516 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23517 that nonconst implies exec.
23519 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23521 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23522 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23523 (inline_edge_summary_vec): Turn into ...
23524 (ipa_call_summaries): ... this one.
23525 (redirect_to_unreachable, edge_set_predicate,
23526 evaluate_properties_for_edge, inline_summary_alloc,
23527 reset_ipa_call_summary, reset_inline_summary,
23528 inline_summary_t::duplicate): Update.
23529 (inline_edge_duplication_hook): Turn to ...
23530 (ipa_call_summary_t::duplicate): ... this one.
23531 (inline_edge_removal_hook): Turn to ...
23532 (ipa_call_summary_t::remove): ... this one.
23533 (dump_inline_edge_summary): Turn to ...
23534 (dump_ipa_call_summary): ... this one.
23535 (estimate_function_body_sizes): Update.
23536 (inline_update_callee_summaries): Update.
23537 (remap_edge_change_prob): Update.
23538 (remap_edge_summaries): Update.
23539 (inline_merge_summary): Update.
23540 (do_estimate_edge_time): Update.
23541 (inline_generate_summary): Update.
23542 (inline_read_section): Update.
23543 (inline_read_summary): Update.
23544 (inline_free_summary): Update.
23545 * ipa-inline.c (can_inline_edge_p): Update.
23546 (compute_inlined_call_time): Update.
23547 (want_inline_small_function_p): Update.
23548 (edge_badness): Update.
23549 (early_inliner): Update.
23550 * ipa-inline.h (inline_edge_summary): Turn to ...
23551 (ipa_call_summary): ... this one.
23552 (ipa_call_summary_t): New class.
23553 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23554 (ipa_call_summaries): New.
23555 (inline_edge_summary): Remove.
23556 (estimate_edge_growth): Update.
23557 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23558 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23559 * ipa-split.c (execute_split_functions): Update.
23560 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23562 2017-05-23 Tom de Vries <tom@codesourcery.com>
23564 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23565 attributes): Document rdrand effective target.
23567 2017-05-23 Tom de Vries <tom@codesourcery.com>
23569 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23570 attributes): Sort alphabetically.
23572 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23574 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23576 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23579 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23580 V2DF/V2DI splat into two separate patterns, one that handles
23581 registers, and the other that only handles memory. Drop support
23582 for splatting from a GPR on ISA 2.07 and then splitting the
23583 splat into direct move and splat.
23584 (vsx_splat_<mode>_reg): Likewise.
23585 (vsx_splat_<mode>_mem): Likewise.
23587 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23589 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23591 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23593 PR middle-end/80809
23594 * omp-low.c (finish_taskreg_remap): New function.
23595 (finish_taskreg_scan): If unit size of ctx->record_type
23596 is non-constant, unshare the size expression and replace
23597 decls in it with possible outer var refs.
23599 PR middle-end/80809
23600 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23601 GOVD_SHARED rather than GOVD_PRIVATE with it.
23602 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23603 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23605 PR middle-end/80853
23606 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23607 as last argument to build_outer_var_ref for pointer bases of array
23608 section reductions.
23610 2017-05-19 Martin Sebor <msebor@redhat.com>
23612 * print-tree.c (print_node): Print DECL_READ_P flag.
23614 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23616 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23617 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23618 * cgraph.c: Likewise.
23619 * cgraphunit.c: Likewise.
23620 * gengtype.c: Likewise.
23621 * ipa-cp.c: Likewise.
23622 * ipa-devirt.c: Likewise.
23623 * ipa-icf.c: Likewise.
23624 * ipa-predicate.c: Likewise.
23625 * ipa-profile.c: Likewise.
23626 * ipa-prop.c: Likewise.
23627 * ipa-split.c: Likewise.
23629 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23630 edge_predicate_pool, dump_inline_hints,
23631 inline_summary::account_size_time, redirect_to_unreachable,
23632 edge_set_predicate, set_hint_predicate,
23633 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23634 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23635 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23636 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23637 ipa_call_summary_t::remove, initialize_growth_caches,
23638 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23639 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23640 mark_modified, unmodified_parm_1, unmodified_parm,
23641 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23642 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23643 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23644 will_be_nonconstant_predicate, record_modified_bb_info,
23645 get_minimal_bb, record_modified, param_change_prob,
23646 phi_result_unknown_predicate, predicate_for_phi_result,
23647 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23648 estimate_function_body_sizes, compute_inline_parameters,
23649 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23650 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23651 inline_update_callee_summaries, remap_edge_change_prob,
23652 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23653 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23654 inline_analyze_function, inline_summary_t::insert,
23655 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23656 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23657 inline_free_summary): Move to ipa-fnsummary.h
23658 (predicate_t): Remove.
23659 * ipa-fnsummary.c: New file.
23660 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23661 (enum inline_hints_vals, inline_hints, agg_position_info,
23662 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23663 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23664 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23665 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23666 inline_read_summary, inline_write_summary, inline_free_summary,
23667 inline_analyze_function, initialize_inline_failed,
23668 inline_merge_summary, inline_update_overall_summary,
23669 compute_inline_parameters): Move to ipa-fnsummary.h
23670 * ipa-fnsummary.h: New file.
23671 * ipa-inline-transform.h: Include ipa-inline.h.
23672 * ipa-inline.c: LIkewise.
23674 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23676 * ipa-inline.c (edge_badness): Use inlined_time instead of
23677 inline_summaries->get.
23679 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23681 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23683 2017-05-22 Nathan Sidwell <nathan@acm.org>
23685 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23686 (fdump-lang): Document 'raw' option.
23687 * dumpfile.h (TDI_tu): Delete.
23688 * dumpfile.c (dump_files): Remove translation-unit.
23689 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23691 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23693 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23694 command option from $(AWK) call.
23695 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23696 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23697 [FORMAT]: Remove handling of variable.
23698 * config/avr/t-multilib: Regenerate.
23700 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23702 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23704 (dump_inline_summary): Do not print self_time.
23705 (estimate_function_body_sizes): Do not set self_time.
23706 (compute_inline_parameters): Likewise.
23707 (inline_read_section, inline_write_summary): Do not stream self_time.
23708 * ipa-inline.h (inline_summary): Drop self_time.
23710 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23712 * ipa-inline-analysis.c (account_size_time): Rename to ...
23713 (inline_summary::account_size_time): ... this one.
23714 (reset_ipa_call_summary): Turn to ...
23715 (ipa_call_summary::reset): ... this one.
23716 (reset_inline_summary): Turn to ...
23717 (inline_summary::reset): ... this one.
23718 (inline_summary_t::remove): Update.
23719 (inline_summary_t::duplicate): Update.
23720 (ipa_call_summary_t::remove): Update.
23721 (dump_inline_summary): Update.
23722 (estimate_function_body_sizes): Update.
23723 (compute_inline_parameters): Update.
23724 (estimate_node_size_and_time): Update.
23725 (inline_merge_summary): Update.
23726 (inline_update_overall_summary): Update.
23727 (inline_read_section): Update.
23728 (inline_write_summary): Update.
23729 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23730 add account_size_time and reset member functions.
23731 (ipa_call_summary): Add reset function.
23732 * ipa-predicate.h (predicate::operator &): Constify.
23734 2017-05-22 Richard Biener <rguenther@suse.de>
23736 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23738 2017-05-19 Jason Merrill <jason@redhat.com>
23740 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23742 2017-05-19 Marek Polacek <polacek@redhat.com>
23745 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23746 TYPE_OVERFLOW_WRAPS checks.
23748 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23750 * tree-core.h (enum omp_clause_default_kind): Add
23751 "OMP_CLAUSE_DEFAULT_PRESENT".
23752 * tree-pretty-print.c (dump_omp_clause): Handle it.
23753 * gimplify.c (enum gimplify_omp_var_data): Add
23754 "GOVD_MAP_FORCE_PRESENT".
23755 (gimplify_adjust_omp_clauses_1): Map it to
23756 "GOMP_MAP_FORCE_PRESENT".
23757 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23759 * gimplify.c (oacc_default_clause): Clarify.
23761 2017-05-19 Nathan Sidwell <nathan@acm.org>
23763 LANG_HOOK_REGISTER_DUMPS
23764 * toplev.c (general_init): Call register dump lang hook.
23765 * doc/invoke.texi: Document -fdump-lang option family.
23766 * dumpfile.c (dump_files): Remove class dump here.
23767 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23768 * dumpfile.h (tree_dump_index): Remove TDI_class.
23769 * langhooks-def.h (lhd_register_dumps): Declare.
23770 (LANG_HOOKS_REGISTER_DUMPS): Define.
23771 (LANG_HOOKS_INITIALIZER): Add it.
23772 * langhooks.c (lhd_register_dumps): Define.
23773 * langhooks.h (struct lang_hooks): Add register_dumps.
23775 2017-05-19 Nathan Sidwell <nathan@acm.org>
23777 * context.h (context::set_passes): New.
23778 * context.c (context::context): Do not create pass manager.
23779 * toplev.c (general_init): Create pass manager here.
23781 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23783 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23784 use this splitter if two add or or instructions would also work for
23785 the constant we want to generate.
23787 2017-05-19 Richard Biener <rguenther@suse.de>
23790 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23791 predicate evaluation.
23793 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23795 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23797 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23800 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23802 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23803 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23804 (inline_edge_summary_vec): Turn into ...
23805 (ipa_call_summaries): ... this one.
23806 (redirect_to_unreachable, edge_set_predicate,
23807 evaluate_properties_for_edge, inline_summary_alloc,
23808 reset_ipa_call_summary, reset_inline_summary,
23809 inline_summary_t::duplicate): Update.
23810 (inline_edge_duplication_hook): Turn to ...
23811 (ipa_call_summary_t::duplicate): ... this one.
23812 (inline_edge_removal_hook): Turn to ...
23813 (ipa_call_summary_t::remove): ... this one.
23814 (dump_inline_edge_summary): Turn to ...
23815 (dump_ipa_call_summary): ... this one.
23816 (estimate_function_body_sizes): Update.
23817 (inline_update_callee_summaries): Update.
23818 (remap_edge_change_prob): Update.
23819 (remap_edge_summaries): Update.
23820 (inline_merge_summary): Update.
23821 (do_estimate_edge_time): Update.
23822 (inline_generate_summary): Update.
23823 (inline_read_section): Update.
23824 (inline_read_summary): Update.
23825 (inline_free_summary): Update.
23826 * ipa-inline.c (can_inline_edge_p): Update.
23827 (compute_inlined_call_time): Update.
23828 (want_inline_small_function_p): Update.
23829 (edge_badness): Update.
23830 (early_inliner): Update.
23831 * ipa-inline.h (inline_edge_summary): Turn to ...
23832 (ipa_call_summary): ... this one.
23833 (ipa_call_summary_t): New class.
23834 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23835 (ipa_call_summaries): New.
23836 (inline_edge_summary): Remove.
23837 (estimate_edge_growth): Update.
23838 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23839 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23840 * ipa-split.c (execute_split_functions): Update.
23841 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23843 2017-05-19 Richard Biener <rguenther@suse.de>
23845 PR middle-end/80764
23846 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23848 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23850 * config/rs6000/rs6000.c (struct machine_function): Add field
23851 fpr_is_wrapped_separately.
23852 (rs6000_get_separate_components): Use 64 components. Handle the
23853 new FPR components.
23854 (rs6000_components_for_bb): Handle the FPR components.
23855 (rs6000_emit_prologue_components): Handle the FPR components.
23856 (rs6000_emit_epilogue_components): Handle the FPR components.
23857 (rs6000_set_handled_components): Handle the FPR components.
23858 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23859 that are already separately shrink-wrapped.
23860 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23861 that are already separately shrink-wrapped.
23863 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23866 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23869 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23870 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23871 cases where the register allocator uses a move and an offsettable
23872 memory operation to/from a FPR register on ISA 2.06/2.07.
23873 (define_peephole2 for Altivec d-form store): Likewise.
23875 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23878 * config/i386/mmx.md (*mov<mode>_internal): Enable
23879 alternatives 11, 12, 13 and 14 also for 32bit targets.
23880 Remove alternatives 15, 16, 17 and 18.
23881 * config/i386/sse.md (vec_concatv2di): Change
23882 alternative (!x, *y) to (x, ?!*Yn).
23884 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23886 * dumpfile.h (enum dump_kind): Remove stray comma.
23888 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23890 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23891 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23892 predicate::num_conditions
23893 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23894 (CHANGED): turn into predicate::changed.
23895 (agg_position_info): Move to ipa-predicate.h
23896 (add_condition, predicate::add_clause, predicate::operator &=,
23897 predicate::or_with, predicate::evaluate, predicate::probability,
23898 dump_condition, dump_clause, predicate::dump,
23899 predicate::remap_after_duplication, predicate::remap_after_inlining,
23900 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23901 (evaluate_conditions_for_known_args): Update.
23902 (set_cond_stmt_execution_predicate): Update.
23903 * ipa-inline.h: Include ipa-predicate.h
23904 (condition, inline_param_summary, conditions, agg_position_info,
23905 predicate): Move to ipa-predicate.h
23906 * ipa-predicate.c: New file.
23907 * ipa-predicate.h: New file.
23909 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23911 * final.c (leaf_function_p): Check we are not in a sequence.
23913 2017-05-18 Martin Liska <mliska@suse.cz>
23915 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23916 * dumpfile.c (dump_register): Use new enum dump_kind.
23917 (get_dump_file_name): Likewise.
23918 (dump_enable_all): Likewise.
23919 (dump_switch_p_1): Likewise.
23920 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23921 * dumpfile.h (enum dump_kind): New enum type.
23922 (struct dump_file_info): Create constructor and
23923 format fields and comments.
23924 * passes.c (pass_manager::register_one_dump_file):
23926 * statistics.c (statistics_early_init): Likewise.
23927 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23928 TDF_TREE with TDF_SLIM.
23929 (gather_memory_references_ref): Likewise.
23931 2017-05-18 Martin Liska <mliska@suse.cz>
23933 * vec.h (struct vnull): Use it.
23935 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23937 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23938 (true_predicate, false_predicate, true_predicate_p,
23939 false_predicate_p): Remove.
23940 (single_cond_predicate, not_inlined_predicate): Turn to member function
23942 (add_condition): Update.
23943 (add_clause): Turn to...
23944 (predicate::add_clause): ... this one; update; allow passing NULL
23946 (and_predicates): Turn to ...
23947 (predicate::operator &=): ... this one.
23948 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23949 (or_predicates): Turn to ...
23950 (predicate::or_with): ... this one.
23951 (evaluate_predicate): Turn to ...
23952 (predicate::evaluate): ... this one.
23953 (predicate_probability): Turn to ...
23954 (predicate::probability): ... this one.
23955 (dump_condition): Update.
23956 (dump_predicate): Turn to ...
23957 (predicate::dump): ... this one.
23958 (account_size_time): Update.
23959 (edge_set_predicate): Update.
23960 (set_hint_predicate): UPdate.
23961 (evaluate_conditions_for_known_args): Update.
23962 (evaluate_properties_for_edge): Update.
23963 (remap_predicate_after_duplication): Turn to...
23964 (predicate::remap_after_duplication): ... this one.
23965 (remap_hint_predicate_after_duplication): Update.
23966 (inline_summary_t::duplicate): UPdate.
23967 (dump_inline_edge_summary): Update.
23968 (dump_inline_summary): Update.
23969 (set_cond_stmt_execution_predicate): Update.
23970 (set_switch_stmt_execution_predicate): Update.
23971 (compute_bb_predicates): Update.
23972 (will_be_nonconstant_expr_predicate): Update.
23973 (will_be_nonconstant_predicate): Update.
23974 (phi_result_unknown_predicate): Update.
23975 (predicate_for_phi_result): Update.
23976 (array_index_predicate): Update.
23977 (estimate_function_body_sizes): Update.
23978 (estimate_node_size_and_time): Update.
23979 (estimate_ipcp_clone_size_and_time): Update.
23980 (remap_predicate): Rename to ...
23981 (predicate::remap_after_inlining): ... this one.
23982 (remap_hint_predicate): Update.
23983 (inline_merge_summary): Update.
23984 (inline_update_overall_summary): Update.
23985 (estimate_size_after_inlining): Update.
23986 (read_predicate): Rename to ...
23987 (predicate::stream_in): ... this one.
23988 (read_inline_edge_summary): Update.
23989 (write_predicate): Rename to ...
23990 (predicate::stream_out): ... this one.
23991 (write_inline_edge_summary): Update.
23992 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23993 (clause_t): Turn to uint32_t
23994 (predicate): Turn to class; implement constructor and operators
23996 (size_time_entry): Update.
23997 (inline_summary): Update.
23998 (inline_edge_summary): Update.
24000 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24002 * fold-const.c (fold_binary_loc): Move transformation...
24003 * match.pd (C - X CMP X): ... here.
24005 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24007 * config/sparc/sparc.c (sparc_option_override): Set function
24008 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24009 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24011 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24013 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24015 2017-05-18 Marek Polacek <polacek@redhat.com>
24018 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24019 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24020 gimple_assign_load_p.
24022 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24024 PR middle-end/80692
24025 * real.c (do_compare): Give decimal_do_compare preference over
24026 comparing just the signs.
24028 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24030 * doc/md.texi (Canonicalization of Instructions): Describe the
24031 canonical form of instructions that inherently set a condition
24034 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24036 PR middle-end/80775
24037 * tree-cfg.c: Move deletion of unreachable case statements to after
24038 the merging of consecutive case labels.
24040 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24042 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24043 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24044 restoring of callee-saved registers.
24046 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24048 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24049 * config/visium/visium.c (single_set_and_flags): Likewise.
24050 * config/visium/visium.md (Substitutions): Likewise.
24052 2017-05-17 Martin Liska <mliska@suse.cz>
24054 * cfg.c: Introduce dump_flags_t type and
24055 use it instead of int type.
24057 * cfghooks.c: Likewise.
24058 * cfghooks.h (struct cfg_hooks): Likewise.
24059 * cfgrtl.c: Likewise.
24060 * cfgrtl.h: Likewise.
24061 * cgraph.c (cgraph_node::get_body): Likewise.
24062 * coretypes.h: Likewise.
24063 * domwalk.c: Likewise.
24064 * domwalk.h: Likewise.
24065 * dumpfile.c (struct dump_option_value_info): Likewise.
24066 (dump_enable_all): Likewise.
24067 (dump_switch_p_1): Likewise.
24068 (opt_info_switch_p): Likewise.
24069 * dumpfile.h (enum tree_dump_index): Likewise.
24070 (struct dump_file_info): Likewise.
24071 * genemit.c: Likewise.
24072 * generic-match-head.c: Likewise.
24073 * gengtype.c (open_base_files): Likewise.
24074 * gimple-pretty-print.c: Likewise.
24075 * gimple-pretty-print.h: Likewise.
24076 * graph.c (print_graph_cfg): Likewise.
24077 * graphite-scop-detection.c (dot_all_sese): Likewise.
24078 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24079 * loop-unroll.c (report_unroll): Likewise.
24080 * passes.c (pass_manager::register_one_dump_file): Likewise.
24081 * print-tree.c: Likewise.
24082 * statistics.c: Likewise.
24083 * tree-cfg.c: Likewise.
24084 * tree-cfg.h: Likewise.
24085 * tree-dfa.c: Likewise.
24086 * tree-dfa.h: Likewise.
24087 * tree-dump.c (dump_function): Likewise.
24088 * tree-dump.h (struct dump_info): Likewise.
24089 * tree-pretty-print.c: Likewise.
24090 * tree-pretty-print.h: Likewise.
24091 * tree-ssa-live.c: Likewise.
24092 * tree-ssa-live.h: Likewise.
24093 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24094 * tree-vect-loop.c: Likewise.
24095 * tree-vect-slp.c: Likewise.
24097 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24098 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24100 PR tree-optimization/80457
24101 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24102 of all arguments to a statement as scalar_to_vec operations.
24103 (vectorizable_call): Adjust call to vect_model_simple_cost for
24105 (vectorizable_conversion): Likewise.
24106 (vectorizable_assignment): Likewise.
24107 (vectorizable_shift): Likewise.
24108 (vectorizable_operation): Likewise.
24109 (vectorizable_comparison): Likewise.
24110 (vect_is_simple_cond): Record the def types for operands.
24111 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24112 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24113 for statement argument count.
24115 2017-05-16 Carl Love <cel@us.ibm.com>
24117 * config/rs6000/rs6000-c: Add support for built-in functions
24118 vector unsigned long long vec_bperm (vector unsigned long long,
24119 vector unsigned char)
24120 vector signed long long vec_mule (vector signed int,
24122 vector unsigned long long vec_mule (vector unsigned int,
24123 vector unsigned int)
24124 vector signed long long vec_mulo (vector signed int,
24126 vector unsigned long long vec_mulo (vector unsigned int,
24127 vector unsigned int)
24128 vector signed char vec_sldw (vector signed char,
24129 vector signed char,
24131 vector unsigned char vec_sldw (vector unsigned char,
24132 vector unsigned char,
24134 vector signed short vec_sldw (vector signed short,
24135 vector signed short,
24137 vector unsigned short vec_sldw (vector unsigned short,
24138 vector unsigned short,
24140 vector signed int vec_sldw (vector signed int,
24143 vector unsigned int vec_sldw (vector unsigned int,
24144 vector unsigned int,
24146 vector signed long long vec_sldw (vector signed long long,
24147 vector signed long long,
24149 vector unsigned long long vec_sldw (vector unsigned long long,
24150 vector unsigned long long,
24152 * config/rs6000/rs6000-c: Add support for built-in functions
24153 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24154 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24155 * doc/extend.texi: Update the built-in documentation for the
24156 new built-in functions.
24158 2017-05-16 Marek Polacek <polacek@redhat.com>
24162 * tree.c (save_expr): Don't fold the expression.
24164 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24166 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24167 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24168 and (?*y,m). Update insn attributes.
24170 2017-05-16 Martin Liska <mliska@suse.cz>
24172 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24173 flags argument of print_gimple_stmt, print_gimple_expr,
24174 print_generic_stmt and print_generic_expr.
24175 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24176 * coretypes.h: Likewise.
24177 * except.c (dump_eh_tree): Likewise.
24178 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24179 * gimple-pretty-print.h: Likewise.
24180 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24181 (backprop::push_to_worklist): Likewise.
24182 (backprop::pop_from_worklist): Likewise.
24183 (backprop::process_use): Likewise.
24184 (backprop::intersect_uses): Likewise.
24185 (note_replacement): Likewise.
24186 * gimple-ssa-store-merging.c
24187 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24188 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24189 (pass_store_merging::execute): Likewise.
24190 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24191 (ssa_base_cand_dump_callback): Likewise.
24192 (dump_incr_vec): Likewise.
24193 (replace_refs): Likewise.
24194 (replace_mult_candidate): Likewise.
24195 (create_add_on_incoming_edge): Likewise.
24196 (create_phi_basis): Likewise.
24197 (insert_initializers): Likewise.
24198 (all_phi_incrs_profitable): Likewise.
24199 (introduce_cast_before_cand): Likewise.
24200 (replace_one_candidate): Likewise.
24201 * gimplify.c (gimplify_expr): Likewise.
24202 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24203 (set_rename): Likewise.
24204 (rename_uses): Likewise.
24205 (copy_loop_phi_nodes): Likewise.
24206 (add_close_phis_to_merge_points): Likewise.
24207 (copy_loop_close_phi_args): Likewise.
24208 (copy_cond_phi_args): Likewise.
24209 (graphite_copy_stmts_from_block): Likewise.
24210 (translate_pending_phi_nodes): Likewise.
24211 * graphite-poly.c (print_pdr): Likewise.
24212 (dump_gbb_cases): Likewise.
24213 (dump_gbb_conditions): Likewise.
24214 (print_scop_params): Likewise.
24215 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24216 (build_cross_bb_scalars_use): Likewise.
24217 (gather_bbs::before_dom_children): Likewise.
24218 * hsa-dump.c (dump_hsa_immed): Likewise.
24219 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24220 (get_replacement_map): Likewise.
24221 * ipa-inline-analysis.c (dump_condition): Likewise.
24222 (estimate_function_body_sizes): Likewise.
24223 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24224 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24225 * ipa-prop.c (ipa_dump_param): Likewise.
24226 (ipa_print_node_jump_functions_for_edge): Likewise.
24227 (ipa_modify_call_arguments): Likewise.
24228 (ipa_modify_expr): Likewise.
24229 (ipa_dump_param_adjustments): Likewise.
24230 (ipa_dump_agg_replacement_values): Likewise.
24231 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24232 * ipa-pure-const.c (check_stmt): Likewise.
24233 (pass_nothrow::execute): Likewise.
24234 * ipa-split.c (execute_split_functions): Likewise.
24235 * omp-offload.c (dump_oacc_loop_part): Likewise.
24236 (dump_oacc_loop): Likewise.
24237 * trans-mem.c (tm_log_emit): Likewise.
24238 (tm_memopt_accumulate_memops): Likewise.
24239 (dump_tm_memopt_set): Likewise.
24240 (dump_tm_memopt_transform): Likewise.
24241 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24242 (print_loop): Likewise.
24243 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24244 (chkp_gather_checks_info): Likewise.
24245 (chkp_get_check_result): Likewise.
24246 (chkp_remove_check_if_pass): Likewise.
24247 (chkp_use_outer_bounds_if_possible): Likewise.
24248 (chkp_reduce_bounds_lifetime): Likewise.
24249 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24250 (chkp_mark_completed_bounds): Likewise.
24251 (chkp_register_incomplete_bounds): Likewise.
24252 (chkp_mark_invalid_bounds): Likewise.
24253 (chkp_maybe_copy_and_register_bounds): Likewise.
24254 (chkp_build_returned_bound): Likewise.
24255 (chkp_get_bound_for_parm): Likewise.
24256 (chkp_build_bndldx): Likewise.
24257 (chkp_get_bounds_by_definition): Likewise.
24258 (chkp_generate_extern_var_bounds): Likewise.
24259 (chkp_get_bounds_for_decl_addr): Likewise.
24260 * tree-chrec.c (chrec_apply): Likewise.
24261 * tree-data-ref.c (dump_data_reference): Likewise.
24262 (dump_subscript): Likewise.
24263 (dump_data_dependence_relation): Likewise.
24264 (analyze_overlapping_iterations): Likewise.
24265 * tree-inline.c (expand_call_inline): Likewise.
24266 (tree_function_versioning): Likewise.
24267 * tree-into-ssa.c (dump_defs_stack): Likewise.
24268 (dump_currdefs): Likewise.
24269 (dump_names_replaced_by): Likewise.
24270 (dump_update_ssa): Likewise.
24271 (update_ssa): Likewise.
24272 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24273 * tree-parloops.c (build_new_reduction): Likewise.
24274 (try_create_reduction_list): Likewise.
24275 (ref_conflicts_with_region): Likewise.
24276 (oacc_entry_exit_ok_1): Likewise.
24277 (oacc_entry_exit_single_gang): Likewise.
24278 * tree-pretty-print.h: Likewise.
24279 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24280 (get_scalar_evolution): Likewise.
24281 (add_to_evolution): Likewise.
24282 (get_loop_exit_condition): Likewise.
24283 (analyze_evolution_in_loop): Likewise.
24284 (analyze_initial_condition): Likewise.
24285 (analyze_scalar_evolution): Likewise.
24286 (instantiate_scev): Likewise.
24287 (number_of_latch_executions): Likewise.
24288 (gather_chrec_stats): Likewise.
24289 (final_value_replacement_loop): Likewise.
24290 (scev_const_prop): Likewise.
24291 * tree-sra.c (dump_access): Likewise.
24292 (disqualify_candidate): Likewise.
24293 (create_access): Likewise.
24294 (reject): Likewise.
24295 (maybe_add_sra_candidate): Likewise.
24296 (create_access_replacement): Likewise.
24297 (analyze_access_subtree): Likewise.
24298 (analyze_all_variable_accesses): Likewise.
24299 (sra_modify_assign): Likewise.
24300 (initialize_constant_pool_replacements): Likewise.
24301 (find_param_candidates): Likewise.
24302 (decide_one_param_reduction): Likewise.
24303 (replace_removed_params_ssa_names): Likewise.
24304 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24305 * tree-ssa-copy.c (dump_copy_of): Likewise.
24306 (copy_prop_visit_cond_stmt): Likewise.
24307 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24308 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24309 (record_equivalences_from_stmt): Likewise.
24310 * tree-ssa-dse.c (compute_trims): Likewise.
24311 (delete_dead_call): Likewise.
24312 (delete_dead_assignment): Likewise.
24313 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24314 (forward_propagate_into_cond): Likewise.
24315 (pass_forwprop::execute): Likewise.
24316 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24317 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24319 (move_computations_worker): Likewise.
24320 (execute_sm): Likewise.
24321 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24322 (remove_exits_and_undefined_stmts): Likewise.
24323 (remove_redundant_iv_tests): Likewise.
24324 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24325 (adjust_iv_update_pos): Likewise.
24326 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24327 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24328 (value_replacement): Likewise.
24329 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24330 * tree-ssa-pre.c (print_pre_expr): Likewise.
24331 (get_representative_for): Likewise.
24332 (create_expression_by_pieces): Likewise.
24333 (insert_into_preds_of_block): Likewise.
24334 (eliminate_insert): Likewise.
24335 (eliminate_dom_walker::before_dom_children): Likewise.
24336 (eliminate): Likewise.
24337 (remove_dead_inserted_code): Likewise.
24338 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24339 * tree-ssa-reassoc.c (get_rank): Likewise.
24340 (eliminate_duplicate_pair): Likewise.
24341 (eliminate_plus_minus_pair): Likewise.
24342 (eliminate_not_pairs): Likewise.
24343 (undistribute_ops_list): Likewise.
24344 (eliminate_redundant_comparison): Likewise.
24345 (update_range_test): Likewise.
24346 (optimize_range_tests_var_bound): Likewise.
24347 (optimize_vec_cond_expr): Likewise.
24348 (rewrite_expr_tree): Likewise.
24349 (rewrite_expr_tree_parallel): Likewise.
24350 (linearize_expr): Likewise.
24351 (break_up_subtract): Likewise.
24352 (linearize_expr_tree): Likewise.
24353 (attempt_builtin_powi): Likewise.
24354 (attempt_builtin_copysign): Likewise.
24355 (transform_stmt_to_copy): Likewise.
24356 (transform_stmt_to_multiply): Likewise.
24357 (dump_ops_vector): Likewise.
24358 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24359 (print_scc): Likewise.
24360 (set_ssa_val_to): Likewise.
24361 (visit_reference_op_store): Likewise.
24362 (visit_use): Likewise.
24363 (sccvn_dom_walker::before_dom_children): Likewise.
24364 (run_scc_vn): Likewise.
24365 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24367 (expr_hash_elt::print): Likewise.
24368 (const_and_copies::pop_to_marker): Likewise.
24369 (const_and_copies::record_const_or_copy_raw): Likewise.
24370 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24371 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24372 (dump_predicates): Likewise.
24373 (find_uninit_use): Likewise.
24374 (warn_uninitialized_phi): Likewise.
24375 (pass_late_warn_uninitialized::execute): Likewise.
24376 * tree-ssa.c (verify_vssa): Likewise.
24377 (verify_ssa): Likewise.
24378 (maybe_optimize_var): Likewise.
24379 * tree-vrp.c (dump_value_range): Likewise.
24380 (dump_all_value_ranges): Likewise.
24381 (dump_asserts_for): Likewise.
24382 (register_edge_assert_for_2): Likewise.
24383 (vrp_visit_cond_stmt): Likewise.
24384 (vrp_visit_switch_stmt): Likewise.
24385 (vrp_visit_stmt): Likewise.
24386 (vrp_visit_phi_node): Likewise.
24387 (simplify_cond_using_ranges_1): Likewise.
24388 (fold_predicate_in): Likewise.
24389 (evrp_dom_walker::before_dom_children): Likewise.
24390 (evrp_dom_walker::push_value_range): Likewise.
24391 (evrp_dom_walker::pop_value_range): Likewise.
24392 (execute_early_vrp): Likewise.
24394 2017-05-16 Richard Biener <rguenther@suse.de>
24396 * dwarf2out.c (loc_list_from_tree_1): Do not create
24397 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24399 2017-05-16 Richard Biener <rguenther@suse.de>
24401 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24403 (note_variable_value_in_expr): If we resolved the decl ref
24404 do not push to the stack.
24406 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24408 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24409 operations in fast-math mode.
24410 (vaddq_f16): Likewise.
24411 (vmul_f16): Likewise.
24412 (vmulq_f16): Likewise.
24413 (vsub_f16): Likewise.
24414 (vsubq_f16): Likewise.
24415 * config/arm/neon.md (add<mode>3): New.
24417 (fma:<VH:mode>3): New. Also remove outdated comment.
24420 2017-05-16 Martin Liska <mliska@suse.cz>
24424 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24425 (odr_types_equivalent_p): Likewise.
24427 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24429 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24431 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24434 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24435 non-interunit SSE move alternatives with '?'.
24436 (zero-extendsidi peephole2): New peephole to skip intermediate
24437 general register in SSE zero-extend sequence.
24439 2017-05-15 Jeff Law <law@redhat.com>
24441 * reorg.c (relax_delay_slots): Create a new variable to hold
24442 the temporary target rather than clobbering TARGET_LABEL.
24444 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24445 missing argument to extract_bit_field call.
24446 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24448 2017-05-15 Martin Liska <mliska@suse.cz>
24451 * gcc.c (process_command): Do not allow empty argument of -o option.
24453 2017-05-15 Renlin Li <renlin.li@arm.com>
24455 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24456 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24457 * config/aarch64/constraints.md (Usf): Add long call check.
24458 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24459 (call_value): Likewise.
24460 (sibcall): Likewise.
24461 (sibcall_value): Likewise.
24463 (call_value_insn): New.
24464 (sibcall_insn): Update rtx pattern.
24465 (sibcall_value_insn): Likewise.
24466 (call_internal): Remove.
24467 (call_value_internal): Likewise.
24468 (sibcall_internal): Likewise.
24469 (sibcall_value_internal): Likewise.
24470 (call_reg): Likewise.
24471 (call_symbol): Likewise.
24472 (call_value_reg): Likewise.
24473 (call_value_symbol): Likewise.
24475 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24478 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24480 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24482 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24483 compatible with CCGOCmode and with CCZmode.
24485 2017-05-14 Martin Sebor <msebor@redhat.com>
24487 PR middle-end/77671
24488 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24489 (gimple_fold_builtin_snprintf): Same.
24490 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24491 (gimple_fold_builtin_snprintf): Same.
24492 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24493 of character types.
24494 (is_call_safe): New function.
24495 (try_substitute_return_value): Call it.
24496 (try_simplify_call): New function.
24497 (pass_sprintf_length::handle_gimple_call): Call it.
24499 2017-05-14 Martin Sebor <msebor@redhat.com>
24501 PR middle-end/80669
24502 * builtins.c (expand_builtin_stpncpy): Simplify.
24504 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24506 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24507 * config/i386/i386.h
24508 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24509 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24510 (struct machine_function): Add new members call_ms2sysv,
24511 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24512 (struct machine_frame_state): New fields sp_realigned and
24513 sp_realigned_offset.
24514 * config/i386/i386.c
24515 (enum xlogue_stub): New enum.
24516 (enum xlogue_stub_sets): New enum.
24517 (class xlogue_layout): New class.
24518 (struct ix86_frame): New fields stack_realign_allocate_offset,
24519 stack_realign_offset and outlined_save_offset. Modify comments to
24520 detail stack layout when using out-of-line stubs.
24521 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24522 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24523 -mcall-ms2sysv-xlogues.
24524 (stub_managed_regs): New static variable.
24525 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24526 registers managed by out-of-line stub.
24527 (disable_call_ms2sysv_xlogues): New function.
24528 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24529 m->call_ms2sysv when appropriate and compute frame layout for
24531 (sp_valid_at, fp_valid_at): New inline functions.
24532 (choose_basereg): New function.
24533 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24535 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24536 Use align parameter of choose_baseaddr to generated aligned SSE movs
24538 (pro_epilogue_adjust_stack): Modify to track
24539 machine_frame_state::sp_realigned.
24540 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24541 (ix86_nsaved_sseregs): Likewise.
24542 (ix86_emit_save_regs): Likewise.
24543 (ix86_emit_save_regs_using_mov): Likewise.
24544 (ix86_emit_save_sse_regs_using_mov): Likewise.
24545 (get_scratch_register_on_entry): Likewise.
24546 (gen_frame_set): New function.
24547 (gen_frame_load): Likewise.
24548 (gen_frame_store): Likewise.
24549 (emit_outlined_ms2sysv_save): Likewise.
24550 (emit_outlined_ms2sysv_restore): Likewise.
24551 (ix86_expand_prologue): Modify stack re-alignment code and call
24552 emit_outlined_ms2sysv_save when appropriate.
24553 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24554 parameter rtx_insn *insn, which allows the function to be used to only
24555 generate the notes.
24556 (ix86_expand_epilogue): Modify validity checks of frame and stack
24557 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24558 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24559 * config/i386/predicates.md
24560 (save_multiple): New predicate.
24561 (restore_multiple): Likewise.
24562 * config/i386/sse.md
24563 (save_multiple<mode>): New pattern.
24564 (save_multiple_realign<mode>): Likewise.
24565 (restore_multiple<mode>): Likewise.
24566 (restore_multiple_and_return<mode>): Likewise.
24567 (restore_multiple_leave_return<mode>): Likewise.
24568 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24570 2017-05-14 Julia Koval <julia.koval@intel.com>
24572 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24573 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24574 (__builtin_ia32_xsetbv): New builtins.
24575 * config/i386/i386.c (ix86_expand_special_args_builtin):
24577 (ix86_expand_builtin): Special expand for new intrinsics.
24578 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24579 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24580 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24582 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24584 * cfganal.c (inverted_post_order_compute): Change argument type
24586 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24587 * df-core.c (rest_of_handle_df_initialize): Adjust.
24588 (rest_of_handle_df_finish): Likewise.
24589 (df_analyze_1): Likewise.
24590 (df_analyze): Likewise.
24591 (loop_inverted_post_order_compute): Change argument to be a vec *.
24592 (df_analyze_loop): Adjust.
24593 (df_get_n_blocks): Likewise.
24594 (df_get_postorder): Likewise.
24595 * df.h (struct df_d): Change field to be a vec.
24596 * lcm.c (compute_laterin): Adjust.
24597 (compute_available): Likewise.
24598 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24599 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24600 * tree-ssa-pre.c (compute_antic): Likewise.
24602 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24604 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24605 (depth_first_search::depth_first_search): Change structure init
24606 function to this constructor.
24607 (depth_first_search::add_bb): Rename function to this member.
24608 (depth_first_search::execute): Likewise.
24609 (flow_dfs_compute_reverse_finish): Adjust.
24611 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24613 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24614 (longest_simple_path): Likewise.
24615 * shrink-wrap.c (spread_components): Likewise.
24616 (disqualify_problematic_components): Likewise.
24617 (emit_common_heads_for_components): Likewise.
24618 (emit_common_tails_for_components): Likewise.
24619 (insert_prologue_epilogue_for_components): Likewise.
24621 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24623 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24626 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24628 * df-core.c (df_set_blocks): Start using auto_bitmap.
24629 (df_compact_blocks): Likewise.
24630 * df-problems.c (df_rd_confluence_n): Likewise.
24631 * df-scan.c (df_insn_rescan_all): Likewise.
24632 (df_process_deferred_rescans): Likewise.
24633 (df_update_entry_block_defs): Likewise.
24634 (df_update_exit_block_uses): Likewise.
24635 (df_entry_block_bitmap_verify): Likewise.
24636 (df_exit_block_bitmap_verify): Likewise.
24637 (df_scan_verify): Likewise.
24638 * lra-constraints.c (lra_constraints): Likewise.
24639 (undo_optional_reloads): Likewise.
24640 (lra_undo_inheritance): Likewise.
24641 * lra-remat.c (calculate_gen_cands): Likewise.
24642 (do_remat): Likewise.
24643 * lra-spills.c (assign_spill_hard_regs): Likewise.
24644 (spill_pseudos): Likewise.
24645 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24646 (bitmap_set_subtract_values): Likewise.
24648 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24650 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24651 management with auto_bitmap.
24652 (fix_inter_tick): Likewise.
24653 (fix_recovery_deps): Likewise.
24654 * ira.c (add_store_equivs): Likewise.
24655 (find_moveable_pseudos): Likewise.
24656 (split_live_ranges_for_shrink_wrap): Likewise.
24657 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24658 (rtx_reuse_manager::seen_def_p): Likewise.
24659 (rtx_reuse_manager::set_seen_def): Likewise.
24660 * print-rtl.h (class rtx_reuse_manager): Likewise.
24662 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24664 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24666 (migrate_btr_def): Likewise.
24667 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24668 * df-core.c (loop_post_order_compute): Likewise.
24669 (loop_inverted_post_order_compute): Likewise.
24670 * hsa-common.h: Likewise.
24671 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24672 * init-regs.c (initialize_uninitialized_regs): Likewise.
24673 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24674 (inline_small_functions): Likewise.
24675 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24676 * ira.c (combine_and_move_insns): Likewise.
24677 (build_insn_chain): Likewise.
24678 * loop-invariant.c (find_invariants): Likewise.
24679 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24680 * predict.c (tree_predict_by_opcode): Likewise.
24681 (predict_paths_leading_to): Likewise.
24682 (predict_paths_leading_to_edge): Likewise.
24683 (estimate_loops_at_level): Likewise.
24684 (estimate_loops): Likewise.
24685 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24686 (spread_components): Likewise.
24687 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24688 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24689 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24690 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24691 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24692 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24693 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24694 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24695 (mark_threaded_blocks): Likewise.
24696 (thread_through_all_blocks): Likewise.
24697 * tree-ssa.c (verify_ssa): Likewise.
24698 (execute_update_addresses_taken): Likewise.
24699 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24701 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24703 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24705 (post_order_compute): Likewise.
24706 (inverted_post_order_compute): Likewise.
24707 (pre_and_rev_post_order_compute_fn): Likewise.
24709 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24711 * genrecog.c (int_set::int_set): Explicitly construct our
24712 auto_vec base class.
24713 * vec.h (auto_vec::auto_vec): New constructor.
24715 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24717 * bitmap.h (class auto_bitmap): New constructor taking
24718 bitmap_obstack * argument.
24720 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24722 * bitmap.h (class auto_bitmap): Change type of m_bits to
24723 bitmap_head, and adjust ctor / dtor and member operators.
24725 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24727 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24728 when returned register mode doesn't match original mode.
24730 2017-05-12 Jeff Law <law@redhat.com>
24731 Jakub Jelinek <jakub@redhat.com>
24733 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24734 we look for cc setter after the compare-elim changes.
24735 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24736 within the vector to match what compare-elim now expects.
24737 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24738 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24740 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24741 after the compare-elim changes.
24742 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24743 the vector to match what compare-elim now expects.
24744 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24745 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24746 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24747 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24748 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24750 * config/visium/visium.c (single_set_and_flags): Fix where
24751 we look for cc setter after the compare-elim changes.
24752 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24753 with the vector to match what compare-elim now expects.
24754 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24755 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24756 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24757 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24758 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24759 (neg<mode>2_insn_set_overflow): Likewise.
24761 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24763 PR middle-end/79794
24764 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24765 maybe_expand_insn call, set ops[0].target. If still set after call,
24766 set alt_rtl. Add extra arg to recursive calls.
24767 (extract_bit_field): Add alt_rtl argument. Pass to
24769 * expmed.h (extract_bit_field): Fix prototype.
24770 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24771 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24772 to extract_bit_field_calls.
24773 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24774 Pass alt_rtl to extract_bit_field calls.
24775 * calls.c (store_unaligned_arguments_into_psuedos)
24776 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24777 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24779 * optabs.h (struct expand_operand): Add target bitfield.
24781 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24783 * compare-elim.c (try_eliminate_compare): Canonicalize
24784 operation with embedded compare to
24785 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24786 (set (reg) (operation)].
24788 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24790 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24793 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24794 cost of adding a carry flag for ADC instruction.
24795 [case MINUS]: Ignore the cost of subtracting a carry flag
24796 for SBB instruction.
24798 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24800 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24802 * config/rs6000/bmiintrin.h: New file.
24803 * config/rs6000/bmi2intrin.h: New file.
24804 * config/rs6000/x86intrin.h: New file.
24806 2017-05-12 Jeff Law <law@redhat.com>
24808 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24811 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24813 PR middle-end/80707
24814 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24816 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24818 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24819 early expansion of vector divide builtins.
24820 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24821 builtins identified as having unsigned arguments.
24823 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24825 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24826 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24827 expansion of vector logical operations (and, andc, or, xor,
24830 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24832 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24833 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24835 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24837 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24838 early GIMPLE expansion of vector multiplies.
24840 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24842 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24843 TARGET_HAVE_MOVT conditional.
24844 (movt splitter): Likewise.
24846 2017-05-12 Richard Biener <rguenther@suse.de>
24848 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24849 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24850 Fold all stmts not inplace.
24852 2017-05-12 Richard Biener <rguenther@suse.de>
24854 PR tree-optimization/80713
24855 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24856 inserted_exprs bit for not removed stmts.
24858 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24860 PR middle-end/69921
24861 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24862 parallelized" attribute for parallelized OpenACC kernels.
24863 * omp-offload.c (execute_oacc_device_lower): Use it.
24865 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24866 Set "oacc kernels" attribute.
24867 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24868 parameter. Adjust all users.
24869 (oacc_fn_attrib_kernels_p): Remove function.
24870 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24871 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24872 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24873 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24874 assert "oacc kernels" attribute is set.
24876 2017-05-11 Carl Love <cel@us.ibm.com>
24878 * config/rs6000/rs6000-c: Add support for built-in functions
24879 vector unsigned char vec_popcnt (vector signed char)
24880 vector unsigned char vec_popcnt (vector unsigned char)
24881 vector unsigned short vec_popcnt (vector signed short)
24882 vector unsigned short vec_popcnt (vector unsigned short)
24883 vector unsigned int vec_popcnt (vector signed int)
24884 vector unsigned int vec_popcnt (vector unsigned int)
24885 vector unsigned long long vec_popcnt (vector signed long long)
24886 vector unsigned long long vec_popcnt (vector unsigned long long)
24887 vector signed long long vec_slo (vector signed long long,
24888 vector signed char)
24889 vector signed long long vec_slo (vector signed long long,
24890 vector unsigned char)
24891 vector unsigned long long vec_slo (vector unsigned long long,
24892 vector signed char)
24893 vector unsigned long long vec_slo (vector unsigned long long,
24894 vector unsigned char)
24895 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24896 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24897 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24898 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24899 * doc/extend.texi: Update the built-in documentation file for the
24900 new built-in functions.
24902 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24904 * attribs.h (sorted_attr_string): Move machine independent
24905 functions for target clone support from the i386 port to common
24906 code. Rename ix86_function_versions to common_function_versions.
24907 Rename make_name to make_unique_name.
24908 (common_function_versions): Likewise.
24909 (make_unique_name): Likewise.
24910 (make_dispatcher_decl): Likewise.
24911 (is_function_default_version): Likewise.
24912 * attribs.c (attr_strcmp): Likewise.
24913 (sorted_attr_string): Likewise.
24914 (common_function_versions): Likewise.
24915 (make_unique_name): Likewise.
24916 (make_dispatcher_decl): Likewise.
24917 (is_function_default_version): Likewise.
24918 * config/i386/i386.c (attr_strcmp): Likewise.
24919 (sorted_attr_string): Likewise.
24920 (ix86_function_versions): Likewise.
24921 (make_name): Likewise.
24922 (make_dispatcher_decl): Likewise.
24923 (is_function_default_version): Likewise.
24924 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24926 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24929 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24930 Account for direct move costs for vec_construct of integer
24933 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24936 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24937 (UNSPEC_STX_ATOMIC): Ditto.
24938 (loaddi_via_sse): New insn.
24939 (storedi_via_sse): Ditto.
24940 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24941 Update corresponding peephole2 patterns.
24942 (atomic_storedi_fpu): Ditto.
24944 2017-05-11 Julia Koval <julia.koval@intel.com>
24946 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24947 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24949 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24950 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24951 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24953 2017-05-11 Nathan Sidwell <nathan@acm.org>
24955 * graphite-poly.c: Include dumpfile.h.
24957 * dumpfle.h (dump_function): Declare here ...
24958 * tree-dump.h (dump_function): ... not here.
24959 * dumpfile.c: #include tree-cfg.h.
24960 (dump_function): Move here from ...
24961 * tree-dump.c (dump_function): ... here.
24962 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24963 * graphite-poly.c: Don't include tree-dump.h.
24964 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24965 * print-tree.c: Likewise.
24966 * stor-layout.c: Likewise.
24967 * tree-nested.c: Likewise.
24969 * dumpfile.c (dump_start): Use TDF_FLAGS.
24970 (dump_enable_all): Fix TDF_KIND check thinko.
24972 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24974 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24975 array entries to represent two legal parameterizations of the
24976 overloaded __builtin_cmpb function, as represented by the
24977 P6_OV_BUILTIN_CMPB constant.
24978 (altivec_resolve_overloaded_builtin): Add special case handling
24979 for the __builtin_cmpb function, as represented by the
24980 P6_OV_BUILTIN_CMPB constant.
24981 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24982 (BU_P6_64BIT_2): New macro.
24983 (BU_P6_OVERLOAD_2): New macro
24984 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24985 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24986 (CMPB): Add overload support to represent both 32-bit and 64-bit
24987 compare-bytes function.
24988 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24989 support for TARGET_CMPB.
24990 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24991 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24992 documentation of the __builtin_cmpb overloaded built-in function.
24994 2017-05-11 Richard Biener <rguenther@suse.de>
24996 PR tree-optimization/80705
24997 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24998 bases are not vectorizable.
25000 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25002 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25003 when counting register pressure.
25005 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25007 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25008 (iv_ca_more_deps): Renamed to ...
25009 (iv_ca_compare_deps): ... this.
25010 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25012 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25014 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25016 (determine_group_iv_costs): ... here.
25017 (find_inv_vars_cb): Record inv var if it's not recorded before.
25019 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25021 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25022 (get_shiftadd_cost): Ditto.
25024 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25026 * tree-ssa-address.c: Include header file.
25027 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25029 (add_to_parts): Refactor.
25030 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25031 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25034 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25036 PR tree-optimization/53090
25037 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25039 (extract_cond_operands): Detect condition with IV on both sides
25040 and return COMP_IV_EXPR_2.
25041 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25042 (rewrite_use_compare): Simplify by removing call to function
25043 extract_cond_operands.
25045 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25047 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25048 (extract_cond_operands): Detect condition comparing against non-
25049 invariant bound and return appropriate enum value.
25050 (find_interesting_uses_cond): Update use of extract_cond_operands.
25051 Handle its return value accordingly.
25052 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25054 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25056 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25057 nonlinear iv_use computation in loop invariant sensitive way.
25059 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25061 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25062 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25064 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25066 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25067 (dump_cand): Support iv_cand.inv_exprs.
25068 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25070 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25073 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25075 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25077 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25078 as local function. Include necessary header files.
25079 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25081 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25083 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25085 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25087 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25088 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25089 RSHIFT_EXPR and BIT_NOT_EXPR.
25091 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25093 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25094 (adjust_setup_cost): New parameter supporting round up adjustment.
25095 (struct address_cost_data): Delete.
25096 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25097 (split_address_cost, ptr_difference_cost): Delete.
25098 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25099 (struct ainc_cost_data): New struct.
25100 (get_address_cost_ainc): New function.
25101 (get_address_cost, get_computation_cost): Reimplement.
25102 (determine_group_iv_cost_address): Record inv_expr for all uses of
25104 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25105 (iv_ca_has_deps): Reimplemented to ...
25106 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25108 (iv_ca_extend): Call iv_ca_more_deps.
25110 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25112 * tree-ssa-address.c (struct mem_address): Move to header file.
25113 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25114 * tree-ssa-address.h (struct mem_address): Move from C file.
25115 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25117 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25119 * tree-affine.h (aff_combination_type): New interface.
25120 (aff_combination_zero_p): Remove static.
25121 (aff_combination_const_p): New interface.
25122 (aff_combination_singleton_var_p): New interfaces.
25124 2017-05-11 Richard Biener <rguenther@suse.de>
25126 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25127 Skip unreachable blocks and destinations.
25128 (eliminate): Move stmt removal and fixup ...
25129 (fini_eliminate): ... here. Skip inserted exprs.
25130 (pass_pre::execute): Move fini_pre after fini_eliminate.
25131 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25132 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25133 PRE to get rid of dead code that has invalid SSA form and
25134 split critical edges again.
25136 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25138 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25140 2017-05-11 Richard Biener <rguenther@suse.de>
25142 * passes.c (execute_function_todo): Verify loops if they are
25143 said to be up-to-date.
25144 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25145 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25147 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25150 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25151 handle calling assemble_external ourself.
25154 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25155 modes with zero size. Enhance comment.
25157 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25159 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25160 built-ins for vec_xl and vec_xst with short and char pointer
25163 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25165 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25166 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25167 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25168 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25169 (_mm_maskz_min_round_ss): New intrinsics.
25170 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25171 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25172 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25173 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25174 (__builtin_ia32_minss_mask_round): New builtins.
25175 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25176 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25177 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25179 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25180 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25182 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25185 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25187 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25188 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25189 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25190 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25191 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25192 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25193 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25194 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25195 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25196 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25197 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25198 (__builtin_ia32_mulss_mask_round): New builtins.
25199 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25200 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25201 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25203 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25204 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25206 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25209 2017-05-10 Julia Koval <julia.koval@intel.com>
25211 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25212 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25213 (_mm256_setr_m128i): New intrinsics.
25215 2017-05-10 Julia Koval <julia.koval@intel.com>
25217 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25218 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25219 (_mm_maskz_rcp14_ss): New intrinsics.
25220 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25221 (__builtin_ia32_rcp14ss_mask): New builtins.
25222 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25224 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25226 PR tree-optimization/51513
25227 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25228 (assert_unreachable_fallthru_edge_p): Use it.
25229 (group_case_labels_stmt): Likewise.
25230 * tree-cfg.h: Prototype it.
25231 * stmt.c: Include cfghooks.h and tree-cfg.h.
25232 (emit_case_dispatch_table) <gap_label>: New local variable.
25233 Use it to fill dispatch table gaps.
25234 Test for default_label before updating probabilities.
25235 (expand_case) <default_label>: Remove unneeded initialization.
25236 Test for unreachable default case statement and remove its edge.
25237 Set default_label accordingly.
25238 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25240 2017-05-10 Carl Love <cel@us.ibm.com>
25242 * config/rs6000/rs6000-c: Add support for built-in functions
25243 vector signed char vec_neg (vector signed char)
25244 vector signed short int vec_neg (vector short int)
25245 vector signed int vec_neg (vector signed int)
25246 vector signed long long vec_neg (vector signed long long)
25247 vector float vec_neg (vector float)
25248 vector double vec_neg (vector double)
25249 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25251 * config/rs6000/altivec.h: Add define for vec_neg
25252 * doc/extend.texi: Update the built-in documentation for the
25253 new built-in functions.
25255 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25257 PR tree-optimization/77644
25258 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25260 2017-05-10 Nathan Sidwell <nathan@acm.org>
25262 * dumpfile.h (TDI_lang_all): New.
25263 (TDF_KIND): New. Renumber others
25264 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25266 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25268 (get_dump_file_name): Adjust suffix generation.
25269 (dump_enable_all): Use TDF_KIND.
25270 * doc/invoke.texi (-fdump-lang-all): Document.
25272 * dumpfile.h: Tabify.
25274 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25277 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25278 Move member access before delete.
25280 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25282 * tree-inline.c (expand_call_inline): Split block at stmt
25285 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25288 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25289 are now unused after splitting mov{sf,sd}_hardfloat.
25290 (f32_lr2): Likewise.
25291 (f32_lm): Likewise.
25292 (f32_lm2): Likewise.
25293 (f32_li): Likewise.
25294 (f32_li2): Likewise.
25295 (f32_lv): Likewise.
25296 (f32_sr): Likewise.
25297 (f32_sr2): Likewise.
25298 (f32_sm): Likewise.
25299 (f32_sm2): Likewise.
25300 (f32_si): Likewise.
25301 (f32_si2): Likewise.
25302 (f32_sv): Likewise.
25303 (f32_dm): Likewise.
25304 (f32_vsx): Likewise.
25305 (f32_av): Likewise.
25306 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25307 For movsf, order stores so the VSX stores occur before the GPR
25308 store which encourages the register allocator to use a traditional
25309 FPR instead of a GPR. For movsd, order the stores so that the GPR
25310 store comes before the VSX stores to allow the power6 to work.
25311 This is due to the power6 not having a 32-bit integer store
25312 instruction from a FPR.
25313 (movsf_hardfloat): Likewise.
25314 (movsd_hardfloat): Likewise.
25316 2017-05-09 Martin Sebor <msebor@redhat.com>
25318 PR translation/80280
25319 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25322 PR translation/80280
25323 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25324 data member added in r247778.
25325 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25327 2017-05-09 Nathan Sidwell <nathan@acm.org>
25329 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25331 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25334 2017-05-09 Marek Polacek <polacek@redhat.com>
25336 * doc/invoke.texi: Fix typo.
25338 2017-05-09 Richard Biener <rguenther@suse.de>
25340 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25341 (vrp_val_is_min): Likewise.
25342 (set_value_range_to_value): Likewise.
25343 (set_value_range_to_nonnegative): Likewise.
25344 (gimple_assign_nonzero_p): Likewise.
25345 (gimple_stmt_nonzero_p): Likewise.
25346 (vrp_int_const_binop): Likewise. Remove unreachable case.
25347 (adjust_range_with_scev): Adjust comments.
25348 (compare_range_with_value): Likewise.
25349 (extract_range_from_phi_node): Likewise.
25350 (test_for_singularity): Likewise.
25352 2017-05-09 Richard Biener <rguenther@suse.de>
25354 * tree-vrp.c (get_single_symbol): Add assert that we don't
25355 get overflowed constants as invariant part.
25356 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25357 checks. Use wi::cmp instead of recursing for integer constants.
25358 (compare_values): Just ignore whether we assumed undefined
25359 overflow instead of failing the compare.
25360 (extract_range_for_var_from_comparison_expr): Add comment before the
25361 TREE_NO_WARNING sets.
25362 (test_for_singularity): Likewise.
25363 (extract_range_from_comparison): Do not disable optimization
25364 when we assumed undefined overflow.
25365 (extract_range_basic): Remove init of unused var.
25367 2017-05-09 Richard Biener <rguenther@suse.de>
25369 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25370 (extract_range_from_multiplicative_op_1): Adjust.
25371 (extract_range_from_binary_expr_1): Use int_const_binop.
25373 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25376 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25377 rs6000_store_data_bypass_p in seven define_bypass directives and
25378 in several comments.
25379 * config/rs6000/rs6000-protos.h: Add prototype for
25380 rs6000_store_data_bypass_p function.
25381 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25382 function implements slightly different (rs6000-specific) semantics
25383 than store_data_bypass_p, returning false rather than aborting
25384 with assertion error when arguments do not satisfy the
25385 requirements of store data bypass.
25386 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25387 rs6000_store_data_bypass_p.
25389 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25391 * config/xtensa/xtensa-protos.h
25392 (xtensa_initial_elimination_offset): New declaration.
25393 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25394 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25395 macro definition, add case for FRAME_POINTER_REGNUM when
25396 FRAME_GROWS_DOWNWARD.
25397 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25398 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25399 xtensa_initial_elimination_offset.
25401 2017-05-08 Nathan Sidwell <nathan@acm.org>
25403 * doc/invoke.texi: Alphabetize -fdump options.
25405 2017-05-08 Martin Sebor <msebor@redhat.com>
25407 PR translation/80280
25408 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25410 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25412 * target.def (compute_frame_layout): New optional target hook.
25413 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25414 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25415 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25417 * reload1.c (verify_initial_elim_offsets): Likewise.
25418 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25419 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25420 (arm_get_frame_offsets): Split up into this ...
25421 (arm_compute_frame_layout): ... and this function.
25423 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25425 * config/aarch64/constraints.md (Usa): New constraint.
25426 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25428 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25430 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25431 with_multilib_list after it has been checked.
25433 2017-05-08 Richard Biener <rguenther@suse.de>
25435 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25436 (bitmap_set_subtract_values): Likewise.
25438 2017-05-08 Richard Biener <rguenther@suse.de>
25440 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25441 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25443 (gimple_stmt_nonzero_warnv_p): Rename to ...
25444 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25446 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25447 (extract_range_basic): Adjust, do not disable propagation on
25448 strict overflow sensitive simplification.
25449 (vrp_visit_cond_stmt): Likewise.
25451 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25453 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25454 body size unconditionally.
25456 2017-05-07 Jeff Law <law@redhat.com>
25459 2017-05-06 Jeff Law <law@redhat.com>
25460 PR tree-optimization/78496
25461 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25464 PR tree-optimization/78496
25465 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25466 (simplify_stmt_using_ranges): Call it.
25467 (vrp_dom_walker::before_dom_children): Extract equivalences
25468 from an ASSERT_EXPR with an equality comparison against a
25471 2017-05-06 Jeff Law <law@redhat.com>
25473 PR tree-optimization/78496
25474 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25477 PR tree-optimization/78496
25478 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25479 (simplify_stmt_using_ranges): Call it.
25480 (vrp_dom_walker::before_dom_children): Extract equivalences
25481 from an ASSERT_EXPR with an equality comparison against a
25484 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25486 * lra-constraints.c (lra_copy_reg_equiv): New function.
25487 (split_reg): Use it to copy equivalence information from the
25488 original register to the spill register.
25490 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25492 PR rtl-optimization/75964
25493 * simplify-rtx.c (simplify_const_relational_operation): Remove
25494 invalid handling of comparisons of integer ABS.
25496 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25498 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25499 initialize to zero.
25500 (init_regs): Remove declaration.
25501 (function_arg_advance_32): Initialize error_p as boolean variable.
25503 2017-05-05 Nathan Sidwell <nathan@acm.org>
25505 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25506 lines. Use for (;;).
25508 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25510 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25511 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25512 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25513 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25514 VF=2 that require versioning.
25516 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25518 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25521 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25523 * diagnostic.h (diagnostic_override_option_index): Convert from
25524 macro to inline function.
25526 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25528 * diagnostic.c (last_module_changed_p): New function.
25529 (set_last_module): New function.
25530 (diagnostic_report_current_module): Convert macro usage to
25531 the above functions.
25532 * diagnostic.h (diagnostic_context::last_module): Strengthen
25533 from const line_map * to const line_map_ordinary *.
25534 (diagnostic_last_module_changed): Delete macro.
25535 (diagnostic_set_last_module): Delete macro.
25537 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25539 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25540 with diagnostic_report_diagnostic.
25541 (diagnostic_n_impl_richloc): Likewise.
25542 * diagnostic.h (report_diagnostic): Delete macro.
25543 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25544 with diagnostic_report_diagnostic.
25545 * substring-locations.c (format_warning_va): Likewise.
25547 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25549 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25550 save/restor of format_spec. Move option-printing code to...
25551 (print_option_information): ...this new function, and
25552 reimplement by simply printing to the pretty_printer,
25553 rather than appending to the format string.
25555 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25557 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25558 handling logic into...
25559 (update_effective_level_from_pragmas): ...this new function.
25561 2017-05-04 Andrew Waterman <andrew@sifive.com>
25563 * config/riscv/riscv.opt (mstrict-align): New option.
25564 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25565 (SLOW_UNALIGNED_ACCESS): Define.
25566 (riscv_slow_unaligned_access): Declare.
25567 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25569 (riscv_slow_unaligned_access): New variable.
25570 (rocket_tune_info): Set slow_unaligned_access to true.
25571 (optimize_size_tune_info): Set slow_unaligned_access to false.
25572 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25573 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25574 (riscv_option_override): Set riscv_slow_unaligned_access.
25575 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25577 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25579 * config/riscv/riscv.md: Unify indentation.
25581 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25586 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25588 (extendsi<mode>2): Add support for doing sign extension via
25589 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25590 don't have ISA 3.0 instructions.
25591 (extendsi<mode>2 splitter): Likewise.
25592 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25593 generate the normal insns since SImode can now go in vector
25594 registers. Disallow the special UNSPECs needed for previous
25595 machines to hide SImode being used. Add new insns
25596 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25597 (fix_trunc<mode>si2_stfiwx): Likewise.
25598 (fix_trunc<mode>si2_internal): Likewise.
25599 (fixuns_trunc<mode>si2): Likewise.
25600 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25601 (fctiw<u>z_<mode>_smallint): Likewise.
25602 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25603 of floating point to 32-bit integer from doing a direct move to
25604 the GPR registers to do a store.
25605 (fctiwz_<mode>): Break long line.
25607 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25609 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25610 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25611 (addr_list, addr_offset_valid_p): New.
25612 (split_address_groups): Check offset validity with above function.
25613 (gt-tree-ssa-loop-ivopts.h): Include header file.
25615 2017-05-05 Nathan Sidwell <nathan@acm.org>
25617 * config.gcc (arm*-*-*): Add missing 'fi'.
25619 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25621 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25622 included in -fopt-info does not matter.
25623 * doc/optinfo.texi (-fopt-info): Fix description of default
25624 behavour. Explicitly say order of options included in -fopt-info
25627 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25629 * config.gcc: Allow combinations of aprofile and rmprofile values for
25630 --with-multilib-list.
25631 * config/arm/t-multilib: New file.
25632 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25633 variables. Remove setting of ISA and floating-point ABI in
25634 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25635 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25636 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25637 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25639 * config/arm/t-rmprofile: Likewise except for the matches changes.
25640 * doc/install.texi (--with-multilib-list): Document the combination of
25641 aprofile and rmprofile values and warn about pitfalls in doing that.
25643 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25645 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25646 (movdi_aarch64): Likewise.
25648 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25650 PR tree-optimization/80632
25651 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25653 (build_arrays): Initialize it for virtual phis.
25654 (fix_phi_nodes): Use it for virtual phis.
25656 PR tree-optimization/80558
25657 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25658 [x, y] op z into [x op, y op z] for op & or | if conditions
25661 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25662 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25665 * config/arm/arm.md (use_literal_pool): Remove.
25666 (64-bit immediate split): No longer takes cost into consideration
25667 if arm_disable_literal_pool is enabled.
25668 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25669 used when arm_disable_literal_pool is enabled.
25670 (arm_max_const_double_inline_cost): Remove use of
25671 arm_disable_literal_pool.
25672 (push_minipool_fix): Add assert.
25673 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25674 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25675 (no_literal_pool_sf_immediate): New.
25677 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25679 PR tree-optimization/80613
25680 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25681 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25683 2017-05-05 Richard Biener <rguenther@suse.de>
25685 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25687 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25689 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25690 of this flag from insn conditions due to removal from r247495.
25692 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25694 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25696 (arm_early_store_addr_dep_ptr): Likewise.
25697 * config/arm/aarch-common-protos.h
25698 (arm_early_load_addr_dep_ptr): Add prototype.
25699 (arm_early_store_addr_dep_ptr): Likewise.
25700 * config/arm/cortex-a53.md: Add new bypasses.
25702 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25704 * tree.c (next_type_uid): Change type to unsigned.
25705 (type_hash_canon): Decrement back next_type_uid if
25706 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25707 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25710 2017-05-04 Martin Sebor <msebor@redhat.com>
25712 * builtins.c: Fix a trivial typo in a comment.
25714 PR middle-end/79234
25715 * builtins.c (check_sizes): Adjust to handle reading past the end.
25716 Avoid printing excessive upper bound of ranges. Use %E to print
25717 tree nodes instead of converting them to %wu.
25718 (expand_builtin_memchr): New function.
25719 (compute_dest_size): Rename...
25720 (compute_objsize): ...to this.
25721 (expand_builtin_memcpy): Adjust.
25722 (expand_builtin_mempcpy): Adjust.
25723 (expand_builtin_strcat): Adjust.
25724 (expand_builtin_strcpy): Adjust.
25725 (check_strncat_sizes): Adjust.
25726 (expand_builtin_strncat): Adjust.
25727 (expand_builtin_strncpy): Adjust and simplify.
25728 (expand_builtin_memset): Adjust.
25729 (expand_builtin_bzero): Adjust.
25730 (expand_builtin_memcmp): Adjust.
25731 (expand_builtin): Handle memcmp.
25732 (maybe_emit_chk_warning): Check strncat just once.
25734 2017-05-04 Martin Sebor <msebor@redhat.com>
25736 PR preprocessor/79214
25737 PR middle-end/79222
25738 PR middle-end/79223
25739 * builtins.c (check_sizes): Add inlining context and issue
25740 warnings even when -Wno-system-headers is set.
25741 (check_strncat_sizes): Same.
25742 (expand_builtin_strncat): Same.
25743 (expand_builtin_memmove): New function.
25744 (expand_builtin_stpncpy): Same.
25745 (expand_builtin): Handle memmove and stpncpy.
25747 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25749 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25750 which is not used any more.
25752 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25754 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25756 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25758 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25759 (cortexa53_tunings): Likewise.
25760 (cortexa57_tunings): Likewise.
25761 (cortexa72_tunings): Likewise.
25762 (cortexa73_tunings): Likewise.
25764 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25766 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25767 Set loop alignment to 8.
25769 2017-05-04 Martin Sebor <msebor@redhat.com>
25771 PR translation/80280
25772 * builtins.c (expand_builtin_object_size): Add missing quoting to
25773 %D and like directives.
25774 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25775 (hsa_type_for_tree_type): Same.
25776 (verify_function_arguments): Same.
25777 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25778 * varasm.c (get_section): Same.
25781 2017-05-04 Martin Sebor <msebor@redhat.com>
25783 PR translation/80280
25784 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25786 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25788 * config/aarch64/aarch64.c (generic_addrcost_table):
25789 Change HI/TI mode setting.
25791 2017-05-04 Martin Jambor <mjambor@suse.cz>
25793 PR tree-optimization/80622
25794 * tree-sra.c (comes_initialized_p): New function.
25795 (build_accesses_from_assign): Only set write lazily when
25796 comes_initialized_p is false.
25797 (analyze_access_subtree): Use comes_initialized_p.
25798 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25799 instead of testing for PARM_DECL.
25801 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25803 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25804 constraint on operand 0 to allow more general addressing modes.
25805 Adjust output template.
25806 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25808 * config/aarch64/aarch64-protos.h
25809 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25810 * config/aarch64/constraints.md (Dp): New address constraint.
25811 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25814 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25816 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25817 update use of estimate_ipcp_clone_size_and_time.
25818 (estimate_local_effects): Update use of
25819 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25820 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25821 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25822 Return nonspecialized time.
25824 2017-05-04 Richard Biener <rguenther@suse.de>
25826 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25827 for the last VUSE which def dominates the PHI. Directly call
25829 (get_continuation_for_phi_1): Remove.
25831 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25833 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25834 to explain the use of truncating division. Cap the number of
25835 iterations to the maximum given by nb_iterations_upper_bound,
25838 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25840 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25841 * configure: Regenerate.
25842 * config.in: Regenerate.
25843 * config/i386/driver-mingw32.c: new file.
25844 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25845 * config.host: Link driver-mingw32.o on MinGW host.
25846 * doc/install.texi: Document new --enable-mingw-wildcard configure
25849 2017-05-04 Marek Polacek <polacek@redhat.com>
25851 PR tree-optimization/80612
25852 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25854 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25855 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25857 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25858 (movt splitter): Likewise.
25859 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25860 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25861 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25862 block for Thumb-1 with MOVT.
25863 (thumb2_legitimate_address_p): Move code block ...
25864 (can_avoid_literal_pool_for_label_p): ... into this new function.
25865 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25867 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25868 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25869 "M-profile targets with the MOVT instruction".
25871 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25873 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25874 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25875 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25877 2017-05-04 Martin Liska <mliska@suse.cz>
25879 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25880 variable cond_code.
25882 2017-05-04 Richard Biener <rguenther@suse.de>
25884 * tree.c (array_at_struct_end_p): Handle arrays at struct
25885 end with flexarrays more conservatively. Refactor and treat
25886 arrays of arrays or aggregates more strict. Fix
25887 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25888 * tree.c (array_at_struct_end_p): Adjust prototype.
25889 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25890 * gimple-fold.c (get_range_strlen): Likewise.
25891 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25893 2017-05-04 Richard Biener <rguenther@suse.de>
25895 PR tree-optimization/31130
25896 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25898 (supports_overflow_infinity): Likewise.
25899 (is_negative_overflow_infinity): Likewise.
25900 (is_positive_overflow_infinity): Likewise.
25901 (is_overflow_infinity): Likewise.
25902 (stmt_overflow_infinity): Likewise.
25903 (overflow_infinity_range_p): Likewise.
25904 (usable_range_p): Remove as always returning true.
25905 (make_overflow_infinity): Remove.
25906 (negative_overflow_infinity): Likewise.
25907 (positive_overflow_infinity): Likewise.
25908 (avoid_overflow_infinity): Likewise.
25909 (set_value_range): Adjust accordingly.
25910 (set_value_range_to_nonnegative): Likewise, remove now unused
25911 overflow_infinity arg.
25912 (vrp_operand_equal_p): Adjust.
25913 (update_value_range): Likewise.
25914 (range_int_cst_singleton_p): Likewise.
25915 (operand_less_p): Likewise.
25916 (compare_values_warnv): Likewise.
25917 (extract_range_for_var_from_comparison_expr): Likewise.
25918 (vrp_int_const_binop): Likewise.
25919 (zero_nonzero_bits_from_vr): Likewise.
25920 (extract_range_from_multiplicative_op_1): Likewise.
25921 (extract_range_from_binary_expr_1): Likewise.
25922 (extract_range_from_unary_expr): Likewise.
25923 (extract_range_from_comparison): Likewise.
25924 (extract_range_basic): Likewise.
25925 (adjust_range_with_scev): Likewise.
25926 (compare_ranges): Likewise.
25927 (compare_range_with_value): Likewise.
25928 (dump_value_range): Likewise.
25929 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25931 (simplify_cond_using_ranges): Adjust.
25933 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25935 * brig-builtins.def: Added a builtin for class_f64.
25936 * builtin-types.def: Added a builtin type needed by class_f64.
25938 2017-05-03 Jason Merrill <jason@redhat.com>
25940 * timevar.def: Add TV_CONSTEXPR.
25942 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25944 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25946 2017-05-03 Martin Jambor <mjambor@suse.cz>
25948 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25949 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25950 * ipa-cp.c (ipcp_propagate_stage): Do not call
25951 ipa_update_after_lto_read.
25952 * ipa-inline.c (ipa_inline): Likewise.
25954 2017-05-03 Martin Jambor <mjambor@suse.cz>
25956 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25957 tag. Added a default constructor and a destructor.
25958 (ipa_edge_args_sum_t): New class;
25959 (ipa_edge_args_sum): Declare.
25960 (ipa_edge_args_vector): Remove declaration.
25961 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25962 (ipa_free_edge_args_substructures): Remove declaration.
25963 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25964 (ipa_edge_args_info_available_for_edge_p): Likewise.
25965 * ipa-prop.c (ipa_edge_args_vector): Removed.
25966 (edge_removal_hook_holder): Likewise.
25967 (edge_duplication_hook_holder): Likewise.
25968 (ipa_edge_args_sum): New variable.
25969 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25970 ipa_edge_args_vector.
25971 (ipa_free_edge_args_substructures): Likewise.
25972 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25973 ipa_edge_args_vector.
25974 (ipa_edge_removal_hook): Turned into method
25975 ipa_edge_args_sum_t::remove.
25976 (ipa_edge_duplication_hook): Turned into method
25977 ipa_edge_args_sum_t::duplicate.
25978 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25979 registering edge hooks.
25980 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25981 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25982 ipa_edge_args_sum instead of ipa_edge_args_vector.
25983 * ipa-profile.c (ipa_profile): Likewise.
25985 2017-05-03 Martin Jambor <mjambor@suse.cz>
25987 * symbol-summary.h (function_summary): New method exists.
25988 (function_summary::symtab_removal): Deallocate through release.
25989 (call_summary): New class.
25990 (gt_ggc_mx): New overload.
25991 (gt_pch_nx): Likewise.
25992 (gt_pch_nx): Likewise.
25994 2017-05-03 Jeff Law <law@redhat.com>
25996 PR tree-optimization/78496
25997 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25998 from simplify_cond_using_ranges. Split off code to walk
25999 backwards through casts into ...
26000 (simplify_cond_using_ranges_2): New function.
26001 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26002 (execute_vrp): After identifying jump threads, call
26003 simplify_cond_using_ranges_2.
26005 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26008 * ipa-inline.h (inline_summary): Add ctor.
26009 (create_ggc): Do not use ggc_cleared_alloc.
26011 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26012 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26014 * gcc.c (handle_braces): Support escaping in switch matching
26016 * doc/invoke.texi (Spec Files): Document it.
26017 Remove superfluous @code markup in items.
26019 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26021 * diagnostic-show-locus.c (struct column_range): New struct.
26022 (get_affected_columns): New function.
26023 (get_printed_columns): New function.
26024 (struct correction): New struct.
26025 (correction::ensure_capacity): New function.
26026 (correction::ensure_terminated): New function.
26027 (struct line_corrections): New struct.
26028 (line_corrections::~line_corrections): New dtor.
26029 (line_corrections::add_hint): New function.
26030 (layout::print_trailing_fixits): Reimplement in terms of the new
26032 (selftest::test_overlapped_fixit_printing): New function.
26033 (selftest::diagnostic_show_locus_c_tests): Call it.
26035 2017-05-03 Nathan Sidwell <nathan@acm.org>
26037 Canonicalize canonical type hashing
26038 * tree.h (type_hash_canon_hash): Declare.
26039 * tree.c (type_hash_list, attribute_hash_list): Move into
26040 type_hash_canon_hash.
26041 (build_type_attribute_qual_variant): Break out hash code calc into
26042 type_hash_canon_hash.
26043 (type_hash_canon_hash): New. Generic type hash computation.
26044 (build_range_type_1, build_array_type_1, build_function_type,
26045 build_method_type_directly, build_offset_type, build_complex_type,
26046 make_vector_type): Call it.
26048 2017-05-03 Richard Biener <rguenther@suse.de>
26050 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26051 When all DRs have unknown misaligned do not always peel
26052 when there is a store but apply the same costing model as if
26053 there were only loads.
26055 2017-05-03 Richard Biener <rguenther@suse.de>
26058 PR tree-optimization/80492
26059 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26060 compare_base_decls returning dont-know properly.
26062 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26064 * config/arm/iterators.md (CCSI): New mode iterator.
26065 (arch): New mode attribute.
26066 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26067 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26068 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26069 code iterator for success result mode.
26070 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26071 the corresponding new insn generators.
26073 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26076 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26077 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26079 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26081 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26082 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26083 (DDR_A): Wrap DDR argument in brackets.
26084 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26085 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26086 (DDR_REVERSED_P): Likewise.
26088 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26090 PR tree-optimization/79472
26091 * tree-switch-conversion.c (struct switch_conv_info): Add
26092 contiguous_range and default_case_nonstandard fields.
26093 (collect_switch_conv_info): Compute contiguous_range and
26094 default_case_nonstandard fields, don't clear final_bb if
26095 contiguous_range and only the default case doesn't have the required
26097 (check_all_empty_except_final): Set default_case_nonstandard instead
26098 of failing if contiguous_range and the default case doesn't have empty
26100 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26101 and only the default case doesn't have the required constants. Skip
26103 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26104 if default_case_nonstandard.
26105 (build_constructors): Build constant 1 just once. Assert that default
26106 values aren't inserted in between cases if contiguous_range. Skip
26108 (build_arrays): Skip virtual phis.
26109 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26110 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26111 Handle virtual phis.
26112 (gen_inbound_check): Handle default_case_nonstandard case.
26113 (process_switch): Adjust check_final_bb caller. Call
26114 gather_default_values with the first non-default case instead of
26115 default case if default_case_nonstandard.
26117 2017-05-02 Nathan Sidwell <nathan@acm.org>
26119 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26120 check. Fix formatting.
26122 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26124 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26125 errors when comparing specialized and unspecialized times.
26127 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26129 * diagnostic-show-locus.c
26130 (layout::should_print_annotation_line_p): Make private.
26131 (layout::print_annotation_line): Make private.
26132 (layout::annotation_line_showed_range_p): Make private.
26133 (layout::show_ruler): Make private.
26134 (layout::print_source_line): Make private. Pass in line and
26135 line_width, rather than calling location_get_source_line. Drop
26137 (layout::print_leading_fixits): New method.
26138 (layout::print_any_fixits): Rename to...
26139 (layout::print_trailing_fixits): ...this, and make private.
26140 Don't print newline fixits.
26141 (diagnostic_show_locus): Move logic for printing one row into...
26142 (layout::print_line): ...this new function. Move the
26143 location_get_source_line call and error-handling from
26144 print_source_line to here. Call print_leading_fixits, and rename
26145 print_any_fixits to print_trailing_fixits.
26146 (selftest::test_fixit_insert_containing_newline): Update now that
26147 newlines are partially supported.
26148 (selftest::test_fixit_insert_containing_newline_2): New test.
26149 (selftest::test_fixit_replace_containing_newline): Update comments.
26150 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26151 * edit-context.c (class added_line): New class.
26152 (class edited_line): Describe newline handling in comment.
26153 (edited_line::actually_edited_p): New method.
26154 (edited_line::print_content): Delete redundant decl.
26155 (edited_line::m_predecessors): New field.
26156 (edited_file::print_content): Call edited_line::print_content.
26157 (edited_file::print_diff): Update to support newlines.
26158 (edited_file::print_diff_hunk): Likewise.
26159 (edited_file::print_run_of_changed_lines): New function.
26160 (edited_file::print_diff_line): Convert to...
26161 (print_diff_line): ...this.
26162 (edited_file::get_effective_line_count): New function.
26163 (edited_line::edited_line): Initialize new field m_predecessors.
26164 (edited_line::~edited_line): Clean up m_predecessors.
26165 (edited_line::apply_fixit): Handle newlines.
26166 (edited_line::get_effective_line_count): New function.
26167 (edited_line::print_content): New function.
26168 (edited_line::print_diff_lines): New function.
26169 (selftest::test_applying_fixits_insert_containing_newline): New
26171 (selftest::test_applying_fixits_replace_containing_newline): New
26173 (selftest::insert_line): New function.
26174 (selftest::test_applying_fixits_multiple_lines): Add example of
26176 (selftest::edit_context_c_tests): Call the new tests.
26178 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26180 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26181 parameter cand. Update dump information.
26182 (get_computation_cost): Update uses.
26184 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26186 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26187 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26188 (get_computation_at, rewrite_use_address): Update use of
26189 get_computation_aff.
26191 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26193 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26194 (get_computation): Delete.
26195 (get_computation_cost): Implement like get_computation_cost_at.
26196 Use get_computation_at.
26197 (get_computation_cost_at): Delete.
26198 (rewrite_use_nonlinear_expr): Use get_computation_at.
26199 (rewrite_use_compare, remove_unused_ivs): Ditto.
26201 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26203 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26205 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26207 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26208 (ivopts_global_cost_for_size): Rename parameter and update uses.
26209 (iv_ca_recount_cost): Update uses.
26210 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26211 candidates seperately in n_invs and n_cands.
26212 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26214 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26216 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26217 (find_inv_vars_cb): New.
26218 (find_depends): Renamed to ...
26219 (find_inv_vars): ... this.
26220 (add_candidate_1, force_var_cost): Call find_inv_vars.
26221 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26223 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26225 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26226 inv_vars. Add inv_exprs.
26227 (struct iv_cand): Rename depends_on to inv_vars.
26228 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26229 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26230 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26231 (dump_cand): Dump inv_vars.
26232 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26233 (record_invariant, find_depends, add_candidate_1): Ditto.
26234 (set_group_iv_cost, force_var_cost): Ditto.
26235 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26236 (get_computation_cost_at, get_computation_cost): Ditto.
26237 (determine_group_iv_cost_generic): Ditto.
26238 (determine_group_iv_cost_address): Ditto.
26239 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26240 (determine_group_iv_costs): Ditto.
26241 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26242 (iv_ca_set_remove_invariants): Renamed to ...
26243 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26244 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26245 (iv_ca_set_add_invariants): Renamed to ...
26246 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26247 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26248 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26249 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26250 (create_new_ivs): Remove useless dump.
26252 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26254 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26256 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26257 (iv_ca_set_no_cp, create_new_iv): Ditto.
26259 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26261 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26263 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26265 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26266 function tree_check2.
26268 2017-05-02 Martin Liska <mliska@suse.cz>
26270 * doc/gcov.texi: Add missing preposition.
26271 * gcov.c (function_info::function_info): Properly fill up
26272 all member variables.
26274 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26276 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26278 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26280 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26282 2017-05-02 Martin Liska <mliska@suse.cz>
26285 * lto-streamer-in.c (lto_read_tree_1): Remove
26286 LTO_STREAMER_DEBUG.
26287 * lto-streamer.c (struct tree_hash_entry): Likewise.
26288 (struct tree_entry_hasher): Likewise.
26289 (tree_entry_hasher::hash): Likewise.
26290 (tree_entry_hasher::equal): Likewise.
26291 (lto_streamer_init): Likewise.
26292 (lto_orig_address_map): Likewise.
26293 (lto_orig_address_get): Likewise.
26294 (lto_orig_address_remove): Likewise.
26295 * lto-streamer.h: Likewise.
26296 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26297 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26299 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26301 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26302 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26303 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26304 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26305 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26306 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26307 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26308 (mm_maskz_sub_ss): New intrinsics.
26309 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26310 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26311 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26312 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26313 (__builtin_ia32_subss_mask_round): New builtins.
26314 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26315 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26316 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26318 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26319 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26321 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26324 2017-05-02 Martin Jambor <mjambor@suse.cz>
26326 PR tree-optimization/78687
26327 * tree-sra.c (access): New field parent.
26328 (process_subtree_disqualification): New function.
26329 (disqualify_candidate): Call it.
26330 (build_accesses_from_assign): Reset write flag if creating an
26332 (build_access_subtree): Fill in parent field and also prpagate
26333 down grp_write flag.
26334 (create_artificial_child_access): New parameter set_grp_write, set
26335 grp_write to its value.
26336 (propagate_subaccesses_across_link): Also propagate grp_write flag
26338 (propagate_all_subaccesses): Push the closest parent back to work
26339 queue if add_access_to_work_queue returned true.
26341 2017-05-02 Richard Biener <rguenther@suse.de>
26343 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26344 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26345 -fstrict-overflow documentation.
26346 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26347 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26348 flag_strict_overflow.
26349 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26350 * lto-opts.c (lto_write_options): Do not stream it.
26351 * lto-wrapper.c (merge_and_complain): Do not handle it.
26352 * opts.c (default_options_table): Do not set -fstrict-overflow.
26353 (finish_options): Likewise do not clear it when sanitizing.
26354 * simplify-rtx.c (simplify_const_relational_operation): Do not
26355 test flag_strict_overflow.
26357 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26359 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26360 using enabled attribute.
26361 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26362 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26363 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26364 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26365 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26366 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26367 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26368 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26369 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26370 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26372 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26374 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26376 2017-05-02 Richard Biener <rguenther@suse.de>
26378 PR tree-optimization/80591
26380 2017-04-10 Richard Biener <rguenther@suse.de>
26382 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26385 2017-05-02 Richard Biener <rguenther@suse.de>
26387 PR tree-optimization/80549
26388 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26389 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26390 headers if they do not have a preheader.
26392 2017-05-02 Martin Liska <mliska@suse.cz>
26395 * common.opt: Fix typo.
26396 * doc/invoke.texi: Likewise.
26398 2017-05-01 Jan Beulich <jbeulich@suse.com>
26400 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26401 swapping, add (x,x,m,x,n) alternative.
26403 2017-05-01 Nathan Sidwell <nathan@acm.org>
26405 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26406 unnecessary unadjusted_alignment check.
26408 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26411 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26413 * gimplify.c (gimplify_cilk_detach): New function.
26414 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26415 * tree-core.h: Document EXPR_CILK_SPAWN.
26416 * tree.h (EXPR_CILK_SPAWN): Define.
26418 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26420 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26421 to use new fixit_hint representation, using the "replace" logic.
26422 (get_line_span_for_fixit_hint): Likewise.
26423 (layout::print_any_fixits): Likewise.
26424 (selftest::test_one_liner_many_fixits): Rename to...
26425 (selftest::test_one_liner_many_fixits_1): ...this, and update
26426 comment and expected output to reflect that the multiple fix-it
26427 hints are now consolidated into one insertion.
26428 (selftest::test_one_liner_many_fixits_2): New test.
26429 (selftest::test_diagnostic_show_locus_one_liner): Update for
26431 (selftest::test_fixit_consolidation): Update for fix-it API
26433 * diagnostic.c (print_parseable_fixits): Likewise.
26434 * edit-context.c (edited_line::m_line_events): Convert from
26435 auto_vec <line_event *> to auto_vec <line_event>.
26436 (class line_event): Convert from abstract base class to a concrete
26437 class, taking over the role of replace_event.
26438 (class insert_event): Delete.
26439 (class replace_event): Rename to class line_event. Convert to
26441 (edit_context::add_fixits): Reimplement.
26442 (edit_context::apply_insert): Delete.
26443 (edit_context::apply_replace): Rename to...
26444 (edit_context::apply_fixit): ...this. Convert to half-open range.
26445 (edited_file::apply_insert): Delete.
26446 (edited_file::apply_replace): Rename to...
26447 (edited_file::apply_fixit): ...this.
26448 (edited_line::~edited_line): Drop deletion of events.
26449 (edited_line::apply_insert): Delete.
26450 (edited_line::apply_replace): Rename to...
26451 (edited_line::apply_fixit): ...this. Convert to half-open range.
26452 Update for change to type of m_line_events.
26453 * edit-context.h (edit_context::apply_insert): Delete.
26454 (edit_context::apply_replace): Rename to...
26455 (edit_context::apply_fixit): ...this.
26457 2017-05-01 Martin Sebor <msebor@redhat.com>
26459 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26462 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26465 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26466 __get_cpuid_max returns 0.
26467 (__get_cpuid_count): Ditto.
26469 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26471 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26472 replacement expression is another instance of one of its arguments.
26474 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26477 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26478 check for stack push/pop autoinc.
26479 * config/i386/i386.c (ix86_agi_dependent): Return false
26480 if the only reason why modified_in_p returned true is that
26481 addr is SP based and set_insn is a push or pop.
26483 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26485 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26488 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26491 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26492 (account_size_time): Use two predicates - exec_pred and
26494 (evaluate_conditions_for_known_args): Compute both clause and
26496 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26497 (inline_summary_t::duplicate): Update.
26498 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26500 (compute_inline_parameters): Likewise.
26501 (estimate_edge_size_and_time): Update caluclation of time.
26502 (estimate_node_size_and_time): Compute both time and nonspecialized
26504 (estimate_ipcp_clone_size_and_time): Update.
26505 (inline_merge_summary): Update.
26506 (do_estimate_edge_time): Update.
26507 (do_estimate_edge_size): Update.
26508 (do_estimate_edge_hints): Update.
26509 (inline_read_section, inline_write_summary): Stream both new predicates.
26510 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26512 (compute_inlined_call_time): Cleanup.
26513 (big_speedup_p): Update.
26514 (edge_badness): Update.
26515 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26516 (size_time_entry): Replace predicate by exec_predicate and
26517 nonconst_predicate.
26518 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26519 (estimate_edge_time): Return also nonspec_time.
26520 (reset_edge_growth_cache): Update.
26522 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26524 PR rtl-optimization/80491
26525 * ifcvt.c (noce_process_if_block): When looking for x setter
26526 with missing else_bb, don't check only the insn right before
26527 cond_earliest, but look for the last insn that x is modified in
26528 within the same bb.
26530 PR rtl-optimization/80491
26531 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26533 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26535 PR tree-optimization/80487
26536 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26538 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26540 PR tree-optimization/79697
26541 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26542 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26543 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26545 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26546 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26548 2017-04-28 Martin Sebor <msebor@redhat.com>
26550 PR tree-optimization/80523
26551 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26552 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26554 (maybe_warn, format_directive, parse_directive): Use new functions.
26555 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26557 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26559 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26561 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26563 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26564 target_header_dir): Set correctly.
26565 * configure: Regenerated.
26566 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26567 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26568 instead of SYSTEM_HEADER_DIR.
26570 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26572 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26573 (estimate_local_effects): Likewise.
26574 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26575 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26576 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26577 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26578 do_estimate_edge_time, estimate_edge_time): Likewise.
26579 * ipa-inline-analysis.c (estimate_node_size_and_time,
26580 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26581 (estimate_time_after_inlining): Remove.
26583 2017-04-28 Martin Liska <mliska@suse.cz>
26585 * doc/gcov.texi: Enhance documentation of gcov.
26587 2017-04-28 Martin Liska <mliska@suse.cz>
26589 * doc/gcov.texi: Sort options in alphabetic order.
26590 * doc/gcov-dump.texi: Likewise.
26591 * doc/gcov-tool.texi: Likewise.
26592 * gcov.c (print_usage): Likewise.
26593 * gcov-dump.c (print_usage): Likewise.
26594 * gcov-tool.c (print_merge_usage_message): Likewise.
26595 (print_rewrite_usage_message): Likewise.
26596 (print_overlap_usage_message): Likewise.
26598 2017-04-28 Martin Liska <mliska@suse.cz>
26600 PR gcov-profile/53915
26601 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26603 2017-04-28 Martin Liska <mliska@suse.cz>
26605 PR gcov-profile/79891
26606 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26607 is marked by compiler as living on a line.
26608 (get_cycles_count): Remove usage of the union.
26609 (output_intermediate_file): Likewise.
26610 (find_source): Fix GNU coding style.
26611 (accumulate_line_counts): Remove old non-all block mode.
26612 (output_lines): Remove usage of the union.
26613 * profile.c (output_location): Include all BBs, even if
26614 belonging to a same line (and file) as a previous BB.
26616 2017-04-28 Martin Liska <mliska@suse.cz>
26618 * gcov.c (process_args): Handle new argument 'w'.
26619 (read_graph_file): Assign ID to BBs.
26620 (output_branch_count): Display BB # if verbose flag is set.
26621 (output_lines): Likewise for arcs.
26622 (print_usage): Add '--verbose' option help.
26623 * doc/gcov.texi: Document --verbose (-w) option.
26625 2017-04-28 Martin Liska <mliska@suse.cz>
26627 * gcov.c (struct block_location_info): New struct.
26628 (process_file): Fill up the new structure.
26629 (read_graph_file): Replace usage of encoding by the newly added
26631 (add_line_counts): Likewise.
26632 (accumulate_line_counts): Remove usage of the union.
26633 (function_info::function_info): New function.
26634 (function_info::~function_info): Likewise.
26635 (process_file): Call delete instead of release_function.
26636 (release_function): Release the function.
26637 (release_structures): Call delete instead of release_function.
26638 (solve_flow_graph): Replace usage of num_blocks.
26639 (find_exception_blocks): Likewise.
26640 (output_lines): Fix GNU coding style.
26642 2017-04-28 Martin Liska <mliska@suse.cz>
26645 * coverage.c (coverage_remove_note_file): New function.
26646 * coverage.h: Declare the function.
26647 * toplev.c (finalize): Clean if an error has been seen.
26649 2017-04-28 Martin Liska <mliska@suse.cz>
26651 PR gcov-profile/80031
26652 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26653 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26654 * gcov.c (read_graph_file): Read just number of blocks.
26655 * profile.c (branch_prob): Do not stream 0 flags per a basic
26658 2017-04-28 Martin Liska <mliska@suse.cz>
26660 * gcov-dump.c (tag_*): Add new argument to declarations.
26661 (dump_gcov_file): Likewise.
26662 (tag_blocks): Add and use new argument depth.
26663 (tag_arcs): Likewise.
26664 (tag_lines): Likewise.
26665 (tag_counters): Likewise.
26666 (tag_summary): Likewise.
26667 (dump_working_sets): Use depth to do a proper indentation.
26669 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26672 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26673 * symtab.c (symtab_node::debug_symtab): Move definition here.
26675 2017-04-28 Richard Biener <rguenther@suse.de>
26677 * lto-streamer.h (LTO_major_version): Bump to 7.
26679 2017-04-28 Richard Biener <rguenther@suse.de>
26681 * tree-vrp.c (assert_info): New struct.
26682 (add_assert_info): New helper.
26683 (register_edge_assert_for_2): Refactor to add asserts to a vector
26685 (register_edge_assert_for_1): Likewise.
26686 (register_edge_assert_for): Likewise.
26687 (finish_register_edge_assert_for): New helper actually registering
26688 asserts where live on edge.
26689 (find_conditional_asserts): Adjust.
26690 (find_switch_asserts): Likewise.
26691 (evrp_dom_walker::try_find_new_range): Generalize.
26692 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26694 2017-04-27 Marek Polacek <polacek@redhat.com>
26697 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26698 arg10 and arg11 to itype.
26700 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26702 * doc/extend.texi (Object Size Checking): Improve grammar.
26704 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26707 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26708 that the logic for permitting reciprocal estimates matches that
26711 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26714 * tree.c (type_cache_hasher::equal): Only compare
26715 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26716 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26717 non-aggregate element types.
26718 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26719 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26721 2017-04-27 Richard Biener <rguenther@suse.de>
26723 PR middle-end/80533
26724 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26725 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26726 keeping a reference to a trailing array.
26728 2017-04-27 Richard Biener <rguenther@suse.de>
26730 PR middle-end/80539
26731 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26732 being in loop-closed SSA form conservatively.
26733 (chrec_fold_multiply_poly_poly): Likewise.
26735 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26737 PR middle-end/79665
26738 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26739 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26741 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26744 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26745 (aarch64_function_arg_alignment): Return unsigned int again, but still
26746 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26747 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26748 Don't emit -Wpsabi note.
26749 (aarch64_function_arg_boundary): Likewise.
26750 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26753 2017-04-26 Nathan Sidwell <nathan@acm.org>
26755 * tree.h (crc32_unsigned_n): Declare.
26756 (crc32_unsigned, crc32_unsigned): Make inline.
26757 * tree.c (crc32_unsigned_bits): Replace with ...
26758 (crc32_unsigned_n): ... this.
26759 (crc32_unsigned, crc32_byte): Remove.
26760 (crc32_string): Remove unnecessary braces.
26762 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26764 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26765 * ipa-inline-analysis.c (MAX_TIME): Remove.
26766 (account_size_time): Use sreal for time.
26767 (dump_inline_summary): Update.
26768 (estimate_function_body_sizes): Update.
26769 (estimate_edge_size_and_time): Update.
26770 (estimate_calls_size_and_time): Update.
26771 (estimate_node_size_and_time): Update.
26772 (inline_merge_summary): Update.
26773 (inline_update_overall_summary): Update.
26774 (estimate_time_after_inlining): Update.
26775 (inline_read_section): Update.
26776 (inline_write_summary): Update.
26777 * ipa-inline.c (compute_uninlined_call_time): Update.
26778 (compute_inlined_call_time): Update.
26779 (recursive_inlining): Update.
26780 (inline_small_functions): Update.
26781 (dump_overall_stats): Update.
26782 * ipa-inline.h: Include sreal.h.
26783 (size_time_entry): Turn time to sreal.
26784 (inline_summary): Turn self_time nad time to sreal.
26786 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26788 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26790 (sreal::stream_out, sreal::stream_in): New.
26791 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26793 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26795 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26798 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26801 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26803 (dimode_scalar_chain::compute_convert_gain): Ditto.
26804 (dimode_scalar_chain::make_vector_copies): Ditto.
26805 (dimode_scalar_chain::convert_reg): Ditto.
26806 (dimode_scalar_chain::convert_insn): Ditto.
26807 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26808 (VI248_AVX512BW_1): New mode iterator.
26809 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26810 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26813 2017-04-25 Martin Sebor <msebor@redhat.com>
26815 PR tree-optimization/80497
26816 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26817 constants are representable in HOST_WIDE_INT.
26818 (parse_directive): Ditto.
26820 2017-04-25 Martin Sebor <msebor@redhat.com>
26823 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26824 (new_zero_array): Adjust signature.
26825 (dom_info::dom_init): Used unsigned rather that size_t.
26826 (dom_info::dom_info): Same.
26828 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26829 Jakub Jelinek <jakub@redhat.com>
26832 * config/arm/arm.c: Include gimple.h.
26833 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26834 returns negative, increment ncrn only if it returned positive.
26835 (arm_needs_doubleword_align): Return int instead of bool,
26836 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26837 members, but if there is any such non-FIELD_DECL
26838 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26839 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26840 returns negative, increment nregs only if it returned positive.
26841 (arm_setup_incoming_varargs): Likewise.
26842 (arm_function_arg_boundary): Emit -Wpsabi note if
26843 arm_needs_doubleword_align returns negative, return
26844 DOUBLEWORD_ALIGNMENT only if it returned positive.
26846 2017-04-25 Marek Polacek <polacek@redhat.com>
26849 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26850 first argument to type.
26852 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26855 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26856 type checks to test for compatibility instead of equality.
26858 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26859 Jakub Jelinek <jakub@redhat.com>
26862 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26864 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26865 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26866 the alignment computation, but return their maximum in warn_alignment.
26867 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26868 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26870 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26871 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26874 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26876 * config/arc/simdext.md (dmpyh): Fix typo.
26878 2017-04-25 Richard Biener <rguenther@suse.de>
26880 PR tree-optimization/80492
26881 * alias.c (compare_base_decls): Handle registers with asm
26882 specification conservatively.
26883 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26884 compare_base_decls returning dont-know properly.
26886 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26888 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26889 (legitimate_offset_address_p): New function.
26890 (arc_legitimate_address_p): Use above function.
26892 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26894 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26896 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26898 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26899 ACCH registers whenever they are available.
26901 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26903 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26904 double regs fix when not used.
26906 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26908 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26910 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26911 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26913 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26915 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26916 when emitting short ADD instructions.
26918 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26920 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26922 (cmpsi_cc_c_insn): Likewise.
26923 (cbranchsi4_scratch): Compute proper instruction length using
26924 compact_hreg_operand.
26925 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26927 2017-04-25 Richard Biener <rguenther@suse.de>
26929 PR middle-end/80509
26930 * passes.c (pass_manager::pass_manager): Initialize
26931 m_name_to_pass_map.
26933 2017-04-25 Richard Biener <rguenther@suse.de>
26935 PR tree-optimization/79201
26936 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26938 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26941 * config/s390/vector.md: Split MEM->GPR vector moves for
26942 non-s_operand addresses.
26944 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26947 * config/s390/predicates.md (reload_const_wide_int_operand): New
26949 * config/s390/s390.md ("movti"): Remove d/P alternative.
26950 ("movti_bigconst"): New pattern definition.
26952 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26955 * s390-protos.h (s390_expand_cs_hqi): Removed.
26956 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26957 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26958 modes as well as CCZ1mode and CCZmode.
26959 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26960 signature of s390_emit_compare_and_swap.
26961 (s390_expand_cs_hqi): Likewise, make static.
26962 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26963 compare-and-swap, in some cases.
26964 (s390_expand_cs): Wrapper function.
26965 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26967 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26968 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26969 patterns for small and large integers. Forbid symref memory operands.
26970 Move expander to s390.c. Require cc register.
26971 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26972 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26973 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26974 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26975 symref memory operands. Remove CC mode and call s390_match_ccmode
26977 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26979 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26981 * config/s390/s390.md (define_peephole2): New peephole to help
26982 combining the load-and-test pattern with volatile memory.
26984 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26986 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26987 with CCZmode for TARGET_Z196.
26989 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26991 PR rtl-optimization/80501
26992 * combine.c (make_compound_operation_int): Set subreg_code to SET
26993 even for AND with mask of the sign bit of mode.
26995 PR rtl-optimization/80500
26996 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26997 sum's initial value.
26999 2017-04-25 Julian Brown <julian@codesourcery.com>
27000 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27002 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27004 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27006 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27008 2017-04-25 Julian Brown <julian@codesourcery.com>
27009 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27011 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27012 (thunderx2t99_sha): New Reservation.
27014 2017-04-25 Julian Brown <julian@codesourcery.com>
27015 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27017 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27018 type for 1-element load.
27020 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27022 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27024 2017-04-24 Martin Jambor <mjambor@suse.cz>
27026 PR tree-optimization/80293
27027 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27028 char arrays not totally scalarizable if it is false.
27029 (analyze_all_variable_accesses): Pass correct value in the new
27030 parameter. Add a statistics counter.
27032 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27034 PR middle-end/79931
27035 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27037 2017-04-24 Richard Biener <rguenther@suse.de>
27039 PR tree-optimization/80494
27040 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27041 out for complex types.
27043 2017-04-24 Richard Biener <rguenther@suse.de>
27045 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27046 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27047 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27048 (DFS): Adjust and never fail.
27049 (sccvn_dom_walker::fail): Remove.
27050 (sccvn_dom_walker::before_dom_children): Adjust.
27051 (run_scc_vn): Likewise and never fail.
27052 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27053 (pass_fre::execute): Likewise.
27055 2017-04-24 Richard Biener <rguenther@suse.de>
27057 PR tree-optimization/79725
27058 * tree-ssa-sink.c (statement_sink_location): Return whether
27059 failure reason was zero uses. Move that check later.
27060 (sink_code_in_bb): Deal with zero uses by removing the stmt
27063 2017-04-24 Richard Biener <rguenther@suse.de>
27066 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27067 pointer-based references.
27069 2017-04-24 Richard Biener <rguenther@suse.de>
27072 * pass_manager.h (pass_manager::operator new): Remove.
27073 (pass_manager::operator delete): Likewise.
27074 * passes.c (pass_manager::operator new): Remove.
27075 (pass_manager::operator delete): Likewise.
27076 (pass_manager::pass_manager): Zero individual pass members.
27078 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27081 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27082 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27083 Check "XEXP (src, 1)" operand here.
27084 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27085 Check "XEXP (src, 1)" operand here.
27086 (dimode_scalar_chain::make_vector_copies): Detect count register
27087 of a shift instruction. Zero extend count register from QImode
27088 to DImode to satisfy vector shift pattern count operand predicate.
27089 Substitute vector shift count operand with a DImode copy.
27090 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27093 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27095 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27096 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27097 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27098 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27099 (UNSPEC_NOREX_MEM): Remove definition.
27101 2017-04-21 Richard Biener <rguenther@suse.de>
27103 PR tree-optimization/79547
27104 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27105 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27106 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27107 without any constraints.
27109 2017-04-21 Richard Biener <rguenther@suse.de>
27111 PR tree-optimization/78847
27112 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27114 2017-04-21 Richard Biener <rguenther@suse.de>
27116 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27117 (build_distinct_type_copy): Likewise.
27118 (build_variant_type_copy): Likewise.
27119 * tree.c (build_qualified_type): Pass down mem-stat info.
27120 (build_distinct_type_copy): Likewise.
27121 (build_variant_type_copy): Likewise.
27123 2017-04-21 Richard Biener <rguenther@suse.de>
27125 PR tree-optimization/80237
27126 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27128 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27129 for a simplified result.
27131 2016-04-21 Richard Biener <rguenther@suse.de>
27133 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27134 sth as strict as a simple_iv but a chrec without symbols and an
27135 operand defined in the loop we are peeling (and not some subloop).
27136 (propagate_constants_for_unrolling): Propagate all constants.
27138 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27141 * config/i386/i386.c (print_reg): Remove assert for disalowed
27142 regno values, call output_operand_lossage instead.
27144 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27147 * config/i386/constraints.md (Yc): New register constraint.
27148 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27149 Use Yc constraint for alternative 2 of operand 0. Remove
27150 preferred_for_speed attribute.
27152 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27154 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27155 lastprivate clauses in SIMT case.
27157 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27159 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27161 2017-04-20 Richard Biener <rguenther@suse.de>
27163 PR tree-optimization/57796
27164 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27165 as N scalar stores.
27166 (vect_model_load_cost): Cost gathers as N scalar loads.
27168 2017-04-20 Richard Biener <rguenther@suse.de>
27170 * ggc-page.c (ggc_allocated_p): Rename to ...
27171 (safe_lookup_page_table_entry): ... this and return the lookup
27173 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27175 2017-04-20 Richard Biener <rguenther@suse.de>
27177 PR tree-optimization/80453
27178 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27179 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27180 from the conditions.
27181 (vn_phi_eq): Pass them down.
27182 (vn_phi_lookup): Record them.
27183 (vn_phi_insert): Likewise.
27185 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27187 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27188 uninitialized variable warning to avoid buffer overrun.
27190 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27193 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27194 is suppressed for '{ 0 }' in C.
27196 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27198 * BASE-VER: Set to 8.0.0.
27200 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27202 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27203 priority .init_array and .fini_array section with SECTION_NOTYPE
27206 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27208 PR middle-end/80423
27209 * tree.h (build_array_type): Add typeless_storage default argument.
27210 * tree.c (type_cache_hasher::equal): Also compare
27211 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27212 (build_array_type): Add typeless_storage argument, set
27213 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27215 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27216 (build_array_type): Likewise. Add typeless_storage argument.
27218 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27219 Jakub Jelinek <jakub@redhat.com>
27221 PR tree-optimization/80426
27222 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27223 operation on symbolic operands, also compute the overflow for the
27224 invariant part when the operation degenerates into a negation.
27226 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27229 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27230 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27233 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27235 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27238 * config/avr/avr.c (tree.h): Include it.
27239 (cgraph.h): Include it.
27240 (avr_encode_section_info): Don't warn for uninitialized progmem
27241 variable if it's just an alias.
27243 2017-04-19 Richard Biener <rguenther@suse.de>
27246 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27247 when needed by AutoPGO.
27249 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27252 * doc/lto.texi: Remove an extra 'that'.
27254 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27256 PR rtl-optimization/80429
27257 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27258 are only used in debug insns.
27260 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27261 Vladimir Makarov <vmakarov@redhat.com>
27263 * config/sparc/predicates.md (input_operand): Add comment. Return
27264 true for any memory operand when LRA is in progress.
27265 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27267 2017-04-18 Jeff Law <law@redhat.com>
27270 * mips.md ({return,simple_return}_internal): Do not overwrite
27273 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27275 PR tree-optimization/80443
27276 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27277 instead of adding 1, subtract -1 and similarly instead of subtracting
27280 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27282 PR rtl-optimization/80357
27283 * haifa-sched.c (tmp_bitmap): New variable.
27284 (model_recompute): Handle duplicate use records.
27285 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27286 (free_global_sched_pressure_data): Free it.
27288 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27291 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27292 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27293 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27294 instead of SYSTEM_HEADER_DIR.
27296 2017-04-18 Jeff Law <law@redhat.com>
27298 PR middle-end/80422
27299 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27300 predecessors after walking up the insn chain.
27302 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27305 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27306 sizetype type into debug info.
27308 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27311 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27312 unneeded test for TARGET_UPPER_REGS_SF.
27313 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27315 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27318 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27319 instead of gsi_after_labels.
27321 2017-04-18 Jeff Law <law@redhat.com>
27323 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27327 2017-04-13 Jeff Law <law@redhat.com>
27328 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27329 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27331 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27334 * config/avr/avr.c (intl.h): Include it.
27335 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27337 2017-04-18 Martin Liska <mliska@suse.cz>
27339 PR gcov-profile/78783
27340 * gcov-tool.c (gcov_output_files): Validate that destination
27341 file is either removed by the tool or by a user.
27343 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27344 Guy Benyei <guybe@mellanox.com>
27346 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27347 block, and do not negate it, the stored id is already negative.
27349 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27351 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27353 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27356 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27357 masks of options that should be turned off if the VSX vector
27358 options are turned off.
27359 (OTHER_P8_VECTOR_MASKS): Likewise.
27360 (OTHER_VSX_VECTOR_MASKS): Likewise.
27361 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27362 rs6000_disable_incompatible_switches to validate no type switches
27364 (rs6000_incompatible_switch): New function to disallow turning on
27365 other vector options if -mno-vsx, -mno-power8-vector, or
27366 -mno-power9-vector are specified.
27368 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27370 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27372 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27374 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27375 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27376 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27377 (ARG_POINTER_CFA_OFFSET): Likewise.
27379 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27381 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27382 conditions to take advantage of various optimizations.
27384 2017-04-13 Jeff Law <law@redhat.com>
27386 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27387 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27388 (zero_extendsidi2_dext): Likewise.
27390 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27393 * fold-const.c (fold_ternary_loc): Revert
27394 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27397 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27399 PR rtl-optimization/80343
27400 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27401 new scratch pseudo.
27403 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27406 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27407 to ubsan_encode_value.
27409 2017-04-13 Jeff Law <law@redhat.com>
27411 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27412 appearing in DEBUG_INSNs.
27414 2017-04-13 Martin Liska <mliska@suse.cz>
27416 PR gcov-profile/80413
27417 * gcov-io.c (gcov_write_string): Copy to buffer just when
27418 allocated size is greater than zero.
27420 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27423 * dwarf2out.c (decls_for_scope): Ignore declarations of
27424 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27426 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27429 * ipa-visibility.c (non_local_p): Fix typos.
27430 (localize_node): When localizing symbol in same comdat group,
27431 dissolve the group only when we know external symbols are going
27433 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27435 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27437 PR tree-optimization/79390
27438 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27439 order does not result in usable sequence, retry with reversed operand
27445 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27446 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27447 op0 instead of fold_convert_loc (loc, type, arg0).
27449 2017-04-12 Jeff Law <law@redhat.com>
27451 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27452 has a delay slot in the generated code.
27454 * config/cris/cris.md (cris_preferred_reload_class): Return
27455 GENNONACR_REGS rather than GENERAL_REGS.
27457 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27460 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27461 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27462 signedness of the result type.
27464 2017-04-12 Richard Biener <rguenther@suse.de>
27465 Jeff Law <law@redhat.com>
27467 PR tree-optimization/80359
27468 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27469 trim stores to TARGET_MEM_REFs.
27471 2017-04-12 Richard Biener <rguenther@suse.de>
27473 PR tree-optimization/79390
27474 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27475 threading case even more.
27477 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27480 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27481 for quad_address_p for TImode, instead of just not indexed_address.
27483 2017-04-12 Richard Biener <rguenther@suse.de>
27484 Bernd Edlinger <bernd.edlinger@hotmail.de>
27486 PR middle-end/79671
27487 * alias.c (component_uses_parent_alias_set_from): Handle
27488 TYPE_TYPELESS_STORAGE.
27489 (get_alias_set): Likewise.
27490 * tree-core.h (tree_type_common): Add typeless_storage flag.
27491 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27492 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27493 for types containing members with TYPE_TYPELESS_STORAGE.
27494 (place_field): Likewise.
27495 (layout_type): Likewise for ARRAY_TYPE.
27496 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27497 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27498 TYPE_TYPELESS_STORAGE.
27499 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27501 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27504 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27505 first argument to type.
27507 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27511 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27512 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27513 (rs6000_expand_binop_builtin): Likewise.
27514 (rs6000_expand_ternop_builtin): Likewise; also add missing
27515 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27516 vshasigma built-ins.
27517 * doc/extend.texi: Document that vec_xxpermdi's third argument
27518 must be a constant.
27520 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27522 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27523 Use shift_const cost parameter when calculating gain of STV shifts.
27525 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27527 PR rtl-optimization/70478
27528 * lra-constraints.c (process_alt_operands): Check memory for
27529 disfavoring memory insn operand.
27531 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27533 PR middle-end/80100
27534 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27535 left shift in unsigned HOST_WIDE_INT type.
27537 PR rtl-optimization/80385
27538 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27539 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27542 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27543 if they have any depend clauses.
27545 2017-04-11 Martin Liska <mliska@suse.cz>
27548 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27549 * ipa-split.c (split_function): Create a local comdat symbol
27550 if caller is in a comdat group.
27552 2017-04-11 Martin Liska <mliska@suse.cz>
27555 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27558 2017-04-11 Martin Sebor <msebor@redhat.com>
27560 PR middle-end/80364
27561 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27562 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27564 (directive::set_width, directive::set_precision, format_character):
27566 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27569 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27572 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27573 conflict, set target->arch_name instead of target->cpu_name.
27575 2017-04-11 Richard Biener <rguenther@suse.de>
27577 PR tree-optimization/80374
27578 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27579 build_zero_cst, remove fold_convertible_p check again.
27581 2017-04-11 Martin Liska <mliska@suse.cz>
27584 * ubsan.c (instrument_object_size): Do not instrument register
27587 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27590 * config/i386/i386-builtin-types.def
27591 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27592 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27593 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27594 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27595 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27596 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27597 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27598 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27599 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27600 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27601 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27602 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27603 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27604 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27605 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27606 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27607 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27608 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27609 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27610 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27611 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27612 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27613 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27614 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27615 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27616 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27617 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27618 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27619 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27620 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27621 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27622 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27623 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27624 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27625 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27626 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27627 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27628 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27629 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27630 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27631 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27632 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27633 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27634 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27635 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27636 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27638 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27639 flag to second_arg_count, handle 4 argument function type _COUNT
27640 aliases, handle second_arg_count on second argument rather than last.
27642 2017-04-10 Jeff Law <law@redhat.com>
27644 PR tree-optimization/80374
27645 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27646 record anything if we can not convert integer_zero_node to the
27649 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27652 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27653 Enhance special handling given to the TARGET_P9_MINMAX option in
27654 relation to certain other options.
27656 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27658 PR tree-optimization/80153
27659 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27660 remove POINTER_PLUS_EXPR's base part directly, rather than through
27663 2017-04-10 Richard Biener <rguenther@suse.de>
27664 Bin Cheng <bin.cheng@arm.com>
27666 PR tree-optimization/80153
27667 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27668 the first element of pointer type aff_tree. Build result expr in
27670 (add_elt_to_tree): Convert to type unconditionally. Remove other
27671 fold_convert calls.
27672 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27673 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27675 2017-04-10 Richard Biener <rguenther@suse.de>
27677 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27680 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27682 PR rtl-optimization/70478
27683 * lra-constraints.c (curr_small_class_check): New.
27684 (update_and_check_small_class_inputs): New.
27685 (process_alt_operands): Update curr_small_class_check. Disfavor
27686 alternative insn memory operands. Check available regs for small
27689 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27692 * config/mips/mips.opt (-mvirt): Update description.
27693 * doc/invoke.texi (-mvirt): Likewise.
27695 2017-04-10 Richard Biener <rguenther@suse.de>
27697 PR middle-end/80362
27698 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27699 looking for NEGATE_EXPR in -A / -B to A / B folding.
27701 2017-04-10 Martin Liska <mliska@suse.cz>
27703 PR gcov-profile/80224
27704 * gcov.c (print_usage): Fix usage string.
27705 (get_gcov_intermediate_filename): Remove.
27706 (output_gcov_file): Use both for normal and intermediate format.
27707 (generate_results): Do not initialize special file for
27708 intermediate format.
27710 2017-04-10 Richard Biener <rguenther@suse.de>
27712 PR tree-optimization/80304
27713 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27716 2017-04-10 Nathan Sidwell <nathan@acm.org>
27719 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27720 (rs6000_init_builtins): Use it.
27722 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27724 * config/arm/arm.md (<mrc>): Add mode to SET source.
27725 (<mrrc>): Likewise.
27727 2017-04-10 Richard Biener <rguenther@suse.de>
27729 PR middle-end/80344
27730 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27732 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27735 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27736 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27737 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27738 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27739 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27740 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27741 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27742 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27743 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27744 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27745 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27746 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27747 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27748 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27749 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27750 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27751 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27752 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27753 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27754 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27755 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27756 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27757 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27759 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27761 PR rtl-optimization/70478
27762 * lra-constraints.c: Reverse the last patch.
27764 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27766 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27767 Add comment for WCHAR_T.
27769 2017-04-08 Martin Liska <mliska@suse.cz>
27772 2017-04-07 Martin Liska <mliska@suse.cz>
27775 * ipa-split.c (split_function): Add function part to a same comdat
27778 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27781 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27783 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27785 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27786 (vec_pairing): ...to this.
27787 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27788 (rs6000_sched_init): Adjust for name change.
27789 (struct rs6000_sched_context): Likewise.
27790 (rs6000_init_sched_context): Likewise.
27791 (rs6000_set_sched_context): Likewise.
27793 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27799 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27801 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27802 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27803 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27805 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27807 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27809 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27811 PR rtl-optimization/70703
27812 * ira-color.c (update_conflict_hard_regno_costs): Use
27813 int64_t instead of HOST_WIDE_INT.
27815 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27817 PR rtl-optimization/70478
27818 * lra-constraints.c (process_alt_operands): Disfavor alternative
27819 insn memory operands.
27821 2017-04-07 Jeff Law <law@redhat.com>
27823 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27824 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27826 2017-04-07 Martin Liska <mliska@suse.cz>
27829 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27830 Show error message instead of an ICE.
27832 2017-04-07 Martin Liska <mliska@suse.cz>
27835 * ipa-split.c (split_function): Add function part to a same comdat
27838 2017-04-07 Richard Biener <rguenther@suse.de>
27840 PR middle-end/80341
27841 * tree.c (get_unwidened): Also handle ! for_type case for
27843 * convert.c (do_narrow): Split out from ...
27844 (convert_to_integer_1): ... here. Do not pass final truncation
27845 type to get_unwidened for TRUNC_DIV_EXPR.
27847 2017-04-07 Richard Biener <rguenther@suse.de>
27849 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27851 (aff_combination_const): Adjust.
27852 (aff_combination_scale): Likewise.
27853 (aff_combination_add_elt): Likewise.
27854 (aff_combination_add_cst): Likewise.
27855 (aff_combination_convert): Likewise.
27856 (add_elt_to_tree): Likewise. Remove unused argument.
27857 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27859 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27861 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27863 * config/arm/arm.c (arm_default_short_enums): Use
27864 ARM_DEFAULT_SHORT_ENUMS.
27865 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27867 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27870 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27871 members with redundant out-of-class redeclaration.
27873 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27876 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27877 * config/i386/i386.md (*zero_extendsidi2):
27878 Add (?*x,*x) and (?*v,*v) alternatives.
27880 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27883 * config/i386/i386.c (ix86_expand_builtin)
27884 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27885 mode from insn data. Convert operands to insn operand mode.
27886 Copy operands that don't satisfy insn predicate to a register.
27888 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27890 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27893 2017-04-06 Richard Biener <rguenther@suse.de>
27895 PR tree-optimization/80334
27896 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27897 preserve alignment of accesses.
27899 2017-04-06 Richard Biener <rguenther@suse.de>
27901 PR tree-optimization/80262
27902 * tree-sra.c (build_ref_for_offset): Preserve address-space
27904 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27905 Drop useless address-space information on MEM_REF offsets.
27907 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27909 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27911 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27913 PR rtl-optimization/70703
27914 * ira-color.c (update_conflict_hard_regno_costs): Use
27915 HOST_WIDE_INT instead of long.
27917 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27920 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27921 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27923 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27924 for x86_64 target. Handle -m3dnowa option.
27926 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27928 PR rtl-optimization/70703
27929 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27930 (update_conflict_hard_regno_costs): Use long instead of unsigned
27931 arithmetic for cost calculation.
27933 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27934 Bernd Edlinger <bernd.edlinger@hotmail.de>
27937 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27940 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27943 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27944 ptr_mode with Pmode throughout.
27945 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27946 into probe_stack_range and use DImode.
27948 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27951 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27952 call_eh_return is true.
27954 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27956 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27957 Initialize last_match_fntype_index.
27959 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27962 * tree-nvr.c: Include internal-fn.h.
27963 (pass_return_slot::execute): Ignore internal calls without
27966 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27967 Richard Biener <rguenther@suse.de>
27970 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27971 captures used multiple times, except for the last use.
27972 * generic-match-head.c: Include gimplify.h.
27974 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27976 PR tree-optimization/79390
27977 * target.h (struct noce_if_info): Declare.
27978 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27979 * target.def (noce_conversion_profitable_p): New target hook.
27980 * ifcvt.h (struct noce_if_info): New type, moved from ...
27981 * ifcvt.c (struct noce_if_info): ... here.
27982 (noce_conversion_profitable_p): Renamed to ...
27983 (default_noce_conversion_profitable_p): ... this. No longer
27985 (noce_try_store_flag_constants, noce_try_addcc,
27986 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27987 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27988 instead of noce_conversion_profitable_p.
27989 * config/i386/i386.c: Include ifcvt.h.
27990 (ix86_option_override_internal): Don't override
27991 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27992 (ix86_noce_conversion_profitable_p): New function.
27993 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27994 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27995 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27996 * doc/tm.texi: Regenerated.
27998 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28000 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28003 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28006 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28007 instructions for small multiply cores.
28009 2017-04-04 Jeff Law <law@redhat.com>
28011 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28013 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28014 that are not set by the loop over the elements.
28016 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28019 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28020 int mode, convert_modes it to mode as unsigned, otherwise use
28021 lowpart_subreg to mode rather than SImode.
28022 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28023 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28024 Use DImode instead of SImode for the shift count operand.
28025 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28028 2017-04-04 Richard Biener <rguenther@suse.de>
28030 PR middle-end/80281
28031 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28032 arithmetic done for the negate or the plus. Simplify.
28033 (A - (-B) -> A + B): Likewise.
28034 * fold-const.c (split_tree): Make sure to not negate pointers.
28036 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28038 PR rtl-optimization/60818
28039 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28040 a compare of comparisons with the thing compared if this results
28041 in a different machine mode.
28043 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28045 * alias.c (base_alias_check): Fix typo in comment.
28046 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28047 * cgraphunit.c (symbol_table::compile): Likewise.
28048 * collect2.c (maybe_run_lto_and_relink): Likewise.
28049 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28050 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28051 * config/avr/avr.c (avr_map_op_t): Likewise.
28052 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28053 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28054 * config/epiphany/epiphany.md (movcc): Likewise.
28055 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28056 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28058 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28059 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28060 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28061 * config/sh/sh.c (sh_rtx_costs): Likewise.
28062 * fold-const.c (fold_truth_andor): Likewise.
28063 * genautomata.c (collapse_flag): Likewise.
28064 * gengtype.h (struct type::u::s): Likewise.
28065 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28066 * input.c (FORMAT_AMOUNT): Likewise.
28067 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28068 (known_aggs_to_agg_replacement_list): Likewise.
28069 * ipa-inline-analysis.c: Likewise.
28070 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28071 * ipa-polymorphic-call.c
28072 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28073 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28074 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28076 * modulo-sched.c (apply_reg_moves): Likewise.
28077 * omp-expand.c (build_omp_regions_1): Likewise.
28078 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28079 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28080 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28081 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28082 * value-prof.c: Likewise.
28083 * var-tracking.c (val_reset): Likewise.
28085 2017-04-03 Richard Biener <rguenther@suse.de>
28087 PR tree-optimization/80275
28088 * fold-const.c (split_address_to_core_and_offset): Handle
28091 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28093 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28094 descriptors is at least equal to that of functions.
28096 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28098 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28100 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28103 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28104 (mov<IMOD4:mode>): New expander.
28105 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28107 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28109 PR rtl-optimization/79405
28110 * fwprop.c (propagations_left): New variable.
28111 (forward_propagate_into): Decrement it.
28112 (fwprop_init): Initialize it.
28113 (fw_prop): If the variable has reached zero, stop propagating.
28114 (fwprop_addr): Ditto.
28116 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28119 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28120 a FUNCTION_DECL, pass it as decl instead of origin to
28123 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28125 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28128 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28131 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28132 TARGET_VSX_SMALL_INTEGER.
28134 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28136 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28137 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28139 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28141 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28142 extraction from odd-numbered MSA register.
28144 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28146 PR middle-end/80173
28147 * expmed.c (store_bit_field_1): Don't attempt to create
28148 a word subreg out of hard registers wider than word if they
28149 have HARD_REGNO_NREGS of 1 for their mode.
28151 PR middle-end/80163
28152 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28153 conversions to integer types wider than word and pointer.
28156 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28157 (rtx_equal_for_cselib_p): Pass 0 to it.
28158 * cselib.c (cselib_hasher::equal): Likewise.
28159 (rtx_equal_for_cselib_1): Add depth argument. If depth
28160 is 128, don't look up VALUE locs and punt. Increment
28161 depth in recursive calls when walking VALUE locs.
28163 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28165 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28166 (make_gcov_file_name): Use the canonical path name for generating
28168 (read_line): Fix handling of files with ascii null bytes.
28170 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28172 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28173 to initialise a vector register instead
28174 of using a const_int.
28176 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28178 PR translation/80189
28179 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28180 diagnostic messages.
28182 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28185 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28186 (dfp_diex_<mode>): Update mode of operand 1.
28187 * doc/extend.texi (dxex, dxexq): Document change to return type.
28188 (diex, diexq): Document change to argument type.
28190 2017-03-30 Martin Jambor <mjambor@suse.cz>
28193 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28194 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28195 it reflects the signature changes performed at the callee side.
28196 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28197 to cgraph_build_function_type_skip_args.
28198 (build_function_decl_skip_args): Adjust call to the above function.
28200 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28203 * config/i386/sse.md
28204 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28205 register as dest whenever it is a MEM not rtx_equal_p to the
28206 corresponding dup operand, and when forcing into reg move the
28207 reg into the memory afterwards.
28208 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28209 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28210 for the force_reg mode.
28211 (avx512vl_vextractf128<mode>): Use register as dest either
28212 always when a MEM, or when it is a MEM not rtx_equal_p to the
28213 corresponding dup operand, or even not when it is a CONST_VECTOR
28214 depending on the mode and lo vs. hi.
28215 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28217 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28218 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28219 Likewise. Require that operands[2] is even.
28220 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28221 Remove extraneous parens. Require that operands[2] is a multiple
28223 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28224 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28225 disallow memory then.
28227 2017-03-30 Richard Biener <rguenther@suse.de>
28229 PR tree-optimization/77498
28230 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28231 to non-constants over backedges.
28233 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28235 PR rtl-optimization/80233
28236 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28237 as last_combined_insn. Do not test for BARRIER_P separately.
28239 2017-03-29 Andreas Schwab <schwab@suse.de>
28242 * calls.c (prepare_call_address): Convert funexp to Pmode before
28243 copying to temp reg.
28245 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28247 PR tree-optimization/80158
28248 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28249 Handle possible future case of more than one alternate
28251 (replace_rhs_if_not_dup): Likewise.
28252 (replace_one_candidate): Likewise.
28254 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28256 PR rtl-optimization/80193
28257 * ira.c (ira): Do not check allocation for LRA.
28259 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28261 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28262 (nvptx_output_simt_exit): Declare.
28263 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28264 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28265 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28266 (nvptx_declare_function_name): ...here. Emit declaration of local
28267 memory space buffer for omp_simt_enter insn.
28268 (nvptx_output_unisimt_switch): New.
28269 (nvptx_output_softstack_switch): New.
28270 (nvptx_output_simt_enter): New.
28271 (nvptx_output_simt_exit): New.
28272 * config/nvptx/nvptx.h (struct machine_function): New fields
28273 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28274 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28275 (UNSPECV_SIMT_EXIT): Ditto.
28276 (omp_simt_enter_insn): New insn.
28277 (omp_simt_enter): New expansion.
28278 (omp_simt_exit): New insn.
28279 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28281 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28282 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28283 (expand_GOMP_SIMT_EXIT): New.
28284 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28285 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28286 (GOMP_SIMT_EXIT): Ditto.
28287 * target-insns.def (omp_simt_enter): New insn.
28288 (omp_simt_exit): Ditto.
28289 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28291 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28292 (lower_rec_input_clauses): Likewise.
28293 (lower_lastprivate_clauses): Handle SIMT privatization.
28295 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28296 (ompdevlow_adjust_simt_enter): New.
28297 (find_simtpriv_var_op): New.
28298 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28299 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28301 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28302 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28303 (copy_decl_for_dup_finish): Ditto.
28305 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28307 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28310 * config/i386/i386.c (ix86_option_override_internal): Always
28311 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28313 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28315 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28317 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28319 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28320 mark new edge's irreducible flag accordign to it.
28321 (vect_do_peeling): Check loop preheader edge's irreducible flag
28322 and pass it to function slpeel_add_loop_guard.
28324 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28326 PR tree-optimization/80218
28327 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28328 Update block frequencies and counts.
28330 2017-03-28 Richard Biener <rguenther@suse.de>
28332 PR tree-optimization/78644
28333 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28334 of a simplification result we may not use it at all.
28336 2017-03-28 Richard Biener <rguenther@suse.de>
28339 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28340 without arguments, generate default definition of a SSA name.
28342 2017-03-28 Richard Biener <rguenther@suse.de>
28344 PR middle-end/80222
28345 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28346 TYPE_REF_CAN_ALIAS_ALL references.
28347 * fold-const.c (fold_indirect_ref_1): Likewise.
28349 2017-03-28 Martin Liska <mliska@suse.cz>
28352 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28353 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28355 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28356 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28358 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28359 (EXTRA_SPECS): Define.
28360 (SUBTARGET_EXTRA_SPECS): Likewise.
28361 (SUBTARGET_CPP_SPEC): Likewise.
28362 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28363 SUBTARGET_EXTRA_SPECS.
28364 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28366 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28368 * config/arc/simdext.md (vst64_insn): Update pattern.
28369 (vld32wh_insn): Likewise.
28370 (vld32wl_insn): Likewise.
28371 (vld64_insn): Likewise.
28372 (vld32_insn): Likewise.
28374 2017-03-28 Marek Polacek <polacek@redhat.com>
28377 * fold-const.c (fold_comparison): Use protected_set_expr_location
28378 instead of SET_EXPR_LOCATION.
28380 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28382 * tree.c (add_expr): Avoid name lookup warning.
28384 2017-03-27 Jeff Law <law@redhat.com>
28386 PR tree-optimization/80216
28387 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28388 function name. Limit recursion depth.
28389 (record_temporary_equivalences): Corresponding changes.
28391 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28393 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28396 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28399 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28401 * cfgcleanup.c (reg_note_cfa_p): New array.
28402 (insns_have_identical_cfa_notes): New function.
28403 (old_insns_match_p): Don't cross-jump in between /f
28404 and non-/f instructions. If both i1 and i2 are frame related,
28405 verify all CFA notes, their order and content.
28407 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28410 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28411 HImode and SImode with zero extend to DImode to one insn.
28412 (bswap<mode>2_extenddi): Likewise.
28413 (bswapsi2_extenddi): Likewise.
28414 (bswaphi2_extendsi): Likewise.
28415 (bswaphi2): Combine bswap HImode and SImode into one insn.
28416 Separate memory insns from swapping register.
28417 (bswapsi2): Likewise.
28418 (bswap<mode>2): Likewise.
28419 (bswaphi2_internal): Delete, no longer used.
28420 (bswapsi2_internal): Likewise.
28421 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28422 store, and gpr<-gpr swap insns.
28423 (bswap<mode>2_store): Likewise.
28424 (bswaphi2_reg): Register only splitter, combine with the splitter.
28425 (bswaphi2 splitter): Likewise.
28426 (bswapsi2_reg): Likewise.
28427 (bswapsi2 splitter): Likewise.
28428 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28429 the insns into load, store, and register/register insns.
28430 (bswapdi2_ldbrx): Likewise.
28431 (bswapdi2_load): Likewise.
28432 (bswapdi2_store): Likewise.
28433 (bswapdi2_reg): Likewise.
28435 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28437 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28438 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28440 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28443 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28445 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28446 special handling for target option conflicts between dform
28447 options (-mpower9-dform, -mpower9-dform-vector,
28448 -mpower9-dform-scalar) and -mno-direct-move.
28450 2017-03-27 Richard Biener <rguenther@suse.de>
28452 PR tree-optimization/80181
28453 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28455 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28457 * config/arc/predicates.md (move_double_src_operand): Replace the
28458 call to move_double_src_operand with a call to address_operand.
28460 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28462 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28463 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28464 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28466 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28468 * config/arc/predicates.md (long_immediate_loadstore_operand):
28469 Consider scaled addresses cases.
28471 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28473 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28474 restored when in interrupt.
28475 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28476 doesn't have delay slot.
28478 2017-03-27 Richard Biener <rguenther@suse.de>
28481 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28482 inlined thunk clones.
28484 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28487 * asan.c (instrument_derefs): Copy over last operand from
28488 original COMPONENT_REF to the new COMPONENT_REF with
28489 DECL_BIT_FIELD_REPRESENTATIVE.
28490 * ubsan.c (instrument_object_size): Likewise.
28492 2017-03-27 Richard Biener <rguenther@suse.de>
28494 PR tree-optimization/80170
28495 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28496 sure DR/SCEV didnt fold in constants we do not see when looking
28497 at the reference base alignment.
28499 2017-03-27 Richard Biener <rguenther@suse.de>
28501 PR middle-end/80171
28502 * gimple-fold.c (fold_ctor_reference): Properly guard against
28503 NULL return value from canonicalize_constructor_val.
28505 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28508 * config/i386/i386.c (ix86_expand_builtin)
28509 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28510 flags reg setting and flags reg using instructions.
28511 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28512 clobbering instructions to zero extend op2.
28514 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28516 * doc/install.texi (Configuration) <--with-aix-soname>:
28517 Update link to AIX ld.
28519 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28521 PR rtl-optimization/80160
28522 PR rtl-optimization/80159
28523 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28524 reg_alternate_class into account.
28526 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28529 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28530 to consider in curr_insn_transform.
28532 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28534 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28535 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28536 and emit_mode_inner.
28538 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28540 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28541 argument to the overloaded builtin variants. Use the new flag to
28542 deprecate certain builtin variants.
28543 * config/s390/s390-builtin-types.def: Add new builtin types.
28544 * config/s390/s390-builtins.h: Support new flags field for
28545 overloaded builtins.
28546 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28547 (s390_macro_to_expand): Enable vector float data type.
28548 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28549 builtins by incrementing the __VEC__ version number.
28550 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28552 (s390_resolve_overloaded_builtin): Emit error messages depending
28553 on the builtin flags.
28554 * config/s390/s390.c (s390_expand_builtin): Support additional
28555 flags argument. Change error message to match the messages
28556 emitted in s390-c.c.
28557 * config/s390/s390.md: New UNSPEC_* constants.
28558 (op_type): Add new instruction types.
28559 * config/s390/vecintrin.h: Add new builtins and test data class
28561 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28562 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28563 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28564 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28565 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28566 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28568 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28569 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28570 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28571 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28573 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28574 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28575 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28576 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28578 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28579 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28580 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28581 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28582 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28583 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28584 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28586 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28587 ("vec_scatter_element<V_HW_4:mode>_DI")
28588 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28589 ("vec_fpint<mode>", "vflls")
28590 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28591 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28592 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28593 ("*vec_cmphe<mode>_cc"): ... these.
28595 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28596 mode constant instead of magic value.
28598 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28600 * config/s390/s390.c (s390_expand_vec_compare): Support other
28601 vector floating point modes than just V2DF.
28602 (s390_expand_vcond): Likewise.
28603 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28604 (s390_cannot_change_mode_class): Prevent mode changes between TF
28605 and V1TF in vector registers.
28606 * config/s390/s390.md (DF, SF): New mode attributes.
28607 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28608 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28609 SFmode support for VRs.
28610 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28612 (VFT, VF_HW): New mode iterators.
28613 (vw, sdx): New mode attributes.
28614 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28615 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28616 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28617 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28618 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28619 also the new vector floating point modes. Renaming to ...
28621 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28622 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28623 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28624 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28625 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28626 ("vec_unordered<mode>"): ... these.
28628 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28629 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28630 ("*vec_extendv2df"): New insn definitions.
28632 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28634 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28635 ("mulditi3_2", "*muldi3_sign"): New patterns.
28636 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28637 rename the pattern definition.
28639 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28641 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28643 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28645 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28647 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28648 instruction if possible.
28649 * config/s390/vector.md (vec_halfnumelts): New mode
28651 ("*vec_vllezlf<mode>"): New pattern.
28653 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28655 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28656 ("popcountv4si2", "popcountv2di2"): Rename to ...
28657 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28658 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28660 ("popcount<mode>2_vxe"): New pattern.
28662 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28664 * common/config/s390/s390-common.c (processor_flags_table): Add
28666 * config.gcc: Add arch12.
28667 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28668 Default to arch12 for unknown CPU model numbers.
28669 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28670 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28671 PROCESSOR_max sanity check.
28672 * config/s390/s390-opts.h (enum processor_type): Add
28674 * config/s390/s390.c (processor_table): Add arch12.
28675 (s390_expand_builtin): Add check for B_VXE flag.
28676 (s390_issue_rate): Add PROCESSOR_ARCH12.
28677 (s390_get_sched_attrmask): Likewise.
28678 (s390_get_unit_mask): Likewise.
28679 (s390_sched_score): Enable z13 scheduling for arch12.
28680 (s390_sched_reorder): Likewise.
28681 (s390_sched_variable_issue): Likewise.
28682 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28684 (s390_tune_attr): Use z13 scheduling also for arch12.
28685 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28686 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28687 (TARGET_VXE_P): New macros.
28688 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28689 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28690 * config/s390/s390.opt: Add arch12 as processor_type.
28692 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28694 * config/s390/s390.md
28695 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28696 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28697 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28699 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28700 Rename expanders to ...
28702 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28703 ("fixuns_truncdddi2_emu"): ... these.
28705 ("fixuns_trunc<mode>si2_emu"): New expander.
28707 ("*fixuns_truncdfdi2_z13"): Rename to ...
28708 ("*fixuns_truncdfdi2_vx"): ... this.
28710 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28712 * config/s390/2964.md: Remove the single element vector compare
28713 instructions which are no longer used.
28714 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28716 (s390_canonicalize_comparison): Remove handling of DFmode
28718 (s390_expand_vec_compare_scalar): Remove function.
28719 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28720 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28722 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28724 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28726 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28728 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28729 will anyway by matched by mov<mode>_64dfp.
28731 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28733 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28734 vlef/vstef. Add missing operand to vleif.
28736 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28738 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28739 pair for all vector types with 64 bit elements.
28740 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28741 * config/s390/vector.md (V_HW_64): ... here.
28742 (V_128_NOSINGLE): New mode iterator.
28743 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28744 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28745 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28746 ("*vec_load_pairv2di"): Change to ...
28747 ("*vec_load_pair<mode>"): ... this one.
28749 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28751 * config/s390/constraints.md: Add comments.
28752 (jKK): Reject element sizes > 8 bytes.
28753 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28755 * config/s390/s390.md: Add the s_operand checks formerly in
28756 s390_split_ok_p to various splitters where they are still
28758 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28759 for 128 bit vectors. Plus two splitters.
28761 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28763 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28766 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28769 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28770 error if the boundary argument is not constant.
28772 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28774 PR rtl-optimization/80112
28775 * loop-doloop.c (doloop_condition_get): Don't check condition
28776 if cmp isn't SET with IF_THEN_ELSE src.
28778 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28780 PR tree-optimization/80158
28781 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28782 replacing a candidate statement, also replace it for the
28783 candidate's alternate interpretation.
28784 (replace_rhs_if_not_dup): Likewise.
28785 (replace_one_candidate): Likewise.
28787 2017-03-24 Richard Biener <rguenther@suse.de>
28789 PR tree-optimization/80167
28790 * graphite-isl-ast-to-gimple.c
28791 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28793 (translate_isl_ast_to_gimple::get_rename): Likewise.
28795 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28797 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28798 handling of certain combinations of target options, including the
28799 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28800 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28802 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28805 * config/arm/arm.md (*load_multiple): Add reload_completed to
28806 matching condition.
28808 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28809 Richard Biener <rguenth@suse.de>
28811 PR tree-optimization/79908
28812 PR tree-optimization/80136
28813 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28814 been cast away, gimplify_and_add suffices.
28816 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28818 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28820 2017-03-23 Richard Biener <rguenther@suse.de>
28822 PR tree-optimization/80032
28823 * gimplify.c (gimple_push_cleanup): Forced unconditional
28824 cleanups still have to go to the conditional_cleanups
28827 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28829 PR tree-optimization/80072
28830 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28832 (next_operand_entry_id): Change type to unsigned int.
28833 (sort_by_operand_rank): Make sure to return the right return value
28834 even if unsigned fields are bigger than INT_MAX.
28835 (struct oecount): Change cnt and id type to unsigned int.
28836 (oecount_hasher::equal): Formatting fix.
28837 (oecount_cmp): Make sure to return the right return value
28838 even if unsigned fields are bigger than INT_MAX.
28839 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28842 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28843 TREE_READONLY on result if writing it more than once.
28846 * doc/invoke.texi (-fsanitize=thread): Document that with
28847 -fnon-call-exceptions atomics are not able to throw
28851 * tsan.c: Include tree-eh.h.
28852 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28853 maybe_clean_or_replace_eh_stmt where needed.
28854 (instrument_memory_accesses): Add cfg_changed argument.
28855 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28856 if it returned true.
28857 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28859 PR rtl-optimization/63191
28860 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28861 wrapper function, moved the whole old content into ...
28862 (ix86_delegitimize_address_1): ... this. New inline function.
28863 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28864 true as last argument instead of ix86_delegitimize_address.
28866 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28868 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28869 cortexa57_branch_cost.
28871 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28873 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28875 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28878 * doc/md.texi (Constraints): Document wA constraint.
28879 * config/rs6000/constraints.md (wA): New.
28880 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28881 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28882 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28883 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28885 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28888 * gimplify.c (is_oacc_declared): New function.
28889 (oacc_default_clause): Use it to set default flags for acc declared
28890 variables inside parallel regions.
28891 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28892 declared variables.
28893 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28894 declare attribute to any decl as necessary.
28896 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28899 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28900 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28901 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28902 (arm_arch_lpae): This.
28903 * config/arm/arm.c (arm_arch7ve): Rename into ...
28904 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28905 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28908 2017-03-22 Martin Liska <mliska@suse.cz>
28911 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28912 error message instead of an ICE.
28914 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28916 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28918 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28920 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28922 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28925 2017-03-21 Martin Sebor <msebor@redhat.com>
28927 * doc/extend.texi: Use "cannot" instead of "can't."
28928 * doc/hostconfig.texi: Same.
28929 * doc/install.texi: Same.
28930 * doc/invoke.texi: Same.
28931 * doc/loop.texi: Same.
28932 * doc/md.texi: Same.
28933 * doc/objc.texi: Same.
28934 * doc/rtl.texi: Same.
28935 * doc/tm.texi: Same.
28936 * doc/tm.texi.in: Same.
28937 * doc/trouble.texi: Same.
28939 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28942 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28943 (collect_checksum_attributes): Set it.
28944 (die_checksum_ordered): Use it.
28946 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28948 PR tree-optimization/79908
28949 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28950 change: For a VA_ARG whose LHS has been cast away, use
28951 force_gimple_operand to construct the side effects.
28953 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28955 PR translation/80001
28956 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28957 more amenable to translation.
28958 (oacc_loop_auto_partitions): Likewise.
28960 2017-03-21 Marek Polacek <polacek@redhat.com>
28961 Martin Sebor <msebor@redhat.com>
28963 PR tree-optimization/80109
28964 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28965 on INTEGRAL_TYPE_P.
28967 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28968 Segher Boessenkool <segher@kernel.crashing.org>
28971 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28972 check reg_used_between_p between insn and one of succ or succ2
28973 depending on if succ is artificial insn not inserted into insn
28976 2017-03-21 Martin Liska <mliska@suse.cz>
28978 PR gcov-profile/80081
28979 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28980 * doc/gcc.texi: Include gcov-dump stuff.
28981 * doc/gcov-dump.texi: New file.
28983 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28985 PR rtl-optimization/79150
28986 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28987 conditional jump, if the jump is the last insn of the loop.
28989 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28990 Richard Biener <rguenth@suse.de>
28992 PR tree-optimization/79908
28993 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28994 been cast away, use force_gimple_operand to construct the side
28997 2017-03-21 Martin Liska <mliska@suse.cz>
28999 PR libfortran/79956
29000 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29003 2017-03-21 Brad Spengler <spender@grsecurity.net>
29006 * plugin.c (htab_hash_plugin): New function.
29007 (add_new_plugin): Use it and adjust.
29008 (parse_plugin_arg_opt): Adjust.
29009 (init_one_plugin): Likewise.
29011 2017-03-21 Richard Biener <rguenther@suse.de>
29013 PR tree-optimization/80032
29014 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29015 if set force the cleanup to happen unconditionally.
29016 (gimplify_target_expr): Push inserted clobbers with force_uncond
29017 to avoid them being removed by control-dependent DCE.
29019 2017-03-21 Richard Biener <rguenther@suse.de>
29021 PR tree-optimization/80122
29022 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29023 va_arg_pack_len when the inlined call stmt requires pack
29025 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29027 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29030 * tsan.c (instrument_builtin_call): If the memory model argument
29031 is not a constant, assume it is valid.
29034 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29037 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29039 PR rtl-optimization/79910
29040 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29041 if its dest is used by an insn before I2 (other than the combined
29042 insns themselves, which are properly handled already).
29044 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29047 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29049 * combine.c (record_used_regs): New static function.
29050 (try_combine): Handle situations where there is an additional
29051 instruction between I2 and I3 which needs to have a LOG_LINK
29055 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29057 * combine.c (try_combine): Delete redundant i1 test. Call
29058 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29060 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29063 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29064 alternatives 13/14.
29066 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29068 PR tree-optimization/80054
29069 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29070 the optimization if a PHI or any of its arguments is not dominated
29071 by the candidate's basis. Use gphi* rather than gimple* as
29073 (replace_profitable_candidates): Clean up a gimple* variable that
29074 should be a gphi* variable.
29076 2017-03-20 Martin Sebor <msebor@redhat.com>
29079 * doc/extend.texi (attribute constructor): Document present limitation.
29081 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29084 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29085 __POWER9_VECTOR__ #ifdef control, change template definition to
29086 use Power9-specific built-in function.
29087 (vec_any_eq): Likewise.
29088 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29089 to control outcomes from this test.
29090 (vector_ae_<mode>p): For VEC_F modes, likewise.
29092 2017-03-20 Ian Lance Taylor <iant@google.com>
29094 * config/i386/i386.c (ix86_function_regparm): Save an extra
29095 register for -fsplit-stack with DECL_STATIC_CHAIN.
29097 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29100 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29101 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29103 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29105 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29107 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29110 2017-03-20 Marek Polacek <polacek@redhat.com>
29113 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29115 2017-03-20 Richard Biener <rguenther@suse.de>
29117 PR tree-optimization/80113
29118 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29119 allocate extra SSA name for PHI def.
29120 (add_close_phis_to_outer_loops): Likewise.
29121 (add_close_phis_to_merge_points): Likewise.
29122 (copy_loop_close_phi_args): Likewise.
29123 (copy_cond_phi_nodes): Likewise.
29125 2017-03-20 Martin Liska <mliska@suse.cz>
29127 PR middle-end/79753
29128 * tree-chkp.c (chkp_build_returned_bound): Do not build
29129 returned bounds for a LHS that's not a BOUNDED_P type.
29131 2017-03-20 Martin Liska <mliska@suse.cz>
29135 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29136 COMPLEX_CST and VECTOR_CST.
29138 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29141 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29142 target operand. A new splitter adds the clobber statement in case
29143 the target operand is dead anyway.
29145 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29147 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29148 to age-old versions of binutils and glibc.
29150 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29152 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29154 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29156 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29158 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29160 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29161 requirement for binutils 2.13.
29163 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29165 * combine.c (try_combine): Delete redundant i1 test. Call
29166 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29168 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29170 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29171 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29173 <riscv64-*-elf>: Re-arrange section
29174 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29175 <riscv32-*-linux>: Likewise.
29176 <riscv64-*-elf>: Likewise
29177 <riscv64-*-linux>: Likewise.
29179 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29182 * aarch64.opt(verbose-cost-dump): Fix typo.
29184 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29187 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29188 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29190 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29192 * reload.c (find_reloads): When reloading a nonoffsettable address,
29193 use RELOAD_OTHER for it and its address reloads.
29195 PR rtl-optimization/79910
29196 * combine.c (record_used_regs): New static function.
29197 (try_combine): Handle situations where there is an additional
29198 instruction between I2 and I3 which needs to have a LOG_LINK
29201 2017-03-17 Jeff Law <law@redhat.com>
29203 PR tree-optimization/71437
29204 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29205 conditional in the hash table first.
29206 (vrp_dom_walker::before_dom_children): Extract condition from
29207 ASSERT_EXPR. Record condition, its inverion and any implied
29208 conditions as well.
29210 2017-03-17 Marek Polacek <polacek@redhat.com>
29211 Markus Trippelsdorf <markus@trippelsdorf.de>
29213 PR tree-optimization/80079
29214 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29217 2017-03-17 Richard Biener <rguenther@suse.de>
29219 PR middle-end/80075
29220 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29221 Properly verify the LHS before the RHS possibly claims to be
29223 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29226 2017-03-17 Martin Jambor <mjambor@suse.cz>
29228 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29229 (List of -O2 options): Likewise.
29230 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29233 2017-03-17 Tom de Vries <tom@codesourcery.com>
29235 * gcov-dump.c (print_usage): Print bug_report_url.
29237 2017-03-17 Richard Biener <rguenther@suse.de>
29239 PR middle-end/80050
29240 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29241 (parser::peek): Likewise.
29243 2017-03-17 Richard Biener <rguenther@suse.de>
29245 PR tree-optimization/80048
29246 * sese.c (free_sese_info): Properly release rename_map and
29247 copied_bb_map elements.
29249 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29251 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29252 Add linked-list forward and backlinks. Insert on
29253 construction, remove on destruction.
29254 (class pass_store_merging): Add m_stores_head field.
29255 (pass_store_merging::terminate_and_process_all_chains):
29256 Iterate over m_stores_head list.
29257 (pass_store_merging::terminate_all_aliasing_chains):
29259 (pass_store_merging::execute): Check for debug stmts first.
29260 Push new chains onto the m_stores_head stack.
29262 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29265 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29266 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29267 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29269 2017-03-16 Jeff Law <law@redhat.com>
29271 PR tree-optimization/71437
29272 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29273 member function. Implementation moved into after_dom_children
29274 member function and into the threader's thread_outgoing_edges
29276 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29277 some code into new thread_outgoing_edges.
29278 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29279 definition. Simplify marker handling (do it here). Assume we always
29280 have the available expression and the const/copies tables.
29281 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29283 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29284 * tree-vrp.c (equiv_stack): No longer file scoped.
29285 (vrp_dom_walker): New class.
29286 (vrp_dom_walker::before_dom_children): New member function.
29287 (vrp_dom_walker::after_dom_children): Likewise.
29288 (identify_jump_threads): Setup domwalker. Use it rather than
29289 walking edges in a random order by hand. Simplify setup/finalization.
29290 (finalize_jump_threads): Remove.
29291 (vrp_finalize): Do not call identify_jump_threads here.
29292 (execute_vrp): Do it here instead and call thread_through_all_blocks
29295 PR tree-optimization/71437
29296 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29298 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29300 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29301 (dom_opt_dom_walker::thread_across_edge): Remove
29302 handle_dominating_asserts argument. All callers changed.
29303 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29304 changes. Remove calls to lhs_of_dominating_assert. Other
29305 uses of handle_dominating_asserts turn into unconditional code
29306 (simplify_control_stmt_condition_1): Likewise.
29307 (simplify_control_stmt_condition): Likewise.
29308 (thread_through_normal_block, thread_across_edge): Likewise.
29309 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29310 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29311 object if it is not an SSA_NAME.
29312 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29313 before calling into the VRP specific simplifiers.
29314 (identify_jump_threads): Remove handle_dominating_asserts
29317 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29320 * tree-diagnostic.c (default_tree_printer): No longer static.
29321 * tree-diagnostic.h (default_tree_printer): New prototype.
29323 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29325 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29326 Change ins into fmov.
29328 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29330 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29331 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29332 Use h_con constraint for operand 1.
29333 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29334 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29336 2017-03-15 Jeff Law <law@redhat.com>
29338 PR tree-optimization/71437
29339 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29340 (record_temporary_equivalences): Use it.
29342 PR tree-optimization/71437
29343 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29344 tree-ssa-scopedtables.
29345 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29346 (record_conditions, record_cond, vuse_eq): Likewise.
29347 (record_edge_info): Adjust to API tweak of record_conditions.
29348 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29349 (record_temporary_equivalences, optimize_stmt): Likewise.
29350 (eliminate_redundant_computations): Likewise.
29351 (record_equivalences_from_stmt): Likewise.
29352 * tree-ssa-scopedtables.c: Include options.h and params.h.
29353 (vuse_eq): New function, moved from tree-ssa-dom.c
29354 (build_and_record_new_cond): Likewise.
29355 (record_conditions): Likewise. Accept vector of conditions rather
29356 than edge_equivalence structure for first argument.
29357 for the first argument.
29358 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29359 from tree-ssa-dom.c.
29360 (avail_exprs_stack::record_cond): Likewise.
29361 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29362 from tree-ssa-dom.c.
29363 (avail_exprs_stack): Add new member functions lookup_avail_expr
29365 (record_conditions): Declare.
29367 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29370 * lra-constraints.c (process_alt_operands): Increase reject for
29371 reloading an input/output operand.
29373 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29376 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29377 insns to convert from signed/unsigned char/short to IEEE 128-bit
29379 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29381 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29384 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29385 subreg of inner mode for values already in registers.
29387 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29389 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29390 iteration reg is used after the loop.
29392 2017-03-14 Martin Sebor <msebor@redhat.com>
29394 PR tree-optimization/79800
29395 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29396 precision in negative-positive range.
29397 (format_floating): Call non-const overload with adjusted precision.
29399 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29402 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29405 2017-03-14 Martin Sebor <msebor@redhat.com>
29407 PR middle-end/80020
29408 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29409 * builtins.def (aligned_alloc): Use it.
29412 * Makefile.in (GTFILES): Add calls.c.
29413 * calls.c: Include "gt-calls.h".
29415 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29417 PR rtl-optimization/79728
29418 * regs.h (struct target_regs): New field
29419 x_contains_allocatable_regs_of_mode.
29420 (contains_allocatable_regs_of_mode): New macro.
29421 * reginfo.c (init_reg_sets_1): Initialize it, and change
29422 contains_reg_of_mode so it includes global regs as well.
29423 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29424 rather than contains_regs_of_mode.
29426 2017-03-14 Martin Liska <mliska@suse.cz>
29428 * doc/invoke.texi: Document options that can't be combined with
29429 -fcheck-pointer-bounds.
29431 2017-03-14 Martin Liska <mliska@suse.cz>
29433 PR middle-end/79831
29434 * doc/invoke.texi (-Wchkp): Document the option.
29436 2017-03-14 Martin Liska <mliska@suse.cz>
29438 * Makefile.in: Install gcov-dump.
29440 2017-03-14 Martin Liska <mliska@suse.cz>
29442 * multiple_target.c (expand_target_clones): Bail out for
29443 an invalid attribute.
29445 2017-03-14 Richard Biener <rguenther@suse.de>
29447 * alias.c (struct alias_set_entry): Pack properly.
29448 * cfgloop.h (struct loop): Likewise.
29449 * cse.c (struct set): Likewise.
29450 * ipa-utils.c (struct searchc_env): Likewise.
29451 * loop-invariant.c (struct invariant): Likewise.
29452 * lra-remat.c (struct cand): Likewise.
29453 * recog.c (struct change_t): Likewise.
29454 * rtl.h (struct address_info): Likewise.
29455 * symbol-summary.h (function_summary): Likewise.
29456 * tree-loop-distribution.c (struct partition): Likewise.
29457 * tree-object-size.c (struct object_size_info): Likewise.
29458 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29459 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29460 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29461 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29462 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29463 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29464 (struct _stmt_vec_info): Likewise.
29466 2017-03-14 Martin Liska <mliska@suse.cz>
29469 * multiple_target.c (create_dispatcher_calls): Check that
29470 a target can create a function dispatcher.
29472 2017-03-14 Martin Liska <mliska@suse.cz>
29475 * multiple_target.c (expand_target_clones): Drop local.local
29476 flag for default implementation.
29478 2017-03-14 Richard Biener <rguenther@suse.de>
29480 PR tree-optimization/80030
29481 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29483 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29485 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29486 gcc_fallthrough() instead of __attribute__((fallthrough));
29488 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29490 * doc/gcc.texi: Remove "up" link to (DIR).
29491 * doc/gccint.texi: Ditto.
29493 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29495 * doc/install.texi (Specific) <avr>: Remove reference to
29498 2017-03-13 Jeff Law <law@redhat.com>
29500 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29501 attribute rather than comments.
29503 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29504 match_scratch operand is highest.
29506 2017-03-13 Martin Liska <mliska@suse.cz>
29508 PR middle-end/78339
29509 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29510 is a CHKP clone, use original declaration.
29512 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29514 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29515 (arc_conditional_register_usage): Use a different allocation order
29516 when optimizing for size.
29517 * common/config/arc/arc-common.c (arc_option_optimization_table):
29518 Section anchors default on when optimizing for size.
29520 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29522 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29524 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29526 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29527 * config/arc/arc.md (cpu_facility): Add cd variant.
29528 (*movqi_insn): Add code density variant.
29529 (*movhi_insn): Likewise.
29530 (*movqi_insn): Likewise.
29531 (*addsi3_mixed): Likewise.
29532 (subsi3_insn): Likewise.
29534 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29536 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29538 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29540 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29541 expressions with MINUS and UNARY ops.
29543 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29546 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29548 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29549 between vec_select and vector argument.
29550 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29551 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29552 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29553 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29554 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29555 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29557 2017-03-13 Richard Biener <rguenther@suse.de>
29560 * params.def (vect-max-peeling-for-alignment): Fix typo.
29562 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29564 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29565 issue that only occurred with binutils below 2.18.
29567 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29569 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29570 refer to binutils 2.11/2.12 minimum.
29572 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29574 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29575 ftp.kernel.org and simplify binutils requirement.
29577 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29579 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29581 (Optimize Options): Ditto. Also remove redundancy.
29583 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29585 PR translation/79848
29586 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29588 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29589 to G_ to avoid double translation.
29591 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29593 PR translation/79923
29594 * auto-profile.c (get_combined_location): Convert leading
29595 character of diagnostics to lower case and remove trailing period.
29596 (read_profile): Likewise for various diagnostics.
29597 * config/arm/arm.c (arm_option_override): Remove trailing period
29598 from various diagnostics.
29599 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29600 (msp430_expand_delay_cycles): Likewise.
29602 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29605 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29606 full command-line argument, rather than just "str".
29607 (aarch64_validate_march): Likewise.
29608 (aarch64_validate_mtune): Likewise.
29610 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29612 PR rtl-optimization/78911
29613 * lra-assigns.c (must_not_spill_p): New function.
29614 (spill_for): Use it.
29616 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29618 PR tree-optimization/79981
29619 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29620 ATOMIC_COMPARE_EXCHANGE ifn result.
29621 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29622 IFN_ATOMIC_COMPARE_EXCHANGE.
29624 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29627 * opts.c (parse_sanitizer_options): Add missing question mark to
29628 "did you mean" message.
29630 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29632 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29634 (VMULEUH_UNS): Likewise.
29635 (VMULOUB_UNS): Likewise.
29636 (VMULOUH_UNS): Likewise.
29637 * config/rs6000/rs6000.c (builtin_function_type): Remove
29638 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29640 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29643 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29644 x with result of extra_parsing_for_operand_code_0.
29645 (function_reader::extra_parsing_for_operand_code_0): Convert
29646 return type from void to rtx, returning x. When reading
29647 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29648 larger size containing struct block_symbol.
29650 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29652 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29653 -mfloat128-hardware without -m64.
29655 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29658 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29659 entries to the case statement that marks unsigned arguments to
29660 overloaded functions.
29662 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29664 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29665 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29667 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29670 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29671 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29673 2017-03-10 Martin Liska <mliska@suse.cz>
29677 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29678 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29681 2017-03-10 Olivier Hainque <hainque@adacore.com>
29683 * tree-switch-conversion (array_value_type): Start by resetting
29684 candidate type to it's main variant.
29686 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29688 PR rtl-optimization/79909
29689 * combine.c (try_combine): Use simplify_replace_rtx on individual
29690 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29691 of the whole CALL_INSN_FUNCTION_USAGE.
29693 PR tree-optimization/79972
29694 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29695 get_range_info on SSA_NAMEs. Formatting fixes.
29697 2017-03-10 Richard Biener <rguenther@suse.de>
29698 Jakub Jelinek <jakub@redhat.com>
29700 PR tree-optimization/77975
29701 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29702 edge to be constant.
29703 (get_val_for): For constant x return it. Formatting fix.
29704 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29705 would use the same bases as the current one.
29707 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29709 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29710 instead of vec_select for V1TImode.
29711 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29713 (VSX_LE_128): Add V1TI to this mode iterator.
29714 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29715 (*vsx_le_perm_store_<mode>): Likewise.
29716 (pre-reload splitter for VSX stores): Likewise.
29717 (post-reload splitter for VSX stores): Likewise.
29718 (*vsx_xxpermdi2_le_<mode>): Likewise.
29719 (*vsx_lxvd2x2_le_<mode>): Likewise.
29720 (*vsx_stxvd2x2_le_<mode>): Likewise.
29722 2017-03-09 Michael Eager <eager@eagercon.com>
29724 Correct failures with --enable-checking=yes,rtl.
29726 * config/microblaze/microblaze.c (microblaze_expand_shift):
29727 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29728 test for const0_rtx.
29729 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29730 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29732 2017-03-09 Richard Biener <rguenther@suse.de>
29734 PR tree-optimization/79977
29735 * graphite-scop-detection.c (scop_detection::merge_sese):
29736 Handle the case of extra exits to blocks dominating the entry.
29738 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29740 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29743 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29745 PR rtl-optimization/79949
29746 * lra-constraints.c (process_alt_operands): Check memory when
29747 trying to predict a cycle. Print about the overall increase.
29749 2017-03-09 Richard Biener <rguenther@suse.de>
29751 PR middle-end/79971
29752 * gimple-expr.c (useless_type_conversion_p): Preserve
29753 TYPE_SATURATING for fixed-point types.
29755 2017-03-09 Richard Biener <rguenther@suse.de>
29758 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29759 alignment of BLKmode params.
29761 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29764 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29765 (VALL_NO_V2Q): Likewise.
29767 * config/aarch64/aarch64-simd.md
29768 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29770 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29771 VALL_NO_V2Q mode iterator.
29772 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29774 2017-03-09 Martin Liska <mliska@suse.cz>
29776 PR tree-optimization/79631
29777 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29778 tree_int_cst_sign_bit just for INTEGER constants.
29780 2017-03-09 Martin Liska <mliska@suse.cz>
29784 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29787 2017-03-09 Marek Polacek <polacek@redhat.com>
29790 * tree.c (inchash::add_expr): Handle TREE_VEC.
29792 2017-03-09 Martin Liska <mliska@suse.cz>
29795 (chkp_narrow_size_and_offset): New function.
29796 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29797 (void chkp_parse_bit_field_ref): New function.
29798 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29799 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29801 2017-03-09 Martin Liska <mliska@suse.cz>
29804 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29805 (chkp_find_bounds_1): Remove gcc_unreachable.
29807 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29810 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29811 BUILT_IN_SYNC*, determine the access type from the size suffix and
29812 always build a MEM_REF with that type. Handle forgotten
29813 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29816 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29817 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29818 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29819 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29820 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29821 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29822 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29823 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29824 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29825 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29826 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29827 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29828 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29829 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29830 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29831 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29832 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29833 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29834 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29835 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29836 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29837 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29838 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29839 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29840 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29841 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29842 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29843 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29844 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29845 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29846 definitions outside of __OPTIMIZE__ guarded section.
29849 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29850 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29851 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29852 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29855 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29857 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29858 ("vfenez<mode>"): Add missing constraints.
29860 2017-03-08 Martin Sebor <msebor@redhat.com>
29863 * config/nds32/nds32.c (nds32_option_override):
29864 Fix misspelled diagnostic.
29866 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29869 * gimplify.c (gimplify_omp_for): Replace index var in outer
29870 taskloop statement with an artificial variable and add
29871 OMP_CLAUSE_PRIVATE clause for it.
29873 2017-03-08 Richard Biener <rguenther@suse.de>
29875 PR tree-optimization/79955
29876 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29877 for accesses that are completely outside of the variable.
29879 2017-03-08 Andrew Haley <aph@redhat.com>
29881 PR tree-optimization/79943
29882 * tree-ssa-loop-split.c (compute_new_first_bound): When
29883 calculating the new upper bound, (END-BEG) should be added, not
29886 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29888 * config/avr/avr.md (setmemhi): Make sure match_dup
29889 operand number comes before match_scratch.
29891 2017-03-08 Richard Biener <rguenther@suse.de>
29893 PR tree-optimization/79920
29894 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29895 with ncopies == 1 to ...
29896 (vect_transform_slp_perm_load): ... here. Properly compute
29897 all element loads by iterating VF times over the group. Do
29898 not handle ncopies (computed in a broken way) in
29899 vect_create_mask_and_perm.
29901 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29904 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29905 is a uniform vector, use uniform_vector_p return value instead of
29906 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29908 2017-03-07 Marek Polacek <polacek@redhat.com>
29910 PR middle-end/79809
29911 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29912 (alloca_call_type): Likewise.
29914 2017-03-07 Martin Liska <mliska@suse.cz>
29916 * gcov.c (process_args): Put comment to correct location.
29918 2017-03-07 Martin Liska <mliska@suse.cz>
29920 PR middle-end/68270
29921 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29922 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29923 (chkp_narrow_bounds_for_field): Likewise.
29924 (chkp_parse_array_and_component_ref): Pass one more argument to
29927 2017-03-07 Richard Biener <rguenther@suse.de>
29929 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29932 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29934 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29935 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29937 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29940 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29941 to end of description.
29942 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29944 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29946 PR rtl-optimization/79901
29947 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29949 (*avx512f_<code><mode>3<mask_name>): ... this.
29950 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29951 iterator instead of VI8_AVX2_AVX512BW.
29953 PR rtl-optimization/79901
29954 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29955 min/max expander, expand it using expand_vec_cond_expr.
29958 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29961 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29964 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29965 to void * for PCH reasons.
29966 * dwarf2out.c (output_loc_operands, output_die): Cast
29967 v.val_vec.array to unsigned char *.
29969 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29972 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29975 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29977 PR rtl-optimization/79571
29978 * lra-constraints.c (process_alt_operands): Calculate static
29979 reject and subtract it from overall when only addresses will be
29982 2017-03-06 Julia Koval <julia.koval@intel.com>
29985 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29986 incoming stack boundary to 128 for 64-bit targets.
29988 2017-03-06 Richard Biener <rguenther@suse.de>
29990 PR tree-optimization/79894
29991 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29992 to NULL after folding it.
29994 2017-03-06 Richard Biener <rguenther@suse.de>
29996 PR tree-optimization/79824
29997 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29998 check disabling peeling for gaps.
30000 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30002 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30003 attributes): Document gettimeofday.
30005 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30007 * config/s390/s390.c (s390_option_override_internal): Set
30008 PARAM_MIN_VECT_LOOP_BOUND
30010 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30012 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30013 * config/s390/s390.md: Likewise.
30015 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30018 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30019 (<avx2_avx512>_perm<mode>): Rename to ...
30020 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30022 (<avx512>_perm<mode>_mask): Rename to ...
30023 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30025 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30026 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30027 instead of VI8F_256_512.
30028 (avx512f_perm<mode>): New define_expand.
30029 (avx512f_perm<mode>_mask): Likewise.
30030 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30031 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30033 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30035 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30036 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30038 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30040 2017-03-06 Martin Liska <mliska@suse.cz>
30043 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30044 when having a SSA NAME w/o VAR_DECL assigned to it.
30046 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30048 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30049 msa_dpsub_<su>_d): Fix MODE for vec_select.
30051 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30053 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30055 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30057 2017-03-06 Richard Biener <rguenther@suse.de>
30059 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30060 * plugin.c (register_plugin_info): Likewise.
30061 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30063 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30065 * config/i386/sse.md (sse_storehps, sse_storelps,
30066 avx_<castmode><avxsizesuffix>_<castmode>,
30067 avx512f_<castmode><avxsizesuffix>_<castmode>,
30068 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30069 in condition that at least one operand is not a MEM.
30071 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30073 PR middle-end/79805
30074 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30075 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30077 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30078 gimple_call_nothrow_p flag based on whether original builtin can throw.
30079 If it can, emit following stmts on the fallthrough edge.
30080 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30081 don't create new bb if inserting just debug stmts on the edge, try to
30082 insert them on the fallthru bb or just reset debug stmts.
30084 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30087 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30088 restore recog_data (including the operand rtxes inside it) around
30089 the call to get_insn_template.
30091 2017-03-03 Martin Sebor <msebor@redhat.com>
30093 PR tree-optimization/79699
30094 * context.c (context::~context): Free MPFR caches to avoid
30095 a memory leak on program exit.
30097 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30099 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30100 Use wide_int::ulow () instead of .elt (0).
30102 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30104 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30105 (*pushxf): Limit oF constraint to 32bit targets and add oC
30106 constraint for 64bit targets.
30107 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30108 (*pushdf): Change rmF constraint to rmC.
30110 2017-03-03 Martin Liska <mliska@suse.cz>
30112 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30113 Remove unused variable.
30115 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30118 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30119 is a memory operand, increase num_memory.
30120 (ix86_expand_args_builtin): Likewise.
30122 2017-03-03 Jan Hubicka <jh@suse.cz>
30125 * ipa-devirt.c (maybe_record_node): Properly handle
30126 __cxa_pure_virtual visibility.
30128 2017-03-03 Martin Liska <mliska@suse.cz>
30130 PR tree-optimization/79803
30131 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30133 (pass_loop_prefetch::execute): Disabled optimization if an
30134 assumption about L1 cache size is not met.
30136 2017-03-03 Martin Liska <mliska@suse.cz>
30138 PR rtl-optimization/79574
30139 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30140 (hash_scan_set): Likewise.
30141 (dump_hash_table): Likewise.
30142 (hoist_code): Likewise.
30144 2017-03-03 Richard Biener <rguenther@suse.de>
30146 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30147 in place of uhigh ().
30148 (fixed_convert_from_real): Likewise.
30150 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30153 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30155 2017-03-03 Richard Biener <rguenther@suse.de>
30157 PR middle-end/79818
30158 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30159 TYPE_OVERFLOW_UNDEFINED check.
30161 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30163 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30165 (vector_ae_<mode>_p): Likewise.
30166 (vector_nez_<mode>_p): Likewise.
30167 (vector_ne_v2di_p): Likewise.
30168 (vector_ae_v2di_p): Likewise.
30169 (vector_ne_<mode>_p): Likewise.
30170 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30172 (vsx_tsqrt<mode>2_fe): Likewise.
30174 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30177 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30179 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30181 PR rtl-optimization/79780
30182 * cprop.c (one_cprop_pass): When second and further conditional trap
30183 in a single basic block is turned into an unconditional trap, turn it
30184 into a deleted note to avoid RTL verification failures.
30186 2017-03-02 Richard Biener <rguenther@suse.de>
30188 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30190 2017-03-02 Richard Biener <rguenther@suse.de>
30192 PR tree-optimization/79345
30194 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30195 param and abort the walk, returning -1 if it is hit.
30196 (walk_aliased_vdefs): Take a limit param and pass it on.
30197 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30198 defaulting to 0 and return a signed int.
30199 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30200 (check_defs): New helper.
30201 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30202 about uninitialized memory.
30203 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30204 bogus uninitialized warning.
30205 (fixed_convert_from_real): Likewise.
30207 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30209 PR tree-optimization/66768
30210 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30211 iv_use if base object can't be determined.
30213 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30215 PR tree-optimization/79345
30216 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30217 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30218 (get_pattern_stats): Initialize it.
30219 * genemit.c (gen_expand): Verify match_scratch numbers come after
30220 match_operand/match_dup numbers.
30221 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30222 match_scratch numbers.
30223 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30225 * config/s390/s390.md (trunctdsd2): Likewise.
30227 2017-03-02 Richard Biener <rguenther@suse.de>
30229 * wide-int.h (wide_int_storage::operator=): Implement in terms
30232 2017-03-02 Richard Biener <rguenther@suse.de>
30234 PR tree-optimization/79777
30235 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30236 the to insert expression to sth existing.
30238 2017-03-01 Martin Sebor <msebor@redhat.com>
30240 PR middle-end/79692
30241 * gimple-ssa-sprintf.c
30242 (directive::known_width_and_precision): New function.
30243 (format_integer): Use it.
30244 (get_mpfr_format_length): Consider the full range of precision
30245 when computing %g output with the # flag. Set the likely byte
30246 count to 3 rather than 1 when precision is indeterminate.
30247 (format_floating): Correct the lower bound of precision.
30249 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30251 * doc/invoke.texi: Document default code model for 64-bit Linux.
30253 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30256 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30257 udiv rather than div since input pattern is unsigned.
30259 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30261 * config/i386/i386.c (print_reg): Warn for values of
30262 unsupported size in integer register.
30264 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30267 * config/rs6000/predicates.md (current_file_function_operand): Do
30268 not allow self calls to be local if the function is replaceable.
30270 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30273 * config/rs6000/altivec.h (vec_ctz and others): Change the
30274 preprocessor macro that controls conditional compilation from
30275 _ARCH_PWR9 to __POWER9_VECTOR__.
30276 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30277 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30278 control (instead of _ARCH_PWR9 control) so that template
30279 definition uses power9-specific function.
30280 (vec_any_eq): Likewise.
30281 (vec_all_ne): Change macro definition to use a power9-specific
30282 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30283 _ARCH_PWR9 control).
30284 (vec_any_eq) Likewise.
30285 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30286 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30287 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30288 support for xvcmpnedp instruction.
30289 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30290 macro expansion so that Power9 implementation of vec_all_ne does
30291 not use the AltiVec predicate framework.
30292 (VCMPNEH_P): Likewise.
30293 (VCMPNEW_P): Likewise.
30294 (VCMPNED_P): Likewise.
30295 (VCMPNEFP_P): Likewise.
30296 (VCMPNEDP_P): Likewise.
30297 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30298 implementation of vec_any_eq to not use AltiVec predicate
30300 (VCMPAEH_P): Likewise.
30301 (VCMPAEW_P): Likewise.
30302 (VCMPAED_P): Likewise.
30303 (VCMPAEFP_P): Likewise.
30304 (VCMPAEDP_P): Likewise.
30305 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30306 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30307 not use the AltiVec predicate framework.
30308 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30309 of vec_any_eq to not use AltiVec predicate framework.
30310 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30311 support for predefined __POWER9_VECTOR__ macro to indicate that
30312 Power9 instruction selection is enabled.
30313 (altivec_overloaded_builtins): Remove extraneous
30314 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30315 function argument types RS6000_BTI_bool_V16QI and
30316 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30317 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30318 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30319 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30320 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30321 Power9 for implementations of vec_cmpne. Change the signature for
30322 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30323 (representing vec_all_ne) to remove the previously described first
30324 argument of type RS6000_BTI_INTSI, as this was an artifact of
30325 reliance on the AltiVec predicate framework, which is no longer
30326 used in the implementation of these functions. Add
30327 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30328 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30329 since, unlike the AltiVec predicate framework implementation, we
30330 do not share function descriptors between vec_alle and vec_anyeq.
30331 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30332 set of modes that receive special treatment even when
30333 TARGET_P9_VECTOR is true. The special treatment emits code that
30334 does not depend on Power9 instructions.
30335 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30336 define_expand to not rely on AltiVec predicate framework.
30337 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30339 (vector_ne_v2di_p): Change this define_expand to not rely on
30340 AltiVec predicate framework.
30341 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30343 (vector_ne_<mode>_p): Change this define_expand to not rely on
30344 AltiVec predicate framework.
30345 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30347 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30348 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30349 define_insn pattern.
30350 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30351 define_insn pattern because the xvcmpne<VSs>. instruction is not
30353 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30354 instruction is not supported.
30356 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30358 * config/nvptx/nvptx.c: Include intl.h.
30360 2017-03-01 Martin Jambor <mjambor@suse.cz>
30363 * ipa-prop.h (ipa_bits): Removed field known.
30364 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30365 to pointers. Adjusted their comments to warn about their sharing.
30366 (ipcp_transformation_summary): Change bits to a vector of pointers.
30367 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30368 (ipa_get_ipa_bits_for_value): Declare.
30369 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30370 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30371 (ipa_bits_hash_table): Likewise.
30372 (ipa_vr_ggc_hash_traits): Likewise.
30373 (ipa_vr_hash_table): Likewise.
30374 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30375 being pointers and vr_known being removed.
30376 (ipa_set_jf_unknown): Likewise.
30377 (ipa_get_ipa_bits_for_value): New function.
30378 (ipa_set_jfunc_bits): Likewise.
30379 (ipa_get_value_range): New overloaded functions.
30380 (ipa_set_jfunc_vr): Likewise.
30381 (ipa_compute_jump_functions_for_edge): Use the above functions to
30382 construct bits and vr parts of jump functions.
30383 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30384 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30386 (ipcp_grow_transformations_if_necessary): Also allocate
30387 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30389 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30390 them. Fix too long lines.
30391 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30392 vr_known being removed.
30393 (ipa_read_jump_function): Use new setter functions to construct bits
30394 and vr parts of jump functions or set them to NULL.
30395 (write_ipcp_transformation_info): Adjust for bits being pointers.
30396 (read_ipcp_transformation_info): Likewise.
30397 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30399 Include gt-ipa-prop.h.
30400 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30402 (ipcp_store_bits_results): Likewise.
30403 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30404 Do not write to existing jump functions but use a temporary instead.
30406 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30409 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30410 attempt to use its first operand as BIT_FIELD_REF base.
30412 2017-03-01 Richard Biener <rguenther@suse.de>
30414 PR middle-end/79721
30415 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30416 interpolating formula in wrapping arithmetic.
30417 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30419 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30421 PR tree-optimization/79734
30422 * tree-vect-generic.c (expand_vector_condition): Optimize
30423 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30424 Handle VEC_COND_EXPR where comparison has different inner width from
30425 type's inner width.
30427 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30429 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30430 markup, and similar issues. Remove @opindex entries for things
30431 that aren't options. Add missing -mmpy-option entries.
30433 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30435 PR tree-optimization/79737
30436 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30437 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30438 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30439 instead of byte_size. Formatting fix.
30440 (shift_bytes_in_array_right): Formatting fix.
30442 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30445 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30446 condition on optimize for the leaf function test.
30448 2017-02-28 Martin Liska <mliska@suse.cz>
30451 * read-rtl-function.c (function_reader::handle_unknown_directive):
30452 Bail out when one uses -flto.
30454 2017-02-28 Martin Liska <mliska@suse.cz>
30456 * common.opt: Replace space with tabular for options of <number>
30458 * config/i386/i386.opt: Show <number> value for
30459 -mlarge-data-threshold.
30460 * opts.c (print_filtered_help): Do not display number in hexadecimal
30463 2017-02-28 Martin Liska <mliska@suse.cz>
30465 * common.opt: Fix --help=option -Q for options which are of
30468 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30470 * config/i386/i386.c (print_reg): Error out for values
30471 of 8-bit size in invalid integer register.
30473 2017-02-28 Martin Sebor <msebor@redhat.com>
30475 PR tree-optimization/79691
30476 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30478 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30481 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30482 gcc_unreachable with output_operand_lossage.
30484 2017-02-28 Richard Biener <rguenther@suse.de>
30486 PR tree-optimization/79740
30487 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30489 (visit_nary_op): Insert the nary into the hashtable if we
30490 pattern-matched sth.
30491 * tree-ssa-pre.c (eliminate_insert): Robustify.
30493 2017-02-28 Richard Biener <rguenther@suse.de>
30495 PR middle-end/79731
30496 * fold-const.c (decode_field_reference): Reject out-of-bound
30499 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30501 * config/i386/i386.c: Include intl.h.
30502 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30503 instead of just cond ? "..." : "...".
30504 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30505 * coverage.c (read_counts_file): Likewise.
30506 * omp-offload.c: Include intl.h.
30507 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30508 of just cond ? "..." : "...".
30509 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30510 of just cond ? "..." : "...".
30512 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30515 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30517 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30518 'tune for' CPU name.
30519 * config/arm/arm-cpu-data.h: Regenerated.
30521 2017-02-28 Richard Biener <rguenther@suse.de>
30523 PR tree-optimization/79732
30524 * tree-inline.c (expand_call_inline): Do not shadow var.
30526 2017-02-28 Richard Biener <rguenther@suse.de>
30528 PR tree-optimization/79723
30529 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30530 address-space properly.
30532 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30534 * doc/optinfo.texi (Optimization groups): Fix option used for
30536 * doc/invoke.texi (-fopt-info): Document "omp".
30537 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30538 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30539 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30540 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30541 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30543 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30545 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30546 OPTGROUP_OMP with "omp".
30548 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30551 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30552 for arithmetic shift of unsigned V2DI.
30554 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30556 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30557 arc/linux.h headers.
30558 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30559 (LINK_SPEC): Likewise.
30560 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30561 (EXTRA_SPECS): Likewise.
30562 (STARTFILE_SPEC): Likewise.
30563 (ENDFILE_SPEC): Likewise.
30564 (LIB_SPEC): Likewise.
30565 (TARGET_SDATA_DEFAULT): Likewise.
30566 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30567 (MULTILIB_DEFAULTS): Likewise.
30568 (DWARF2_UNWIND_INFO): Likewise.
30569 * config/arc/big.h: New file.
30570 * config/arc/elf.h: Likewise.
30571 * config/arc/linux.h: Likewise.
30572 * config/arc/t-uClibc: Remove.
30574 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30576 PR tree-optimization/77536
30577 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30578 (tree_transform_and_unroll_loop): Use above function to compute the
30579 estimated niter of unrolled loop and use it when scaling profile.
30580 Also use count info rather than frequency if it's non-zero.
30581 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30582 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30583 (vect_transform_loop): Call above function.
30585 2017-02-27 Richard Biener <rguenther@suse.de>
30587 PR tree-optimization/45397
30588 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30589 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30590 (visit_nary_op): Add pattern matching for CSEing sign-changed
30591 or truncated operations with wider ones.
30593 2017-02-27 Richard Biener <rguenther@suse.de>
30595 PR tree-optimization/79690
30596 * tree-vect-stmts.c (vectorizable_store): Use vector type
30597 built from the DR with address-space.
30599 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30601 * doc/invoke.texi (Optimize Options): Refine the description
30602 of asan-use-after-return.
30604 2017-02-25 Alan Modra <amodra@gmail.com>
30606 PR rtl-optimization/79584
30607 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30608 base, not ad->base_term, the reg within base. Remove assertion
30609 that ad->base == ad->base_term. Replace gen_int_mode using
30610 bogus mode with const0_rtx.
30612 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30614 PR middle-end/79396
30615 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30616 FMA_EXPR like tcc_binary or tcc_unary.
30618 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30621 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30623 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30624 (output_loc_operands): Handle DW_OP_call_ref and
30625 DW_OP_GNU_variable_value.
30626 (struct variable_value_struct): New type.
30627 (struct variable_value_hasher): Likewise.
30628 (variable_value_hash): New variable.
30629 (string_types): Remove.
30630 (copy_loc_descr): New function.
30631 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30632 (prepend_loc_descr_to_each): New function.
30633 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30634 instead of add_loc_descr_to_each if the first argument is single
30635 location list and the second has multiple.
30636 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30637 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30638 when looking for variable value which doesn't have other location info.
30639 (loc_list_from_tree): Formatting fix.
30640 (gen_array_type_die): Simplify DW_AT_string_length handling.
30641 (adjust_string_types): Remove.
30642 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30643 string_types. Call resolve_variable_values.
30644 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30645 (resolve_addr_in_expr): Likewise. Add A argument.
30646 (copy_deref_exprloc): Remove deref argument. Adjust for the
30647 original expression being DW_OP_GNU_variable_value with optionally
30648 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30649 optionally after it.
30650 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30651 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30652 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30653 (variable_value_hasher::hash, variable_value_hasher::equal): New
30655 (resolve_variable_value_in_expr, resolve_variable_value,
30656 resolve_variable_values, note_variable_value_in_expr,
30657 note_variable_value): New functions.
30658 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30661 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30664 * opts.h (handle_generated_option): Add GENERATED_P argument.
30665 * opts-common.c (handle_option): Adjust function comment.
30666 (handle_generated_option): Add GENERATED_P argument, pass it to
30668 (control_warning_option): Pass false to handle_generated_option
30670 * opts.c (maybe_default_option): Pass true to handle_generated_option
30672 * optc-gen.awk: Likewise.
30674 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30676 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30677 a REG, look at the REG it is a SUBREG of.
30678 (splitter for cmpeqsi_t): Ditto.
30680 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30682 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30683 the special USEs with the pattern of the insn, not the insn itself.
30685 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30688 * doc/invoke.texi: Document -mload-store-pairs.
30690 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30691 Sandra Loosemore <sandra@codesourcery.com>
30693 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30694 argument isn't a CONST_INT.
30695 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30696 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30697 (nios2_validate_compare): Bypass alternate compare logic if *op2
30698 is not a CONST_INT.
30699 (ldstwm_operation_p): Return false if first_base is not a REG or
30700 if first_offset is not a CONST_INT.
30702 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30704 * config/cris/cris.md: Use correct operand in a define_peephole2.
30706 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30708 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30710 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30712 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30713 this_insn if it is an INSN or JUMP_INSN.
30714 (force_offsettable): Look at base, not at addr.
30715 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30716 on things that aren't necessarily CONST_INTs.
30718 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30720 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30721 -mfpmath=sse is the default also for x86-32 targets with SSE2
30722 instruction set when @option{-ffast-math} is enabled
30724 2017-02-24 Jeff Law <law@redhat.com>
30726 PR rtl-optimizatoin/79286
30727 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30730 2017-02-24 Richard Biener <rguenther@suse.de>
30732 PR tree-optimization/79389
30733 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30736 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30738 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30739 function comment to reflect reality.
30740 (loop_exits_before_overflow): Fix typo in function description.
30742 2017-02-24 Richard Biener <rguenther@suse.de>
30744 PR tree-optimization/79389
30745 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30746 properly that a threading opportunity exists. Detect conditional
30747 copy/constant propagation opportunities.
30749 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30751 * config/visium/visium.md (type): Add trap.
30752 (b): New mode attribute.
30753 (*btst): Rename into...
30754 (*btst<mode>): ...this and adjust.
30755 (*cbranchsi4_btst_insn): Rename into...
30756 (*cbranch<mode>4_btst_insn): ...this and adjust.
30757 (trap): New define_insn.
30759 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30761 PR tree-optimization/79389
30762 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30763 (noce_reversed_cond_code): New function.
30764 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30765 reversed_comparison_code. Formatting fix.
30766 (noce_try_store_flag): Test rev_cond != NULL in addition to
30767 reversed_comparison_code.
30768 (noce_try_store_flag_constants): Likewise.
30769 (noce_try_store_flag_mask): Likewise.
30770 (noce_try_addcc): Use rev_cond if non-NULL instead of
30771 reversed_comparison_code.
30772 (noce_try_cmove_arith): Likewise. Formatting fixes.
30773 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30774 (noce_find_if_block): Initialize rev_cond.
30775 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30776 instead of false as last argument never attempt to reverse it
30779 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30781 PR tree-optimization/79663
30782 * tree-predcom.c (combine_chains): Process refs in reverse order
30783 only for ZERO length chains, and add explaining comment.
30785 2017-02-23 Jeff Law <law@redhat.com>
30787 PR tree-optimization/79578
30788 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30789 in call to operand_equal_p.
30791 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30794 * config/i386/cpuid.h: Fix another undefined behavior.
30796 2017-02-23 Richard Biener <rguenther@suse.de>
30798 PR tree-optimization/79683
30799 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30800 vector types for data-refs.
30802 2017-02-23 Martin Liska <mliska@suse.cz>
30804 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30806 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30808 PR middle-end/79665
30809 * internal-fn.c (get_range_pos_neg): Moved to ...
30810 * tree.c (get_range_pos_neg): ... here. No longer static.
30811 * tree.h (get_range_pos_neg): New prototype.
30812 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30813 are known to be in between 0 and signed maximum inclusive, try to
30814 expand both unsigned and signed divmod and use the cheaper one from
30817 2017-02-22 Jeff Law <law@redhat.com>
30819 PR tree-optimization/79578
30820 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30821 to compare base operands.
30823 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30826 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30827 gpc_reg_operand instead of fpr_reg_operand.
30829 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30831 * config/mips/mips.c (mips_return_in_memory): Force FP
30832 vector types to be returned in memory for o32 ABI.
30834 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30836 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30837 instead of DW_TAG_member for static data member declarations and don't
30838 set no_linkage_name for static inline data members.
30839 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30842 2017-02-22 Martin Liska <mliska@suse.cz>
30844 * doc/invoke.texi: Replace inequality signs with square brackets
30847 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30850 * lra-constraints.c (simplify_operand_subreg): Handle
30851 WORD_REGISTER_OPERATIONS targets.
30853 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30856 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30857 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30858 elimination by swapping fld*.
30860 2017-02-22 Richard Biener <rguenther@suse.de>
30862 PR tree-optimization/79673
30863 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30864 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30865 irrelevant address-space qualifiers and avoiding a
30866 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30868 2017-02-22 Richard Biener <rguenther@suse.de>
30870 PR tree-optimization/79666
30871 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30872 to not symbolically negate if that may introduce undefined
30875 2017-02-22 Martin Liska <mliska@suse.cz>
30878 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30879 * data-streamer-out.c (streamer_write_gcov_count_stream):
30881 * value-prof.c (stream_out_histogram_value): Make assert more
30882 precise based on type of counter.
30884 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30887 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30888 Use nonimmediate_operand instead of memory_operand for operand 1.
30889 (float-extend standard_x87sse_constant_load splitter): Ditto.
30891 2017-02-21 Jeff Law <law@redhat.com>
30893 PR tree-optimization/79621
30894 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30895 blocks with edges to themselves.
30897 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30900 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30901 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30902 Use gimple_call_builtin_p.
30905 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30906 on temporarily removed DEBUG_INSNs.
30908 PR tree-optimization/79649
30909 * tree-loop-distribution.c (classify_partition): Give up on
30910 non-generic address space loads/stores.
30912 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30914 * doc/loop.texi (Loop manipulation): Remove nonexistent
30915 tree_ssa_loop_version from the documentation.
30916 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30918 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30921 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30922 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30923 * config/rs6000/rs6000.c: Include except.h.
30924 (rs6000_expand_split_stack_prologue): Call
30925 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30927 2017-02-21 Martin Jambor <mjambor@suse.cz>
30930 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30931 have been analyzed.
30933 2017-02-21 Martin Jambor <mjambor@suse.cz>
30935 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30936 for backward compatibility only.
30937 * doc/invoke.texi (Option Summary): Remove all references to
30938 -fipa-cp-alignment.
30940 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30944 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30946 * lra-constraints.c (curr_insn_transform): Handle
30947 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30949 2017-02-21 Martin Liska <mliska@suse.cz>
30951 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30953 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30955 PR translation/79638
30956 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30958 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30961 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30962 (arm_function_ok_for_sibcall): Return false for an indirect call by
30963 descriptor if all the argument registers are used.
30964 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30965 alignment of the function.
30967 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30969 PR tree-optimization/61441
30970 * simplify-rtx.c (simplify_const_unary_operation): For
30971 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30972 the sNaN unmodified.
30974 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30976 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30977 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30978 instead of SYSTEM_HEADER_DIR.
30980 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30981 Martin Liška <mliska@suse.cz>
30983 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30984 Fix typos and grammar, use active voice, and clarify.
30986 2017-02-20 Marek Polacek <polacek@redhat.com>
30988 PR middle-end/79537
30989 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30992 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30994 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30997 * config/i386/i386.c (ix86_expand_builtin): Handle
30998 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30999 ix86_builtins_isa[fcode].isa as a requirement of those
31000 flags and any other flag in the bitmask.
31001 (ix86_init_mmx_sse_builtins): Use 0 instead of
31002 ~OPTION_MASK_ISA_64BIT as mask.
31003 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31004 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31005 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31006 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31008 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31011 * lra-constraints.c (split_reg): Check requested split mode
31012 is supported by the register.
31014 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31016 * lra-constraints.c (simplify_operand_subreg): Remove early
31019 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31022 * lra-constraints.c (curr_insn_transform): Tighten condition
31023 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31025 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31028 * lra-constraints.c (curr_insn_transform): Handle
31029 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31031 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31034 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31036 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31038 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31041 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31044 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31046 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31049 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31051 * final.c (last_columnnum, override_columnnum): New variables.
31052 (final_start_function): Set last_columnnum, pass it to begin_prologue
31053 hook and pass 0 to dwarf2out_begin_prologue.
31054 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31055 to source_line debug hook.
31056 (notice_source_line): Compute last_columnnum and for debug_column_info
31057 return true on column changes.
31058 * debug.h (struct gcc_debug_hooks): Add column argument to
31059 source_line and begin_prologue hooks.
31060 (debug_nothing_int_charstar_int_bool): Remove prototype.
31061 (debug_nothing_int_int_charstar,
31062 debug_nothing_int_int_charstar_int_bool): New prototypes.
31063 (dwarf2out_begin_prologue): Add column argument.
31064 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31065 begin_prologue hooks.
31066 (debug_nothing_int_charstar_int_bool): Remove.
31067 (debug_nothing_int_int_charstar,
31068 debug_nothing_int_int_charstar_int_bool): New functions.
31069 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31070 through to dwarf2out_source_line.
31071 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31072 (dwarf2out_source_line): Add column argument, emit it if requested.
31073 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31075 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31076 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31077 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31078 through to dwarf2out_begin_prologue.
31079 (vmsdbgout_source_line): Add column argument, pass it through to
31080 dwarf2out_source_line.
31081 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31082 dbxout_source_line caller.
31083 (dbxout_source_line): Add column argument.
31085 * common.opt (gno-column-info, gcolumn-info): New options.
31086 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31087 (check_die): Also test for multiple DW_AT_decl_column attributes.
31088 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31089 DW_AT_decl_column if requested.
31090 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31092 (gen_variable_die): Likewise.
31093 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31094 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31097 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31098 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31099 (ix86_handle_option): Handle OPT_m3dnowa.
31100 * doc/invoke.texi (-m3dnowa): Document.
31101 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31102 -m3dnowa instead of -m3dnow -march=athlon.
31105 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31106 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31108 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31111 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31112 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31113 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31114 generator for vsx_xxpermdi_<mode>_be.
31115 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31116 force big-endian semantics.
31117 (vsx_xxpermdi_<mode>_be): New define_expand with same
31118 implementation as previous version of vsx_xxpermdi_<mode>.
31120 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31122 PR tree-optimization/79327
31123 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31124 variable, its initialization and use.
31126 2017-02-17 Julia Koval <julia.koval@intel.com>
31128 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31129 (OPTION_MASK_ISA_PKU_UNSET): New.
31130 (ix86_handle_option): Handle -mrdpid.
31131 * config/i386/cpuid.h (bit_RDPID): New.
31132 * config/i386/driver-i386.c (host_detect_local_cpu):
31133 Detect RDPID feature.
31134 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31135 * config/i386/i386-c.c (ix86_target_macros_internal):
31137 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31138 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31139 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31140 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31141 * config/i386/i386.md (define_insn "rdpid"): New.
31142 * config/i386/i386.opt Add -mrdpid.
31143 * config/i386/immintrin.h (_rdpid_u32): New.
31145 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31147 PR rtl-optimization/79541
31148 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31149 instead of transforming it into USE.
31151 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31153 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31154 If HONOR_SNANS (SFmode) force the input to a register.
31155 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31156 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31157 an frsp or similar insn.
31159 2017-02-17 Martin Liska <mliska@suse.cz>
31161 PR rtl-optimization/79577
31162 * params.def (selsched-max-sched-times): Increase minimum to 1.
31164 2017-02-17 Martin Liska <mliska@suse.cz>
31166 PR rtl-optimization/79574
31167 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31169 2017-02-17 Martin Liska <mliska@suse.cz>
31171 PR tree-optimization/79529
31172 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31173 ssa_defined_default_def_p to handle cases which are implicitly
31175 * tree-ssa.c (ssa_defined_default_def_p): New function.
31176 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31177 which are implicitly defined.
31178 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31180 2017-02-17 Richard Biener <rguenther@suse.de>
31182 PR middle-end/79576
31183 * params.def (max-ssa-name-query-depth): Limit to 10.
31185 2017-02-17 Richard Biener <rguenther@suse.de>
31187 PR tree-optimization/79552
31188 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31191 2017-02-17 Richard Biener <rguenther@suse.de>
31194 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31196 2017-02-17 Marek Polacek <polacek@redhat.com>
31198 PR middle-end/79536
31199 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31200 (fold_negate_expr): New wrapper.
31202 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31204 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31205 Correct terminology and de-emphasize pre-standard behavior.
31207 2017-02-16 Alan Modra <amodra@gmail.com>
31209 PR rtl-optimization/79286
31210 * ira.c (def_dominates_uses): New function.
31211 (update_equiv_regs): Don't create an equivalence for insns that
31212 may trap where the register def does not dominate the use.
31214 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31216 PR rtl-optimization/78127
31217 * lra.c (lra): Call lra_eliminate before finish the loop after
31220 2017-02-16 Richard Biener <rguenther@suse.de>
31222 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31224 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31225 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31226 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31227 (isl_val_int_from_wi): New function.
31228 (extract_affine_gmp): Rename to ...
31229 (extract_affine_wi): ... this, take a widest_int.
31230 (extract_affine_int): Just wrap extract_affine_wi.
31231 (add_param_constraints): Use isl_val_int_from_wi.
31232 (add_loop_constraints): Likewise, and extract_affine_wi.
31234 2017-02-15 Jeff Law <law@redhat.com>
31236 PR middle-end/79521
31237 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31238 ira_init_register_move_cost_if_necessary.
31240 2017-02-15 Martin Sebor <msebor@redhat.com>
31242 PR middle-end/32003
31243 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31244 removed in a prior commit.
31246 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31248 PR tree-optimization/79347
31249 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31250 counters during peeling.
31252 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31254 * Makefile.in (site.exp): Remove "set ISLVER".
31256 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31259 * real.c (real_from_integer): Call real_convert even for decimal.
31261 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31264 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31266 2017-02-14 Andrew Pinski <apinski@cavium.com>
31268 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31269 cores and change the partno/implementer to be correct.
31270 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31271 the 'B" as the implementer.
31272 * config/aarch64/aarch64-tune.md: Regenerate.
31274 2017-02-14 Carl Love <cel@us.ibm.com>
31276 * config/rs6000/rs6000.c: Add case statement entry to make the
31277 xvcvuxdsp built-in argument unsigned.
31278 * config/rs6000/vsx.md: Fix the source and return operand types so they
31279 match the instruction definitions from the ISA document. Fix typo
31280 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31283 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31286 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31287 member early_clobber_alts.
31288 * lra-lives.c (reg_early_clobber_p): New.
31289 (process_bb_lives): Use it.
31290 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31291 (debug_operand_data): Initialize early_clobber_alts.
31292 (setup_operand_alternative): Set up early_clobber_alts.
31293 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31294 alternatives to new_insn_reg.
31295 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31297 (lra_update_insn_regno_info): Pass the new arg.
31299 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31301 PR middle-end/79505
31302 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31303 (new_oacc_loop_raw): Don't clear already cleared fields.
31306 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31307 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31308 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31310 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31313 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31315 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31318 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31319 the extra instruction to the right place to store 128-bit constant
31322 2017-02-14 Martin Sebor <msebor@redhat.com>
31324 PR middle-end/79448
31325 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31326 warning for strings of unknown length.
31328 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31330 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31332 2017-02-14 Jeff Law <law@redhat.com>
31335 * ira-costs.c (scan_one_insn): Initialize register move costs
31336 for pseudos seen in USE/CLOBBER insns.
31338 PR tree-optimization/79095
31339 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31340 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31341 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31342 if the operands are known to be not equal, then the resulting range
31344 (intersect_ranges): If the new range is ~[0,0] and the old range is
31345 wide, then prefer ~[0,0].
31346 * tree-vrp.c (overflow_comparison_p_1): New function.
31347 (overflow_comparison_p): New function.
31348 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31349 if NAME is used in an overflow test.
31350 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31351 overflow check that can be expressed as an equality test, then adjust
31352 ops to be that equality test.
31354 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31356 * config/s390/s390-builtin-types.def: Remove flags argument.
31357 * config/s390/s390.c (s390_init_builtins): Likewise.
31359 2017-02-14 Martin Liska <mliska@suse.cz>
31361 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31362 vector. Fix trailing white spaces.
31364 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31366 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31369 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31371 PR rtl-optimization/68664
31372 * config/arm/arm.c (arm_sched_can_speculate_insn):
31373 New function. Declare prototype.
31374 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31376 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31378 PR rtl-optimization/68664
31379 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31381 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31383 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31385 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31386 max skip bytes for function, loop and jump.
31388 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31390 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31391 ABS_EXPR for gimple dump.
31393 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31396 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31398 PR tree-optimization/79408
31399 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31400 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31401 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31402 also if rhs1 is INTEGER_CST.
31404 2017-02-14 Richard Biener <rguenther@suse.de>
31406 PR middle-end/79432
31407 * tree-into-ssa.c (insert_phi_nodes): When the function can
31408 have abnormal edges rewrite SSA names with broken use-def
31409 dominance out of SSA and register them for PHI insertion.
31411 2017-02-13 Martin Sebor <msebor@redhat.com>
31413 PR middle-end/79496
31414 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31415 clearing info.nowrite flag when snprintf size argument is a range.
31417 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31419 * cprop.c (cprop_jump): Add missing space in string literal.
31420 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31421 (get_constraint_for_component_ref): Likewise.
31422 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31423 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31424 * lra-constraints.c (process_alt_operands): Likewise.
31425 * ipa-inline.c (inline_small_functions): Likewise.
31426 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31427 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31428 * trans-mem.c (diagnose_tm_1_op): Likewise.
31429 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31430 (grid_parallel_clauses_gridifiable): Likewise.
31432 * config/nvptx/mkoffload.c (process): Add space in between
31435 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31436 "MOD4_SSE_REGS" and "ALL_REGS".
31438 * spellcheck.c (test_data): Add , in between "foo" and "food".
31440 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31443 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31444 boundary crossing check and subsequent code generation agree.
31446 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31448 * config/aarch64/aarch64.c (has_memory_op): Delete.
31449 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31452 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31454 PR rtl-optimization/79388
31455 PR rtl-optimization/79450
31456 * combine.c (distribute_notes): When removing TEM_INSN for which
31457 corresponding dest has last value recorded, invalidate that last
31460 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31462 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31463 of explicit '@'. Add missing assembly comment marker on branch costs
31466 2017-02-13 Nathan Sidwell <nathan@acm.org>
31468 * gengtype-lex.l (<in_struct>): Add '/'.
31470 2017-02-13 Martin Liska <mliska@suse.cz>
31473 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31475 2017-02-13 Richard Biener <rguenther@suse.de>
31477 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31479 * configure: Re-generate.
31480 * config.in: Likewise.
31481 * graphite-dependences.c: Simplify as if
31482 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31483 * graphite-isl-ast-to-gimple.c: Likewise.
31484 * graphite-optimize-isl.c: Likewise.
31485 * graphite-poly.c: Likewise.
31486 * graphite-sese-to-poly.c: Likewise.
31487 * graphite.h: Likewise.
31488 * toplev.c: Include isl/version.h and use isl_version () for
31489 printing the ISL version.
31490 * doc/install.texi: Update ISL requirement.
31492 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31494 * doc/standards.texi (Standards): Update reference to
31497 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31499 * doc/extend.texi (Named Address Spaces): sourceware.org now
31501 * doc/install.texi (Binaries): Ditto.
31504 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31506 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31507 standard terminology "stringize"/"stringizing" throughout.
31508 * doc/cppinternals.texi: Likewise.
31510 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31512 * doc/extend.texi: Fix some spelling mistakes and typos.
31513 * doc/invoke.texi: Likewise.
31515 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31518 * params.def (inline-min-speedup) Change from 10 to 8.
31520 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31522 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31525 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31528 * ipa-inline-analysis.c (get_minimal_bb): New function.
31529 (record_modified): Use it.
31530 (remap_edge_change_prob): Handle also ancestor functions.
31532 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31534 * doc/contrib.texi (Contributors): Remove broken link into
31535 the Mauve CVS repository.
31537 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31539 PR middle-end/79454
31540 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31541 result computation whenever lhs doesn't have vector mode, not
31542 just when it has BLKmode.
31544 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31546 * doc/makefile.texi (profiledbootstrap): Refer to the
31547 installation instructions only in textual form.
31549 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31552 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31554 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31556 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31557 (Specific): Update mingw-w64 reference.
31559 (Specific): Remove broken link to Renesas RX processor.
31561 2017-02-10 Richard Biener <rguenther@suse.de>
31563 * toplev.c (process_options): Do not mention obsolete graphite
31564 options when printing sorry message about missing graphite support.
31565 Mention -floop-nest-optimize.
31567 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31569 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31570 (vtst_p16): Likewise.
31571 (vtstq_p8): Likewise.
31572 (vtstq_p16): Likewise.
31574 (vtstq_p64): Likewise.
31575 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31576 (vset_lane_p64): New.
31577 (vsetq_lane_p64): New.
31579 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31581 PR tree-optimization/79411
31582 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31583 stmt operands are SSA_NAMEs used in abnormal phis.
31584 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31587 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31590 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31593 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31595 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31597 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31599 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31603 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31604 not arbitrary TREE_CONSTANT.
31607 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31608 "omp declare target link" attribute unless is_global_var.
31609 * omp-offload.c (find_link_var_op): Likewise.
31611 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31612 Chung-Lin Tang <cltang@codesourcery.com>
31614 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31616 (gimplify_adjust_omp_clauses): Don't delete TILE.
31617 (gimplify_omp_for): Deal with TILE.
31618 * internal-fn.c (expand_GOACC_TILE): New function.
31619 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31621 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31622 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31624 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31625 avoid DIV for outermost collapse var.
31626 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31627 Remove out of date comments, fix whitespace.
31628 * omp-general.c (omp_extract_for_data): Deal with tiling.
31629 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31630 adjust OLF_DIM_BASE value.
31631 (struct omp_for_data): Add tiling field.
31632 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31633 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31634 for auto loops. Remove default auto determining, moved to
31635 oacc_loop_fixed_partitions.
31636 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31637 stmts, add e_mask field.
31638 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31639 (oacc_thread_numbers): Use oacc_dim_call.
31640 (oacc_xform_tile): New.
31641 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31642 (finish_oacc_loop): Adjust for ifns vector.
31643 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31644 add case for GOACC_TILE fns.
31645 (oacc_loop_xform_loop): Delete.
31646 (oacc_loop_process): Iterate over call list directly, and add
31647 handling for GOACC_TILE fns.
31648 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31650 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31651 vector partitioning to outer loops. Assign 2 partitions to loops
31652 when available. Add TILE handling.
31653 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31654 (execite_oacc_device_lower): Process GOACC_TILE fns,
31655 ignore unknown specs.
31656 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31657 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31658 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31660 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31662 * configure.ac (ACX_BUGURL): Update.
31663 * configure: Regenerate.
31665 2017-02-09 Richard Biener <rguenther@suse.de>
31667 PR tree-optimization/69823
31668 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31669 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31671 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31673 * config/arc/arc-c.def: Add __NPS400__ definition.
31674 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31675 (TARGET_NPS400): Define.
31677 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31679 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31681 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31682 pointer, arch_info.
31683 (arc_cpu_types): Fill the arch_info field with a pointer into the
31684 arc_arch_types table.
31685 (arc_selected_cpu): Declare.
31686 * config/arc/arc.c (arc_selected_cpu): Make global.
31687 (arc_selected_arch): Delete.
31688 (arc_base_cpu): Delete.
31689 (arc_override_options): Remove references to deleted variables,
31690 update access to arch information.
31691 (ARC_OPT): Update access to arch information.
31692 (ARC_OPTX): Likewise.
31693 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31694 (TARGET_ARC600): Update access to arch information.
31695 (TARGET_ARC601): Likewise.
31696 (TARGET_ARC700): Likewise.
31697 (TARGET_EM): Likewise.
31698 (TARGET_HS): Likewise.
31699 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31702 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31705 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31706 condition/operands for integer GE/LE/GEU/LEU operations.
31708 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31710 PR translation/79397
31711 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31714 2017-02-08 Martin Jambor <mjambor@suse.cz>
31717 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31718 whether allocation happened.
31719 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31720 nothing was allocated.
31722 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31724 PR tree-optimization/79408
31725 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31726 constant, but SSA_NAME with a known integer range, use the minimum
31727 of that range instead of op1 to determine if modulo can be replaced
31728 with its first operand.
31730 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31732 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31734 2017-02-08 Richard Biener <rguenther@suse.de>
31736 PR tree-optimization/71824
31737 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31738 Check all loops contained in the merged region.
31740 2017-02-07 Andrew Pinski <apinski@cavium.com>
31742 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31744 2017-02-07 Andrew Pinski <apinski@cavium.com>
31746 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31747 (thunderxt88): Likewise.
31748 (thunderxt81): Disable LSE and change v8.1 to v8.
31749 (thunderxt83): Likewise.
31751 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31752 Richard Biener <rguenther@suse.de>
31754 PR middle-end/79399
31755 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31756 type from int to size_t.
31757 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31759 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31761 PR rtl-optimization/79386
31762 * cprop.c (bypass_conditional_jumps): Initialize
31763 bypass_last_basic_block already before splitting bbs after
31764 unconditional traps...
31765 (bypass_conditional_jumps): ... rather than here.
31768 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31769 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31770 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31771 fix -masm=intel patterns.
31773 2017-02-07 Richard Biener <rguenther@suse.de>
31775 PR tree-optimization/79256
31776 PR middle-end/79278
31777 * builtins.c (get_object_alignment_2): Use min_align_of_type
31778 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31779 and ADJUST_FIELD_ALIGN.
31781 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31783 * doc/tm.texi: Regenerate.
31784 * stor-layout.c (layout_decl): Adjust.
31785 (update_alignment_for_field): Likewise.
31786 (place_field): Likewise.
31787 (min_align_of_type): Likewise.
31788 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31789 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31790 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31791 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31792 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31793 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31794 * config/i386/i386.c (x86_field_alignment): Likewise.
31795 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31796 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31797 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31798 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31799 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31800 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31804 2017-01-30 Richard Biener <rguenther@suse.de>
31806 PR tree-optimization/79256
31807 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31808 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31811 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31813 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31814 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31815 builtins to SImode and emit a zero-extend, if necessary.
31817 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31819 * docs/invoke.texi (RISC-V Options): Alphabetize.
31821 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31823 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31826 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31828 * config/riscv/riscv.c: New file.
31829 * common/config/riscv/riscv-common.c: Likewise.
31830 * config.gcc: Likewise.
31831 * config/riscv/constraints.md: Likewise.
31832 * config/riscv/elf.h: Likewise.
31833 * config/riscv/generic.md: Likewise.
31834 * config/riscv/linux.h: Likewise.
31835 * config/riscv/multilib-generator: Likewise.
31836 * config/riscv/peephole.md: Likewise.
31837 * config/riscv/pic.md: Likewise.
31838 * config/riscv/predicates.md: Likewise.
31839 * config/riscv/riscv-builtins.c: Likewise.
31840 * config/riscv/riscv-c.c: Likewise.
31841 * config/riscv/riscv-ftypes.def: Likewise.
31842 * config/riscv/riscv-modes.def: Likewise.
31843 * config/riscv/riscv-opts.h: Likewise.
31844 * config/riscv/riscv-protos.h: Likewise.
31845 * config/riscv/riscv.h: Likewise.
31846 * config/riscv/riscv.md: Likewise.
31847 * config/riscv/riscv.opt: Likewise.
31848 * config/riscv/sync.md: Likewise.
31849 * config/riscv/t-elf-multilib: Likewise.
31850 * config/riscv/t-linux: Likewise.
31851 * config/riscv/t-linux-multilib: Likewise.
31852 * config/riscv/t-riscv: Likewise.
31853 * configure.ac: Likewise.
31854 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31855 Waterman as RISC-V maintainers.
31856 * doc/install.texi: Add RISC-V entries.
31857 * doc/invoke.texi: Add RISC-V options section.
31858 * doc/md.texi: Add RISC-V constraints section.
31859 * configure: Regenerated.
31861 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31864 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31865 false values to be constant vectors with all 0 or all 1 bits set.
31866 (vcondu<mode><mode>): Likewise.
31867 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31869 (fpmask_comparison_operator): Update comment.
31870 (vecint_comparison_operator): New predicate.
31871 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31872 vector conditionals when the true and false values are constant
31873 vectors with all 0 bits or all 1 bits set.
31875 2017-02-06 Martin Sebor <msebor@redhat.com>
31877 PR tree-optimization/79376
31878 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31880 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31882 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31883 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31884 to simplify split condition.
31886 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31888 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31889 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31892 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31894 PR rtl-optimization/68664
31895 * target.def (can_speculate_insn): New hook.
31896 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31897 * doc/tm.texi: Regenerate.
31898 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31899 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31900 (rs6000_sched_can_speculate_insn): New function.
31902 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31904 PR tree-optimization/79284
31905 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31906 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31907 vectorizable_mask_load_store, vectorizable_operation,
31908 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31909 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31910 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31911 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31912 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31913 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31914 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31915 is_gimple_assign (stmt). Replace another such test with
31916 is_gimple_assign (stmt).
31918 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31921 * config/avr/avr.c (rtl-iter.h): Include it.
31922 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31923 (avr_legitimate_combined_insn): ...and implementation.
31925 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31927 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31928 * config/s390/s390.c (s390_const_operand_ok)
31929 (s390_canonicalize_comparison, s390_extract_part)
31930 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31931 (s390_contiguous_bitmask_p, s390_rtx_costs)
31932 (legitimize_pic_address): Likewise.
31933 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31934 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31935 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31936 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31937 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31939 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31941 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31942 REGNO($0) == REGNO($1).
31944 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31946 * config/s390/linux.h(SIZE_TYPE): Add comment.
31948 2017-02-06 Julian Brown <julian@codesourcery.com>
31949 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31950 Virendra Pathak <virendra.pathak@broadcom.com>
31952 * config/aarch64/aarch64-cores.def: Change the scheduler
31954 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31955 * config/aarch64/thunderx2t99.md: New file.
31957 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31959 * doc/standards.texi (Go Language): Update link to language
31962 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31964 * tree-eh.c (lower_resx): Sanitize profile.
31965 (cleanup_empty_eh_move_lp): Likewise.
31967 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31970 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31972 * cfgloopmanip.h (loop_version): Update prototype.
31973 * modulo-sched.c (sms_schedule): Update call of loop_version.
31974 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31975 * tree-parloops.c (gen_parallel_loop): Likewise.
31976 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31977 * tree-ssa-loop-split.c (split_loop): Likewise.
31978 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31979 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31981 2017-02-05 Martin Liska <mliska@suse.cz>
31984 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31986 (print_operand_address): Initialize a struct to zero.
31988 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31990 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31991 garbage collector only in textual form.
31993 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31995 * doc/extend.texi (x86 specific memory model extensions for
31996 transactional memory): Simplify a phrase.
31998 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32001 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32002 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32003 (atomic_storedi_1): Likewise.
32005 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32007 PR tree-optimization/79338
32008 * tree-parloops.c (gather_scalar_reductions): Don't call
32009 vect_analyze_loop_form for loop->inner before destroying loop's
32012 2017-02-03 Martin Sebor <msebor@redhat.com>
32014 PR tree-optimization/79327
32015 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32016 when precision has resulted in leading zeros.
32017 (format_integer): Adjust the likely counter to assume an unknown
32018 argument that may be zero is non-zero.
32020 2017-02-03 Jason Merrill <jason@redhat.com>
32023 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32024 avoid copying non-taken branch.
32026 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32028 PR tree-optimization/79340
32029 * tree-vect-loop.c (vectorizable_reduction): Release
32030 vec_defs elements after safe_splicing them into other vectors.
32033 PR tree-optimization/79327
32034 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32035 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32037 (format_integer): Use wide_int_to_tree instead of build_int_cst
32038 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32039 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32040 of shortest and longest sequence.
32042 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32044 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32045 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32047 2017-02-03 Walter Lee <walt@tilera.com>
32050 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32051 after initial stackframe link reg save.
32052 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32054 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32057 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32058 wu for stxssp alternative.
32060 2017-02-03 Martin Sebor <msebor@redhat.com>
32062 PR tree-optimization/79352
32063 * gimple-fold.c (get_range_strlen): Add argument.
32064 (get_range_strlen): Change return type to bool.
32065 (get_maxval_strlen): Pass in a dummy argument.
32066 * gimple-fold.h (get_range_strlen): Change return type to bool.
32067 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32068 * tree.h (array_at_struct_end_p): Add argument.
32069 * tree.c (array_at_struct_end_p): Handle it.
32071 2017-02-03 Martin Liska <mliska@suse.cz>
32074 * multiple_target.c (create_dispatcher_calls): Redirect edge
32075 from a caller of a dispatcher.
32076 (expand_target_clones): Make the clones local.
32077 (ipa_target_clone): Do both target clones and resolvers.
32078 (ipa_dispatcher_calls): Remove the pass.
32079 (pass_dispatcher_calls::gate): Likewise.
32080 (make_pass_dispatcher_calls): Likewise.
32081 * passes.def (pass_target_clone): Put as very first IPA early
32084 2017-02-03 Martin Liska <mliska@suse.cz>
32086 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32087 in case of a function with ifunc attribute.
32089 2017-02-03 Martin Liska <mliska@suse.cz>
32091 * cgraph.c (cgraph_node::dump): Dump function version info.
32092 * symtab.c (symtab_node::dump_base): Add missing new line.
32094 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32096 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32097 (ifcombine_ifandif): Use it.
32099 2017-02-03 Martin Liska <mliska@suse.cz>
32101 * doc/invoke.texi: Document default value for
32102 use-after-scope-direct-emission-threshold.
32104 2017-02-03 Martin Liska <mliska@suse.cz>
32106 PR tree-optimization/79339
32107 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32108 (format_floating): Likewise.
32110 2017-02-03 Martin Liska <mliska@suse.cz>
32113 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32115 (ipa_node_params_t::remove): Likewise.
32116 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32117 initialization from removed ipa_node_params_t::insert.
32118 (ipa_node_params::~ipa_node_params): Move from removed
32119 ipa_node_params_t::release.
32120 * symbol-summary.h (symbol_summary::m_released): New member.
32121 Do not release a summary twice. Do not allow to call finalizer
32122 for types of a summary that live in GGC memory.
32124 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32126 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32129 2017-02-02 Martin Sebor <msebor@redhat.com>
32131 PR middle-end/79275
32132 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32133 (format_string): Tighten up the range of output for non-constant
32134 strings and correct the expected range for wide non-constant strings.
32136 2017-02-02 Martin Sebor <msebor@redhat.com>
32138 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32140 PR middle-end/32003
32141 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32143 (-fdump-tree-@var): Add to index and document how to come up
32144 with pass-specific option and dump file names.
32145 (-fdump-passes): Clarify where to look for output.
32147 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32149 PR middle-end/77445
32150 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32151 statistics of the analyzed path; allow threading for speed when
32152 any of BBs along the path are optimized for speed.
32154 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32156 PR middle-end/78468
32157 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32158 settings of the virtual registers.
32161 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32163 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32164 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32167 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32169 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32170 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32172 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32174 * config/s390/s390.md: Add missing comments with the expanded
32176 * config/s390/vector.md: Likewise.
32177 * config/s390/vx-builtins.md: Likewise.
32179 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32182 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32183 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32184 conditions on a single line.
32186 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32188 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32189 __S390_VX__ to __VX__.
32191 2017-02-01 Andrew Pinski <apinski@cavium.com>
32193 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32194 stmt_info to record_stmt_cost.
32195 (vect_get_known_peeling_cost): Pass stmt_info if known to
32197 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32198 cpu_vector_cost field into
32199 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32200 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32201 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32202 splitting of scalar_stmt_cost and vec_stmt_cost.
32203 (thunderx_vector_cost): Likewise.
32204 (cortexa57_vector_cost): LIkewise.
32205 (exynosm1_vector_cost): Likewise.
32206 (xgene1_vector_cost): Likewise.
32207 (thunderx2t99_vector_cost): Improve after the splitting of the two
32209 (aarch64_builtin_vectorization_cost): Update for the splitting of
32210 scalar_stmt_cost and vec_stmt_cost.
32212 2017-02-01 Torvald Riegel <triegel@redhat.com>
32213 Richard Henderson <rth@redhat.com>
32215 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32216 conditional on existance of a fast atomic load.
32217 * optabs-query.c (can_atomic_load_p): New function.
32218 * optabs-query.h (can_atomic_load_p): Declare it.
32219 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32220 no fast atomic load is available for the particular size of access.
32221 (expand_atomic_compare_and_swap): Likewise.
32222 (expand_atomic_load): Likewise.
32223 (expand_atomic_store): Likewise.
32224 (expand_atomic_fetch_op): Likewise.
32225 * testsuite/lib/target-supports.exp
32226 (check_effective_target_sync_int_128): Remove x86 because it provides
32227 no fast atomic load.
32228 (check_effective_target_sync_int_128_runtime): Likewise.
32230 2017-02-01 Richard Biener <rguenther@suse.de>
32232 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32233 (graphite_transform_loops): Provide opt-info for optimized nests.
32234 * tree-parloop.c (parallelize_loops): Provide opt-info for
32235 parallelized loops.
32237 2017-02-01 Richard Biener <rguenther@suse.de>
32239 PR middle-end/79315
32240 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32241 was not set before.
32243 2017-02-01 Richard Biener <rguenther@suse.de>
32245 PR tree-optimization/71824
32246 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32247 Verify the loops are valid in the merged SESE region.
32248 (scop_detection::can_represent_loop_1): Check analyzing the
32249 evolution of the number of iterations in the region succeeds.
32251 2017-01-31 Ian Lance Taylor <iant@golang.org>
32253 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32254 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32256 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32258 PR preprocessor/79210
32259 * input.c (get_substring_ranges_for_loc): Replace line_width
32260 assertion with error-handling.
32262 2017-01-31 Richard Biener <rguenther@suse.de>
32264 PR tree-optimization/77318
32265 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32266 (create_pw_aff_from_tree): Take loop parameter.
32267 (add_condition_to_pbb): Pass loop of the condition to
32268 create_pw_aff_from_tree.
32270 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32272 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32273 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32275 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32279 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32281 (convert_int_to_float128): Likewise.
32282 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32283 (convert_int_to_float128): Likewise.
32284 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32285 (UNSPEC_IEEE128_CONVERT): Likewise.
32286 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32287 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32288 Use local variables for IBM extended format.
32289 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32290 (fix_trunc<mode>si2_fprs): Likewise.
32291 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32292 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32293 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32294 to know that we can now have integers of all sizes in vector
32296 (fix<uns>_<mode>di2_hw): Likewise.
32297 (float<uns>_<mode>si2_hw): Likewise.
32298 (fix_<mode>si2_hw): Likewise.
32299 (fixuns_<mode>si2_hw): Likewise.
32300 (float<uns>_<mode>di2_hw): Likewise.
32301 (float_<mode>di2_hw): Likewise.
32302 (float_<mode>si2_hw): Likewise.
32303 (floatuns_<mode>di2_hw): Likewise.
32304 (floatuns_<mode>si2_hw): Likewise.
32305 (xscvqp<su>wz_<mode>): Delete, no longer used.
32306 (xscvqp<su>dz_<mode>): Likewise.
32307 (xscv<su>dqp_<mode>): Likewise.
32308 (ieee128_mfvsrd_64bit): Likewise.
32309 (ieee128_mfvsrd_32bit): Likewise.
32310 (ieee128_mfvsrwz): Likewise.
32311 (ieee128_mtvsrw): Likewise.
32312 (ieee128_mtvsrd_64bit): Likewise.
32313 (ieee128_mtvsrd_32bit): Likewise.
32315 2017-01-31 Martin Liska <mliska@suse.cz>
32318 * ipa-prop.c (ipa_free_all_node_params): Call release method
32319 instead of ~sumbol_summary to not to trigger double times
32322 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32324 PR tree-optimization/71691
32325 * bitmap.h (class auto_bitmap): New.
32326 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32327 is_maybe_undefined instead of ssa_undefined_value_p.
32329 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32331 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32332 __S390_ARCH_LEVEL__ to __ARCH__.
32334 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32336 PR tree-optimization/79267
32337 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32338 if should_remove_lhs_p is true.
32340 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32343 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32344 (add_alignment_attribute): New.
32345 (base_type_die): Add alignment attribute.
32346 (subrange_type_die): Likewise.
32347 (modified_type_die): Likewise.
32348 (gen_array_type_die): Likewise.
32349 (gen_descr_array_type_die: Likewise.
32350 (gen_enumeration_type_die): Likewise.
32351 (gen_subprogram_die): Likewise.
32352 (gen_variable_die): Likewise.
32353 (gen_field_die): Likewise.
32354 (gen_ptr_to_mbr_type_die): Likewise.
32355 (gen_struct_or_union_type_die): Likewise.
32356 (gen_subroutine_type_die): Likewise.
32357 (gen_typedef_die): Likewise.
32358 (base_type_cmp): Compare alignment attribute.
32360 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32363 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32364 (setb_unsigned) New pattern for setb with CCUNS.
32365 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32366 subfc./subfe sequence to avoid overflow problems. Generate a
32367 shorter sequence with cmpld/setb for power9.
32368 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32369 for generating subfc. instruction.
32370 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32371 now uses this instruction.
32373 2017-01-30 Ian Lance Taylor <iant@google.com>
32376 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32377 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32379 2017-01-30 Martin Sebor <msebor@redhat.com>
32381 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32382 Move constant to the right of a relational operator.
32383 (get_mpfr_format_length, format_character, format_string): Ditto.
32384 (should_warn_p, maybe_warn): Same.
32386 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32388 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32391 * asan.c (get_translation_unit_decl): Remove function.
32392 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32394 2017-01-30 Martin Liska <mliska@suse.cz>
32396 PR gcov-profile/79259
32397 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32398 -fprofile-generate.
32400 2017-01-30 Martin Liska <mliska@suse.cz>
32403 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32404 Initialize variables with NULL value.
32406 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32409 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32411 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32413 2017-01-30 Richard Biener <rguenther@suse.de>
32415 PR tree-optimization/79276
32416 * tree-vrp.c (process_assert_insertions): Properly adjust common
32417 when removing a duplicate.
32419 2017-01-30 Richard Biener <rguenther@suse.de>
32421 PR tree-optimization/79256
32422 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32423 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32425 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32427 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32430 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32431 ("*r<noxa>sbg_<mode>_sll_bitmask")
32432 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32433 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32434 Use contiguous_bitmask_nowrap_operand.
32436 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32439 * config/rs6000/altivec.h (vec_xl): Revise #define.
32440 (vec_xst): Likewise.
32442 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32444 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32446 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32448 PR rtl-optimization/79194
32449 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32450 traps before call to bypass_conditional_jumps.
32452 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32454 PR tree-optimization/71374
32455 * lra-constraints.c (check_conflict_input_operands): New.
32456 (match_reload): Use it.
32458 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32461 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32462 account to calculate conflict_set.
32464 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32466 PR rtl-optimization/78559
32467 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32468 other_insn in combine.
32470 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32472 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32473 uint16_type_node for BT_UINT16.
32475 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32477 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32478 "RTL Tests" to menu.
32479 (GIMPLE Tests): New node.
32480 (RTL Tests): New node.
32482 2017-01-27 Richard Biener <rguenther@suse.de>
32484 PR tree-optimization/79245
32485 * tree-loop-distribution.c (distribute_loop): Apply cost
32486 modeling also to detected patterns.
32488 2017-01-27 Richard Biener <rguenther@suse.de>
32490 PR tree-optimization/71433
32491 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32492 (compare_assert_loc): New function.
32493 (process_assert_insertions): Sort and optimize assert locations
32494 to remove duplicates and push down identical assertions on
32495 edges to their destination block.
32497 2017-01-27 Richard Biener <rguenther@suse.de>
32499 PR tree-optimization/79244
32500 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32501 out SSA names even if abnormal.
32503 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32505 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32506 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32507 instead of MPFR_RNDN.
32509 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32512 * arm.c (arm_option_override): Don't call build_target_option_node
32513 until after doing all option overrides.
32514 (arm_valid_target_attribute_tree): Likewise.
32516 2017-01-27 Martin Liska <mliska@suse.cz>
32518 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32519 for {cd}tors and C++ {cd}tors.
32521 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32523 * config/s390/s390.md ("*setmem_long_and")
32524 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32526 2017-01-26 Martin Sebor <msebor@redhat.com>
32528 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32531 2017-01-26 Martin Sebor <msebor@redhat.com>
32533 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32534 HAVE_DFmode before using XFmode or DFmode.
32535 (parse_directive): Avoid using the z length modifier to avoid
32536 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32538 PR middle-end/78703
32539 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32540 to accept adjustment as an array.
32541 (get_int_range): New function.
32542 (struct directive): Make width and prec arrays.
32543 (directive::set_width, directive::set_precision): Call get_int_range.
32544 (format_integer, format_floating): Handle width and precision ranges.
32545 (format_string, parse_directive): Same.
32547 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32550 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32551 comdat_type_p set, just clone them, but keep the children in the
32555 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32556 which have direct callers with -fvar-tracking-assignments enabled
32558 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32559 inside of type units.
32561 2017-01-26 Martin Sebor <msebor@redhat.com>
32563 PR middle-end/78703
32564 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32566 (struct format_result): Replace number_chars, number_chars_min,
32567 and number_chars_max with a single member of struct result_range.
32569 (format_result::operator+=): Adjust.
32570 (struct fmtresult): Remove bounded. Handle likely and unlikely
32572 (fmtresult::adjust_for_width_or_precision): New function.
32573 (fmtresult:type_max_digits): New function.
32574 (bytes_remaining): Handle likely and unlikely counters.
32575 (min_bytes_remaining): Remove.
32576 (format_percent): Simplify.
32577 (format_integer, format_floating): Set likely and unlikely counters.
32578 (get_string_length, format_character, format_string): Same.
32579 (format_plain, should_warn_p): New function.
32580 (maybe_warn): Call should_warn_p. Update diagnostic messages
32581 and handle those for all directives, including plain strings.
32582 (format_directive): Handle likely and unlikely counters.
32583 Remove unnecessary quoting from diagnostics. Add an informational
32585 (add_bytes): Remove.
32586 (pass_sprintf_length::compute_format_length): Simplify.
32587 (try_substitute_return_value): Handle likely and unlikely counters.
32589 2017-01-26 Carl Love <cel@us.ibm.com>
32591 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32592 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32594 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32597 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32598 endianess for subregs into account.
32599 * lra-constraints.c (lra_constraints): Do risky transformations
32600 always on the first iteration.
32601 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32602 last_call_used_reg_set.
32603 (process_bb_lives): Define and use last_call_used_reg_set.
32604 * lra.c (lra): Always continue after lra_constraints on the first
32607 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32609 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32611 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32613 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32615 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32616 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32617 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32618 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32619 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32620 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32621 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32622 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32623 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32625 2017-01-26 Marek Polacek <polacek@redhat.com>
32628 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32629 for the third operand.
32631 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32633 PR middle-end/79236
32634 * omp-low.c (struct omp_context): Add simt_stmt field.
32635 (scan_omp_for): Return omp_context *.
32636 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32637 context to the _simt_ SIMD stmt.
32638 (lower_omp_for): For combined SIMD with sibling _simt_
32639 SIMD, make sure to use the same decls in _looptemp_
32640 clauses as in the sibling.
32642 2017-01-26 David Sherwood <david.sherwood@arm.com>
32644 PR middle-end/79212
32645 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32648 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32651 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32652 emit fld b; fld a; if possible.
32654 * brig-builtins.def: Update copyright years.
32655 * config/arm/arm_acle_builtins.def: Update copyright years.
32657 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32660 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32661 constraint instead of o for the stxsd instruction.
32663 2017-01-25 Carl Love <cel@us.ibm.com>
32665 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32666 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32668 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32670 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32672 2017-01-25 Richard Biener <rguenther@suse.de>
32674 PR tree-optimization/69264
32675 * target.def (vector_alignment_reachable): Improve documentation.
32676 * doc/tm.texi: Regenerate.
32677 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32679 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32680 earlier changes with respect to TYPE_USER_ALIGN.
32681 (vector_alignment_reachable_p): Likewise. Improve dumping.
32683 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32686 * config/arm/arm.md (xordi3): Force constant operand into a register
32689 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32691 * doc/invoke.texi (-fstore-merging): Correct default optimization
32692 levels at which it is enabled.
32693 (-O): Move -fstore-merging from list to...
32696 2017-01-25 Richard Biener <rguenther@suse.de>
32699 * omp-expand.c: Include debug.h.
32700 (expand_omp_taskreg): Make sure to generate early debug before
32701 outlining anything from a function.
32702 (expand_omp_target): Likewise.
32703 (grid_expand_target_grid_body): Likewise.
32705 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32708 * asan.c (get_translation_unit_decl): New function.
32709 (asan_add_global): Extract modules file name from globals
32710 TRANSLATION_UNIT_DECL name.
32712 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32715 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32716 for long calls with APCS frame and VFP.
32718 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32720 * cfg.c (original_copy_tables_initialized_p): New function.
32721 * cfg.h (original_copy_tables_initialized_p): New decl.
32722 * cfgrtl.c (relink_block_chain): Guard the call to
32723 free_original_copy_tables with a call to
32724 original_copy_tables_initialized_p.
32725 * cgraph.h (symtab_node::native_rtl_p): New decl.
32726 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32727 (symtab_node::needed_p): Don't assert for early assembly output
32728 for __RTL functions.
32729 (cgraph_node::finalize_function): Set "force_output" for __RTL
32731 (cgraph_node::analyze): Bail out early for __RTL functions.
32732 (analyze_functions): Update assertion to support __RTL functions.
32733 (cgraph_node::expand): Bail out early for __RTL functions.
32734 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32736 * function.h (struct function): Update comment for field
32738 * gimple-expr.c: Include "tree-pass.h".
32739 (gimple_has_body_p): Return false for __RTL functions.
32740 * Makefile.in (OBJS): Add run-rtl-passes.o.
32741 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32743 (gcc::pass_manager::get_clean_slate): New accessor.
32744 * passes.c: Include "insn-addr.h".
32745 (should_skip_pass_p): Add logging. Update logic for running
32746 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32747 property-provider override so it is only done for gimple passes.
32749 (skip_pass): New function.
32750 (execute_one_pass): Call skip_pass when skipping passes.
32751 * read-md.c (md_reader::read_char): Support filtering
32752 the input to a subset of line numbers.
32753 (md_reader::md_reader): Initialize fields
32754 m_first_line and m_last_line.
32755 (md_reader::read_file_fragment): New function.
32756 * read-md.h (md_reader::read_file_fragment): New decl.
32757 (md_reader::m_first_line): New field.
32758 (md_reader::m_last_line): New field.
32759 * read-rtl-function.c (function_reader::create_function): Only
32760 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32761 curr_properties. Set DECL_INITIAL to a dummy block.
32762 (read_rtl_function_body_from_file_range): New function.
32763 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32765 * run-rtl-passes.c: New file.
32766 * run-rtl-passes.h: New file.
32768 2017-01-24 Jeff Law <law@redhat.com>
32770 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32773 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32775 PR tree-optimization/79159
32776 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32777 (record_nonwrapping_iv): Improve boundary using above function if no
32778 value range information.
32780 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32781 Martin Jambor <mjambor@suse.cz>
32783 * brig-builtins.def: New file.
32784 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32785 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32786 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32787 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32788 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32789 * builtin-types.def (BT_INT8): New.
32790 (BT_INT16): Likewise.
32791 (BT_UINT8): Likewise.
32792 (BT_UINT16): Likewise.
32793 (BT_FN_ULONG): Likewise.
32794 (BT_FN_UINT_INT): Likewise.
32795 (BT_FN_UINT_ULONG): Likewise.
32796 (BT_FN_UINT_LONG): Likewise.
32797 (BT_FN_UINT_PTR): Likewise.
32798 (BT_FN_ULONG_PTR): Likewise.
32799 (BT_FN_INT8_FLOAT): Likewise.
32800 (BT_FN_INT16_FLOAT): Likewise.
32801 (BT_FN_UINT32_FLOAT): Likewise.
32802 (BT_FN_UINT16_FLOAT): Likewise.
32803 (BT_FN_UINT8_FLOAT): Likewise.
32804 (BT_FN_UINT64_FLOAT): Likewise.
32805 (BT_FN_UINT16_UINT32): Likewise.
32806 (BT_FN_UINT32_UINT16): Likewise.
32807 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32808 (BT_FN_INT_PTR_INT): Likewise.
32809 (BT_FN_UINT_PTR_UINT): Likewise.
32810 (BT_FN_LONG_PTR_LONG): Likewise.
32811 (BT_FN_ULONG_PTR_ULONG): Likewise.
32812 (BT_FN_VOID_UINT64_UINT64): Likewise.
32813 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32814 (BT_FN_INT8_INT8_INT8): Likewise.
32815 (BT_FN_INT16_INT16_INT16): Likewise.
32816 (BT_FN_INT_INT_INT): Likewise.
32817 (BT_FN_UINT_FLOAT_UINT): Likewise.
32818 (BT_FN_FLOAT_UINT_UINT): Likewise.
32819 (BT_FN_ULONG_UINT_UINT): Likewise.
32820 (BT_FN_ULONG_UINT_PTR): Likewise.
32821 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32822 (BT_FN_UINT_UINT_UINT): Likewise.
32823 (BT_FN_VOID_UINT_PTR): Likewise.
32824 (BT_FN_UINT_UINT_PTR: Likewise.
32825 (BT_FN_UINT32_UINT64_PTR): Likewise.
32826 (BT_FN_INT_INT_UINT_UINT): Likewise.
32827 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32828 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32829 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32830 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32831 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32832 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32833 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32834 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32835 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32836 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32837 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32838 * doc/frontends.texi: List BRIG FE.
32839 * doc/install.texi (Testing): Add BRIG tesring requirements.
32840 * doc/invoke.texi (Overall Options): Mention BRIG.
32841 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32843 2017-01-24 Richard Biener <rguenther@suse.de>
32845 PR translation/79208
32846 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32848 2017-01-24 Martin Jambor <mjambor@suse.cz>
32851 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32852 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32853 and known_contexts.
32855 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32857 PR middle-end/79123
32858 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32859 casts from signed to unsigned really don't have a range.
32861 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32863 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32864 GMP_RNDx for compatiblity.
32866 2017-01-24 Martin Liska <mliska@suse.cz>
32869 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32870 that would prevent us to call alloca with -1 as argument.
32872 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32874 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32875 Avoid -Wformat-security warning.
32877 2017-01-23 Andrew Pinski <apinski@cavium.com>
32879 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32882 2017-01-23 Martin Sebor <msebor@redhat.com>
32884 PR middle-end/78703
32885 * gimple-ssa-sprintf.c (warn_level): New global.
32886 (format_integer): Use it here and throughout the rest of the file.
32887 Use the same switch to compute sign as base.
32888 (maybe_warn): New function.
32889 (format_directive): Factor out warnings into maybe_warn.
32890 Add debugging output. Use warn_level.
32891 (add_bytes): Use warn_level.
32892 (pass_sprintf_length::compute_format_length): Add debugging output.
32893 (try_substitute_return_value): Same.
32894 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32896 PR middle-end/78703
32897 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32898 (struct fmtresult, format_integer, format_floating): Adjust.
32899 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32900 (get_string_length, format_string,format_directive): Same.
32901 (pass_sprintf_length::compute_format_length): Same.
32902 (try_substitute_return_value): Simplify slightly.
32904 PR middle-end/78703
32905 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32906 (fmtresult::operator+=): Outlined.
32907 (struct fmtresult): Add ctors.
32908 (struct conversion_spec): Rename...
32909 (struct directive): ...to this. Add and remove data members.
32910 (directive::set_width, directive::set_precision): New functions.
32911 (format_percent): Use fmtresult ctor.
32912 (get_width_and_precision): Remove.
32913 (format_integer): Make naming changes. Avoid computing width and
32915 (format_floating): Same. Adjust indentation.
32916 (format_character, format_none): New functions.
32917 (format_string): Moved character handling to format_character.
32918 (format_directive): Remove arguments, change return type.
32919 (parse_directive): New function.
32920 (pass_sprintf_length::compute_format_length): Move directive
32921 parsing to parse_directive.
32923 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32925 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32926 (assign_assembler_name_if_needed): ... this.
32927 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32928 (assign_assembler_name_if_needed): ... this.
32929 (free_lang_data_in_cgraph): Adjust callers.
32930 * cgraphunit.c (cgraph_node::analyze): Likewise.
32931 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32934 2017-01-23 Richard Biener <rguenther@suse.de>
32936 PR tree-optimization/79088
32937 PR tree-optimization/79188
32938 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32939 resetting loop bounds after last path deletion. Reset loop
32940 bounds of the target loop, make code match the comments.
32941 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32942 Make sure loops need no fixups.
32944 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32946 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32947 exponent support with double type for first argument.
32948 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32949 type returned by __builtin_vec_extract_sig,
32950 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32951 functions from "vector int" to "vector unsigned int" or from
32952 "vector long long int" to "vector unsigned long long int".
32953 Changed type returned by __builtin_vec_extract_exp,
32954 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32955 functions from "vector int" to "vector unsigned int" or from
32956 "vector long long int" to "vector unsigned long long int".
32957 Changed return type of __builtin_vec_test_data_class,
32958 __builtin_vec_test_data_class_sp, and
32959 __builtin_vec_test_data_class_dp from "vector int" to
32960 "vector bool int" or from "vector long long int" to "vector bool
32961 long long int" and changed second argument type from "unsigned
32962 int" to "int". Added new overloaded function forms "vector float
32963 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32964 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32965 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32966 double, vector unsigned long long int)" and "vector double
32967 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32968 long int)". Changed return type of
32969 __builtin_scalar_test_data_class and
32970 __builtin_scalar_test_data_class_sp and
32971 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32972 int" and changed second argument from "unsigned int" to "int".
32973 Changed type returned by __builtin_scalar_test_neg,
32974 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32975 from "int" to "bool int". Added new overloaded function form
32976 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32977 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32978 exponent double-precision with floating point first argument.
32979 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32980 documentation of scalar_test_data_class, scalar_test_neg,
32981 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32982 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32983 vec_test_data_class built-in functions to reflect refinements in
32984 their type signatures.
32986 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32988 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32990 (aarch64_elf_asm_destructor): Likewise.
32992 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32994 PR rtl-optimization/78634
32995 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32996 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32997 * ifcvt.c (noce_try_cmove): Add missing cost check.
32999 PR rtl-optimization/71724
33000 * combine.c (if_then_else_cond): Look for situations where it is
33001 beneficial to undo the work of one of the recursive calls.
33003 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33005 PR tree-optimization/70754
33006 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33007 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33008 combined stmt before it if not NULL.
33009 (combine_chains): Process refs reversely and compute dominance point
33012 2017-01-23 Martin Liska <mliska@suse.cz>
33014 PR tree-optimization/79196
33015 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33016 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33018 (strlen_optimize_stmt): Call the renamed function.
33020 2017-01-23 Michael Matz <matz@suse.de>
33022 PR tree-optimization/78384
33023 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33025 2017-01-23 Richard Biener <rguenther@suse.de>
33027 PR tree-optimization/79186
33028 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33029 both incoming edges before moving an assert.
33031 2017-01-23 Martin Jambor <mjambor@suse.cz>
33033 * ipa-prop.c (load_from_param_1): Removed.
33034 (load_from_unmodified_param): Bits from load_from_param_1 put back
33036 (load_from_param): Removed.
33037 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33038 with stmt. Reverted back to use of load_from_unmodified_param.
33040 2017-01-23 Martin Jambor <mjambor@suse.cz>
33043 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33044 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33045 field a pointer to garbage collected vector, mark lattices and
33046 ipcp_orig_node with GTY((skip)).
33047 (ipa_get_param_count): Adjust to descriptors being a pointer.
33048 (ipa_get_param): Likewise.
33049 (ipa_get_type): Likewise.
33050 (ipa_get_param_move_cost): Likewise.
33051 (ipa_set_param_used): Likewise.
33052 (ipa_get_controlled_uses): Likewise.
33053 (ipa_set_controlled_uses): Likewise.
33054 (ipa_is_param_used): Likewise.
33055 (ipa_node_params_t): Move into garbage collector. New methods insert
33057 (ipa_node_params_sum): Annotate wth GTY(()).
33058 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33060 (ipa_load_from_parm_agg): Adjust declaration.
33061 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33062 * ipa-profile.c (ipa_profile): Likewise.
33063 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33064 (ipa_populate_param_decls): Make descriptors parameter garbage
33066 (ipa_dump_param): Adjust to descriptors being a pointer.
33067 (ipa_alloc_node_params): Likewise.
33068 (ipa_initialize_node_params): Likewise.
33069 (load_from_param_1): Make descriptors parameter garbage collected.
33070 (load_from_unmodified_param): Likewise.
33071 (load_from_param): Likewise.
33072 (ipa_load_from_parm_agg): Likewise.
33073 (ipa_node_params::~ipa_node_params): Removed.
33074 (ipa_free_all_node_params): Remove call to delete operator.
33075 (ipa_node_params_t::insert): New.
33076 (ipa_node_params_t::remove): Likewise.
33077 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33078 copy known_csts and known_contexts vectors.
33079 (ipa_read_node_info): Adjust to descriptors being a pointer.
33080 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33082 (ipcp_transform_function): Make descriptors variable garbage
33085 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33087 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33088 * config/i386/avx512dqintrin.h: Ditto.
33089 * config/i386/avx512fintrin.h: Ditto.
33090 * config/i386/i386.c: Handle new builtins.
33091 * config/i386/i386-builtin.def: Add new builtins.
33092 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33093 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33095 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33096 Martin Liska <mliska@suse.cz>
33098 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33099 * asan.c (asan_expand_poison_ifn): Support stores and use
33100 appropriate ASAN report function.
33101 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33102 * internal-fn.def (ASAN_POISON_USE): Declare.
33103 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33104 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33105 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33106 ASAN_POISON calls w/o LHS.
33107 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33108 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33109 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33110 * gimplify.c (asan_poison_variables): Add attribute
33111 use_after_scope_memory to variables that really needs to live
33113 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33114 having the attribute.
33116 2017-01-23 Martin Liska <mliska@suse.cz>
33118 * asan.c (create_asan_shadow_var): New function.
33119 (asan_expand_poison_ifn): Likewise.
33120 * asan.h (asan_expand_poison_ifn): New declaration.
33121 * internal-fn.c (expand_ASAN_POISON): Likewise.
33122 * internal-fn.def (ASAN_POISON): New builtin.
33123 * sanopt.c (pass_sanopt::execute): Expand
33124 asan_expand_poison_ifn.
33125 * tree-inline.c (copy_decl_for_dup_finish): Make function
33127 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33128 * tree-ssa.c (is_asan_mark_p): New function.
33129 (execute_update_addresses_taken): Rewrite local variables
33130 (identified just by use-after-scope as addressable) into SSA.
33132 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33134 * doc/install.texi (Specific): opensource.apple.com uses https
33135 now. Remove trailing slash.
33137 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33139 * README.Portability: Remove note on an Irix compatibility issue.
33141 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33143 * gcov.c (INCLUDE_ALGORITHM): Define.
33144 (INCLUDE_VECTOR): Define.
33145 No longer include <vector> and <algorithm> directly.
33147 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33149 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33151 * doc/invoke.texi (Code Gen Options): Ditto.
33153 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33156 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33158 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33160 rtl-optimization/79125
33161 * cprop.c (local_cprop_pass): Handle cases where we make an
33162 unconditional trap.
33164 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33168 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33169 read from, for big endian.
33171 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33173 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33174 register pauth builtins for LP64 only.
33176 2017-01-20 Marek Polacek <polacek@redhat.com>
33179 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33182 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33184 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33186 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33187 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33188 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33190 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33193 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33194 in big-endian mode when they are not a single duplicated value.
33196 2017-01-20 Richard Biener <rguenther@suse.de>
33198 * BASE-VER: Bump to 7.0.1.
33200 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33202 * omp-low.c (omplow_simd_context): New struct. Use it...
33203 (lower_rec_simd_input_clauses): ...here and...
33204 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33205 references to idx, lane, max_vf, is_simt.
33207 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33209 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33212 2017-01-20 Martin Jambor <mjambor@suse.cz>
33214 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33215 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33217 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33218 (GTFILES): Rename hsa.c to hsa-common.c.
33219 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33220 * hsa-dump.c: Likewise.
33221 * hsa-gen.c: Likewise.
33222 * hsa-regalloc.c: Likewise.
33223 * ipa-hsa.c: Likewise.
33224 * omp-expand.c: Likewise.
33225 * omp-low.c: Likewise.
33226 * toplev.c: Likewise.
33228 2017-01-20 Marek Polacek <polacek@redhat.com>
33231 * doc/invoke.texi: Document -Wduplicated-branches.
33232 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33233 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33234 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33235 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33236 return 0 only when not OEP_LEXICOGRAPHIC.
33237 (fold_build_cleanup_point_expr): Use the expression
33238 location when building CLEANUP_POINT_EXPR.
33239 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33240 * tree.c (add_expr): Handle error_mark_node.
33242 2017-01-20 Martin Liska <mliska@suse.cz>
33245 * tree-profile.c (init_ic_make_global_vars): Do not call
33247 (gimple_init_gcov_profiler): Likewise.
33249 2017-01-20 Martin Liska <mliska@suse.cz>
33252 * cgraph.h (maybe_create_reference): Remove argument and
33254 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33256 * ipa-cp.c (create_specialized_node): Likewise.
33257 * symtab.c (symtab_node::maybe_create_reference): Handle
33258 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33260 2017-01-20 Martin Liska <mliska@suse.cz>
33262 * read-rtl-function.c (function_reader::create_function): Use
33263 build_decl instread of build_decl_stat.
33265 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33267 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33268 * config/i386/avx512dqintrin.h: Ditto.
33269 * config/i386/avx512fintrin.h: Ditto.
33270 * config/i386/i386-builtin-types.def: Add new types.
33271 * config/i386/i386.c: Handle new types.
33272 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33273 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33274 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33275 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33276 (__builtin_ia32_kshiftridi): New.
33277 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33279 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33283 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33284 define to rs6000_init_stack_protect_guard.
33285 (rs6000_init_stack_protect_guard): New function.
33287 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33288 Yunqiang Su <yunqiang.su@imgtec.com>
33290 * config.gcc (supported_defaults): Add madd4.
33291 (with_madd4): Add validation.
33292 (all_defaults): Add madd4.
33293 * config/mips/mips.opt (mmadd4): New option.
33294 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33296 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33298 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33299 (ISA_HAS_FUSED_MADD4): Likewise.
33300 * doc/invoke.texi (-mmadd4): Document the new option.
33301 * doc/install.texi (--with-madd4): Document the new option.
33303 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33305 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33306 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33307 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33308 (aarch64_init_pauth_hint_builtins): New.
33309 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33310 (aarch64_expand_builtin): Expand new builtins.
33312 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33314 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33315 * combine-stack-adj.c (no_unhandled_cfa): Handle
33316 REG_CFA_TOGGLE_RA_MANGLE.
33317 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33318 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33319 info for return address signing.
33320 (aarch64_expand_epilogue): Likewise.
33322 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33324 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33325 * config/aarch64/aarch64-protos.h
33326 (aarch64_return_address_signing_enabled): New declaration.
33327 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33329 (aarch64_expand_prologue): Sign return address before it's pushed onto
33331 (aarch64_expand_epilogue): Authenticate return address fetched from
33333 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33334 (aarch64_attributes): New function attributes for "sign-return-address".
33335 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33336 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33337 ("*do_return"): Generate combined instructions according to key index.
33338 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33339 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33341 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33342 * config/aarch64/aarch64.opt (msign-return-address=): New.
33343 * doc/extend.texi (AArch64 Function Attributes): Documents
33344 "sign-return-address=".
33345 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33347 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33349 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33350 overall option summary.
33352 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33354 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33355 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33356 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33357 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33359 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33361 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33362 -mpower9-minmax by default for -mcpu=power9.
33363 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33364 128-bit floating point.
33366 2017-01-20 Alan Modra <amodra@gmail.com>
33368 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33369 optimizing for size.
33371 2017-01-20 Alan Modra <amodra@gmail.com>
33374 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33375 for strcmp and strncmp from corresponding builtin decl.
33377 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33379 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33380 instead of i386/rtems-64.h.
33381 * config/i386/rtems-64.h: Remove.
33383 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33387 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33389 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33391 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33393 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33394 Change int to HOST_WIDE_INT.
33395 * config/aarch64/aarch64-protos.h
33396 (aarch64_simd_gen_const_vector_dup): Likewise.
33397 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33399 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33401 * langhooks-def.h (lhd_type_for_size): New decl.
33402 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33403 * langhooks.c (lhd_type_for_size): New function, taken from
33406 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33408 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33409 define_bypass for CR latency.
33410 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33411 (power9-alu2): Add define_bypass for CR latency.
33413 (power9-mul): Update insn latency.
33414 (power9-mul-compare): Update insn latency, bypass latency and remove
33417 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33419 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33421 * config/aarch64/aarch64.md
33422 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33423 aarch64_nopcrelative_literal_loads.
33424 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33426 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33428 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33429 TARGET_LOONGSON_3A.
33430 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33432 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33435 * config.gcc (supported_defaults): Add lxc1-sxc1.
33436 (with_lxc1_sxc1): Add validation.
33437 (all_defaults): Add lxc1-sxc1.
33438 * config/mips/mips.opt (mlxc1-sxc1): New option.
33439 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33441 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33442 __mips_no_lxc1_sxc1.
33443 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33444 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33445 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33447 2017-01-19 Richard Biener <rguenther@suse.de>
33449 PR tree-optimization/72488
33450 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33451 sure to restore SSA info.
33452 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33454 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33456 PR rtl-optimization/79121
33457 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33458 of the inner type when shifting an extended value.
33460 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33463 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33464 interposable aliases.
33466 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33469 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33470 Use the evmergelohi instruction.
33471 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33472 (mov_sitf_e500_subreg8_2_be): Likewise.
33473 (mov_sitf_e500_subreg12_2_le): Likewise.
33474 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33475 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33476 (mov_sitf_e500_subreg8_2_le): Likewise.
33477 (mov_sitf_e500_subreg12_2_be): Likewise.
33479 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33481 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33482 attribute from vecsimple to vecperm.
33483 (altivec_vbpermq2): Likewise.
33485 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33488 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33490 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33491 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33492 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33493 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33494 case where N arg is SIZE_MAX.
33495 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33496 (cmpstrsi): Add pattern.
33498 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33500 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33501 __builtin_vec_revb builtins.
33502 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33503 built-in functions to support generation of the ISA 3.0 XXBR<x>
33504 vector byte reverse instructions.
33505 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33506 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33507 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33508 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33509 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33510 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33511 (P9V_BUILTIN_VEC_REVB): Likewise.
33512 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33513 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33514 (p9_xxbrq_v16qi): Likewise.
33515 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33516 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33517 (p9_xxbrh_v8hi): Likewise.
33518 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33519 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33520 vec_revb built-in functions.
33522 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33524 PR rtl-optimization/78952
33525 * config/i386/i386.md (any_extract): New code iterator.
33526 (*insvqi_2): Use any_extract for source operand.
33527 (*insvqi_3): Use any_shiftrt for source operand.
33529 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33531 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33533 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33535 2017-01-18 Matthias Klose <doko@ubuntu.com>
33537 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33539 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33541 * config/rs6000/altivec.h (vec_bperm): Change #define.
33542 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33543 (altivec_vbpermq2): New define_insn.
33544 (altivec_vbpermd): Likewise.
33545 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33546 function interface.
33547 (VBPERMD): Likewise.
33548 (VBPERM): New polymorphic function interface.
33549 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33550 Add entries for P9V_BUILTIN_VEC_VBPERM.
33551 * doc/extend.texi: Add interfaces for vec_bperm.
33553 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33555 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33556 first letter of error messages.
33557 (s390_resolve_overloaded_builtin): Likewise.
33558 * config/s390/s390.c (s390_expand_builtin): Likewise.
33559 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33560 (s390_valid_target_attribute_inner_p): Likewise.
33561 * config/s390/s390.md ("tabort"): Likewise.
33563 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33565 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33566 (ISA_AVOID_DIV_HILO): New macro.
33567 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33568 (ISA_HAS_DDIV): Likewise.
33570 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33572 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33574 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33576 * doc/invoke.texi (fabi-version): Spelling fix.
33578 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33581 * doc/invoke.texi (fabi-version): Mention mangling fix for
33584 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33587 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33589 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33592 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33595 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33597 (stack_protect_test): Ditto.
33598 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33599 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33601 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33602 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33603 -mstack-protector-guard-offset=.
33604 (RS/6000 and PowerPC Options): Ditto.
33606 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33608 * config/i386/i386.h (MASK_CLASS_P): New define.
33609 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33610 there are no registers from different register sets also when
33611 mask registers are used. Update function comment.
33612 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33613 to (*k/*r) and (*k/*km) alternatives.
33615 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33617 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33618 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33619 (EH_RETURN_HANDLER_RTX): New define.
33620 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33621 Force frame pointer in EH return functions.
33622 (aarch64_expand_epilogue): Add barrier for eh_return.
33623 (aarch64_final_eh_return_addr): Remove.
33624 (aarch64_eh_return_handler_rtx): New function.
33625 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33627 (aarch64_eh_return_handler_rtx): New prototype.
33629 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33631 * config/rs6000/altivec.h (vec_rlmi): New #define.
33632 (vec_vrlnm): Likewise.
33633 (vec_rlnm): Likewise.
33634 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33635 (UNSPEC_VRLNM): Likewise.
33636 (VIlong): New mode iterator.
33637 (altivec_vrl<VI_char>mi): New define_insn.
33638 (altivec_vrl<VI_char>nm): Likewise.
33639 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33641 (VRLDNM): Likewise.
33642 (RLNM): New polymorphic function entry.
33643 (VRLWMI): New monomorphic function entry.
33644 (VRLDMI): Likewise.
33645 (RLMI): New polymorphic function entry.
33646 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33647 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33648 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33651 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33654 * dwarf2out.c (field_byte_offset): Restore the
33655 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33656 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33657 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33660 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33663 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33665 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33668 * dwarf2out.c (add_data_member_location_attribute): For constant
33669 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33670 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33671 DW_AT_byte_size attributes.
33673 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33675 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33676 after forcing to constant memory when the code model is medium.
33678 2017-01-17 Julia Koval <julia.koval@intel.com>
33681 * config/i386/avx512fintrin.h
33682 (_mm512_i32gather_ps): Change __addr type to void const*.
33683 (_mm512_mask_i32gather_ps): Ditto.
33684 (_mm512_i32gather_pd): Ditto.
33685 (_mm512_mask_i32gather_pd): Ditto.
33686 (_mm512_i64gather_ps): Ditto.
33687 (_mm512_mask_i64gather_ps): Ditto.
33688 (_mm512_i64gather_pd): Ditto.
33689 (_mm512_mask_i64gather_pd): Ditto.
33690 (_mm512_i32gather_epi32): Ditto.
33691 (_mm512_mask_i32gather_epi32): Ditto.
33692 (_mm512_i32gather_epi64): Ditto.
33693 (_mm512_mask_i32gather_epi64): Ditto.
33694 (_mm512_i64gather_epi32): Ditto.
33695 (_mm512_mask_i64gather_epi32): Ditto.
33696 (_mm512_i64gather_epi64): Ditto.
33697 (_mm512_mask_i64gather_epi64): Ditto.
33698 (_mm512_i32scatter_ps): Change __addr type to void*.
33699 (_mm512_mask_i32scatter_ps): Ditto.
33700 (_mm512_i32scatter_pd): Ditto.
33701 (_mm512_mask_i32scatter_pd): Ditto.
33702 (_mm512_i64scatter_ps): Ditto.
33703 (_mm512_mask_i64scatter_ps): Ditto.
33704 (_mm512_i64scatter_pd): Ditto.
33705 (_mm512_mask_i64scatter_pd): Ditto.
33706 (_mm512_i32scatter_epi32): Ditto.
33707 (_mm512_mask_i32scatter_epi32): Ditto.
33708 (_mm512_i32scatter_epi64): Ditto.
33709 (_mm512_mask_i32scatter_epi64): Ditto.
33710 (_mm512_i64scatter_epi32): Ditto.
33711 (_mm512_mask_i64scatter_epi32): Ditto.
33712 (_mm512_i64scatter_epi64): Ditto.
33713 (_mm512_mask_i64scatter_epi64): Ditto.
33714 * config/i386/avx512pfintrin.h
33715 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33716 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33717 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33718 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33719 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33720 (_mm512_prefetch_i32scatter_ps): Ditto.
33721 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33722 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33723 (_mm512_prefetch_i64scatter_pd): Ditto.
33724 (_mm512_prefetch_i64scatter_ps): Ditto.
33725 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33726 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33727 * config/i386/avx512vlintrin.h
33728 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33729 (_mm_mmask_i32gather_ps): Ditto.
33730 (_mm256_mmask_i32gather_pd): Ditto.
33731 (_mm_mmask_i32gather_pd): Ditto.
33732 (_mm256_mmask_i64gather_ps): Ditto.
33733 (_mm_mmask_i64gather_ps): Ditto.
33734 (_mm256_mmask_i64gather_pd): Ditto.
33735 (_mm_mmask_i64gather_pd): Ditto.
33736 (_mm256_mmask_i32gather_epi32): Ditto.
33737 (_mm_mmask_i32gather_epi32): Ditto.
33738 (_mm256_mmask_i32gather_epi64): Ditto.
33739 (_mm_mmask_i32gather_epi64): Ditto.
33740 (_mm256_mmask_i64gather_epi32): Ditto.
33741 (_mm_mmask_i64gather_epi32): Ditto.
33742 (_mm256_mmask_i64gather_epi64): Ditto.
33743 (_mm_mmask_i64gather_epi64): Ditto.
33744 (_mm256_i32scatter_ps): Change __addr type to void*.
33745 (_mm256_mask_i32scatter_ps): Ditto.
33746 (_mm_i32scatter_ps): Ditto.
33747 (_mm_mask_i32scatter_ps): Ditto.
33748 (_mm256_i32scatter_pd): Ditto.
33749 (_mm256_mask_i32scatter_pd): Ditto.
33750 (_mm_i32scatter_pd): Ditto.
33751 (_mm_mask_i32scatter_pd): Ditto.
33752 (_mm256_i64scatter_ps): Ditto.
33753 (_mm256_mask_i64scatter_ps): Ditto.
33754 (_mm_i64scatter_ps): Ditto.
33755 (_mm_mask_i64scatter_ps): Ditto.
33756 (_mm256_i64scatter_pd): Ditto.
33757 (_mm256_mask_i64scatter_pd): Ditto.
33758 (_mm_i64scatter_pd): Ditto.
33759 (_mm_mask_i64scatter_pd): Ditto.
33760 (_mm256_i32scatter_epi32): Ditto.
33761 (_mm256_mask_i32scatter_epi32): Ditto.
33762 (_mm_i32scatter_epi32): Ditto.
33763 (_mm_mask_i32scatter_epi32): Ditto.
33764 (_mm256_i32scatter_epi64): Ditto.
33765 (_mm256_mask_i32scatter_epi64): Ditto.
33766 (_mm_i32scatter_epi64): Ditto.
33767 (_mm_mask_i32scatter_epi64): Ditto.
33768 (_mm256_i64scatter_epi32): Ditto.
33769 (_mm256_mask_i64scatter_epi32): Ditto.
33770 (_mm_i64scatter_epi32): Ditto.
33771 (_mm_mask_i64scatter_epi32): Ditto.
33772 (_mm256_i64scatter_epi64): Ditto.
33773 (_mm256_mask_i64scatter_epi64): Ditto.
33774 (_mm_i64scatter_epi64): Ditto.
33775 (_mm_mask_i64scatter_epi64): Ditto.
33776 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33777 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33778 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33779 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33780 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33781 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33782 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33783 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33784 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33785 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33786 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33787 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33788 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33789 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33790 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33791 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33792 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33793 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33794 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33795 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33796 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33797 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33798 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33799 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33800 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33801 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33802 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33803 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33804 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33805 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33806 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33807 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33808 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33809 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33810 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33811 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33812 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33813 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33814 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33815 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33816 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33817 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33818 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33819 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33820 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33821 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33822 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33823 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33824 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33825 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33826 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33827 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33828 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33829 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33830 definitions accordingly.
33832 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33833 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33836 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33839 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33842 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33843 hard regs for inner regno.
33845 2017-01-17 Martin Liska <mliska@suse.cz>
33848 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33849 assumption and add comment.
33851 2017-01-17 Nathan Sidwell <nathan@acm.org>
33853 * ipa-visibility.c (localize_node): New function, broken out of ...
33854 (function_and_variable_visibility): ... here. Call it.
33856 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33858 PR middle-end/77445
33859 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33860 correctly set frequency of oudgoing edge.
33861 (duplicate_thread_path): Fix profile updating.
33863 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33866 * configure.ac: Add GCC_BASE_VER.
33867 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33868 version from BASE-VER file.
33869 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33870 (gcc.o): Depend on $(BASEVER).
33871 * common.opt (dumpfullversion): New option.
33872 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33873 * doc/invoke.texi: Document -dumpfullversion.
33874 * doc/install.texi: Document --with-gcc-major-version-only.
33875 * configure: Regenerated.
33877 2017-01-17 Richard Biener <rguenther@suse.de>
33879 PR tree-optimization/71433
33880 * tree-vrp.c (register_new_assert_for): Merge same asserts
33881 on all incoming edges.
33882 (process_assert_insertions_for): Handle insertions at the
33885 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33887 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33888 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33890 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33893 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33896 2017-01-17 Alan Modra <amodra@gmail.com>
33899 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33900 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33901 symbolic stack limit when pic.
33903 2017-01-16 Martin Sebor <msebor@redhat.com>
33905 PR tree-optimization/78608
33906 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33908 2017-01-16 Jeff Law <law@redhat.com>
33911 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33912 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33913 for several include directories that may be relative to sysroot.
33914 * config/i386/x-mingw32 (gplus_includedir): Define.
33915 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33916 (native_system_includedir): Likewise.
33917 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33918 override if TARGET_SYSTEM_ROOT is defined.
33919 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33921 PR tree-optimization/79090
33922 PR tree-optimization/33562
33923 PR tree-optimization/61912
33924 PR tree-optimization/77485
33925 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33926 and computed trims into the dump file.
33928 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33930 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33932 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33935 * gimplify.c (gimplify_init_constructor): If want_value and
33936 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33940 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33941 sequence. Formatting fixes.
33942 (doloop_optimize): Formatting fixes.
33945 * gcc.c (debug_level_greater_than_spec_func): New function.
33946 (static_spec_functions): Add debug-level-gt spec function.
33947 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33949 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33950 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33951 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33952 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33953 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33954 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33956 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33958 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33959 QImode fixups to general and mask registers only.
33961 2017-01-16 Carl Love <cel@us.ibm.com>
33963 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33964 for built-in functions
33965 vector signed char vec_nabs (vector signed char)
33966 vector signed short vec_nabs (vector signed short)
33967 vector signed int vec_nabs (vector signed int)
33968 vector signed long long vec_nabs (vector signed long long)
33969 vector float vec_nabs (vector float)
33970 vector double vec_nabs (vector double)
33971 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33973 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33974 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33975 * doc/extend.texi: Update the documentation file for the new built-in
33978 2017-01-16 Martin Sebor <msebor@redhat.com>
33980 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33983 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33985 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33986 UNSPEC_VSX__XXSPLTD to require special splat handling.
33988 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33991 * system.h: Poison strndup.
33993 2017-01-16 Alan Modra <amodra@gmail.com>
33996 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33999 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34001 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34003 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34005 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34006 call recog here. Assert that INSN_CODE (insn) is non-negative.
34008 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34011 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34013 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34014 in the currently scheduled RTL fragment.
34016 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34018 PR rtl-optimization/78751
34019 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34022 2017-01-14 Jeff Law <law@redhat.com>
34024 PR tree-optimization/79090
34025 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34026 variable length stores.
34027 (compute_trims): Delete dead assignment to *trim_tail.
34028 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34031 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34033 PR rtl-optimization/78626
34034 PR rtl-optimization/78727
34035 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34036 of a block, and split such blocks after everything else is finished.
34038 2017-01-14 Alan Modra <amodra@gmail.com>
34041 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34042 target legitimate_combined_insn.
34043 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34044 (rs6000_legitimate_combined_insn): New function.
34045 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34047 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34048 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34049 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34051 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34053 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34055 2017-01-13 Jeff Law <law@redhat.com>
34057 PR tree-optimization/33562
34058 PR tree-optimization/61912
34059 PR tree-optimization/77485
34060 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34062 (delete_dead_assignment): Likewise.
34063 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34064 statement to delete_dead_call and delete_dead_assignment.
34066 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34069 * substring-locations.c (format_warning_va): Strengthen case 1 so
34070 that both endpoints of the substring must be within the format
34071 range for just the substring to be printed.
34073 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34075 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34076 * config/i386/i386.c (ix86_target_string): Add missing options
34077 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34078 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34079 flags_other and ix86_target_other to flags2_other. Display unknown
34081 (ix86_valid_target_attribute_inner_p): Add missing options and
34082 reorder options by implied ISAs, as in ix86_target_string.
34084 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34086 * hash-table.h (hash_table::too_empty_p): New function.
34087 (hash_table::expand): Use it.
34088 (hash_table::traverse): Likewise.
34089 (hash_table::empty_slot): Use sizeof (value_type) instead of
34090 sizeof (PTR) to convert bytes to elements. Shrink the table
34091 if the current size is excessive for the current number of
34094 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34096 * ira-costs.c (record_reg_classes): Break from the inner loop
34097 early once alt_fail is known to be true. Update outer loop
34098 handling accordingly.
34100 2017-01-13 Jeff Law <law@redhat.com>
34102 * tree-ssa-dse.c (decrement_count): New function.
34103 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34104 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34105 when we know the partially dead statement is a mem* function.
34107 PR tree-optimization/61912
34108 PR tree-optimization/77485
34109 * tree-ssa-dse.c: Include expr.h.
34110 (maybe_trim_constructor_store): New function.
34111 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34113 PR tree-optimization/33562
34114 PR tree-optimization/61912
34115 PR tree-optimization/77485
34116 * doc/invoke.texi: Document new dse-max-object-size param.
34117 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34118 * tree-ssa-dse.c: Include params.h.
34119 (dse_store_status): New enum.
34120 (initialize_ao_ref_for_dse): New, partially extracted from
34122 (valid_ao_ref_for_dse, normalize_ref): New.
34123 (setup_live_bytes_from_ref, compute_trims): Likewise.
34124 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34125 (maybe_trim_partially_dead_store): Likewise.
34126 (maybe_trim_complex_store): Likewise.
34127 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34128 Track what bytes live from the original store. Return tri-state
34129 for dead, partially dead or live.
34130 (dse_dom_walker): Add constructor, destructor and new private members.
34131 (delete_dead_call, delete_dead_assignment): New extracted from
34133 (dse_optimize_stmt): Make a member of dse_dom_walker.
34134 Use initialize_ao_ref_for_dse.
34136 PR tree-optimization/33562
34137 PR tree-optimization/61912
34138 PR tree-optimization/77485
34139 * sbitmap.h (bitmap_count_bits): Prototype.
34140 (bitmap_clear_range, bitmap_set_range): Likewise.
34141 * sbitmap.c (bitmap_clear_range): New function.
34142 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34144 2017-01-13 Martin Liska <mliska@suse.cz>
34147 * function.c (set_cfun): Add new argument force.
34148 * function.h (set_cfun): Likewise.
34149 * ipa-inline-transform.c (inline_call): Use the function when
34150 strict alising from is dropped for function we inline to.
34152 2017-01-13 Richard Biener <rguenther@suse.de>
34154 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34155 for dumping GIMPLE INTEGER_CSTs.
34157 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34159 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34160 to 201112L since C++17.
34162 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34165 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34166 if -fsanitize=kernel-address is present.
34168 2017-01-13 Richard Biener <rguenther@suse.de>
34170 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34171 as _Literal ( type ) number in case usual suffixes do not
34172 preserve all information.
34174 2017-01-13 Richard Biener <rguenther@suse.de>
34176 PR tree-optimization/77283
34177 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34178 and ssa-iterators.h.
34179 (is_feasible_trace): Implement a cost model based on joiner
34182 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34185 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34186 char or short to __float128/_Float128 directly.
34188 2017-01-12 Martin Sebor <msebor@redhat.com>
34190 to -Wformat-overflow.
34191 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34192 (min_bytes_remaining): Same.
34193 (get_string_length): Same.
34194 (format_string): Same.
34195 (format_directive): Same.
34197 (pass_sprintf_length::handle_gimple_call): Same.
34199 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34201 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34202 info.nowrite calls with no lhs that can't throw. Return bool
34203 whether gsi_remove has been called or not.
34204 (pass_sprintf_length::handle_gimple_call): Return bool whether
34205 try_substitute_return_value called gsi_remove. Formatting fix.
34206 (pass_sprintf_length::execute): Don't use gsi_remove if
34207 handle_gimple_call returned true.
34210 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34211 be removed due to side-effects, don't remove following barrier nor
34212 turn the successor edge into fallthru edge.
34214 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34217 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34218 element-reversing loads and stores as not swappable.
34220 2017-01-12 Nathan Sidwell <nathan@acm.org>
34221 Nicolai Stange <nicstange@gmail.com>
34223 * combine.c (try_combine): Don't ignore result of overlap checking
34224 loop. Combine overlap & asm check into single loop.
34226 2017-01-12 Richard Biener <rguenther@suse.de>
34228 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34229 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34231 2017-01-12 Richard Biener <rguenther@suse.de>
34233 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34234 and TS_TARGET_OPTION directly derive from TS_BASE.
34235 * tree-core.h (tree_optimization_option): Derive from tree_base.
34236 (tree_target_option): Likewise.
34238 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34240 * config/i386/i386.c (memory_address_length): Increase len
34241 only when rip_relative_addr_p returns false.
34243 2017-01-11 Julia Koval <julia.koval@intel.com>
34245 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34246 (OPTION_MASK_ISA_SGX_SET): New.
34247 (ix86_handle_option): Handle OPT_msgx.
34248 * config.gcc: Added sgxintrin.h.
34249 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34250 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34251 * config/i386/i386.c (ix86_target_string): Add -msgx.
34253 (ix86_option_override_internal): Handle new options.
34254 (ix86_valid_target_attribute_inner_p): Add sgx.
34255 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34256 * config/i386/i386.opt: Add msgx.
34257 * config/i386/sgxintrin.h: New file.
34258 * config/i386/x86intrin.h: Add sgxintrin.h.
34260 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34263 * fold-const.c (maybe_nonzero_address): Return 1 for function
34265 (tree_single_nonzero_warnv_p): Don't handle function local objects
34269 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34272 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34275 * opts.c: Include "spellcheck.h"
34276 (struct string_fragment): New struct.
34277 (struct edit_distance_traits<const string_fragment &>): New
34279 (get_closest_sanitizer_option): New function.
34280 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34282 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34284 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34286 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34287 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34288 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34289 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34290 for initial die_offset if dwarf_split_debug_info.
34291 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34292 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34294 (output_skeleton_debug_sections): Formatting fix. Use
34295 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34296 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34298 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34300 * config/arm/cortex-a53.md: Add bypasses for
34301 cortex_a53_r2f_cvt.
34302 (cortex_a53_r2f): Only use for transfers.
34303 (cortex_a53_f2r): Likewise.
34304 (cortex_a53_r2f_cvt): Add reservation for conversions.
34305 (cortex_a53_f2r_cvt): Likewise.
34307 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34309 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34310 to all inlined functions, change static to extern.
34312 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34315 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34317 (arm_assemble_integer): Likewise.
34319 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34321 * config.gcc: Use new awk script to check CPU, FPU and architecture
34322 parameters for --with-... options.
34323 * config/arm/parsecpu.awk: New file
34324 * config/arm/arm-cpus.in: New file.
34325 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34327 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34329 * config/arm/t-arm: Update dependency rules.
34330 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34331 of processing .def files.
34332 * config/arm/genopt.sh: Deleted.
34333 * config/arm/gentune.sh: Deleted.
34334 * config/arm/arm-cores.def: Deleted.
34335 * config/arm/arm-arches.def: Deleted.
34336 * config/arm/arm-fpus.def: Deleted.
34337 * config/arm/arm-tune.md: Regenerated.
34338 * config/arm/arm-tables.opt: Regenerated.
34339 * config/arm/arm-cpu.h: New generated file.
34340 * config/arm/arm-cpu-data.h: New generated file.
34341 * config/arm/arm-cpu-cdata.h: New generated file.
34343 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34346 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34348 (input_varpool_node): Unpack dynamically_initialized bit.
34350 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34352 PR rtl-optimization/79032
34353 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34354 the alignment of the adjusted memory reference against that of MODE,
34355 instead of the alignment of the original memory reference.
34357 2017-01-11 Martin Jambor <mjambor@suse.cz>
34359 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34361 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34362 decorated functions.
34364 2017-01-11 Richard Biener <rguenther@suse.de>
34366 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34367 set range/nonnull info for PHI results. Do not set it on
34368 stmts marked for removal.
34370 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34372 * expr.c (store_field): In the bitfield case, fetch the return value
34373 from the registers before applying a single big-endian adjustment.
34374 Always do a final load for a BLKmode value not larger than a word.
34376 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34379 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34380 that we correctly handle column numbers greater than
34381 LINE_MAP_MAX_COLUMN_NUMBER.
34383 2017-01-10 Martin Sebor <msebor@redhat.com>
34385 PR middle-end/78245
34386 * gimple-ssa-sprintf.c (get_destination_size): Call
34387 {init,fini}object_sizes.
34388 * tree-object-size.c (addr_object_size): Adjust.
34389 (pass_through_call): Adjust.
34390 (pass_object_sizes::execute): Adjust.
34391 * tree-object-size.h (fini_object_sizes): Declare.
34393 2017-01-10 Martin Sebor <msebor@redhat.com>
34395 PR tree-optimization/78775
34396 * builtins.c (get_size_range): Move...
34397 * calls.c: ...to here.
34398 (alloc_max_size): Accept zero argument.
34399 (operand_signed_p): Remove.
34400 (maybe_warn_alloc_args_overflow): Call get_size_range.
34401 * calls.h (get_size_range): Declare.
34403 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34405 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34406 from TI's devices.csv file as of September 2016.
34407 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34409 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34411 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34412 * doc/invoke.texi: Likewise.
34413 * doc/md.texi: Likewise.
34414 * doc/objc.texi: Likewise.
34416 2017-01-10 Joshua Conner <joshconner@google.com>
34418 * config/arm/fuchsia-elf.h: New file.
34419 * config/fuchsia.h: New file.
34420 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34421 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34423 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34425 2016-01-10 Richard Biener <rguenther@suse.de>
34427 PR tree-optimization/79034
34428 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34429 Propagate out degenerate PHIs in the joiner.
34431 2017-01-10 Martin Liska <mliska@suse.cz>
34433 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34434 (sort_congruence_classes_by_decl_uid): Likewise.
34435 (sort_congruence_class_groups_by_decl_uid): Likewise.
34436 (sem_item_optimizer::merge_classes): Sort class, groups in these
34437 classes and members in the groups by DECL_UID of declarations.
34438 This would make merge operations stable.
34440 2017-01-10 Martin Liska <mliska@suse.cz>
34442 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34443 usage of m_classes_vec.
34444 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34445 (sem_item_optimizer::get_group_by_hash): Likewise.
34446 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34447 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34448 (sem_item_optimizer::verify_classes): Likewise.
34449 (sem_item_optimizer::process_cong_reduction): Likewise.
34450 (sem_item_optimizer::dump_cong_classes): Likewise.
34451 (sem_item_optimizer::merge_classes): Likewise.
34452 * ipa-icf.h (congruence_class_hash): Rename from
34453 congruence_class_group_hash. Remove declaration of m_classes_vec.
34455 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34457 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34458 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34459 * config.gcc: Add avx512vpopcntdqintrin.h.
34460 * config/i386/avx512vpopcntdqintrin.h: New.
34461 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34462 * config/i386/i386-builtin-types.def: Add new types.
34463 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34464 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34465 __builtin_ia32_vpopcountq_v8di_mask): New.
34466 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34467 __AVX512VPOPCNTDQ__.
34468 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34469 (PTA_AVX512VPOPCNTDQ): Define.
34470 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34471 TARGET_AVX512VPOPCNTDQ_P): Define.
34472 * config/i386/i386.opt: Add mavx512vpopcntdq.
34473 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34474 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34476 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34478 PR middle-end/77484
34479 * predict.def (PRED_CALL): Set to 67.
34481 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34483 * expr.c (store_field): In the bitfield case, if the value comes from
34484 a function call and is of an aggregate type returned in registers, do
34485 not modify the field mode; extract the value in all cases if the mode
34486 is BLKmode and the size is not larger than a word.
34488 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34491 * config/i386/cpuid.h: Fix undefined behavior.
34493 2017-01-04 Jeff Law <law@redhat.com>
34495 PR tree-optimization/79007
34496 PR tree-optimization/67955
34497 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34498 conservative for pt.null when flag_non_call_exceptions is on.
34500 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34502 PR translation/79019
34503 PR translation/79020
34504 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34505 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34506 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34508 * config/avr/avr.opt (maccumulate-args): Likewise.
34509 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34510 * common.opt (freport-bug): Likewise.
34511 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34512 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34513 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34514 translatable string.
34515 * config/i386/i386.c (function_value_32): Likewise.
34516 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34517 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34519 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34520 * common/config/msp430/msp430-common.c (msp430_handle_option):
34522 * symtab.c (symtab_node::verify_base): Likewise.
34523 * opts.c (set_debug_level): Likewise.
34524 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34525 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34526 missing whitespace to translatable strings.
34527 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34528 * config/sh/superh.h: Likewise.
34529 * config/i386/xopintrin.h: Likewise.
34530 * config/i386/znver1.md: Likewise.
34531 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34532 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34533 * double-int.h (struct double_int): Likewise.
34534 * double-int.c (div_and_round_double): Likewise.
34535 * wide-int.cc: Likewise.
34536 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34537 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34538 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34539 (crossjumps_occurred): ... this.
34540 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34543 PR tree-optimization/78899
34544 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34545 returning bool return struct loop *, NULL for failure and the new
34547 (versionable_outer_loop_p): Don't version outer loop if it has
34548 dont_vectorized bit set.
34549 (tree_if_conversion): When versioning outer loop, ensure
34550 tree_if_conversion is performed also on the inner loop of the
34551 non-vectorizable outer loop copy.
34552 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34553 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34554 prevent vectorization of it.
34555 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34556 the outer loop vectorization of the non-scalar version is attempted
34557 before vectorization of the inner loop in scalar version. If
34558 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34559 vectorization of its inner loop.
34560 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34561 has 2 inner loops, rename also on edges from bb whose single pred
34562 is outer_loop->header. Fix typo in function comment.
34564 2017-01-09 Martin Sebor <msebor@redhat.com>
34567 * asan.c (asan_emit_stack_protection): Increase local buffer size
34568 to avoid snprintf truncation warning.
34570 2017-01-09 Andrew Pinski <apinski@cavium.com>
34572 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34573 to reference thunderx2t99 for the tuning structure
34574 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34576 (thunderx2t99_extra_costs): This.
34577 * config/aarch64/aarch64-tune.md: Regenerate.
34578 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34579 (vulcan_addrcost_table): This.
34580 (vulcan_regmove_cost): Rename to ...
34581 (thunderx2t99_regmove_cost): This.
34582 (vulcan_vector_cost): Rename to ...
34583 (thunderx2t99_vector_cost): this.
34584 (vulcan_branch_cost): Rename to ...
34585 (thunderx2t99_branch_cost): This.
34586 (vulcan_tunings): Rename to ...
34587 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34588 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34590 2017-01-09 Martin Jambor <mjambor@suse.cz>
34594 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34595 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34596 (propagate_vr_accross_jump_function): Use the above function for all
34597 value range computations for pass-through jump functions and type
34598 converasion from explicit value range values.
34599 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34600 parameters from TYPE_ARG_TYPES.
34601 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34602 (ipa_write_node_info): Stream type of the actual argument.
34603 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34605 2017-01-09 Martin Liska <mliska@suse.cz>
34608 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34609 (lookup_compiler): Do not show error message with have_E.
34611 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34613 PR tree-optimization/78938
34614 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34615 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34616 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34617 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34620 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34622 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34625 2017-01-09 Richard Biener <rguenther@suse.de>
34627 PR tree-optimization/78997
34628 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34629 name condition properly.
34631 2017-01-09 Richard Biener <rguenther@suse.de>
34634 * dwarf2out.c (is_cxx): New overload with context.
34635 (is_naming_typedef_decl): Use it.
34637 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34639 * invoke.texi (Option Summary): Correct spacing in option lists
34640 and add line breaks to fix over-long lines.
34642 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34644 PR middle-end/17660
34646 * extend.texi (Common Variable Attributes): Add xref to GCC
34647 Internals manual to explain mode attribute keywords.
34649 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34652 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34653 and Preprocessor Options.
34654 (Options for Linking): Document -pthread here....
34655 (RS/6000 and PowerPC Options): ...not here.
34656 (Solaris 2 Options): ...or here.
34657 * doc/cppopts.texi: Document -pthread.
34659 2017-01-08 Martin Sebor <msebor@redhat.com>
34661 PR middle-end/77708
34662 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34663 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34664 New member functions.
34665 (format_directive): Used them.
34667 (pass_sprintf_length::handle_gimple_call): Same.
34668 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34669 to avoid truncation for any argument.
34670 (extract_affine_mul): Same.
34671 * tree.c (get_file_function_name): Same.
34673 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34675 PR middle-end/77484
34676 * predict.def (PRED_INDIR_CALL): Set to 86.
34678 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34680 PR preprocessor/54124
34681 * doc/cppopts.texi: Reformat -d subtable to list the full name
34682 of the options. Add cross-reference to the docs for the general
34683 compiler -d options.
34684 * doc/invoke.texi (Developer Options): Add cross-reference to the
34685 preprocessor-specific -d option documentation.
34687 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34689 PR preprocessor/13498
34690 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34691 redudant material, and reflect new command-line options.
34692 (System Headers): Likewise.
34694 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34696 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34697 -isystem, and -idirafter. Copy-edit.
34698 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34699 default for -ftrack-macro-expansion. Delete obsolete and
34700 badly-formatted implementation details about -fdebug-cpp output.
34701 * doc/cppwarnopts.texi: Copy-edit.
34703 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34706 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34707 that the transition from a max line width >= 1<<10 to narrower
34708 lines works correctly.
34710 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34712 * doc/options.texi (PerFunction): New.
34713 * opt-functions.awk (switch_flags): Map both Optimization and
34714 PerFunction to CL_OPTIMIZATION.
34715 * opth-gen.awk: Test for PerFunction flag along with
34717 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34718 it only when the latter is present. Skip those that don't in
34719 the hash function generator.
34720 * common.opt (fvar-tracking): Mark as PerFunction instead of
34722 (fvar-tracking-assignments): Likewise.
34723 (fvar-tracking-assignments-toggle): Likewise.
34724 (fvar-tracking-uninit): Likewise.
34726 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34728 PR translation/79018
34729 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34732 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34735 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34736 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34737 TARGET_LONG_JUMP_TABLE_OFFSETS.
34738 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34739 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34740 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34741 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34742 * config/m68k/m68k.md (tablejump expander): Likewise.
34743 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34744 TARGET_LONG_JUMP_TABLE_OFFSETS.
34745 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34746 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34748 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34749 David Holsgrove <david.holsgrove@xilinx.com>
34751 * common/config/microblaze/microblaze-common.c
34752 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34753 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34755 * config/microblaze/microblaze.c (microblaze_must_save_register)
34756 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34758 (microblaze_eh_return): New function.
34759 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34760 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34761 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34762 * config/microblaze/microblaze.md (eh_return): New pattern.
34764 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34766 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34767 GCC_DIAGNOSTIC_STRINGIFY): Define.
34769 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34771 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34773 * config/arm/arm.md (<mcrr>): New.
34775 * config/arm/arm.c (arm_arch5te): New.
34776 (arm_option_override): Set arm_arch5te.
34777 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34779 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34780 (arm_mcrr_qualifiers): ... this. New.
34781 (MRRC_QUALIFIERS): Define to...
34782 (arm_mrrc_qualifiers): ... this. New.
34783 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34785 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34786 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34787 (MRRCI, mrrc, MRRC): New.
34788 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34789 VUNSPEC_MRRC2): New.
34791 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34793 * config/arm/arm.md (<mcr>): New.
34795 * config/arm/arm.c (arm_coproc_builtin_available): Add
34796 support for mcr, mrc, mcr2 and mrc2.
34797 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34798 (arm_mcr_qualifiers): ... this. New.
34799 (MRC_QUALIFIERS): Define to ...
34800 (arm_mrc_qualifiers): ... this. New.
34801 (MCR_QUALIFIERS): Define to ...
34802 (arm_mcr_qualifiers): ... this. New.
34803 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34805 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34806 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34807 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34808 VUNSPEC_MRC2): New.
34810 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34812 * config/arm/arm.md (*ldc): New.
34816 * config/arm/arm.c (arm_coproc_builtin_available): Add
34817 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34818 (arm_coproc_ldc_stc_legitimate_address): New.
34819 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34820 'qualifier_const_pointer'.
34821 (LDC_QUALIFIERS): Define to...
34822 (arm_ldc_qualifiers): ... this. New.
34823 (STC_QUALIFIERS): Define to...
34824 (arm_stc_qualifiers): ... this. New.
34825 * config/arm/arm-protos.h
34826 (arm_coproc_ldc_stc_legitimate_address): New.
34827 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34828 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34829 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34830 stc2, stcl, stc2l): New.
34831 * config/arm/constraints.md (Uz): New.
34832 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34833 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34834 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34835 VUNSPEC_STC2L): New.
34837 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34839 * config/arm/arm.md (<cdp>): New.
34840 * config/arm/arm.c (neon_const_bounds): Rename this ...
34841 (arm_const_bounds): ... this.
34842 (arm_coproc_builtin_available): New.
34843 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34844 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34845 (CDP_QUALIFIERS): Define to...
34846 (arm_cdp_qualifiers): ... this. New.
34848 (arm_expand_builtin_args): Add case for 6 arguments.
34849 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34850 (arm_const_bounds): ... this.
34851 (arm_coproc_builtin_available): New.
34852 * config/arm/arm_acle.h (__arm_cdp): New.
34854 * config/arm/arm_acle_builtins.def (cdp): New.
34856 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34857 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34858 'arm_const_bounds'.
34859 * config/arm/types.md (coproc): New.
34860 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34861 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34862 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34863 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34865 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34867 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34868 (UBINOP_QUALIFIERS): New.
34870 (acle_builtin_data): New. Change comment.
34871 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34872 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34873 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34874 arm_acle_builtins.def.
34875 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34876 (arm_init_acle_builtins): New.
34877 (CRC32_BUILTIN): Remove.
34878 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34879 crc32cb, crc32ch and crc32cw.
34880 (arm_init_crc32_builtins): Remove.
34881 (arm_init_builtins): Use arm_init_acle_builtins rather
34882 than arm_init_crc32_builtins.
34883 (arm_expand_acle_builtin): New.
34884 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34885 * config/arm/arm_acle_builtins.def: New.
34887 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34889 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34890 (arm_builtin_datum): ... this.
34891 (arm_init_neon_builtin): Rename to ...
34892 (arm_init_builtin): ... this. Add a new parameters PREFIX
34893 and USE_SIG_IN_NAME.
34894 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34895 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34896 'arm_builtin_datum'.
34897 (arm_init_vfp_builtins): Likewise.
34898 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34899 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34900 (arm_expand_neon_args): Rename to ...
34901 (arm_expand_builtin_args): ... this. Rename builtin_arg
34902 enum values and differentiate between ARG_BUILTIN_MEMORY
34903 and ARG_BUILTIN_NEON_MEMORY.
34904 (arm_expand_neon_builtin_1): Rename to ...
34905 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34906 values, arm_expand_builtin_args and add bool parameter NEON.
34907 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34908 (arm_expand_vfp_builtin): Likewise.
34909 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34911 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34913 PR middle-end/77484
34914 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34915 * predict.c (tree_estimate_probability_bb): Reverse direction of
34916 polymorphic call predictor.
34918 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34920 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34921 (determine_pass_name_match): ...this new function and...
34922 (should_skip_pass_p): ...this new function.
34924 2017-01-06 Nathan Sidwell <nathan@acm.org>
34926 * ipa-visibility.c (function_and_variable_visibility): Reformat
34927 comments and long lines. Remove extrneous if.
34928 * symtab.c (symtab_node::make_decl_local): Fix code format.
34929 (symtab_node::set_section_for_node): Fix comment typo.
34931 2017-01-06 Martin Liska <mliska@suse.cz>
34934 * lra-constraints.c: Rename invariant to lra_invariant.
34935 * predict.c (set_even_probabilities): Initialize e to NULL.
34937 2017-01-05 Martin Sebor <msebor@redhat.com>
34939 PR tree-optimization/78910
34940 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34941 (format_integer): Correct off-by-one error in the handling
34942 of precision with negative numbers in signed conversions..
34944 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34946 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34948 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34950 PR tree-optimization/71016
34951 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34952 factor_out_conditional_conversion. Formatting fix.
34953 (factor_out_conditional_conversion): Add cond_stmt argument.
34954 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34955 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34958 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34960 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34961 read-rtl-function.o, and selftest-rtl.o.
34962 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34963 (selftest::aarch64_test_loading_full_dump): New function.
34964 (selftest::aarch64_run_selftests): New function.
34965 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34966 selftest::aarch64_run_selftests.
34967 * config/i386/i386.c
34968 (selftest::ix86_test_loading_dump_fragment_1): New function.
34969 (selftest::ix86_test_loading_call_insn): New function.
34970 (selftest::ix86_test_loading_full_dump): New function.
34971 (selftest::ix86_test_loading_unspec): New function.
34972 (selftest::ix86_run_selftests): Call the new functions.
34973 * emit-rtl.c (maybe_set_max_label_num): New function.
34974 * emit-rtl.h (maybe_set_max_label_num): New decl.
34975 * function.c (instantiate_decls): Guard call to
34976 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34977 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34979 * gensupport.c (gen_reader::gen_reader): Pass "false"
34980 for new "compact" param of rtx_reader.
34981 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34982 rather than an empty string for NULL strings.
34983 * read-md.c: Potentially include config.h rather than bconfig.h.
34984 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34985 (have_error): New global, copied from errors.c.
34986 (md_reader::read_name): Rename to...
34987 (md_reader::read_name_1): ...this, adding "out_loc" param,
34988 and converting "missing name or number" to returning false, rather
34990 (md_reader::read_name): Reimplement in terms of read_name_1.
34991 (md_reader::read_name_or_nil): New function.
34992 (md_reader::read_string): Handle "(nil)" by returning NULL.
34993 (md_reader::md_reader): Add new param "compact".
34994 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34995 (md_reader::read_file): New method.
34996 * read-md.h (md_reader::md_reader): Add new param "compact".
34997 (md_reader::read_file): New method.
34998 (md_reader::is_compact): New accessor.
34999 (md_reader::read_name): Convert return type from void to file_location.
35000 (md_reader::read_name_or_nil): New decl.
35001 (md_reader::read_name_1): New decl.
35002 (md_reader::m_compact): New field.
35003 (noop_reader::noop_reader): Pass "false" for new "compact" param
35005 (rtx_reader::rtx_reader): Add new "compact" param.
35006 (rtx_reader::read_rtx_operand): Make virtual and convert return
35007 type from void to rtx.
35008 (rtx_reader::read_until): New decl.
35009 (rtx_reader::handle_any_trailing_information): New virtual function.
35010 (rtx_reader::postprocess): New virtual function.
35011 (rtx_reader::finalize_string): New virtual function.
35012 (rtx_reader::m_in_call_function_usage): New field.
35013 (rtx_reader::m_reuse_rtx_by_id): New field.
35014 * read-rtl-function.c: New file.
35015 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35016 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35017 (selftest::verify_three_block_rtl_cfg): New decl.
35018 * read-rtl-function.h: New file.
35019 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35020 For host, include function.h, memmodel.h, and emit-rtl.h.
35021 (one_time_initialization): New function.
35022 (struct compact_insn_name): New struct.
35023 (compact_insn_names): New array.
35024 (find_code): Handle insn codes in compact dumps.
35025 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35026 (bind_subst_iter_and_attr): Likewise.
35027 (add_condition_to_string): Likewise.
35028 (add_condition_to_rtx): Likewise.
35029 (apply_attribute_uses): Likewise.
35030 (add_current_iterators): Likewise.
35031 (apply_iterators): Likewise.
35032 (initialize_iterators): Guard usage of apply_subst_iterator with
35033 #ifdef GENERATOR_FILE.
35034 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35035 (md_reader::read_mapping): Likewise.
35036 (add_define_attr_for_define_subst): Likewise.
35037 (add_define_subst_attr): Likewise.
35038 (read_subst_mapping): Likewise.
35039 (check_code_iterator): Likewise.
35040 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35042 (one_time_initialization): New function.
35043 (rtx_reader::read_until): New method.
35044 (read_flags): New function.
35045 (parse_reg_note_name): New function.
35046 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35047 Handle reuse_rtx ids.
35048 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35049 Add parsing support for RTL dumps, mirroring the special-cases in
35050 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35051 values, and calling handle_any_trailing_information.
35052 (rtx_reader::read_rtx_operand): Convert return type from void
35053 to rtx, returning return_rtx. Handle case 'e'. Call
35054 finalize_string on XSTR and XTMPL fields.
35055 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35056 "(nil)" values were omitted. Call the postprocess vfunc on the
35058 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35059 class ctor. Initialize m_in_call_function_usage. Call
35060 one_time_initialization.
35061 * rtl-tests.c (selftest::test_uncond_jump): Call
35062 set_new_first_and_last_insn.
35063 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35064 * selftest-rtl.c: New file.
35065 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35066 (selftest::get_insn_by_uid): New decl.
35067 * selftest-run-tests.c (selftest::run_tests): Call
35068 read_rtl_function_c_tests.
35069 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35070 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35073 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35075 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35076 operands in a special way. Assert that pos+len <= mode precision.
35078 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35080 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35081 3 argument Alias with unlimited for the negative form.
35082 (fno-vect-cost-model): Removed.
35084 2017-01-05 Martin Liska <mliska@suse.cz>
35086 * hsa-gen.c (gen_hsa_divmod): New function.
35087 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35089 2017-01-05 Martin Liska <mliska@suse.cz>
35092 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35095 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35097 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35098 small constant length operands.
35100 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35102 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35103 between loop iterations.
35105 2017-01-05 Martin Liska <mliska@suse.cz>
35108 * gimplify.c (gimplify_decl_expr): Compare to
35109 asan_poisoned_variables instread of checking flags.
35110 (gimplify_target_expr): Likewise.
35111 (gimplify_expr): Likewise.
35112 (gimplify_function_tree): Conditionally initialize
35113 asan_poisoned_variables.
35115 2017-01-04 Jeff Law <law@redhat.com>
35117 PR tree-optimizatin/78812
35118 * rtl.h (contains_mem_rtx_p): Prototype.
35119 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35120 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35121 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35122 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35123 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35125 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35127 * input.c (assert_char_at_range): Default-initialize actual_range.
35129 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35131 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35132 to match the caller.
35134 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35136 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35137 insns after final jump in test to emit dummy move.
35139 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35141 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35142 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35144 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35146 * multiple_target.c (create_dispatcher_calls): Init e_next.
35147 * tree-ssa-loop-split.c (split_loop): Init border.
35148 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35151 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35156 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35157 (altivec_register_operand): Do not return true if the operand
35158 contains a SUBREG mixing SImode and SFmode.
35159 (vsx_register_operand): Likewise.
35160 (vsx_reg_sfsubreg_ok): New predicate.
35161 (vfloat_operand): Do not return true if the operand contains a
35162 SUBREG mixing SImode and SFmode.
35163 (vint_operand): Likewise.
35164 (vlogical_operand): Likewise.
35165 (gpc_reg_operand): Likewise.
35166 (int_reg_operand): Likewise.
35167 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35168 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35169 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35171 (rs6000_emit_move_si_sf_subreg): New helper function.
35172 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35173 fixup SUBREGs involving SImode and SFmode.
35174 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35175 numbers for the new peephole2 optimization.
35176 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35177 the GLIBC math library that do AND/IOR/XOR operations on single
35178 precision floating point.
35179 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35180 target macros to say whether we need to avoid SUBREGs mixing
35182 (TARGET_ALLOW_SF_SUBREG): Likewise.
35183 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35184 (UNSPEC_SI_FROM_SF): Likewise.
35185 (iorxor): Change spacing.
35186 (and_ior_xor): New iterator for AND, IOR, and XOR.
35187 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35188 (movdi_from_sf_zero_ext): Likewise.
35189 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35190 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35191 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35192 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35193 (fms<mode>4): Likewise.
35194 (fnma<mode>4): Likewise.
35195 (fnms<mode>4): Likewise.
35196 (nfma<mode>4): Likewise.
35197 (nfms<mode>4): Likewise.
35199 2017-01-04 Marek Polacek <polacek@redhat.com>
35202 * doc/invoke.texi: Document -Wpointer-compare.
35204 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35206 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35209 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35210 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35213 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35216 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35217 documentation of the powerpc_popcntb_ok attribute.
35218 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35219 code to issue warning messages if a requested CPU configuration is
35220 not supported by the binary (assembler and loader) toolchain.
35221 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35222 made to define a built-in function that has been disabled.
35223 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35224 made to define a built-in function that has been disabled.
35225 (altivec_init_builtins): Add comment explaining why definition
35226 of the DST built-in functions is not preceded by an assertion
35227 check. Add assertions to prevent ICE if attempts are made to
35228 define an altivec predicate or an abs* built-in function that has
35230 (htm_init_builtins): Add comment explaining why definition of the
35231 htm built-in functions is not preceded by an assertion check.
35233 2017-01-04 Jeff Law <law@redhat.com>
35235 PR tree-optimizatin/67955
35236 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35237 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35238 the points-to solution does not include pt_null. Use DECL_PT_UID
35241 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35243 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35244 Use gen_int_mode instead of gen_lopwart for const_int operands.
35246 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35248 PR tree-optimization/71563
35249 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35250 out of range value - has low bits known to be zero.
35252 2017-01-04 Alan Modra <amodra@gmail.com>
35254 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35255 * configure: Regenerate.
35256 * config.in: Regenerate.
35258 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35261 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35262 a substring of the message, but strcmp with the whole message. Ifdef
35263 ENABLE_NLS, translate the message first using dgettext.
35265 2017-01-03 Jeff Law <law@redhat.com>
35267 PR tree-optimizatin/78856
35268 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35269 (mark_threaded_blocks): Remove code to truncate thread paths that
35270 cross multiple loop headers. Instead invalidate the cached loop
35271 iteration information and handle case of a thread path walking
35272 into an irreducible region.
35274 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35277 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35278 assertions. Add support for doing the signbit if the IEEE 128-bit
35279 floating point value is in a GPR.
35280 * config/rs6000/rs6000.md (Fsignbit): Delete.
35281 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35282 Update the length attribute if the value is in a GPR.
35283 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35284 the sign or zero extension instruction, since the value is always 0/1.
35285 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35288 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35289 extracting SImode to a GPR register so that we can generate a
35290 store, limit the vector to be in a traditional Altivec register
35291 for the vextuwrx instruction.
35293 2017-01-03 Ian Lance Taylor <iant@google.com>
35295 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35297 2017-01-03 Martin Sebor <msebor@redhat.com>
35299 PR tree-optimization/78696
35300 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35301 precision. Use MPFR for %f for greater fidelity. Correct handling
35303 (pass_sprintf_length::compute_format_length): Set width and precision
35304 specified by asrerisk to void_node for vararg functions.
35305 (try_substitute_return_value): Adjust dump output.
35307 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35309 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35311 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35313 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35314 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35315 -mlra/-mno-lra was passed to the compiler.
35317 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35319 PR rtl-optimization/65618
35320 * emit-rtl.c (try_split): Move initialization of "before" and
35321 "after" to just before the call to emit_insn_after_setloc.
35323 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35325 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35327 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35329 * dwarf2out.c (gen_enumeration_type_die): When
35330 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35332 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35334 PR tree-optimization/78965
35335 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35336 Change first argument from const call_info & to call_info &. For %n
35337 set info.nowrite to false.
35339 PR middle-end/78901
35340 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35341 possibly throwing calls.
35343 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35344 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35345 and fns handling, rather than in a separate case SSA_NAME.
35347 2017-01-02 Jeff Law <law@redhat.com>
35349 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35350 fixes for first_period and second_period variables.
35352 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35355 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35356 (*insvqi_1): New insn pattern.
35357 (*insvqi_1_mem_rex64): Ditto.
35358 (*insvqi_2): Ditto.
35359 (*insvqi_3): Rename from *insvqi.
35361 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35363 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35365 * doc/cfg.texi (Edges): Remove reference to Java.
35366 (Maintaining the CFG): Ditto.
35368 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35370 PR middle-end/77674
35371 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35372 transparent aliases.
35374 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35376 PR middle-end/77484
35377 * predict.def (PRED_CALL): Update hitrate.
35378 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35379 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35380 into direct/indirect/polymorphic variants.
35382 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35384 Update copyright years.
35386 * gcc.c (process_command): Update copyright notice dates.
35387 * gcov-dump.c (print_version): Ditto.
35388 * gcov.c (print_version): Ditto.
35389 * gcov-tool.c (print_version): Ditto.
35390 * gengtype.c (create_file): Ditto.
35391 * doc/cpp.texi: Bump @copying's copyright year.
35392 * doc/cppinternals.texi: Ditto.
35393 * doc/gcc.texi: Ditto.
35394 * doc/gccint.texi: Ditto.
35395 * doc/gcov.texi: Ditto.
35396 * doc/install.texi: Ditto.
35397 * doc/invoke.texi: Ditto.
35399 Copyright (C) 2017 Free Software Foundation, Inc.
35401 Copying and distribution of this file, with or without modification,
35402 are permitted in any medium without royalty provided the copyright
35403 notice and this notice are preserved.