1 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3 * i386.c (has_dispatch): Disable for Ryzen.
5 2017-10-08 Olivier Hainque <hainque@adacore.com>
7 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
10 (thumb_set_return_address): Likewise.
12 2017-10-08 Olivier Hainque <hainque@adacore.com>
14 * common/config/arm/arm-common.c (arm_except_unwind_info):
15 Handle DWARF2_UNWIND_INFO.
17 2017-10-07 Michael Collison <michael.collison@arm.com>
19 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
22 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
24 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
25 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
26 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
27 defined, force the creation of a new block for a dispatch label.
29 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
31 * invoke.texi (Wsuggest-attribute=cold): Document.
32 * common.opt (Wsuggest-attribute=cold): New
33 * ipa-pure-const.c (warn_function_cold): New function.
34 * predict.c (compute_function_frequency): Use it.
35 * predict.h (warn_function_cold): Declare.
37 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
39 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
42 2017-10-06 Martin Liska <mliska@suse.cz>
44 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
45 keyword for member functions.
46 (struct sanopt_tree_couple): New struct.
47 (struct sanopt_tree_couple_hash): New function.
48 (struct sanopt_ctx): Add new hash_map.
49 (has_dominating_ubsan_ptr_check): New function.
50 (record_ubsan_ptr_check_stmt): Likewise.
51 (maybe_optimize_ubsan_ptr_ifn): Likewise.
52 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
53 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
55 2017-10-06 Sudakshina Das <sudi.das@arm.com>
58 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
59 aarch64_simd_valid_immediate on CONST_VECTORs.
60 (aarch64_reg_or_bic_imm): Likewise.
62 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
64 PR rtl-optimization/82396
65 * haifa-sched.c (ready_sort_real): Disable qsort checking.
67 2017-10-06 Sebastian Pop <sebpop@gmail.com>
69 * graphite-dependences.c (scop_get_reads): Move code to...
70 (scop_get_must_writes): Move code to...
71 (scop_get_may_writes): Move code to...
72 (scop_get_reads_and_writes): ... here.
73 (scop_get_dependences): Call scop_get_reads_and_writes.
75 2017-10-06 Jakub Jelinek <jakub@redhat.com>
77 PR tree-optimization/82434
78 * fold-const.h (can_native_encode_type_p,
79 can_native_encode_string_p): Remove.
80 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
81 don't encode anything, just return what would be otherwise returned.
82 (native_encode_fixed, native_encode_complex, native_encode_vector):
84 (native_encode_string): Likewise. Inline by hand
85 can_native_encode_string_p.
86 (can_native_encode_type_p): Remove.
87 (can_native_encode_string_p): Remove.
88 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
89 STRING_CSTs using can_native_encode_string_p, test all
90 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
91 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
92 argument from native_encode_expr.
93 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
94 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
97 2017-10-06 Richard Biener <rguenther@suse.de>
99 PR tree-optimization/82397
100 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
101 operand_equal_p but rely on data_ref_compare_tree for detecting
103 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
104 to match up with dr_group_sort_cmp.
106 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
109 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
111 * config/s390/s390-builtin-types.def: Regenerate.
113 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
116 * config/s390/s390-builtin-types.def: Regenerate.
117 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
118 Change flag from B_VXE to B_VX.
119 (s390_vec_min_dbl): Remove B_VXE flag.
121 2017-10-06 Richard Biener <rguenther@suse.de>
123 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
124 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
125 translate_isl_ast_to_gimple::is_valid_rename,
126 translate_isl_ast_to_gimple::get_rename,
127 translate_isl_ast_to_gimple::get_def_bb_for_const,
128 translate_isl_ast_to_gimple::get_new_name,
129 translate_isl_ast_to_gimple::collect_all_ssa_names,
130 translate_isl_ast_to_gimple::copy_loop_phi_args,
131 translate_isl_ast_to_gimple::collect_all_ssa_names,
132 translate_isl_ast_to_gimple::copy_loop_phi_args,
133 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
134 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
135 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
136 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
137 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
138 translate_isl_ast_to_gimple::copy_cond_phi_args,
139 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
140 translate_isl_ast_to_gimple::edge_for_new_close_phis,
141 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
142 translate_isl_ast_to_gimple::rename_uses,
143 translate_isl_ast_to_gimple::rename_all_uses): Remove.
144 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
145 (set_rename_for_each_def): Likewise.
146 (graphite_copy_stmts_from_block): Handle debug stmt resetting
147 here. Handle rewriting SCEV analyzable uses here.
148 (copy_bb_and_scalar_dependences): Generate code for PHI
150 (graphite_regenerate_ast_isl): Adjust.
151 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
152 (add_write, add_read): New functions.
153 (build_cross_bb_scalars_def): Use it and simplify.
154 (build_cross_bb_scalars_use): Likewise.
155 (graphite_find_cross_bb_scalar_vars): Inline into...
156 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
157 simulating out-of-SSA. Compute liveout and add dependencies.
158 (build_scops): Force an empty entry block.
159 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
161 (sese_build_liveouts): Declare.
162 (sese_trivially_empty_bb_p): Likewise.
163 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
164 compute liveout and debug_liveout.
165 (sese_bad_liveouts_use): Remove.
166 (sese_reset_debug_liveouts_bb): Likewise.
167 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
168 (sese_build_liveouts): Build liveout and debug_liveout and store
170 (new_sese_info): Adjust.
171 (free_sese_info): Likewise.
172 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
173 do not build liveout here.
174 (move_sese_in_condition): Adjust region entry.
175 (scev_analyzable_p): Match up with chrec_apply requirements.
176 (sese_trivially_empty_bb_p): New.
177 * tree-into-ssa.c (get_reaching_def): Properly support generating
178 default-defs for incremental rewrite of anonymous names.
180 2017-10-06 Richard Biener <rguenther@suse.de>
182 * graphite-sese-to-poly.c (extract_affine): For casts increasing
183 precision do not perform modulo reduction.
185 2017-10-06 Richard Biener <rguenther@suse.de>
187 PR tree-optimization/82436
188 * tree-vect-slp.c (vect_supported_load_permutation_p): More
189 conservatively choose the vectorization factor when checking
190 whether we can perform the required load permutation.
191 (vect_transform_slp_perm_load): Assert when we may not fail.
193 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
195 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
196 message for incompatible -msdata=* and -mcall-* options.
198 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
200 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
201 for post-reload scheduling.
203 2017-10-05 Tamar Christina <tamar.christina@arm.com>
205 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
207 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
209 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
210 to improve monte carlo in scimark.
212 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
214 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
215 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
216 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
217 pentium4_cost, nocona_cost): Set reassociation width to 1.
218 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
219 width to 2 for fp operations and 1 otherwise.
220 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
222 (atom_cost): Set reassociation width to 2.
223 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
224 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
225 (core_cost): Set fp reassociation width to 4 and vector to 2.
226 (ix86_reassociation_width): Rewrite using cost table; special case
227 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
228 and TARGET_AVX128_OPTIMAL.
229 * i386.h (processor_costs): Add
230 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
231 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
232 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
233 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
234 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
235 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
237 2017-10-05 Nathan Sidwell <nathan@acm.org>
239 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
241 2017-10-05 Tamar Christina <tamar.christina@arm.com>
243 * config/arm/arm.c (arm_test_fpu_data): New.
244 (arm_run_selftests): Call arm_test_fpu_data.
246 2017-10-04 Nathan Sidwell <nathan@acm.org>
248 * toplev.c (toplev::main): Remove excess parens on pretty_printer
250 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
252 2017-10-04 Sudakshina Das <sudi.das@arm.com>
254 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
255 check type for aarch64_simd_valid_immediate.
256 (aarch64_output_simd_mov_immediate): Update prototype.
257 (aarch64_simd_valid_immediate): Update prototype.
258 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
259 support for ORR-immediate.
260 (and<mode>3): modified pattern to add support for BIC-immediate.
261 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
262 now checks for valid immediate for BIC and ORR based on new enum
264 (aarch64_output_simd_mov_immediate): Function now used to output
265 BIC/ORR imm as well based on new enum argument.
266 * config/aarch64/constraints.md (Do): New vector immediate constraint.
268 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
269 (aarch64_reg_or_bic_imm): Likewise.
271 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
273 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
274 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
276 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
280 PR rtl-optimization/82396
281 * haifa-sched.c (autopref_multipass_init): Simplify
283 (autopref_rank_data): Simplify sort order.
284 * sched-int.h (autopref_multipass_data_): Remove
285 multi_mem_insn_p, min_offset and max_offset.
287 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
289 * doc/sourcebuild.texi: Document vect_peeling_profitable.
291 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
293 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
296 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
298 * doc/sourcebuild.texi: Document vect_long_mult.
300 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
302 PR tree-optimization/82413
303 * fold-const.c (build_range_check): Use widest_int when comparing
304 the maximum ETYPE value with HIGH.
306 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
308 PR rtl-optimization/82396
309 * haifa-sched.c (autopref_multipass_init): Simplify
311 (autopref_rank_data): Simplify sort order.
312 * sched-int.h (autopref_multipass_data_): Remove
313 multi_mem_insn_p, min_offset and max_offset.
315 2017-10-04 Jakub Jelinek <jakub@redhat.com>
317 PR tree-optimization/82381
318 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
319 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
322 PR tree-optimization/82374
323 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
324 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
325 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
326 current_function_decl to the new decl.
328 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
330 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
331 helper macro for IEEE float128 hardware built-in functions.
332 (SQRTF128_ODD): Add built-in functions with the round-to-odd
334 (TRUNCF128_ODD): Likewise.
335 (ADDF128_ODD): Likewise.
336 (SUBF128_ODD): Likewise.
337 (MULF128_ODD): Likewise.
338 (DIVF128_ODD): Likewise.
339 (FMAF128_ODD): Likewise.
340 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
341 UNSPEC_TRUNC_ROUND_TO_ODD.
342 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
343 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
344 floating point round to odd instructions.
345 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
346 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
347 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
348 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
349 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
350 (trunc<mode>sf2_hw): Change the truncate with round to odd
351 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
352 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
353 to odd hardware instructions.
354 (sub<mode>3_odd): Likewise.
355 (mul<mode>3_odd): Likewise.
356 (div<mode>3_odd): Likewise.
357 (sqrt<mode>2_odd): Likewise.
358 (fma<mode>4_odd): Likewise.
359 (fms<mode>4_odd): Likewise.
360 (nfma<mode>4_odd): Likewise.
361 (nfms<mode>4_odd): Likewise.
362 (trunc<mode>df2_odd): Change the truncate with round to odd
363 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
365 * doc/extend.texi (PowerPC built-in functions): Update documentation
366 for existing IEEE float128-bit built-in functions. Add built-in
367 functions that generate the IEEE 128-bit floating point round to
370 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
372 PR rtl-optimization/77729
373 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
374 to (X&(C1&~C2))|C2 transformations.
376 2017-10-03 Martin Jambor <mjambor@suse.cz>
378 PR tree-optimization/82363
379 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
380 mismatch, mark lacc written regardless of racc.
382 2017-10-03 Jakub Jelinek <jakub@redhat.com>
384 PR tree-optimization/82381
385 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
386 stmt_to_insert nor wheather SSA_NAMEs are default defs.
387 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
388 fallthrough into reassoc_stmt_dominates_stmt_p.
391 * combine.c (combine_instructions): Don't combine in unreachable
394 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
397 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
398 function to not use the have_cpu variable. Do not set cpu_index,
399 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
401 (rs6000_valid_attribute_p): Remove duplicate initializations of
402 old_optimize and func_optimize.
403 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
404 (rs6000_activate_target_options): Make global.
405 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
408 2017-10-02 Jakub Jelinek <jakub@redhat.com>
410 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
411 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
412 Set *poffset to 0 and *psize and *pmax_size to -1 if
413 *poffset + *psize overflows in HOST_WIDE_INT.
415 PR tree-optimization/82387
416 PR tree-optimization/82388
417 PR tree-optimization/82389
418 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
419 instead of live_bytes non-NULL.
421 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
424 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
427 2017-10-02 Richard Biener <rguenther@suse.de>
429 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
430 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
431 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
433 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
435 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
436 requested precision matches the type's.
437 * calls.c (alloc_max_size): Calculate the new candidate size as
438 a widest_int and use wi::to_widest when comparing it with the
439 current candidate size.
440 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
441 zero rather than integer_zero_node.
442 * match.pd: Check for a no-op conversion before using wi::add
443 rather than after. Use tree_to_uhwi when summing small shift
444 counts into an unsigned int.
446 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
447 Alan Hayward <alan.hayward@arm.com>
448 David Sherwood <david.sherwood@arm.com>
451 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
452 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
453 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
456 2017-10-02 Richard Biener <rguenther@suse.de>
458 PR tree-optimization/82355
459 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
460 a mapping for the enclosing loop but avoid generating one for
462 (copy_bb_and_scalar_dependences): Remove premature codegen
463 error on PHIs in blocks duplicated into multiple places.
464 * graphite-scop-detection.c
465 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
466 in the region use it as loop and nest to analyze the DR in.
467 (try_generate_gimple_bb): Likewise.
468 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
469 (add_loop_constraints): For blocks in a loop not in the region
470 create a dimension with a single iteration.
471 * sese.h (gbb_loop_at_index): Remove assert.
473 2017-10-01 Kevin Buettner <kevinb@redhat.com>
475 * omp-expand.c (adjust_context_scope): New function.
476 (expand_parallel_call): Call adjust_context_scope.
478 2017-10-01 Jeff Law <law@redhat.com>
480 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
481 dom_opt_dom_walker class with direct access to private members.
482 Add comments. Call test_for_singularity.
483 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
484 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
485 m_dummy_cond anymore.
486 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
488 (pass_dominator:execute): Build the dummy_cond here and pass it
489 to the dom_opt_dom_walker ctor.
490 (test_for_singularity): New function.
492 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
493 Maya Rashish <coypu@sdf.org>
495 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
496 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
497 (alpha*-*-netbsd*) Use nbsd_tm_file.
498 (arm*-*-netbsdelf*) Likewise.
499 (i[34567]86-*-netbsdelf*) Likewise.
500 (x86_64-*-netbsd*) Likewise.
501 (mips*-*-netbsd*) Likewise.
502 (powerpc-*-netbsd*) Likewise.
503 (sh*-*-netbsd*) Likewise.
504 (sparc-*-netbsdelf*) Likewise.
505 (sparc64-*-netbsd*) Likewise.
506 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
508 (vax-*-netbsdelf*) Likewise.
509 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
510 (UINT_FAST8_TYPE) Likewise.
511 (INT_FAST16_TYPE) Check CHAR_FAST16.
512 (UINT_FAST16_TYPE) Likewise.
514 2017-09-30 Jakub Jelinek <jakub@redhat.com>
517 * config/i386/i386.md
518 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
519 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
520 *divmodsi4_zext_2): New define_insn_and_split.
521 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
522 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
523 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
524 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
525 New define_insn_and_split.
526 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
527 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
528 operands[1] having DImode when mode is SImode.
530 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
531 always SImode for DIV and MOD in REG_EQUAL notes.
533 2017-09-29 Yury Gribov <tetra2005@gmail.com>
536 * match.pd: Fix handling of NaNs in pattern.
538 2017-09-29 Jeff Law <law@redhat.com>
540 * sbitmap.c (bitmap_bit_in_range_p): New function.
541 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
542 * tree-ssa-dse.c (live_bytes_read): New function.
543 (dse_classify_store): Ignore reads of dead bytes.
545 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
546 typos and whitespace errors.
547 * config/i386/predicates.md (address_no_seg_operand): Likewise.
548 * config/s390/s390.c (s390_emit_prologue): Likewise.
550 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
553 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
554 with a symbol for LRA.
556 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
558 PR rtl-optimization/82338
559 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
561 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
563 * genmodes.c (calc_wider_mode): Suppress qsort macro.
564 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
565 (qsort_chk): Declare.
566 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
567 (qsort_chk): New function.
569 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
571 PR tree-optimization/82337
572 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
573 phi definition if the PHI result appears in an abnormal PHI.
574 (find_basis_for_base_expr): Don't record a basis if the LHS of the
575 basis appears in an abnormal PHI.
577 2017-09-29 Richard Biener <rguenther@suse.de>
579 * graphite-isl-ast-to-gimple.c
580 (translate_isl_ast_to_gimple::set_codegen_error): New function.
581 (binary_op_to_tree): Use it.
582 (get_rename_from_scev): Likewise.
583 (copy_loop_phi_nodes): Likewise.
584 (copy_bb_and_scalar_dependences): Likewise.
585 (translate_pending_phi_nodes): Likewise.
587 2017-09-29 Jakub Jelinek <jakub@redhat.com>
590 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
591 for movabsq $(i32 << shift), r64.
593 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
595 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
596 index when encoding %esp as %rsp to avoid 0x67 prefix.
598 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
600 * config/i386/i386.md (*movsf_internal, *movdf_internal):
601 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
603 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
605 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
606 Extensions with more than 16 double VFP registers.
607 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
608 to_clear_mask and all code related to it. Replace the remaining
609 entry by a sbitmap and adapt code accordingly.
611 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
613 * brig-builtins.def: Change pure attributes to const.
615 2017-09-28 Joseph Myers <joseph@codesourcery.com>
617 * config.gcc (default_gnu_indirect_function): Default to yes for
618 sparc*-*-linux* with glibc.
620 2017-09-28 Joseph Myers <joseph@codesourcery.com>
622 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
623 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
624 when creating .init_array and .fini_array sections with priority
627 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
630 * config/aarch64/aarch64.c
631 (aarch64_builtin_support_vector_misalignment): Always return false
632 when misalignment is unknown.
634 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
636 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
637 this function to return false if the definition used by the swap
638 instruction is artificial, or if the memory address from which the
639 constant value is loaded is not represented by a base address held
640 in a register or if the base address register is a frame or stack
641 pointer. Additionally, return false if the base address of the
642 loaded constant is a SYMBOL_REF but is not considered to be a
644 (replace_swapped_load_constant): New function.
645 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
646 loaded constant vector with a load of a swapped constant vector.
648 2017-09-27 Carl Love <cel@us.ibm.com>
650 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
651 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
652 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
655 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
657 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
658 first, always call autopref_rank_data otherwise.
660 2017-09-27 Richard Biener <rguenther@suse.de>
662 * graphite-scop-detection.c (find_scop_parameters): Move
663 loop bound handling ...
664 (gather_bbs::before_dom_children): ... here, avoiding the need
665 to build scop_info->loop_nest.
666 (record_loop_in_sese): Remove.
667 * sese.h (sese_info_t::loop_nest): Remove.
668 * sese.c (new_sese_info): Do not allocate loop_nest.
669 (free_sese_info): Do not free loop_nest.
671 2017-09-27 Jakub Jelinek <jakub@redhat.com>
674 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
675 lhs from calls if the lhs has addressable type.
677 2017-09-27 Richard Biener <rguenther@suse.de>
679 * graphite.h (scop::max_alias_set): New member.
680 * graphite-scop-detection.c: Remove references to non-existing
682 (build_alias_sets): Record the maximum alias set used for drs.
683 (build_scops): Support zero as unlimited for
684 --param graphite-max-arrays-per-scop.
685 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
687 (build_poly_sr_1): ... here. Compute alias set based on the
688 maximum alias set used for drs rather than
689 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
691 2017-09-27 Richard Biener <rguenther@suse.de>
693 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
694 --param loop-block-tile-size=0 to disable tiling.
696 2017-09-27 Richard Biener <rguenther@suse.de>
698 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
699 (graphite-max-nb-scop-params): Document special value zero.
700 * domwalk.h (dom_walker::STOP): New symbolical constant.
701 (dom_walker::dom_walker): Add optional parameter for bb to
703 (dom_walker::~dom_walker): Declare.
704 (dom_walker::before_dom_children): Document STOP return value.
705 (dom_walker::m_user_bb_to_rpo): New member.
706 (dom_walker::m_bb_to_rpo): Likewise.
707 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
708 mapping here if not provided by the user.
709 (dom_walker::~dom_walker): Free bb to RPO mapping if not
710 provided by the user.
711 (dom_walker::STOP): Define.
712 (dom_walker::walk): Do not compute bb to RPO mapping here.
713 Support STOP return value from before_dom_children to stop
715 * graphite-optimize-isl.c (optimize_isl): If the schedule
716 is the same still generate code if -fgraphite-identity
717 or -floop-parallelize-all are given.
718 * graphite-scop-detection.c: Include cfganal.h.
719 (gather_bbs::gather_bbs): Get and pass through bb to RPO
721 (gather_bbs::before_dom_children): Return STOP for BBs
723 (build_scops): Compute bb to RPO mapping and pass it to
724 the domwalk. Treat --param graphite-max-nb-scop-params=0
725 as not limiting the number of params.
726 * graphite.c (graphite_initialize): Remove limit on the
727 number of basic-blocks in a function.
728 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
729 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
732 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
734 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
735 Adjust code to eliminate needing to do the shift right 32-bits
736 operation after XSCVDPSPN.
738 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
740 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
741 ((X / Y) != 0 -> X >= Y): Likewise.
743 2017-09-26 Carl Love <cel@us.ibm.com>
745 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
746 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
747 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
748 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
749 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
750 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
751 definitions and overloading.
752 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
753 statement for P9V_BUILTIN_XST_LEN_R.
754 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
755 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
756 define_expand and define_insn for the instructions and builtins.
757 * doc/extend.texi: Update the built-in documentation file for the new
759 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
760 define_insn for the instructions
762 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
765 * gcc/config/netbsd-protos.h: New file.
766 * gcc/config/netbsd.c: New file.
767 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
768 * gcc/config/t-netbsd: New file.
769 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
770 (tmake_file) Add t-netbsd.
771 (extra_objs) Add netbsd.o.
773 2017-09-26 Janus Weil <janus@gcc.gnu.org>
777 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
779 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
781 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
782 sign extension from a vector register to a GPR by doing a 32-bit
783 direct move and then an EXTSW.
784 (extendsi<mode>2 splitter): Likewise.
785 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
786 right or vector extract after doing XSCVDPSPN. Use
787 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
789 (movdi_from_sf_zero_ext): Likewise.
790 (reload_gpr_from_vsxsf): Likewise.
791 (p8_mfvsrd_4_disf): Delete, no longer used.
792 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
793 then needing to move the SFmode to a GPR to use the XSCVDPSP
794 instruction instead of FRSP and XSCVDPSPN.
795 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
796 it is adjacent to the other XSCVSPDP insns.
797 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
798 SFmode to be in traditional Altivec registers.
799 (vsx_xscvdpspn): Eliminate useless alternative constraint.
800 (vsx_xscvspdpn): Likewise.
801 (vsx_xscvspdpn_scalar): Likewise.
803 2017-09-26 Martin Jambor <mjambor@suse.cz>
805 * tree-sra.c (compare_access_positions): Put integral types first,
806 stabilize sorting of integral types, remove conditions putting
807 non-full-precision integers last.
808 (sort_and_splice_var_accesses): Disable scalarization if a
809 non-integert would be represented by a non-full-precision integer.
811 2017-09-26 Joseph Myers <joseph@codesourcery.com>
813 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
814 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
815 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
816 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
817 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
818 conditionals inside the function instead of around it. Call
819 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
820 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
822 2017-09-26 Richard Biener <rguenther@suse.de>
824 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
826 (scop_detection::build_scop_breadth): ... this. Removed.
827 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
828 (scop_detection::harmful_stmt_in_bb): Likewise.
829 (scop_detection::graphite_can_represent_stmt): Likewise.
830 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
831 (scop_detection::can_represent_loop): Remove recursion, fold in ...
832 (scop_detection::can_represent_loop_1): ... this. Removed.
833 (scop_detection::harmful_loop_in_region): Simplify after inlining
834 the above and remove more quadraticness.
835 (build_scops): Adjust.
836 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
839 2017-09-26 Jakub Jelinek <jakub@redhat.com>
842 * config/i386/i386.c (ix86_print_operand_address_as): Only test
843 REGNO (base) == SP_REG if base is a REG.
846 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
847 if it is different SSA_NAME.
848 (optimize_range_tests_cmp_bitwise): New function.
849 (optimize_range_tests): Call it.
851 2017-09-26 Richard Biener <rguenther@suse.de>
853 PR tree-optimization/82321
854 * graphite.c (canonicalize_loop_closed_ssa): Properly check
855 for the def being inside the loop.
857 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
859 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
861 * config/s390/s390-builtins.def: Fix constraint on op4.
863 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
865 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
866 independent expanders.
867 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
868 ("vec_ordered", "vec_unordered"): New expanders.
870 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
872 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
875 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
877 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
878 vec_unpacks_lo_v16qi.
879 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
881 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
883 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
884 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
885 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
887 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
889 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
891 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
893 ("*vec_slb<mode>"): New insn pattern.
894 ("vec_shr_<mode>"): New expander.
895 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
896 and force the shift count operand to V16QImode.
897 ("vec_srb<mode>"): Set shift count mode to V16QI.
899 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
901 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
902 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
903 ("vec_widen_smult_hi_<mode>"): New expander definitions.
905 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
908 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
910 2017-09-26 Richard Biener <rguenther@suse.de>
912 PR tree-optimization/82320
913 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
916 2017-09-25 Jeff Law <law@redhat.com>
918 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
919 prototype for new argument.
920 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
921 mostly extracted from rs6000_emit_allocate_stack.
922 (rs6000_emit_probe_stack_range_stack_clash): New function.
923 (rs6000_emit_allocate_stack): Call
924 rs6000_emit_probe_stack_range_stack_clash as needed.
925 (rs6000_emit_probe_stack_range): Add additional argument
926 to call to gen_probe_stack_range{si,di}.
927 (output_probe_stack_range): New.
928 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
929 (output_probe_stack_range_stack_clash): New.
930 (rs6000_emit_prologue): Emit notes into dump file as requested.
931 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
932 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
933 Add additional operand and pass it to output_probe_stack_range.
935 2017-09-25 Bin Cheng <bin.cheng@arm.com>
937 PR tree-optimization/82163
938 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
939 (checking_verify_loop_closed_ssa): New parameter.
940 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
941 (check_loop_closed_ssa_stmt): Delete.
942 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
943 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
944 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
947 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
949 * brig-builtins.def: Treat HSAIL barrier builtins as
950 setjmp/longjump style functions.
952 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
954 * target.def (constant_alignment): New hook.
955 * defaults.h (CONSTANT_ALIGNMENT): Delete.
956 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
957 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
958 * doc/tm.texi: Regenerate.
959 * targhooks.h (default_constant_alignment): Declare.
960 (constant_alignment_word_strings): Likewise.
961 * targhooks.c (default_constant_alignment): New function.
962 (constant_alignment_word_strings): Likewise.
963 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
964 instead of CONSTANT_ALIGNMENT.
965 * varasm.c (align_variable, get_variable_align, build_constant_desc)
966 (force_const_mem): Likewise.
967 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
968 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
969 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
970 (TARGET_CONSTANT_ALIGNMENT): Redefine.
971 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
973 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
974 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
975 constant_alignment_word_strings.
976 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
977 (CONSTANT_ALIGNMENT): Likewise.
978 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
979 (arm_constant_alignment): New function.
980 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
981 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
982 constant_alignment_word_strings.
983 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
984 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
985 constant_alignment_word_strings.
986 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
987 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
988 (cris_constant_alignment): New function.
989 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
990 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
991 (epiphany_constant_alignment): New function.
992 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
993 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
994 constant_alignment_word_strings.
995 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
996 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
997 constant_alignment_word_strings.
998 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
999 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1000 constant_alignment_word_strings.
1001 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1002 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1003 * config/i386/i386.c (ix86_constant_alignment): Make static.
1004 Use the same interface as the target hook.
1005 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1006 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1007 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1008 constant_alignment_word_strings.
1009 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1010 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1011 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1012 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1013 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1014 constant_alignment_word_strings.
1015 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1016 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1017 constant_alignment_word_strings.
1018 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1019 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1020 constant_alignment_word_strings.
1021 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1022 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1024 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1025 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1026 * config/mips/mips.c (mips_constant_alignment): New function.
1027 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1028 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1029 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1030 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1031 (mmix_constant_alignment): Make static. Use the same interface
1033 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1034 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1035 constant_alignment_word_strings.
1036 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1037 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1038 constant_alignment_word_strings.
1039 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1040 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1041 constant_alignment_word_strings.
1042 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1043 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1044 (rs6000_constant_alignment): New function.
1045 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1046 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1047 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1048 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1049 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1050 (rs6000_constant_alignment): New function.
1051 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1052 * config/s390/s390.c (s390_constant_alignment): New function.
1053 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1054 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1055 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1056 constant_alignment_word_strings.
1057 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1058 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1059 (sparc_constant_alignment): New function.
1060 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1061 * config/spu/spu.c (spu_constant_alignment): New function.
1062 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1063 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1064 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1065 constant_alignment_word_strings.
1066 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1067 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1068 constant_alignment_word_strings.
1069 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1070 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1071 constant_alignment_word_strings.
1072 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1073 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1074 (visium_constant_alignment): New function.
1075 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1076 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1077 (xtensa_constant_alignment): New function.
1078 * system.h (CONSTANT_ALIGNMENT): Poison.
1080 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1082 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1083 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1084 (rs6000_builtin_valid_without_lhs): New helper function.
1085 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1086 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1088 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1090 * target.h (vec_perm_indices): Use unsigned short rather than
1092 (auto_vec_perm_indices): Likewise.
1093 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1094 Use unsigned int rather than unsigned char.
1095 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1097 2017-09-25 Richard Biener <rguenther@suse.de>
1099 * cfgloop.h (sort_sibling_loops): Declare.
1100 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1101 (sort_sibling_loops): New function sorting the sibling loop list
1103 * graphite.c (graphite_transform_loops): Sort sibling loops.
1105 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1107 * target.def (vec_perm_const_ok): Change sel parameter to
1109 * optabs-query.c (can_vec_perm_p): Update accordingly.
1110 * doc/tm.texi: Regenerate.
1111 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1112 auto_vec_perm_indices and remove separate nelt field.
1113 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1114 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1115 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1116 (aarch64_expand_vec_perm_const): Update accordingly.
1117 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1118 to vec_perm_indices.
1119 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1120 auto_vec_perm_indices and remove separate nelt field.
1121 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1122 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1123 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1125 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1126 to vec_perm_indices.
1127 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1128 sel to vec_perm_indices.
1129 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1130 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1131 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1133 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1136 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1139 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1140 on the FUNCTION_DECL function context if it has a DIE that is a
1143 2017-09-25 Richard Biener <rguenther@suse.de>
1145 PR tree-optimization/82285
1146 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1149 2017-09-25 Tom de Vries <tom@codesourcery.com>
1153 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1156 2017-09-25 Richard Biener <rguenther@suse.de>
1158 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1159 ISL errors other than isl_error_quota happen. Dump if the
1160 schedule is the same.
1161 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1162 errors instead of aborting inside ISL.
1164 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1167 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1168 of libgcc_eh for m64.
1169 * config/i386/darwin64.h: Likewise.
1171 2017-09-25 Richard Biener <rguenther@suse.de>
1174 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1175 attribute for incomplete types nor twice for complete ones.
1177 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1180 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1181 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1184 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1187 * config/i386/i386.opt (mprefer-avx256): Use
1188 ix86_target_flags variable.
1189 * config/i386/i386.c (ix86_target_string): Move
1190 -mprefer-avx256 to flag2_opts.
1192 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1195 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1196 and x != -1 | y != -1 into (x & y) != -1.
1198 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1200 * config.gcc: Add new case statement to set
1201 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1202 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1203 s390x-*-linux* case statements. Added aarch64 to the list of
1204 supported architectures.
1206 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1208 PR tree-optimization/82289
1209 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1210 STMT_VINFO_RELEVANT_P.
1212 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1213 Alan Hayward <alan.hayward@arm.com>
1214 David Sherwood <david.sherwood@arm.com>
1216 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1217 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1218 (extract_range_from_binary_expr_1): Don't call
1219 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1221 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1222 Alan Hayward <alan.hayward@arm.com>
1223 David Sherwood <david.sherwood@arm.com>
1225 * target.def (preferred_vector_alignment): New hook.
1226 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1228 * doc/tm.texi: Regenerate.
1229 * targhooks.h (default_preferred_vector_alignment): Declare.
1230 * targhooks.c (default_preferred_vector_alignment): New function.
1231 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1233 (DR_TARGET_ALIGNMENT): New macro.
1234 (aligned_access_p): Update commentary.
1235 (vect_known_alignment_in_bytes): New function.
1236 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1238 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1239 Calculate the misalignment based on the target alignment rather than
1241 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1242 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1243 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1244 the target alignment, rather than masking the element misalignment
1245 with the number of elements in a vector. Also use the target
1246 alignment when calculating the maximum number of peels.
1247 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1248 instead of TYPE_ALIGN_UNIT.
1249 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1250 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1251 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1252 (vect_create_data_ref_ptr): Likewise.
1253 (vect_setup_realignment): Realign by ANDing with
1254 -DR_TARGET_MISALIGNMENT.
1255 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1256 the number of peels based on DR_TARGET_ALIGNMENT.
1257 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1258 with the guaranteed alignment boundary when deciding whether
1260 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1261 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1262 (ensure_base_align): Remove stmt_info parameter. Get the
1263 target base alignment from DR_TARGET_ALIGNMENT.
1264 (vectorizable_store): Update call accordingly. Interpret
1265 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1267 (vectorizable_load): Likewise.
1269 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1270 Alan Hayward <alan.hayward@arm.com>
1271 David Sherwood <david.sherwood@arm.com>
1273 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1274 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1275 (vect_enhance_data_refs_alignment): Likewise.
1277 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1279 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1280 error. Only quit immediately if parsing is complete.
1281 (BEGIN): Initialize fatal_err and parse_done.
1282 (begin fpu, end fpu): Check number of arguments.
1283 (begin arch, end arch): Likewise.
1284 (begin cpu, end cpu): Likewise.
1285 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1286 (optalias): Likewise.
1288 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1290 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1291 * config/arm/arm-isa.h: Delete. Move definitions to ...
1292 * arm-cpus.in: ... here. Use new feature and fgroup values.
1293 * config/arm/arm.c (arm_option_override): Use lower case for feature
1295 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1296 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1297 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1299 (print_isa_bits_for): New function.
1300 (gen_isa): New function.
1301 (gen_comm_data): Use print_isa_bits_for.
1302 (define feature): New keyword.
1303 (define fgroup): New keyword.
1304 * config/arm/t-arm (TM_H): Remove.
1305 (GTM_H): Add arm-isa.h.
1306 (arm-isa.h): Add rule to generate file.
1307 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1308 case for feature bit names.
1310 2017-09-22 Richard Biener <rguenther@suse.de>
1312 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1314 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1315 print of no dependency loops ...
1316 * graphite.c (graphite_transform_loops): ... here.
1317 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1319 (same_close_phi_node, remove_duplicate_close_phi,
1320 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1321 (canonicalize_loop_closed_ssa): ... here and simplify.
1322 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1323 (optimize_isl): Use dump_printf_loc to tell when we stopped
1324 optimizing because of an ISL timeout.
1326 2017-09-22 Richard Biener <rguenther@suse.de>
1328 PR tree-optimization/82291
1329 * tree-if-conv.c (predicate_mem_writes): Make sure to
1330 remove writes in blocks predicated with false.
1332 2017-09-22 Richard Biener <rguenther@suse.de>
1334 * sese.c: Include cfganal.h.
1335 (if_region_set_false_region): Remove.
1336 (create_if_region_on_edge): Likewise.
1337 (move_sese_in_condition): Re-implement without destroying
1340 2017-09-22 Richard Biener <rguenther@suse.de>
1342 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1343 Verify both BBs contain loop PHI nodes before dispatching to
1345 (graphite_regenerate_ast_isl): Do not recompute dominators,
1346 do not verify three times. Restructure for clarity.
1347 * graphite-scop-detection.c (same_close_phi_node,
1348 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1349 defined_in_loop_p, canonicalize_loop_closed_ssa,
1350 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1351 checking and SSA rewrite, move to ...
1352 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1353 (graphite_initialize): Do not pass in ctx, do not reset the
1354 SCEV cache, compute only dominators.
1355 (graphite_transform_loops): Allocate ISL ctx after
1356 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1357 Maintain post-dominators only around build_scops.
1358 * sese.c (if_region_set_false_region): Make static. Free
1359 and recompute dominators.
1360 (move_sese_in_condition): Assert we don't get called with
1361 post-dominators computed.
1362 * sese.h (if_region_set_false_region): Remove.
1364 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1366 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1367 mode attribute for TARGET_AVX512VL.
1369 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1371 * config/i386/i386.opt (mprefer-avx256): New option.
1372 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1374 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1375 for TARGET_PREFER_AVX256.
1376 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1378 2017-09-21 Jeff Law <law@redhat.com>
1380 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1381 Fix dump output if the only stack space is for pushed registers.
1383 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1385 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1388 2017-09-21 Martin Sebor <msebor@redhat.com>
1391 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1392 code (in C++) or code that triggers warnings.
1394 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1396 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1398 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1400 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1401 * sched-rgn.c: Ditto.
1402 * sel-sched-ir.c: Ditto.
1404 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1406 * toplev.h (set_random_seed): Adjust return type.
1407 * toplev.c (init_local_tick): Move eager initialization of random_seed
1408 to get_random_seed. Adjust comment.
1409 (init_random_seed): Inline to get_random_seed, delete.
1410 (get_random_seed): Initialize random_seed lazily.
1411 (set_random_seed): Do not return previous value.
1412 (print_switch_value): Do not call get_random_seed.
1414 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1416 * cgraph.c (delete_function_version): New, broken out from...
1417 (cgraph_node::delete_function_version): ...here. Rename to
1418 cgraph_node::delete_function_version_by_decl. Update all uses.
1419 (cgraph_node::remove): Call delete_function_version.
1421 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1424 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1425 VAR_DECLs to which addressable non-volatile parameters are mapped
1426 and for id->retvar after the return value assignment. Clear
1427 id->retval and id->retbnd after inlining.
1429 2017-09-21 Richard Biener <rguenther@suse.de>
1431 PR tree-optimization/82276
1432 PR tree-optimization/82244
1433 * tree-vrp.c (build_assert_expr_for): Set
1434 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1436 (remove_range_assertions): Revert earlier change.
1438 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1441 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1443 2017-09-21 Richard Biener <rguenther@suse.de>
1445 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1446 Restore valid IL after code generation errors.
1447 * graphite.c (graphite_transform_loops): Diagnose code
1448 generation issues as MSG_MISSED_OPTIMIZATION and continue
1449 with processing SCOPs.
1451 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1452 Alan Hayward <alan.hayward@arm.com>
1453 David Sherwood <david.sherwood@arm.com>
1455 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1456 rather than choosing between plus_constant and gen_rtx_<CODE>.
1457 * expr.c (emit_push_insn): Likewise.
1458 (expand_expr_real_2): Likewise.
1460 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1461 Alan Hayward <alan.hayward@arm.com>
1462 David Sherwood <david.sherwood@arm.com>
1464 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1466 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1467 Alan Hayward <alan.hayward@arm.com>
1468 David Sherwood <david.sherwood@arm.com>
1470 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1471 calling tree_to_uhwi.
1473 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1474 Alan Hayward <alan.hayward@arm.com>
1475 David Sherwood <david.sherwood@arm.com>
1477 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1478 INTEGER_CST rather than a negative test for ADDR_EXPR.
1480 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1485 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1487 2017-09-21 Richard Biener <rguenther@suse.de>
1489 PR tree-optimization/71351
1490 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1491 graphite_create_new_loop_guard): Remove, fold remaining parts
1493 (translate_isl_ast_node_for): ... here and simplify.
1495 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1498 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1499 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1500 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1501 alternative always use QI mode, for -Os imov (=R,R) alternative
1502 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1505 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1506 Jeff Law <law@redhat.com>
1508 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1509 (allocate_stack_space): New function, partially extracted from
1511 (s390_emit_prologue): Track offset to most recent stack probe.
1512 Code to allocate space moved into allocate_stack_space.
1513 Dump actions when no stack is allocated.
1514 (s390_prologue_plus_offset): New function.
1515 (s390_emit_stack_probe): Likewise.
1517 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1519 * common.opt (Wa, Wl, Wp, g, gz=): Add
1521 (gno-column-info): Remove.
1522 (gcolumn-info): Drop RejectNegative.
1524 (gno-record-gcc-switches): Remove.
1525 (grecord-gcc-switches): Drop RejectNegative.
1526 (gno-split-dwarf): Remove.
1527 (gsplit-dwarf): Drop RejectNegative.
1528 (gno-strict-dwarf): Remove.
1529 (gstrict-dwarf): Drop RejectNegative.
1530 * config/darwin.opt (gfull, gused): Add RejectNegative.
1531 * dwarf2out.c (gen_producer_string): Drop
1532 gno-record-gcc-switches handler.
1533 * optc-gen.awk: Add g to prefixes with negative forms.
1534 * opts-common.c (remapping_prefix_p): New.
1535 (find_opt): Check it.
1536 (generate_canonical_option): Test g prefix.
1537 (option_map): Add -gno- mapping.
1538 (add_misspelling_candidates): Check remapping_prefix_p.
1540 2017-09-20 Jeff Law <law@redhat.com>
1542 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1543 thinko in stack clash protection support.
1545 * explow.c (compute_stack_clash_protection_loop_data): Use
1546 CONST_INT_P instead of explicit test. Verify object is a
1547 CONST_INT_P before looking at INTVAL.
1548 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1549 instead of explicit test.
1551 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1554 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1555 address instead of to r1 and r11.
1557 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1559 * config.gcc: Support "knm".
1560 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1561 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1563 * config/i386/i386.c (m_KNM): Define.
1564 (processor_target_table): Add "knm".
1566 (ix86_option_override_internal): Add "knm".
1567 (ix86_issue_rate): Add PROCESSOR_KNM.
1568 (ix86_adjust_cost): Ditto.
1569 (ia32_multipass_dfa_lookahead): Ditto.
1570 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1571 (fold_builtin_cpu): Add M_INTEL_KNM.
1572 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1573 (processor_type): Add PROCESSOR_KNM.
1574 * config/i386/x86-tune.def: Add m_KNM.
1575 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1577 2017-09-20 Richard Biener <rguenther@suse.de>
1579 PR tree-optimization/80213
1580 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1581 are allowed in empty BBs as well.
1582 (canonicalize_loop_closed_ssa): Also look for other complex
1584 (scop_detection::get_sese): Include the loop-closed PHI block
1586 (scop_detection::merge_sese): Remove code adding extra blocks.
1587 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1588 (build_scops): Assert the final returned scop is invalid.
1590 2017-09-20 Richard Biener <rguenther@suse.de>
1592 PR tree-optimization/82264
1593 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1595 (vn_phi_lookup): Likewise.
1596 (vn_phi_insert): Likewise.
1598 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1600 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1601 that fits into uhwi or shwi, add DW_AT_const_value regardless
1602 of early_dwarf without going through RTL, using add_AT_unsigned
1605 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1606 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1607 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1608 (DEBUG_MACRO_SECTION): Likewise.
1609 (DEBUG_LTO_MACRO_SECTION): Likewise.
1610 (DEBUG_STR_DWO_SECTION): Likewise.
1611 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1612 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1613 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1614 (DEBUG_LTO_LINE_STR_SECTION): Define.
1615 (init_sections_and_labels): Initialize debug_line_str_section
1616 variable. Initialize debug_loc_section for -gdwarf-5 to
1617 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1619 2017-09-20 Richard Biener <rguenther@suse.de>
1621 * graphite-sese-to-poly.c (extract_affine): Properly handle
1622 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1624 2017-09-20 Richard Biener <rguenther@suse.de>
1626 PR tree-optimization/81373
1627 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1628 Force SESE live-out defs to be handled even if they are
1631 2017-09-19 Jeff Law <law@redhat.com>
1633 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1634 nothing for stack adjustments with REG_STACK_CHECK.
1635 * sched-deps.c (parse_add_or_inc): Reject insns with
1636 REG_STACK_CHECK from dependency breaking.
1637 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1638 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1639 * reg-notes.def (STACK_CHECK): New note.
1641 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1642 (ix86_expand_prologue): Dump stack clash info as needed.
1643 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1645 * function.c (dump_stack_clash_frame_info): New function.
1646 * function.h (dump_stack_clash_frame_info): Prototype.
1647 (enum stack_clash_probes): New enum.
1649 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1650 flag_stack_clash_protection.
1651 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1652 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1653 (arm_frame_pointer_required): Likewise.
1654 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1655 (ia64_expand_prologue): Likewise.
1656 * config/mips/mips.c (mips_expand_prologue): Likewise.
1657 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1658 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1659 (sparc_flat_expand_prologue): Likewise.
1660 * config/spu/spu.c (spu_expand_prologue): Likewise.
1662 * explow.c: Include "params.h".
1663 (anti_adjust_stack_and_probe_stack_clash): New function.
1664 (get_stack_check_protect): Likewise.
1665 (compute_stack_clash_protection_loop_data): Likewise.
1666 (emit_stack_clash_protection_loop_start): Likewise.
1667 (emit_stack_clash_protection_loop_end): Likewise.
1668 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1669 Use anti_adjust_stack_and_probe_stack_clash.
1670 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1671 (emit_stack_clash_protection_loop_start): Likewise.
1672 (emit_stack_clash_protection_loop_end): Likewise.
1673 * rtl.h (get_stack_check_protect): Prototype.
1674 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1675 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1676 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1678 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1680 * doc/tm.texi: Rebuilt.
1681 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1682 get_stack_check_protect.
1683 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1684 * config/arm/arm.c (arm_expand_prologue): Likewise.
1685 (arm_frame_pointer_required): Likewise.
1686 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1687 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1688 * config/mips/mips.c (mips_expand_prologue): Likewise.
1689 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1690 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1691 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1692 (sparc_flat_expand_prologue): Likewise.
1694 * common.opt (-fstack-clash-protection): New option.
1695 * flag-types.h (enum stack_check_type): Note difference between
1696 -fstack-check= and -fstack-clash-protection.
1697 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1698 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1699 * toplev.c (process_options): Issue warnings/errors for cases
1700 not handled with -fstack-clash-protection.
1701 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1702 (-fstack-check): Note additional problem with -fstack-check=generic.
1703 Note that -fstack-check is primarily for Ada and refer users
1704 to -fstack-clash-protection for stack-clash-protection.
1705 Document new params for stack clash protection.
1707 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1709 * config/i386/i386.c (ix86_split_long_move): Do not handle
1710 address used for LEA in a special way.
1712 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1714 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1716 2017-09-19 Martin Sebor <msebor@redhat.com>
1719 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1720 of incompatible types.
1722 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1724 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1725 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1726 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1727 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1729 2017-09-19 Richard Biener <rguenther@suse.de>
1731 PR tree-optimization/82244
1732 * tree-vrp.c (remove_range_assertions): Do not propagate
1733 a constant to abnormals but replace the assert with a copy.
1735 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1737 PR rtl-optimization/57878
1738 PR rtl-optimization/68988
1739 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1740 avoidance test involving non_reload_pseudos. Move frequency test
1741 below the general fragmentation avoidance test.
1743 2017-09-19 Richard Biener <rguenther@suse.de>
1745 PR tree-optimization/69728
1746 * graphite-sese-to-poly.c (schedule_error): New global.
1747 (add_loop_schedule): Handle empty domain by failing the
1749 (build_original_schedule): Handle schedule_error.
1751 2017-09-19 Richard Biener <rguenther@suse.de>
1753 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1754 Do not iterate to sibling loops but only to siblings of inner
1757 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1760 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1761 that the registers are different.
1763 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1765 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1766 to processor_model and "amdfam17h" to arch_names_table.
1767 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1769 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1772 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1774 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1775 Alan Hayward <alan.hayward@arm.com>
1776 David Sherwood <david.sherwood@arm.com>
1778 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1780 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1782 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1783 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1784 vect_schedule_slp_instance.
1785 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1786 Update call to vect_slp_analyze_node_operations. Simplify return
1788 (vect_slp_analyze_bb_1): Update call accordingly.
1789 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1790 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1791 (vect_schedule_slp): Update call accordingly.
1793 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1794 Alan Hayward <alan.hayward@arm.com>
1795 David Sherwood <david.sherwood@arm.com>
1797 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
1798 with types that aren't in fact scalar.
1800 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1802 * tree-vect-slp.c (vect_record_max_nunits): New function,
1804 (vect_build_slp_tree_1): ...here.
1805 (vect_build_slp_tree_2): Call it for phis too.
1807 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1809 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1810 to vect_get_vec_def_for_operand when getting the mask operand.
1812 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1813 Alan Hayward <alan.hayward@arm.com>
1814 David Sherwood <david.sherwood@arm.com>
1816 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1819 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1820 Alan Hayward <alan.hayward@arm.com>
1821 David Sherwood <david.sherwood@arm.com>
1823 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1824 calculation for vector booleans.
1826 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1827 Alan Hayward <alan.hayward@arm.com>
1828 David Sherwood <david.sherwood@arm.com>
1830 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1832 (vect_transform_stmt): ...here.
1833 (vect_analyze_stmt): Use it instead of calling
1834 vectorizable_live_operation directly.
1836 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1838 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1839 non-SIMT targets in acc vector loops.
1841 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1843 * configure.ac: Add arc and check if assembler supports gdwarf2.
1844 * configure: Regenerate.
1846 2017-09-18 Richard Biener <rguenther@suse.de>
1848 PR tree-optimization/82220
1849 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1850 epilogue niters from the min_profitable_iters compute.
1852 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1855 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1856 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1857 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1859 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1862 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1863 switching to a new text section.
1865 2017-09-18 Richard Biener <rguenther@suse.de>
1867 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1869 (build_alias_set): Reject aliases with no access function.
1871 2017-09-18 Richard Biener <rguenther@suse.de>
1873 PR tree-optimization/79622
1874 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1876 (build_cross_bb_scalars_use): Likewise.
1878 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1880 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1882 2017-09-18 Alan Modra <amodra@gmail.com>
1885 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1886 stack_pointer_rtx for count 0. Update comments. Break up
1887 large rtl expression.
1889 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1891 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1892 Increase to 20 bytes.
1893 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1894 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1895 or avx version of the stub.
1897 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1900 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1901 compute the minimum stack alignment. Also update preferred stack
1902 boundary for leaf functions.
1904 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1906 PR tree-optimization/82228
1907 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1910 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1912 * common/config/nds32/nds32-common.c
1913 (nds32_option_optimization_table): Refine formatting.
1914 (nds32_option_optimization_table): Use -fsched-pressure and
1915 -fomit-frame-pointer for specific optimization level.
1917 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1919 * config/nds32/nds32.c: Refine formatting and comments.
1920 * config/nds32/nds32.h: Likewise.
1921 * config/nds32/nds32.md: Likewise.
1922 * config/nds32/nds32-cost.c: Likewise.
1923 * config/nds32/nds32-isr.c: Likewise.
1924 * config/nds32/nds32-md-auxiliary.c: Likewise.
1925 * config/nds32/nds32-multiple.md: Likewise.
1926 * config/nds32/nds32-predicates.c: Likewise.
1928 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1929 Jakub Jelinek <jakub@redhat.com>
1931 Add support for -std=c++2a.
1932 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1934 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1936 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1939 * doc/extend.texi (Common Function Attributes): Add
1940 references to ARM, AArch64, and S/390 specific attributes.
1941 (Function Specific Option Pragmas): Add AArch64 and S/390
1942 to list of back ends that support the target pragma.
1944 2017-09-15 Nathan Sidwell <nathan@acm.org>
1946 * doc/standards.texi: Fix C++17 description. Update URLs for
1949 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1951 * common.opt (Wcast-align=strict): New warning option.
1952 * doc/invoke.texi: Document -Wcast-align=strict.
1954 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1956 * cgraph.h (cgraph_thunk_info): Add comments.
1957 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1958 assert for VIRTUAL_* arguments stricter.
1960 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1962 PR tree-optimization/71026
1963 * match.pd: Move RDIV patterns from fold-const.c
1964 * fold-const.c (distribute_real_division): Removed.
1965 (fold_binary_loc): Remove calls to distribute_real_divison.
1967 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1969 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1970 c++1z and gnu++1z as deprecated. Change other references to
1971 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1972 Change -Wc++1z-compat to -Wc++17-compat.
1973 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1974 * dwarf2out.c (highest_c_language): Handle C++17.
1975 (gen_compile_unit_die): Likewise.
1977 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1979 PR rtl-optimization/82192
1980 * combine.c (make_extraction): Don't look through non-paradoxical
1981 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1984 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1985 Alan Hayward <alan.hayward@arm.com>
1986 David Sherwood <david.sherwood@arm.com>
1988 * target.def (function_arg_offset): New hook.
1989 * targhooks.h (default_function_arg_offset): Declare.
1990 * targhooks.c (default_function_arg_offset): New function.
1991 * function.c (locate_and_pad_parm): Use
1992 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1993 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1994 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1995 * doc/tm.texi: Regenerate.
1996 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1997 * config/spu/spu.c (spu_function_arg_offset): New function.
1998 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1999 * system.h (FUNCTION_ARG_OFFSET): Poison.
2001 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2002 Alan Hayard <alan.hayward@arm.com>
2003 David Sherwood <david.sherwood@arm.com>
2005 * target.def (truly_noop_truncation): New hook.
2006 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2007 than TRULY_NOOP_TRUNCATION.
2008 * hooks.h (hook_bool_uint_uint_true): Declare.
2009 * hooks.c (hook_bool_uint_uint_true): New function.
2010 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2011 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2012 * doc/tm.texi: Regenerate.
2013 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2014 rather than TRULY_NOOP_TRUNCATION in comments.
2015 (simplify_comparison): Likewise.
2016 (record_truncated_value): Likewise.
2017 * expmed.c (extract_bit_field_1): Likewise.
2018 (extract_split_bit_field): Likewise.
2019 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2020 instead of TRULY_NOOP_TRUNCATION.
2021 * function.c (assign_parm_setup_block): Likewise.
2022 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2023 * rtlhooks.c: Include target.h.
2024 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2025 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2026 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2027 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2028 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2029 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2030 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2031 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2032 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2033 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2034 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2035 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2036 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2037 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2038 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2039 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2040 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2041 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2042 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2043 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2044 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2045 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2046 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2047 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2048 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2049 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2050 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2051 rather than TRULY_NOOP_TRUNCATION in comments.
2052 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2053 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2054 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2055 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2056 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2057 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2058 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2059 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2060 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2061 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2062 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2063 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2064 rather than TRULY_NOOP_TRUNCATION in comments.
2065 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2066 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2067 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2068 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2069 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2070 TRULY_NOOP_TRUNCATION condition.
2071 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2072 (TRULY_NOOP_TRUNCATION): Delete.
2073 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2074 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2075 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2076 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2077 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2078 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2079 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2080 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2081 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2082 rather than TRULY_NOOP_TRUNCATION in comments.
2083 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2084 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2085 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2086 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2087 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2088 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2090 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2093 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2094 (*cmp_ior): Likewise.
2095 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2096 (*ior_scc_scc_cmp): Likewise.
2097 (*and_scc_scc): Likewise.
2098 (*and_scc_scc_cmp): Likewise.
2100 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2101 Alan Hayard <alan.hayward@arm.com>
2102 David Sherwood <david.sherwood@arm.com>
2104 * target.def (can_change_mode_class): New hook.
2105 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2106 (hard_regno_nregs): Likewise.
2107 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2108 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2109 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2110 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2111 (LOAD_EXTEND_OP): Update accordingly.
2112 * doc/tm.texi: Regenerate.
2113 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2114 CANNOT_CHANGE_MODE_CLASS.
2115 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2116 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2117 * combine.c (simplify_set): Update accordingly.
2118 * emit-rtl.c (validate_subreg): Likewise.
2119 * recog.c (general_operand): Likewise.
2120 * regcprop.c (mode_change_ok): Likewise.
2121 * reload1.c (choose_reload_regs): Likewise.
2122 (inherit_piecemeal_p): Likewise.
2123 * rtlanal.c (simplify_subreg_regno): Likewise.
2124 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2125 instead of CANNOT_CHANGE_MODE_CLASS.
2126 (reload_cse_simplify_operands): Likewise.
2127 * reload.c (push_reload): Use targetm.can_change_mode_class
2128 instead of CANNOT_CHANGE_MODE_CLASS.
2129 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2130 REG_CANNOT_CHANGE_MODE_P.
2131 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2132 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2133 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2134 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2135 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2136 (arm_can_change_mode_class): New function.
2137 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2138 than CANNOT_CHANGE_MODE_CLASS in comments.
2139 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2140 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2141 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2142 (ix86_can_change_mode_class): ...this new function, inverting the
2143 sense of the return value.
2144 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2145 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2146 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2147 (ia64_can_change_mode_class): New function.
2148 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2149 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2150 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2151 (m32c_can_change_mode_class): ...this new function, inverting the
2152 sense of the return value.
2153 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2154 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2155 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2156 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2157 (mips_can_change_mode_class): ...this new function, inverting the
2158 sense of the return value.
2159 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2160 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2161 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2162 (msp430_can_change_mode_class): New function.
2163 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2164 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2165 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2166 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2167 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2168 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2169 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2170 (pa_cannot_change_mode_class): Replace with...
2171 (pa_can_change_mode_class): ...this new function, inverting the
2172 sense of the return value.
2173 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2174 than CANNOT_CHANGE_MODE_CLASS in comments.
2175 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2176 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2177 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2178 (pdp11_cannot_change_mode_class): Replace with...
2179 (pdp11_can_change_mode_class): ...this new function, inverting the
2180 sense of the return value.
2181 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2182 * config/powerpcspe/powerpcspe-protos.h
2183 (rs6000_cannot_change_mode_class_ptr): Delete.
2184 * config/powerpcspe/powerpcspe.c
2185 (rs6000_cannot_change_mode_class_ptr): Delete.
2186 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2187 (rs6000_option_override_internal): Assign to
2188 targetm.can_change_mode_class instead of
2189 rs6000_cannot_change_mode_class_ptr.
2190 (rs6000_cannot_change_mode_class): Replace with...
2191 (rs6000_can_change_mode_class): ...this new function, inverting the
2192 sense of the return value.
2193 (rs6000_debug_cannot_change_mode_class): Replace with...
2194 (rs6000_debug_can_change_mode_class): ...this new function.
2195 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2196 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2197 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2198 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2199 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2201 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2202 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2203 (rs6000_option_override_internal): Assign to
2204 targetm.can_change_mode_class instead of
2205 rs6000_cannot_change_mode_class_ptr.
2206 (rs6000_cannot_change_mode_class): Replace with...
2207 (rs6000_can_change_mode_class): ...this new function, inverting the
2208 sense of the return value.
2209 (rs6000_debug_cannot_change_mode_class): Replace with...
2210 (rs6000_debug_can_change_mode_class): ...this new function.
2211 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2212 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2213 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2214 (s390_can_change_mode_class): ...this new function, inverting the
2215 sense of the return value.
2216 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2217 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2218 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2219 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2220 (sh_cannot_change_mode_class): Replace with...
2221 (sh_can_change_mode_class): ...this new function, inverting the
2222 sense of the return value.
2223 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2224 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2225 (sparc_can_change_mode_class): New function.
2226 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2227 * config/spu/spu.c (spu_can_change_mode_class): New function.
2228 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2229 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2230 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2231 (visium_can_change_mode_class): New function.
2232 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2234 2017-09-15 Richard Biener <rguenther@suse.de>
2236 PR tree-optimization/82217
2237 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2238 but not undefined case.
2240 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2243 * postreload.c (reload_cse_simplify_operands): Skip
2244 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2246 2017-09-15 Richard Biener <rguenther@suse.de>
2248 PR tree-optimization/68823
2249 * graphite-scop-detection.c (build_alias_set): If we have a
2250 possible dependence check whether we can handle them by just
2251 looking at the DRs DR_ACCESS_FNs.
2252 (build_scops): If build_alias_set fails, fail the SCOP.
2254 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2256 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2257 to support float128 built-in functions that require the ISA 3.0
2259 (BU_FLOAT128_3_HW): Likewise.
2260 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2262 (FMAF128): Likewise.
2264 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2265 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2266 floating point instructions.
2267 (rs6000_invalid_builtin): Likewise.
2268 (rs6000_builtin_mask_names): Likewise.
2269 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2270 (RS6000_BTM_FLOAT128_HW): Likewise.
2271 (RS6000_BTM_COMMON): Likewise.
2272 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2274 * doc/extend.texi (RS/6000 built-in functions): Document the
2275 IEEE 128-bit floating point square root and fused multiply-add
2278 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2280 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2281 reg (r2) isn't in the set of registers defined in the prologue.
2283 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2284 Alan Hayward <alan.hayward@arm.com>
2285 David Sherwood <david.sherwood@arm.com>
2287 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2288 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2289 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2290 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2291 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2293 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2294 max_vectorization_factor.
2295 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2297 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2298 Alan Hayward <alan.hayward@arm.com>
2299 David Sherwood <david.sherwood@arm.com>
2301 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2302 (vect_worthwhile_without_simd_p): Declare.
2303 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2304 (vectorizable_reduction): Use it.
2305 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2306 (vectorizable_operation): Likewise.
2308 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2309 Alan Hayward <alan.hayward@arm.com>
2310 David Sherwood <david.sherwood@arm.com>
2312 * tree-vectorizer.h (vect_get_num_copies): New function.
2313 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2314 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2315 (vectorizable_induction): Likewise.
2316 (vectorizable_live_operation): Likewise.
2317 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2318 (vectorizable_bswap): Likewise.
2319 (vectorizable_call): Likewise.
2320 (vectorizable_conversion): Likewise.
2321 (vectorizable_assignment): Likewise.
2322 (vectorizable_shift): Likewise.
2323 (vectorizable_operation): Likewise.
2324 (vectorizable_store): Likewise.
2325 (vectorizable_load): Likewise.
2326 (vectorizable_condition): Likewise.
2327 (vectorizable_comparison): Likewise.
2328 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2330 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2331 Alan Hayward <alan.hayward@arm.com>
2332 David Sherwood <david.sherwood@arm.com>
2334 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2335 of vect_init_vector.
2337 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2338 Alan Hayward <alan.hayward@arm.com>
2339 David Sherwood <david.sherwood@arm.com>
2341 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2342 an inline wrapper that provides a location.
2343 (gimple_build_vector): Likewise.
2344 * gimple-fold.c (gimple_build_vector_from_val): New function.
2345 (gimple_build_vector): Likewise.
2346 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2347 functions to build the initial value. Always return a gimple value.
2348 (get_initial_defs_for_reduction): Likewise. Only compute
2350 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2351 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2352 (vectorizable_induction): Use gimple_build_vector rather than
2355 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2356 Alan Hayward <alan.hayward@arm.com>
2357 David Sherwood <david.sherwood@arm.com>
2359 * target.h (vec_perm_indices): New typedef.
2360 (auto_vec_perm_indices): Likewise.
2361 * optabs-query.h: Include target.h
2362 (can_vec_perm_p): Take a vec_perm_indices *.
2363 * optabs-query.c (can_vec_perm_p): Likewise.
2364 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2365 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2366 * tree-vect-generic.c (lower_vec_perm): Likewise.
2367 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2368 (vect_grouped_load_supported): Likewise.
2369 (vect_shift_permute_load_chain): Likewise.
2370 (vect_permute_store_chain): Use auto_vec_perm_indices.
2371 (vect_permute_load_chain): Likewise.
2372 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2373 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2374 Update uses of can_vec_perm_p.
2375 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2376 mode with a number of elements. Take a vec_perm_indices *.
2377 (vect_create_epilog_for_reduction): Update accordingly.
2378 Use auto_vec_perm_indices.
2379 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2380 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2381 (vect_transform_slp_perm_load): Likewise.
2382 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2383 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2384 (vect_gen_perm_mask_checked): Likewise.
2385 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2386 (vect_gen_perm_mask_checked): Likewise.
2387 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2388 (vectorizable_store): Likewise.
2389 (vectorizable_load): Likewise.
2390 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2391 (vectorizable_bswap): Likewise.
2393 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2394 Alan Hayward <alan.hayward@arm.com>
2395 David Sherwood <david.sherwood@arm.com>
2397 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2398 * tree.c (build_vector): Likewise.
2399 (build_vector_from_ctor): Update accordingly.
2400 (build_vector_from_val): Likewise.
2401 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2402 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2403 * tree-vect-generic.c (add_rshift): Likewise.
2404 (expand_vector_divmod): Likewise.
2405 (optimize_vector_constructor): Likewise.
2406 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2407 (vect_transform_slp_perm_load): Likewise.
2408 (vect_schedule_slp_instance): Likewise.
2409 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2410 (vectorizable_call): Likewise.
2411 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2412 * expmed.c (make_tree): Likewise.
2413 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2414 a vector passed to build_vector.
2415 (fold_convert_const): Likewise.
2416 (exact_inverse): Likewise.
2417 (fold_ternary_loc): Likewise.
2418 (fold_relational_const): Likewise.
2419 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2420 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2421 (const_unop): Likewise. Store the reduction accumulator in a
2422 variable rather than an array.
2423 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2424 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2425 the new vector, rather than constructing it after the input arrays.
2426 (native_interpret_vector): Use auto_vec<tree> when building
2427 a vector passed to build_vector. Add elements in order.
2428 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2429 auto_vec<tree> when building a vector passed to build_vector.
2430 (vect_create_epilog_for_reduction): Likewise.
2431 (vectorizable_induction): Likewise.
2432 (get_initial_def_for_reduction): Likewise. Fix indentation of
2434 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2436 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2437 passed to build_vector.
2439 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2440 Alan Hayward <alan.hayward@arm.com>
2441 David Sherwood <david.sherwood@arm.com>
2443 * tree-core.h (tree_base::u): Add an "nelts" field.
2444 (tree_vector): Use VECTOR_CST_NELTS as the length.
2445 * tree.c (tree_size): Likewise.
2446 (make_vector): Initialize VECTOR_CST_NELTS.
2447 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2448 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2449 TYPE_VECTOR_SUBPARTS.
2450 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2451 as the number of units, setting it from VECTOR_CST_NELTS.
2452 (const_vector_from_tree): Likewise.
2453 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2454 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2455 (fold_negate_expr_1): Likewise.
2456 (fold_convert_const): Likewise.
2457 (const_binop): Likewise. Differentiate the number of output and
2459 (const_unop): Likewise.
2460 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2461 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2462 in cases that did the opposite.
2464 2017-09-14 Richard Biener <rguenther@suse.de>
2466 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2469 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2471 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2473 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2476 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2477 if and where to split a bb, except for splitting before debug insn
2478 sequences followed by non-label real insn. Delete debug insns
2479 in between basic blocks.
2481 * combine.c (make_compound_operation_int): Formatting fixes.
2483 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2484 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2485 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2486 * config/sol2.h (LINK_EH_SPEC): Likewise.
2487 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2488 * config/s390/linux.h (LINK_SPEC): Likewise.
2489 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2490 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2491 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2492 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2493 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2494 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2495 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2496 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2498 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2500 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2502 (ENDFILE_LINUX_SPEC): Likewise.
2503 (LINK_EH_SPEC): Likewise.
2504 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2505 (LINK_OS_LINUX_SPEC32): Likewise.
2506 (LINK_OS_LINUX_SPEC64): Likewise.
2507 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2508 (LINK_OS_LINUX_SPEC): Likewise.
2510 2017-09-13 Martin Liska <mliska@suse.cz>
2513 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2514 CASE_HIGH is NULL_TREE.
2516 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2517 Alan Hayward <alan.hayward@arm.com>
2518 David Sherwood <david.sherwood@arm.com>
2520 * target.def (secondary_memory_needed): New hook.
2521 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2522 instead of SECONDARY_MEMORY_NEEDED.
2523 (secondary_memory_needed_mode): Likewise.
2524 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2525 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2526 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2527 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2528 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2529 * doc/tm.texi: Regenerate.
2530 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2531 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2532 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2533 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2534 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2535 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2536 mode argument first and change the reg_class arguments to reg_class_t.
2537 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2538 Make static. Update the call to inline_secondary_memory_needed.
2539 (ix86_register_move_cost): Update the call to
2540 inline_secondary_memory_needed.
2541 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2542 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2544 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2545 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2547 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2548 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2549 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2550 and match hook interface. Add comment from mips.h.
2551 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2552 * config/mmix/mmix.md (truncdfsf2): Refer to
2553 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2555 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2556 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2557 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2558 (pa_secondary_memory_needed): New function.
2559 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2560 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2561 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2562 (pdp11_secondary_memory_needed): Make static and match hook interface.
2563 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2564 * config/powerpcspe/powerpcspe-protos.h
2565 (rs6000_secondary_memory_needed_ptr): Delete.
2566 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2568 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2569 (rs6000_option_override_internal): Assign to
2570 targetm.secondary_memory_needed rather than
2571 rs6000_secondary_memory_needed_ptr.
2572 (rs6000_secondary_memory_needed): Match hook interface.
2573 (rs6000_debug_secondary_memory_needed): Likewise.
2574 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2575 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2576 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2577 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2578 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2579 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2581 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2582 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2583 (rs6000_option_override_internal): Assign to
2584 targetm.secondary_memory_needed rather than
2585 rs6000_secondary_memory_needed_ptr.
2586 (rs6000_secondary_memory_needed): Match hook interface.
2587 (rs6000_debug_secondary_memory_needed): Likewise.
2588 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2589 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2590 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2591 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2592 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2593 (sparc_secondary_memory_needed): New function.
2594 * lra-constraints.c (check_and_process_move): Refer to
2595 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2597 (curr_insn_transform): Likewise.
2598 (process_alt_operands): Use targetm.secondary_memory_needed
2599 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2600 (check_secondary_memory_needed_p): Likewise.
2601 (choose_split_class): Likewise.
2602 * reload.c: Unconditionally include code that was previously
2603 conditional on SECONDARY_MEMORY_NEEDED.
2604 (push_secondary_reload): Use targetm.secondary_memory_needed
2605 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2606 (push_reload): Likewise.
2607 * reload1.c: Unconditionally include code that was previously
2608 conditional on SECONDARY_MEMORY_NEEDED.
2609 (choose_reload_regs): Use targetm.secondary_memory_needed
2610 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2611 (gen_reload): Likewise.
2612 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2614 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2615 Alan Hayward <alan.hayward@arm.com>
2616 David Sherwood <david.sherwood@arm.com>
2618 * target.def (secondary_memory_needed_mode): New hook:
2619 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2620 * targhooks.h (default_secondary_memory_needed_mode): New function.
2621 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2622 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2623 * doc/tm.texi: Regenerate.
2624 * lra-constraints.c (check_and_process_move): Use
2625 targetm.secondary_memory_needed_mode instead of
2626 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2627 (curr_insn_transform): Likewise.
2628 * reload.c (get_secondary_mem): Likewise.
2629 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2630 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2632 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2633 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2634 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2635 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2636 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2638 * config/powerpcspe/powerpcspe-protos.h
2639 (rs6000_secondary_memory_needed_mode): Delete.
2640 * config/powerpcspe/powerpcspe.c
2641 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2642 (rs6000_secondary_memory_needed_mode): Make static.
2643 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2644 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2646 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2648 (rs6000_secondary_memory_needed_mode): Make static.
2649 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2650 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2651 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2652 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2653 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2655 (sparc_secondary_memory_needed_mode): New function.
2656 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2658 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2660 * config/aarch64/constraints.md (Umq): New constraint.
2661 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2663 (mov<mode>): Update condition.
2665 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2667 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2668 when bitposition is the same.
2670 2017-09-13 Richard Biener <rguenther@suse.de>
2672 * dwarf2out.c (output_die_symbol): Remove.
2673 (output_die): Do not output a DIEs symbol.
2675 2017-09-13 Richard Biener <rguenther@suse.de>
2678 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2679 default-def to avoid breaking iterator update with the weird
2680 interaction with cgraph_update_edges_for_call_stmt_node.
2682 2017-09-13 Richard Biener <rguenther@suse.de>
2684 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2685 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2686 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2687 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2689 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2691 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2692 Disable pc relative literal load irrespective of
2693 TARGET_FIX_ERR_A53_84341 for default.
2695 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2697 * config/sparc/sparc.c (output_return): Output the source location of
2698 the insn in the delay slot, if any.
2699 (output_sibcall): Likewise.
2701 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2704 * common.opt (-static-pie): New alias.
2705 (shared): Negate static-pie.
2706 (-no-pie): Update help text.
2708 (static-pie): New option.
2709 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2710 -static-pie support.
2711 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2712 (LINK_EH_SPEC): Likewise.
2713 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2714 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2715 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2716 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2717 (init_gcc_specs): Likewise.
2718 (init_spec): Likewise.
2719 (display_help): Update help message for -pie.
2720 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2723 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2725 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2726 (movdi_aarch64): Likewise.
2727 (movti_aarch64): Likewise.
2729 2017-09-12 Simon Wright <simon@pushface.org>
2732 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2733 calculation of the minor version, always output as 0.
2735 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2738 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2739 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2740 on it early, rather than manual conversion late. For
2741 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2742 instead of performing manual conversion.
2744 2017-09-12 Carl Love <cel@us.ibm.com>
2746 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2747 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2749 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2750 VMULOSW): Add definitions.
2751 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2752 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2753 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2754 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2755 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2757 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2759 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2761 (movti_aarch64): Likewise.
2762 (movdf_aarch64): Likewise.
2763 (movtf_aarch64): Likewise.
2764 (load_pairdi): Likewise.
2765 (store_pairdi): Likewise.
2766 (load_pairdf): Likewise.
2767 (store_pairdf): Likewise.
2768 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2769 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2770 (ldr_got_small_<mode>): Likewise.
2771 (ldr_got_small_28k_<mode>): Likewise.
2772 (ldr_got_tiny): Likewise.
2773 * config/aarch64/iterators.md (ldst_sz): New.
2774 (ldpstp_sz): Likewise.
2775 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2777 (thunderx_load): Split load_8 to load_16.
2778 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2780 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2781 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2782 (xgene1_store_pair): Split store_8 to store_16.
2783 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2784 (falkor_st_0_st_sd): Split store_8 to store_16.
2786 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2788 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2789 and store1/2/3/4 to store_4/8/12/16.
2790 * config/aarch64/aarch64.md: Update for rename.
2791 * config/arm/arm.md: Likewise.: Likewise.
2792 * config/arm/arm.c: Likewise.
2793 * config/arm/thumb1.md: Likewise.
2794 * config/arm/thumb2.md: Likewise.
2795 * config/arm/vfp.md: Likewise.
2796 * config/arm/arm-generic.md: Likewise.
2797 * config/arm/arm1020e.md: Likewise.
2798 * config/arm/arm1026ejs.md: Likewise.
2799 * config/arm/arm1136jfs.md: Likewise.
2800 * config/arm/arm926ejs.md: Likewise.
2801 * config/arm/cortex-a15.md: Likewise.
2802 * config/arm/cortex-a17.md: Likewise.
2803 * config/arm/cortex-a5.md: Likewise.
2804 * config/arm/cortex-a53.md: Likewise.
2805 * config/arm/cortex-a57.md: Likewise.
2806 * config/arm/cortex-a7.md: Likewise.
2807 * config/arm/cortex-a8.md: Likewise.
2808 * config/arm/cortex-a9.md: Likewise.
2809 * config/arm/cortex-m4.md: Likewise.
2810 * config/arm/cortex-m7.md: Likewise.
2811 * config/arm/cortex-r4.md: Likewise.
2812 * config/arm/exynos-m1.md: Likewise.
2813 * config/arm/fa526.md: Likewise.
2814 * config/arm/fa606te.md: Likewise.
2815 * config/arm/fa626te.md: Likewise.
2816 * config/arm/fa726te.md: Likewise.
2817 * config/arm/fmp626.md: Likewise.
2818 * config/arm/iwmmxt.md: Likewise.
2819 * config/arm/ldmstm.md: Likewise.
2820 * config/arm/marvell-pj4.md: Likewise.
2821 * config/arm/xgene1.md: Likewise.
2822 * config/aarch64/thunderx.md: Likewise.
2823 * config/aarch64/thunderx2t99.md: Likewise.
2824 * config/aarch64/falkor.md: Likewise.
2826 2017-09-12 Martin Liska <mliska@suse.cz>
2828 * attribs.c (private_lookup_attribute): New function.
2829 * attribs.h (private_lookup_attribute): Declared here.
2830 (lookup_attribute): Called from this place.
2832 2017-09-12 Richard Biener <rguenther@suse.de>
2834 PR tree-optimization/82157
2835 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2836 stmts with side-effects.
2838 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2839 Alan Hayward <alan.hayward@arm.com>
2840 David Sherwood <david.sherwood@arm.com>
2842 * target.def (hard_regno_nregs): New hook.
2843 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2844 * targhooks.h (default_hard_regno_nregs): Declare.
2845 * targhooks.c (default_hard_regno_nregs): New function.
2846 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2847 (TARGET_HARD_REGNO_NREGS): ...this hook.
2848 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2849 (CLASS_MAX_NREGS): Likewise.
2850 * doc/tm.texi: Regenerate.
2851 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2852 instead of HARD_REGNO_NREGS.
2853 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2854 HARD_REGNO_NREGS in the comment.
2855 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2856 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2857 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2858 Return an unsigned int.
2859 (TARGET_HARD_REGNO_NREGS): Redefine.
2860 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2861 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2862 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2863 (arc_hard_regno_nregs): New function.
2864 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2865 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2866 (arm_hard_regno_nregs): New function.
2867 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2868 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2869 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2870 (TARGET_HARD_REGNO_NREGS): Redefine.
2871 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2872 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2873 (HARD_REGNO_NREGS): Delete.
2874 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2875 (cr16_hard_regno_nregs): New function.
2876 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2877 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2878 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2879 (cris_hard_regno_nregs): New function.
2880 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2881 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2882 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2883 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2884 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2885 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2886 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2887 (frv_hard_regno_nregs): Make static. Take and return an
2889 (frv_class_max_nregs): Remove outdated copy of documentation.
2890 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2891 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2892 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2893 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2894 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2895 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2896 (TARGET_HARD_REGNO_NREGS): Redefine.
2897 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2898 (CLASS_MAX_NREGS): Update comment.
2899 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2900 (ia64_hard_regno_nregs): New function.
2901 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2902 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2903 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2904 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2905 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2907 (m32c_hard_regno_nregs): Likewise. Make static.
2908 (TARGET_HARD_REGNO_NREGS): Redefine.
2909 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2910 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2911 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2912 (m68k_hard_regno_nregs): New function.
2913 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2914 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2915 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2916 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2917 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2918 Take and return an unsigned int.
2919 (TARGET_HARD_REGNO_NREGS): Redefine.
2920 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2921 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2922 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2923 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2924 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2925 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2926 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2927 (msp430_hard_regno_nregs): Make static. Take and return an
2929 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2930 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2931 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2932 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2933 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2934 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2935 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2936 (TARGET_HARD_REGNO_NREGS): Redefine.
2937 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2938 (PA_HARD_REGNO_NREGS): ...this.
2939 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2940 (PA_HARD_REGNO_NREGS): ...this.
2941 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2942 (pa_hard_regno_nregs): New function.
2943 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2944 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2945 (pdp11_hard_regno_nregs): New function.
2946 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2947 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2948 (rs6000_hard_regno_nregs_hook): New function.
2949 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2950 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2951 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2952 Take and return an unsigned int. Move earlier in file.
2953 (TARGET_HARD_REGNO_NREGS): Redefine.
2954 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2955 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2956 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2957 (rl78_hard_regno_nregs): Make static. Take and return an
2959 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2960 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2961 (rs6000_hard_regno_nregs_hook): New function.
2962 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2963 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2964 (TARGET_HARD_REGNO_NREGS): Redefine.
2965 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2966 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2967 instead of HARD_REGNO_NREGS.
2968 (s390_hard_regno_nregs): New function.
2969 (s390_hard_regno_mode_ok): Add comment from s390.h.
2970 (TARGET_HARD_REGNO_NREGS): Redefine.
2971 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2972 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2973 (sh_hard_regno_nregs): New function.
2974 (sh_pass_in_reg_p): Use it.
2975 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2976 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2977 (sparc_hard_regno_nregs): New function.
2978 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2979 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2980 (spu_function_arg_advance): Use it, supplying a valid register number.
2981 (TARGET_HARD_REGNO_NREGS): Redefine.
2982 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2983 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2984 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2985 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2986 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2987 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2988 (CLASS_MAX_NREGS): Remove copy of old documentation.
2989 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2990 (visium_hard_regno_nregs): New function.
2991 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2992 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2993 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2994 xtensa_hard_regno_nregs): New function.
2995 * system.h (HARD_REGNO_NREGS): Poison.
2997 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2999 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3000 hard_regno_nregs instead of HARD_REGNO_NREGS.
3001 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3002 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3003 (c6x_expand_epilogue): Likewise.
3004 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3005 (frv_read_iacc_argument): Likewise.
3006 * config/sh/sh.c: Include regs.h.
3007 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3008 (regs_used): Likewise.
3009 (output_stack_adjust): Likewise.
3010 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3011 * expmed.c: Include regs.h.
3012 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3013 * ree.c: Include regs.h.
3014 (combine_reaching_defs): Use hard_regno_nregs instead of
3016 (add_removable_extension): Likewise.
3018 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3020 * regs.h (hard_regno_nregs): Turn into a function.
3021 (end_hard_regno): Update accordingly.
3022 * caller-save.c (setup_save_areas): Likewise.
3023 (save_call_clobbered_regs): Likewise.
3024 (replace_reg_with_saved_mem): Likewise.
3025 (insert_restore): Likewise.
3026 (insert_save): Likewise.
3027 * combine.c (can_change_dest_mode): Likewise.
3028 (move_deaths): Likewise.
3029 (distribute_notes): Likewise.
3030 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3031 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3032 (rs6000_split_multireg_move): Likewise.
3033 (rs6000_register_move_cost): Likewise.
3034 (rs6000_memory_move_cost): Likewise.
3035 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3036 (rs6000_split_multireg_move): Likewise.
3037 (rs6000_register_move_cost): Likewise.
3038 (rs6000_memory_move_cost): Likewise.
3039 * cselib.c (cselib_reset_table): Likewise.
3040 (cselib_lookup_1): Likewise.
3041 * emit-rtl.c (set_mode_and_regno): Likewise.
3042 * function.c (aggregate_value_p): Likewise.
3043 * ira-color.c (setup_profitable_hard_regs): Likewise.
3044 (check_hard_reg_p): Likewise.
3045 (calculate_saved_nregs): Likewise.
3046 (assign_hard_reg): Likewise.
3047 (improve_allocation): Likewise.
3048 (calculate_spill_cost): Likewise.
3049 * ira-emit.c (modify_move_list): Likewise.
3050 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3051 (ira_hard_reg_in_set_p): Likewise.
3052 * ira.c (setup_reg_mode_hard_regset): Likewise.
3053 (clarify_prohibited_class_mode_regs): Likewise.
3054 (check_allocation): Likewise.
3055 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3056 (lra_setup_reg_renumber): Likewise.
3057 (setup_try_hard_regno_pseudos): Likewise.
3058 (spill_for): Likewise.
3059 (assign_hard_regno): Likewise.
3060 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3061 * lra-constraints.c (in_class_p): Likewise.
3062 (lra_constraint_offset): Likewise.
3063 (simplify_operand_subreg): Likewise.
3064 (lra_constraints): Likewise.
3065 (split_reg): Likewise.
3066 (split_if_necessary): Likewise.
3067 (invariant_p): Likewise.
3068 (inherit_in_ebb): Likewise.
3069 * lra-lives.c (process_bb_lives): Likewise.
3070 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3071 (get_hard_regs): Likewise.
3072 (do_remat): Likewise.
3073 * lra-spills.c (assign_spill_hard_regs): Likewise.
3074 * mode-switching.c (create_pre_exit): Likewise.
3075 * postreload.c (reload_combine_recognize_pattern): Likewise.
3076 * recog.c (peep2_find_free_register): Likewise.
3077 * regcprop.c (kill_value_regno): Likewise.
3078 (set_value_regno): Likewise.
3079 (copy_value): Likewise.
3080 (maybe_mode_change): Likewise.
3081 (find_oldest_value_reg): Likewise.
3082 (copyprop_hardreg_forward_1): Likewise.
3083 * regrename.c (check_new_reg_p): Likewise.
3084 (regrename_do_replace): Likewise.
3085 * reload.c (push_reload): Likewise.
3086 (combine_reloads): Likewise.
3087 (find_dummy_reload): Likewise.
3088 (operands_match_p): Likewise.
3089 (find_reloads): Likewise.
3090 (find_equiv_reg): Likewise.
3091 (reload_adjust_reg_for_mode): Likewise.
3092 * reload1.c (count_pseudo): Likewise.
3093 (count_spilled_pseudo): Likewise.
3094 (find_reg): Likewise.
3095 (clear_reload_reg_in_use): Likewise.
3096 (free_for_value_p): Likewise.
3097 (allocate_reload_reg): Likewise.
3098 (choose_reload_regs): Likewise.
3099 (reload_adjust_reg_for_temp): Likewise.
3100 (emit_reload_insns): Likewise.
3101 (delete_output_reload): Likewise.
3102 * rtlanal.c (subreg_get_info): Likewise.
3103 * sched-deps.c (sched_analyze_reg): Likewise.
3104 * sel-sched.c (init_regs_for_mode): Likewise.
3105 (mark_unavailable_hard_regs): Likewise.
3106 (choose_best_reg_1): Likewise.
3107 (verify_target_availability): Likewise.
3108 * valtrack.c (dead_debug_insert_temp): Likewise.
3109 * var-tracking.c (track_loc_p): Likewise.
3110 (emit_note_insn_var_location): Likewise.
3111 * varasm.c (make_decl_rtl): Likewise.
3112 * reginfo.c (choose_hard_reg_mode): Likewise.
3113 (init_reg_modes_target): Refer directly to
3114 this_target_regs->x_hard_regno_nregs.
3116 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3118 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3119 instead of hard_regno_nregs.
3121 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3123 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3124 end_hard_regno instead of hard_regno_nregs.
3125 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3126 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3127 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3128 * ira-color.c (improve_allocation): Likewise.
3129 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3130 * lra-lives.c (mark_regno_live): Likewise.
3131 (mark_regno_dead): Likewise.
3132 * lra-remat.c (operand_to_remat): Likewise.
3133 * lra.c (collect_non_operand_hard_regs): Likewise.
3134 * postreload.c (reload_combine_note_store): Likewise.
3135 (move2add_valid_value_p): Likewise.
3136 * reload.c (regno_clobbered_p): Likewise.
3138 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3140 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3142 * config/v850/v850.c (v850_reorg): Likewise.
3143 * reload.c (refers_to_regno_for_reload_p): Likewise.
3144 (find_equiv_reg): Likewise.
3145 * reload1.c (reload_reg_reaches_end_p): Likewise.
3147 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3149 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3151 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3152 * config/arm/arm.c (output_move_neon): Likewise.
3153 (arm_attr_length_move_neon): Likewise.
3154 (neon_split_vcombine): Likewise.
3155 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3156 (c6x_mark_reg_written): Likewise.
3157 (c6x_dwarf_register_span): Likewise.
3158 * config/i386/i386.c (ix86_save_reg): Likewise.
3159 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3160 (rws_access_reg): Likewise.
3161 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3162 * mode-switching.c (create_pre_exit): Likewise.
3163 * ree.c (combine_reaching_defs): Likewise.
3164 (add_removable_extension): Likewise.
3165 * regcprop.c (find_oldest_value_reg): Likewise.
3166 (copyprop_hardreg_forward_1): Likewise.
3167 * reload.c (reload_inner_reg_of_subreg): Likewise.
3168 (push_reload): Likewise.
3169 (combine_reloads): Likewise.
3170 (find_dummy_reload): Likewise.
3171 (reload_adjust_reg_for_mode): Likewise.
3172 * reload1.c (find_reload_regs): Likewise.
3173 (forget_old_reloads_1): Likewise.
3174 (reload_reg_free_for_value_p): Likewise.
3175 (reload_adjust_reg_for_temp): Likewise.
3176 (emit_reload_insns): Likewise.
3177 (delete_output_reload): Likewise.
3178 * sel-sched.c (choose_best_reg_1): Likewise.
3179 (choose_best_pseudo_reg): Likewise.
3181 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3182 Alan Hayward <alan.hayward@arm.com>
3183 David Sherwood <david.sherwood@arm.com>
3185 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3186 * target.def (slow_unaligned_access): New hook.
3187 * targhooks.h (default_slow_unaligned_access): Declare.
3188 * targhooks.c (default_slow_unaligned_access): New function.
3189 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3190 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3191 * doc/tm.texi: Regenerate.
3192 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3193 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3194 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3196 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3197 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3199 (rs6000_slow_unaligned_access): New function.
3200 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3201 (expand_block_compare): Likewise.
3202 (expand_strn_compare): Likewise.
3203 (rs6000_rtx_costs): Likewise.
3204 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3205 (riscv_slow_unaligned_access): Likewise.
3206 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3207 (riscv_slow_unaligned_access_p): ...this and make static.
3208 (riscv_option_override): Update accordingly.
3209 (riscv_slow_unaligned_access): New function.
3210 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3211 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3212 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3213 (rs6000_slow_unaligned_access): New function.
3214 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3215 (rs6000_rtx_costs): Likewise.
3216 * config/rs6000/rs6000-string.c (expand_block_compare)
3217 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3218 of SLOW_UNALIGNED_ACCESS.
3219 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3220 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3221 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3222 of SLOW_UNALIGNED_ACCESS.
3223 * expmed.c (simple_mem_bitfield_p): Likewise.
3224 * expr.c (alignment_for_piecewise_move): Likewise.
3225 (emit_group_load_1): Likewise.
3226 (emit_group_store): Likewise.
3227 (copy_blkmode_from_reg): Likewise.
3228 (emit_push_insn): Likewise.
3229 (expand_assignment): Likewise.
3230 (store_field): Likewise.
3231 (expand_expr_real_1): Likewise.
3232 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3233 * lra-constraints.c (simplify_operand_subreg): Likewise.
3234 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3235 * gimple-ssa-store-merging.c: Likewise in block comment at start
3237 * tree-ssa-strlen.c: Include target.h.
3238 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3239 of SLOW_UNALIGNED_ACCESS.
3240 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3242 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3244 PR rtl-optimization/82185
3245 * expmed.c (emit_store_flag_int): Only test tem if it has been
3248 2017-09-12 Richard Biener <rguenther@suse.de>
3251 * match.pd ((FTYPE) N CMP CST): Fix typo.
3253 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3255 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3257 (mips_near_type_p): Add 'short_call' attribute as a synonym
3259 * doc/extend.texi (short_call): Document new function attribute.
3261 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3264 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3265 assertion check that in the condition.
3266 (get_atomic_generic_size): Likewise. Before testing if parameter
3267 has pointer type, if it has array type, call for C++
3268 default_conversion to perform array-to-pointer conversion.
3270 2017-09-12 Richard Biener <rguenther@suse.de>
3272 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3273 for operations we cannot scalarize.
3275 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3277 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3278 vectors heap vectors. Clean up comments.
3279 Make visited_bbs a reference.
3280 (profitable_jump_thread_path): Make GC
3281 vectors heap vectors. Clean up comments.
3283 (convert_and_register_jump_thread_path): Make GC vectors heap
3285 (check_subpath_and_update_thread_path): Same. Clean up comments.
3286 Make visited_bbs a reference.
3287 (handle_phi): Abstract common code to to
3288 register_jump_thread_path_if_profitable.
3289 Rename VAR_BB to DEF_BB.
3291 Make GC vectors heap vectors.
3292 Make visited_bbs a reference.
3293 (handle_assignment): Same.
3294 (register_jump_thread_path_if_profitable): New.
3295 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3297 Make GC vectors heap vectors. Clean up comments.
3298 Make visited_bbs a reference.
3299 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3300 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3303 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3306 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3307 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3309 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3311 Revert r251800 and r251799.
3313 2017-09-11 Martin Jambor <mjambor@suse.cz>
3316 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3317 arguments in advance.
3318 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3319 use it to find predecessor edges.
3320 (naive_outof_ssa): Collect vector of predecessors.
3322 2017-09-08 Jason Merrill <jason@redhat.com>
3324 PR c++/70029 - ICE with ref-qualifier and -flto
3325 * langhooks.h (struct lang_hooks_for_types): Add
3326 copy_lang_qualifiers.
3327 * attribs.c (build_type_attribute_qual_variant): Use it.
3328 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3330 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3331 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3333 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3336 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3337 (*mulsi3_sp64): New instruction.
3338 (mulsi3): New expander.
3340 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3342 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3344 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3346 * sancov.c: Include memmodel.h.
3348 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3351 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3354 2017-09-07 Carl Love <cel@us.ibm.com>
3356 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3357 the sldi instruction.
3359 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3361 * sancov.c: Include tm_p.h.
3363 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3366 * output.h (switch_to_other_text_partition): New declaration.
3367 * varasm.c (switch_to_other_text_partition): New function.
3368 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3369 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3370 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3371 to the other text partition before emitting LCL label and switch back
3372 after emitting the word after it.
3374 2017-09-07 Richard Biener <rguenther@suse.de>
3376 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3377 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3378 critical edges here, after loop init.
3379 (pass_data_pre): Remove PROP_no_crit_edges flags.
3380 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3381 for valueization of call args to avoid leaking VN_TOP.
3382 (visit_use): Assert we do not visit default defs.
3383 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3384 Use error_mark_node to more easily detect leaking VN_TOP.
3385 All default-defs are varying, not VN_TOP. Mark them visited.
3386 (run_scc_vn): Make code match comment.
3388 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3390 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3391 OPTION_MASK_FLOAT128_KEYWORD.
3392 (POWERPC_MASKS): Likewise.
3393 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3394 support for the -mfloat128-type option, and make -mfloat128
3395 default on PowerPC Linux systems. Define or undefine
3396 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3397 Define __float128 to be __ieee128 if IEEE 128-bit support is
3398 enabled, or undefine it.
3399 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3400 Delete defining __FLOAT128_TYPE__.
3401 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3402 -mfloat128-type option and make -mfloat128 default on PowerPC
3404 (TARGET_FLOAT128_TYPE): Likewise.
3405 (-mfloat128-type): Likewise.
3406 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3407 Delete the -mfloat128-type option and make -mfloat128 default on
3408 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3409 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3410 128-bit floating point is enabled. Change tests from using
3411 -mfloat128-type to -mfloat128.
3412 (rs6000_mangle_type): Use the correct mangling for the __float128
3413 type even if normal long double is restricted to 64-bits.
3414 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3416 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3417 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3418 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3420 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3421 documentation for -mfloat128.
3423 2017-09-06 Olivier Hainque <hainque@adacore.com>
3425 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3427 2017-09-06 Wish Wu <wishwu007@gmail.com>
3428 Jakub Jelinek <jakub@redhat.com>
3430 * asan.c (initialize_sanitizer_builtins): Add
3431 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3432 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3433 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3434 BT_FN_VOID_UINT64_PTR variables.
3435 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3436 (BT_FN_VOID_UINT16_UINT16): Likewise.
3437 (BT_FN_VOID_UINT32_UINT32): Likewise.
3438 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3439 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3440 (BT_FN_VOID_UINT64_PTR): Likewise.
3441 * common.opt (flag_sanitize_coverage): New variable.
3442 (fsanitize-coverage=trace-pc): Remove.
3443 (fsanitize-coverage=): Add.
3444 * flag-types.h (enum sanitize_coverage_code): New enum.
3445 * fold-const.c (fold_range_test): Disable non-short-circuit
3446 optimization if flag_sanitize_coverage.
3447 (fold_truth_andor): Likewise.
3448 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3449 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3450 (coverage_sanitizer_opts): New array.
3451 (get_closest_sanitizer_option): Add OPTS argument, handle also
3452 OPT_fsanitize_coverage_.
3453 (parse_sanitizer_options): Adjusted to also handle
3454 OPT_fsanitize_coverage_.
3455 (common_handle_option): Add OPT_fsanitize_coverage_.
3456 * sancov.c (instrument_comparison, instrument_switch): New function.
3457 (sancov_pass): Add trace-cmp support.
3458 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3459 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3460 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3461 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3462 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3463 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3464 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3465 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3466 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3467 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3469 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3471 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3472 error. Only quit immediately if parsing is complete.
3473 (BEGIN): Initialize fatal_err and parse_done.
3474 (begin fpu, end fpu): Check number of arguments.
3475 (begin arch, end arch): Likewise.
3476 (begin cpu, end cpu): Likewise.
3477 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3478 (optalias): Likewise.
3480 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3482 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3483 * config/arm/arm-isa.h: Delete. Move definitions to ...
3484 * arm-cpus.in: ... here. Use new feature and fgroup values.
3485 * config/arm/arm.c (arm_option_override): Use lower case for feature
3487 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3488 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3489 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3491 (print_isa_bits_for): New function.
3492 (gen_isa): New function.
3493 (gen_comm_data): Use print_isa_bits_for.
3494 (define feature): New keyword.
3495 (define fgroup): New keyword.
3496 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3497 (arm-isa.h): Add rule to generate file.
3498 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3499 case for feature bit names.
3501 2017-09-06 Richard Biener <rguenther@suse.de>
3503 * tree-ssa-pre.c (NECESSARY): Remove.
3504 (create_expression_by_pieces): Do not touch pass-local flags.
3505 (insert_into_preds_of_block): Likewise.
3506 (do_pre_regular_insertion): Likewise.
3507 (eliminate_insert): Likewise.
3508 (eliminate_dom_walker::before_dom_children): Likewise.
3509 (fini_eliminate): Do not look at inserted_exprs.
3510 (mark_operand_necessary): Remove.
3511 (remove_dead_inserted_code): Replace with simple work-list
3512 algorithm based on inserted_exprs and SSA uses.
3513 (pass_pre::execute): Re-order fini_eliminate and
3514 remove_dead_inserted_code.
3516 2017-09-06 Olivier Hainque <hainque@adacore.com>
3518 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3519 for VxWorks 7. Adjust surrounding comments.
3521 2017-09-06 Richard Biener <rguenther@suse.de>
3523 * gimple-ssa-strength-reduction.c
3524 (find_candidates_dom_walker::before_dom_children): Also allow
3527 2017-09-06 Richard Biener <rguenther@suse.de>
3529 PR tree-optimization/82108
3530 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3531 for gap in the non-permutation SLP case.
3533 2017-09-06 Martin Jambor <mjambor@suse.cz>
3535 PR tree-optimization/82078
3536 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3537 add_access_to_work_queue...
3538 (build_accesses_from_assign): ...here.
3539 (propagate_all_subaccesses): Make sure racc is the group
3540 representative, if there is one.
3542 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3545 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3548 2017-09-06 Richard Biener <rguenther@suse.de>
3550 * gimple-ssa-strength-reduction.c
3551 (find_candidates_dom_walker::before_doom_children): Use a
3552 type and not a mode check.
3554 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3557 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3559 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3561 2017-09-05 Jeff Law <law@redhat.com>
3563 PR tree-optimization/64910
3564 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3565 cases where we have 3 or more operands.
3567 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3570 * omp-low.c (lower_omp_for): Recompute tree invariant if
3571 gimple_omp_for_initial/final is ADDR_EXPR.
3574 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3575 into gimple val before gimplification fo the COND_EXPR.
3577 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3579 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3580 REGION_COPY argument.
3581 (thread_through_all_blocks): Remove unused argument to
3582 duplicate_thread_path.
3584 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3585 Alan Hayward <alan.hayward@arm.com>
3586 David Sherwood <david.sherwood@arm.com>
3588 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3589 Take a scalar_mode rather than a machine_mode.
3590 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3591 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3592 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3593 (aarch64_gen_adjusted_ldpstp): Likewise.
3594 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3596 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3597 Alan Hayward <alan.hayward@arm.com>
3598 David Sherwood <david.sherwood@arm.com>
3600 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3601 Take a scalar_int_mode instead of a machine_mode.
3602 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3603 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3604 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3605 (aarch64_simd_attr_length_rglist): Delete.
3606 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3607 a scalar_int_mode instead of a machine_mode.
3608 (aarch64_add_offset): Likewise.
3609 (aarch64_internal_mov_immediate): Likewise
3610 (aarch64_add_constant_internal): Likewise.
3611 (aarch64_add_constant): Likewise.
3612 (aarch64_movw_imm): Likewise.
3613 (aarch64_rtx_arith_op_extract_p): Likewise.
3614 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3615 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3616 Remove assert that the mode isn't a vector.
3617 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3618 (aarch64_expand_mov_immediate): Update calls after above changes.
3619 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3620 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3621 (aarch64_move_imm): Likewise.
3622 (aarch64_can_const_movi_rtx_p): Likewise.
3623 (aarch64_strip_extend): Likewise.
3624 (aarch64_extr_rtx_p): Likewise.
3625 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3626 a CONST_INT when the mode parameter is VOIDmode.
3627 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3629 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3631 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3632 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3633 (bitwise_type_for_mode): Update accordingly.
3635 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3637 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3638 * stor-layout.c (mode_for_size_tree): Likewise.
3639 (mode_for_array): Update accordingly.
3640 (layout_decl): Likewise.
3641 (compute_record_mode): Likewise. Only set the mode once.
3643 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3645 * target.def (get_mask_mode): Change return type to opt_mode.
3647 * doc/tm.texi: Regenerate.
3648 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3649 * targhooks.c (default_get_mask_mode): Likewise.
3650 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3651 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3652 targetm.get_mask_mode.
3653 * tree.c (build_truth_vector_type): Likewise.
3655 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3657 * machmode.h (mode_for_vector): Return an opt_mode.
3658 * stor-layout.c (mode_for_vector): Likewise.
3659 (mode_for_int_vector): Update accordingly.
3660 (layout_type): Likewise.
3661 * config/i386/i386.c (emit_memmov): Likewise.
3662 (ix86_expand_set_or_movmem): Likewise.
3663 (ix86_expand_vector_init): Likewise.
3664 (ix86_get_mask_mode): Likewise.
3665 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3667 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3668 * expmed.c (extract_bit_field_1): Likewise.
3669 * expr.c (expand_expr_real_2): Likewise.
3670 * optabs-query.c (can_vec_perm_p): Likewise.
3671 (can_vec_mask_load_store_p): Likewise.
3672 * optabs.c (expand_vec_perm): Likewise.
3673 * targhooks.c (default_get_mask_mode): Likewise.
3674 * tree-vect-stmts.c (vectorizable_store): Likewise.
3675 (vectorizable_load): Likewise.
3676 (get_vectype_for_scalar_type_and_size): Likewise.
3678 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3680 * machmode.h (mode_for_int_vector): New function.
3681 * stor-layout.c (mode_for_int_vector): Likewise.
3682 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3683 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3684 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3685 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3686 (s390_expand_vcond): Likewise.
3688 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3690 * machmode.h (opt_machine_mode): New type.
3691 (opt_mode<T>): Allow construction from anything that can be
3693 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3694 (mode_for_size): Return an opt_machine_mode.
3695 * stor-layout.c (mode_for_size): Likewise.
3696 (mode_for_size_tree): Update call accordingly.
3697 (bitwise_mode_for_mode): Likewise.
3698 (make_fract_type): Likewise.
3699 (make_accum_type): Likewise.
3700 * caller-save.c (replace_reg_with_saved_mem): Update call
3702 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3703 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3704 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3705 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3706 * expmed.c (extract_bit_field_1): Likewise.
3707 * reload.c (get_secondary_mem): Likewise.
3708 * varasm.c (assemble_integer): Likewise.
3709 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3712 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3714 * machmode.h (decimal_float_mode_for_size): New function.
3715 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3716 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3717 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3718 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3719 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3720 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3722 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3724 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3725 (get_builtin_sync_mode): Likewise.
3726 (expand_ifn_atomic_compare_exchange): Likewise.
3727 (expand_builtin_atomic_clear): Likewise.
3728 (expand_builtin_atomic_test_and_set): Likewise.
3729 (fold_builtin_atomic_always_lock_free): Likewise.
3730 * calls.c (compute_argument_addresses): Likewise.
3731 (emit_library_call_value_1): Likewise.
3732 (store_one_arg): Likewise.
3733 * combine.c (combine_instructions): Likewise.
3734 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3735 * config/arm/arm.c (arm_function_value): Likewise.
3736 (aapcs_allocate_return_reg): Likewise.
3737 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3738 * config/i386/i386.c (construct_container): Likewise.
3739 (ix86_gimplify_va_arg): Likewise.
3740 (ix86_expand_sse_cmp): Likewise.
3741 (emit_memmov): Likewise.
3742 (emit_memset): Likewise.
3743 (expand_small_movmem_or_setmem): Likewise.
3744 (ix86_expand_pextr): Likewise.
3745 (ix86_expand_pinsr): Likewise.
3746 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3747 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3749 * config/mips/mips.c (mips_function_value_1) Likewise.
3750 (mips_block_move_straight): Likewise.
3751 (mips_expand_ins_as_unaligned_store): Likewise.
3752 * config/powerpcspe/powerpcspe.c
3753 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3754 (rs6000_darwin64_record_arg_flush): Likewise.
3755 * config/rs6000/rs6000.c
3756 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3757 (rs6000_darwin64_record_arg_flush): Likewise.
3758 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3759 (sparc_function_value_1): Likewise.
3760 * config/spu/spu.c (adjust_operand): Likewise.
3761 (spu_emit_branch_or_set): Likewise.
3762 (arith_immediate_p): Likewise.
3763 * emit-rtl.c (gen_lowpart_common): Likewise.
3764 * expr.c (expand_expr_real_1): Likewise.
3765 * function.c (assign_parm_setup_block): Likewise.
3766 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3767 * reload1.c (alter_reg): Likewise.
3768 * stor-layout.c (mode_for_vector): Likewise.
3769 (layout_type): Likewise.
3771 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3773 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3774 (spu_convert_move): Likewise.
3775 * lower-subreg.c (resolve_simple_move): Likewise.
3777 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3780 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3781 define_insn to a define_expand.
3782 (altivec_vsum2sws_direct): New define_insn.
3783 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3785 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3787 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3790 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3795 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3796 realign stack if stack alignment needed is less than incoming
3799 2017-09-05 Marek Polacek <polacek@redhat.com>
3802 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3805 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3807 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3809 2017-09-05 Richard Biener <rguenther@suse.de>
3811 PR tree-optimization/82084
3812 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3814 2017-09-05 Richard Biener <rguenther@suse.de>
3816 PR tree-optimization/82102
3817 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3819 2017-09-05 Martin Liska <mliska@suse.cz>
3821 PR tree-optimization/82032
3822 * tree-cfg.c (generate_range_test): New function.
3823 * tree-cfg.h (generate_range_test): Declared here.
3824 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3825 (cleanup_control_expr_graph): Use it.
3826 * tree-switch-conversion.c (try_switch_expansion): Remove
3828 (emit_case_nodes): Use generate_range_test.
3830 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3833 * config/i386/i386.md (*<btsc><mode>_mask): Add
3834 TARGET_USE_BT to insn constraint.
3835 (*btr<mode>_mask): Ditto.
3837 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3839 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3840 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3842 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3845 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3846 TARGET_NEON and TARGET_IWMMXT.
3847 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3848 TARGET_NEON and TARGET_IWMMXT.
3849 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3851 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3853 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3854 (ix86_rewrite_tls_address): Ditto.
3855 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3856 (ix86_rewrite_tls_address_1): Ditto.
3857 (ix86_rewrite_tls_address): Ditto.
3858 * config/i386/predicates.md (tls_address_pattern): New predicate.
3859 * config/i386/i386.md (TLS address splitter): New splitter.
3861 2017-09-04 Richard Biener <rguenther@suse.de>
3863 PR tree-optimization/82084
3864 * fold-const.h (can_native_encode_string_p): Declare.
3865 * fold-const.c (can_native_encode_string_p): Factor out from ...
3866 (native_encode_string): ... here.
3867 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3868 vectorizing stores from constants we later cannot handle.
3870 2017-09-04 Marek Polacek <polacek@redhat.com>
3873 * doc/invoke.texi: Update -Wtautological-compare documentation.
3875 2017-09-04 Jeff Law <law@redhat.com>
3877 PR tree-optimization/64910
3878 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3879 swap the first and last operand if the last is a constant.
3881 2017-09-04 Marek Polacek <polacek@redhat.com>
3884 * convert.c (do_narrow): When sanitizing signed integer overflows,
3885 bail out for signed types.
3886 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3888 2017-09-04 Richard Biener <rguenther@suse.de>
3890 PR tree-optimization/82060
3891 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3892 Move devirtualization after stmt folding and before EH/AB/noreturn
3893 cleanup to get the stmt refs canonicalized. Use a bool instead
3894 of gimple_modified_p since that doesn't work for NOPs. Schedule
3895 NOPs generated by folding for removal.
3897 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3898 Alan Hayward <alan.hayward@arm.com>
3899 David Sherwood <david.sherwood@arm.com>
3901 * coretypes.h (pad_direction): New enum.
3902 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3903 (FUNCTION_ARG_PADDING): Likewise.
3904 * target.def (function_arg_padding): New hook.
3905 * targhooks.h (default_function_arg_padding): Declare.
3906 * targhooks.c (default_function_arg_padding): New function.
3907 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3908 (TARGET_FUNCTION_ARG_PADDING): ...this.
3909 * doc/tm.texi: Regenerate.
3910 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3911 instead of direction.
3912 (compute_argument_addresses): Likewise.
3913 (load_register_parameters): Likewise.
3914 (emit_library_call_value_1): Likewise.
3915 (store_one_arg): Use targetm.calls.function_arg_padding instead
3916 of FUNCTION_ARG_PADDING.
3917 (must_pass_in_stack_var_size_or_pad): Likewise.
3918 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3919 (emit_group_store): Likewise.
3920 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3921 instead of FUNCTION_ARG_PADDING.
3922 (emit_push_insn): Likewise, and propagate enum change throughout
3924 * function.h (direction): Delete.
3925 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3927 * function.c (assign_parm_find_stack_rtl): Likewise.
3928 (assign_parm_setup_block_p): Likewise.
3929 (assign_parm_setup_block): Likewise.
3930 (gimplify_parameters): Likewise.
3931 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3932 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3934 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3935 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3936 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3937 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3938 (aarch64_function_arg_padding): ...this new function.
3939 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3940 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3941 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3942 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3943 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3944 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3945 (arm_pad_arg_upward): Replace with...
3946 (arm_function_arg_padding): ...this new function.
3947 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3949 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3950 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3951 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3952 (ia64_hpux_function_arg_padding): Replace with...
3953 (ia64_function_arg_padding): ...this new function. Use pad_direction
3954 instead of direction. Check for TARGET_HPUX.
3955 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3956 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3957 (iq2000_function_arg_padding): New function.
3958 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3959 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3960 (mips_function_arg_padding): ...this new function.
3961 (mips_pad_reg_upward): Update accordingly.
3962 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3963 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3964 targetm.calls.function_arg_padding.
3965 (FUNCTION_ARG_PADDING): Delete.
3966 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3967 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3968 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3969 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3970 (nios2_block_reg_padding): Return pad_direction instead of direction.
3971 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3972 instead of direction.
3973 (nios2_function_arg_padding): Likewise. Make static.
3974 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3975 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3976 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3977 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3978 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3979 (pa_function_arg_padding): Make static. Return pad_direction instead
3981 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3982 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3983 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3984 instead of direction. Use targetm.calls.function_arg_padding.
3985 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3986 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3987 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3988 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3989 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3991 (function_arg_padding): Rename to...
3992 (rs6000_function_arg_padding): ...this. Make static. Return
3993 pad_direction instead of direction.
3994 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3995 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3996 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3997 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3998 instead of direction. Use targetm.calls.function_arg_padding.
3999 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4000 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4001 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4002 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4003 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4004 (function_arg_padding): Rename to...
4005 (rs6000_function_arg_padding): ...this. Make static. Return
4006 pad_direction instead of direction.
4007 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4008 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4009 * config/s390/s390.c (s390_function_arg_padding): New function.
4010 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4011 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4012 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4013 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4014 (function_arg_padding): Rename to...
4015 (sparc_function_arg_padding): ...this. Make static. Return
4016 pad_direction instead of direction.
4017 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4018 * config/spu/spu.c (spu_function_arg_padding): New function.
4019 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4020 * system.h (FUNCTION_ARG_PADDING): Poison.
4022 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4023 Alan Hayward <alan.hayward@arm.com>
4024 David Sherwood <david.sherwood@arm.com>
4026 * target.def (modes_tieable_p): New hook.
4027 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4028 (TARGET_MODES_TIEABLE_P): ...this.
4029 * doc/tm.texi.in: Regenerate.
4030 * hooks.h (hook_bool_mode_mode_true): Declare.
4031 * hooks.c (hook_bool_mode_mode_true): New function.
4032 * combine.c (subst): Use targetm.modes_tieable_p instead of
4034 * dse.c (find_shift_sequence): Likewise.
4035 * expmed.c (extract_low_bits): Likewise.
4036 * lower-subreg.c: Include target.h.
4037 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4039 * rtlanal.c (rtx_cost): Likewise.
4040 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4041 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4042 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4043 (TARGET_MODES_TIEABLE_P): Redefine.
4044 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4045 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4046 (TARGET_MODES_TIEABLE_P): Redefine.
4047 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4048 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4049 (arc_modes_tieable_p): New function.
4050 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4051 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4052 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4053 (arm_modes_tieable_p): Make static.
4054 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4055 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4056 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4057 (TARGET_MODES_TIEABLE_P): Redefine.
4058 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4059 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4060 (TARGET_MODES_TIEABLE_P): Redefine.
4061 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4062 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4063 (cr16_modes_tieable_p): New function.
4064 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4065 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4066 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4067 (TRULY_NOOP_TRUNCATION): Update comment.
4068 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4069 (TRULY_NOOP_TRUNCATION): Update comment.
4070 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4071 (frv_modes_tieable_p): New function.
4072 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4073 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4074 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4075 (TARGET_MODES_TIEABLE_P): Redefine.
4076 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4077 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4078 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4079 (TARGET_MODES_TIEABLE_P): Redefine.
4080 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4081 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4082 (ia64_modes_tieable_p): New function.
4083 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4084 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4085 (iq2000_modes_tieable_p): New function.
4086 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4087 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4088 (lm32_modes_tieable_p): New function.
4089 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4090 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4091 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4092 (TARGET_MODES_TIEABLE_P): Redefine.
4093 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4094 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4095 (m32r_modes_tieable_p): New function.
4096 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4097 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4098 (m68k_modes_tieable_p): New function.
4099 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4100 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4101 (mcore_modes_tieable_p): New function.
4102 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4103 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4105 (TARGET_MODES_TIEABLE_P): Redefine.
4106 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4107 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4108 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4109 (TARGET_MODES_TIEABLE_P): Redefine.
4110 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4111 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4112 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4113 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4114 (mn10300_modes_tieable_p): ...this and make static.
4115 (TARGET_MODES_TIEABLE_P): Redefine.
4116 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4117 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4118 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4119 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4120 (msp430_modes_tieable_p): Make static.
4121 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4122 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4123 (TARGET_MODES_TIEABLE_P): Redefine.
4124 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4125 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4126 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4127 (TARGET_MODES_TIEABLE_P): Redefine.
4128 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4129 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4130 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4131 (TARGET_MODES_TIEABLE_P): Redefine.
4132 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4133 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4134 (pdp11_modes_tieable_p): New function.
4135 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4136 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4137 (rs6000_modes_tieable_p): New function.
4138 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4139 * config/powerpcspe/powerpcspe.md: Update comment.
4140 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4141 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4142 (TARGET_MODES_TIEABLE_P): Redefine.
4143 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4144 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4145 (rl78_modes_tieable_p): New function.
4146 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4147 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4148 (rs6000_modes_tieable_p): New function.
4149 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4150 * config/rs6000/rs6000.md: Update comment.
4151 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4152 * config/rx/rx.c (rx_modes_tieable_p): New function.
4153 (TARGET_MODES_TIEABLE_P): Redefine.
4154 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4155 * config/s390/s390.c (s390_modes_tieable_p): New function.
4156 (TARGET_MODES_TIEABLE_P): Redefine.
4157 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4158 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4159 (sh_modes_tieable_p): New function.
4160 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4161 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4162 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4163 (sparc_modes_tieable_p): Make static.
4164 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4165 * config/spu/spu.c (spu_modes_tieable_p): New function.
4166 (TARGET_MODES_TIEABLE_P): Redefine.
4167 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4168 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4169 (TARGET_MODES_TIEABLE_P): Redefine.
4170 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4171 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4172 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4173 * config/v850/v850.c (v850_modes_tieable_p): New function.
4174 (TARGET_MODES_TIEABLE_P): Redefine.
4175 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4176 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4177 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4178 (visium_modes_tieable_p): New function.
4179 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4180 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4181 (xtensa_modes_tieable_p): New function.
4182 * system.h (MODES_TIEABLE_P): Poison.
4184 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4185 Alan Hayward <alan.hayward@arm.com>
4186 David Sherwood <david.sherwood@arm.com>
4188 * target.def (hard_regno_mode_ok): New hook.
4189 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4190 (TARGET_HARD_REGNO_MODE_OK): ...this.
4191 * doc/tm.texi.in: Regenerate.
4192 * hooks.h (hook_bool_uint_mode_true): Declare.
4193 * hooks.c (hook_bool_uint_mode_true): New function.
4194 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4196 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4197 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4198 instead of HARD_REGNO_MODE_OK.
4199 * caller-save.c: Include target.h.
4200 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4202 * combine.c (can_combine_p): Likewise.
4203 (combinable_i3pat): Likewise.
4204 (can_change_dest_mode): Likewise.
4205 * expr.c (init_expr_target): Likewise.
4206 (convert_move): Likewise.
4207 (convert_modes): Likewise.
4208 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4209 (setup_prohibited_mode_move_regs): Likewise.
4210 * ira.h (target_ira): Likewise.
4211 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4212 * lra-constraints.c (process_alt_operands): Likewise.
4213 (split_reg): Likewise.
4214 * recog.c (peep2_find_free_register): Likewise.
4215 * ree.c (combine_reaching_defs): Likewise.
4216 * regcprop.c (maybe_mode_change): Likewise.
4217 * reginfo.c (init_reg_sets_1): Likewise.
4218 (choose_hard_reg_mode): Likewise.
4219 (simplifiable_subregs): Likewise.
4220 * regrename.c (check_new_reg_p): Likewise.
4221 * reload.c (find_valid_class): Likewise.
4222 (find_valid_class_1): Likewise.
4223 (reload_inner_reg_of_subreg): Likewise.
4224 (push_reload): Likewise.
4225 (combine_reloads): Likewise.
4226 (find_dummy_reload): Likewise.
4227 (find_reloads): Likewise.
4228 * reload1.c (find_reg): Likewise.
4229 (set_reload_reg): Likewise.
4230 (allocate_reload_reg): Likewise.
4231 (choose_reload_regs): Likewise.
4232 (reload_adjust_reg_for_temp): Likewise.
4233 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4234 (simplify_subreg_regno): Likewise.
4235 * sel-sched.c (init_regs_for_mode): Likewise.
4236 * varasm.c (make_decl_rtl): Likewise.
4237 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4238 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4240 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4241 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4242 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4243 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4244 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4245 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4246 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4247 (arc_mode_class): Delete.
4248 (HARD_REGNO_MODE_OK): Delete.
4249 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4250 (arc_hard_regno_mode_ok): Rename old array to...
4251 (arc_hard_regno_mode_ok_modes): ...this.
4252 (arc_conditional_register_usage): Update accordingly.
4253 (arc_mode_class): Make static.
4254 (arc_hard_regno_mode_ok): New function.
4255 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4256 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4257 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4258 (arm_hard_regno_mode_ok): Make static.
4259 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4261 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4262 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4263 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4265 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4266 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4267 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4268 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4269 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4270 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4271 * config/bfin/predicates.md (valid_reg_operand): Use
4272 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4273 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4274 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4275 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4276 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4277 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4278 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4279 (cr16_hard_regno_mode_ok): Make static and return a bool.
4280 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4281 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4282 (cris_hard_regno_mode_ok): New function.
4283 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4284 (epiphany_mode_class): Delete.
4285 (HARD_REGNO_MODE_OK): Delete.
4286 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4287 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4288 (hard_regno_mode_ok): Rename to...
4289 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4290 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4291 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4293 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4294 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4295 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4296 (frv_hard_regno_mode_ok): Make static and return a bool.
4297 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4299 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4300 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4301 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4302 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4304 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4305 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4306 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4307 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4309 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4310 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4311 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4312 (ia64_hard_regno_mode_ok): New function.
4313 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4314 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4315 (iq2000_hard_regno_mode_ok): New function.
4316 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4317 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4318 (lm32_hard_regno_mode_ok): New function.
4319 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4320 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4321 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4322 instead of HARD_REGNO_MODE_OK.
4323 (m32c_hard_regno_ok): Rename to...
4324 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4325 (m32c_cannot_change_mode_class): Update accordingly.
4326 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4327 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4328 (m32r_mode_class): Delete.
4329 (HARD_REGNO_MODE_OK): Delete.
4330 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4331 (m32r_hard_regno_mode_ok): Rename to...
4332 (m32r_hard_regno_modes): ...this.
4333 (m32r_mode_class): Make static.
4334 (m32r_hard_regno_mode_ok): New function.
4335 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4336 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4337 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4338 (m68k_hard_regno_mode_ok): Make static.
4339 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4340 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4341 (mcore_hard_regno_mode_ok): New function.
4342 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4343 (HARD_REGNO_MODE_OK): Delete.
4344 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4346 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4347 (microblaze_hard_regno_mode_ok): New function.
4348 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4349 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4350 (mips_hard_regno_mode_ok): Delete.
4351 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4352 (mips_hard_regno_mode_ok_p): ...this and make static.
4353 (mips_hard_regno_mode_ok_p): Rename to...
4354 (mips_hard_regno_mode_ok_uncached): ...this.
4355 (mips_hard_regno_mode_ok): New function.
4356 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4357 of HARD_REGNO_MODE_OK.
4358 (mips_option_override): Update after above name changes.
4359 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4360 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4361 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4362 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4363 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4364 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4365 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4366 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4367 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4368 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4369 (msp430_hard_regno_mode_ok): Make static and return a bool.
4370 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4371 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4372 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4374 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4375 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4376 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4377 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4378 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4379 (PA_HARD_REGNO_MODE_OK): ...this
4380 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4381 (PA_HARD_REGNO_MODE_OK): ...this.
4382 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4383 (pa_hard_regno_mode_ok): New function.
4384 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4385 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4386 (pdp11_hard_regno_mode_ok): New function.
4387 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4388 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4390 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4392 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4393 (rs6000_hard_regno_mode_ok): Rename to...
4394 (rs6000_hard_regno_mode_ok_uncached): ...this.
4395 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4396 (rs6000_hard_regno_mode_ok): New function.
4397 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4398 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4399 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4400 (riscv_hard_regno_mode_ok): ...this and make static.
4401 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4402 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4403 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4404 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4405 (rl78_hard_regno_mode_ok): Make static and return bool.
4406 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4407 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4409 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4410 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4411 (rs6000_hard_regno_mode_ok): Rename to...
4412 (rs6000_hard_regno_mode_ok_uncached): ...this.
4413 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4414 (rs6000_hard_regno_mode_ok): New function.
4415 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4416 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4418 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4419 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4420 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4421 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4422 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4423 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4424 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4425 (sh_hard_regno_mode_ok): Make static.
4426 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4427 instead of HARD_REGNO_MODE_OK.
4428 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4429 (sparc_mode_class): Delete.
4430 (HARD_REGNO_MODE_OK): Delete.
4431 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4432 (hard_regno_mode_classes): Make static.
4433 (sparc_mode_class): Likewise.
4434 (sparc_hard_regno_mode_ok): New function.
4435 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4436 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4437 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4439 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4440 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4441 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4442 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4443 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4444 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4445 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4446 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4447 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4448 (visium_hard_regno_mode_ok): New function.
4449 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4450 instead of HARD_REGNO_MODE_OK.
4451 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4452 (HARD_REGNO_MODE_OK): Delete.
4453 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4454 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4455 (xtensa_option_override): Update accordingly.
4456 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4457 (xtensa_hard_regno_mode_ok): New function.
4458 * system.h (HARD_REGNO_MODE_OK): Poison.
4460 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4461 Alan Hayward <alan.hayward@arm.com>
4462 David Sherwood <david.sherwood@arm.com>
4464 * target.def (hard_regno_call_part_clobbered): New hook.
4465 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4466 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4467 * doc/tm.texi: Regenerate.
4468 * hooks.h (hook_bool_uint_mode_false): Declare.
4469 * hooks.c (hook_bool_uint_mode_false): New function.
4470 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4471 * cselib.c (cselib_process_insn): Use
4472 targetm.hard_regno_call_part_clobbered instead of
4473 HARD_REGNO_CALL_PART_CLOBBERED.
4474 * ira-conflicts.c (ira_build_conflicts): Likewise.
4475 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4476 * lra-constraints.c (need_for_call_save_p): Likewise.
4477 * lra-lives.c: Include target.h.
4478 (check_pseudos_live_through_calls): Use
4479 targetm.hard_regno_call_part_clobbered instead of
4480 HARD_REGNO_CALL_PART_CLOBBERED.
4481 * regcprop.c: Include target.h.
4482 (copyprop_hardreg_forward_1): Use
4483 targetm.hard_regno_call_part_clobbered instead of
4484 HARD_REGNO_CALL_PART_CLOBBERED.
4485 * reginfo.c (choose_hard_reg_mode): Likewise.
4486 * regrename.c (check_new_reg_p): Likewise.
4487 * reload.c (find_equiv_reg): Likewise.
4488 * reload1.c (emit_reload_insns): Likewise.
4489 * sched-deps.c (deps_analyze_insn): Likewise.
4490 * sel-sched.c (init_regs_for_mode): Likewise.
4491 (mark_unavailable_hard_regs): Likewise.
4492 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4493 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4494 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4496 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4497 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4498 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4500 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4502 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4503 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4504 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4506 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4507 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4508 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4510 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4511 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4513 * config/powerpcspe/powerpcspe.c
4514 (rs6000_hard_regno_call_part_clobbered): New function.
4515 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4516 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4517 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4519 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4520 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4521 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4523 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4524 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4525 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4527 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4528 Alan Hayward <alan.hayward@arm.com>
4529 David Sherwood <david.sherwood@arm.com>
4531 * rtl.h (subreg_memory_offset): Declare.
4532 * emit-rtl.c (subreg_memory_offset): New function.
4533 * expmed.c (store_bit_field_1): Use it.
4534 * expr.c (undefined_operand_subword_p): Likewise.
4535 * simplify-rtx.c (simplify_subreg): Likewise.
4537 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4539 PR rtl-optimization/57448
4542 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4543 using atomic_load pattern.
4544 (expand_atomic_store): Likewise.
4546 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4549 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4550 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4551 handling. Use replace_call_with_value with NULL instead of
4552 gsi_replace, unlink_stmt_vdef and release_defs.
4554 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4557 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4559 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4562 * rtl.h (emit_library_call_value_1): Declare.
4563 (emit_library_call): Replace declaration with a series of overloads.
4564 Remove the parameter count argument.
4565 (emit_library_call_value): Likewise.
4566 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4567 with an "rtx_mode_t *".
4568 (emit_library_call_value): Delete.
4569 (emit_library_call): Likewise.
4570 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4571 (asan_emit_allocas_unpoison): Likewise.
4572 * builtins.c (expand_builtin_powi): Likewise.
4573 (expand_asan_emit_allocas_unpoison): Likewise.
4574 * cfgexpand.c (expand_main_function): Likewise.
4575 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4576 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4577 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4578 * config/arm/arm.c (arm_trampoline_init): Likewise.
4579 (arm_call_tls_get_addr): Likewise.
4580 (arm_expand_divmod_libfunc): Likewise.
4581 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4582 (smulsi3_highpart): Likewise.
4583 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4584 (c6x_expand_compare): Likewise.
4585 (c6x_expand_movmem): Likewise.
4586 * config/frv/frv.c (frv_trampoline_init): Likewise.
4587 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4588 (ix86_expand_divmod_libfunc): Likewise.
4589 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4590 (ia64_expand_compare): Likewise.
4591 (ia64_profile_hook): Likewise.
4592 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4593 (nonlocal_goto): Likewise.
4594 (restore_stack_nonlocal): Likewise.
4595 * config/m32r/m32r.c (block_move_call): Likewise.
4596 (m32r_trampoline_init): Likewise.
4597 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4598 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4599 (m68k_call_m68k_read_tp): Likewise.
4600 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4601 (microblaze_expand_divide): Likewise.
4602 * config/mips/mips.h (mips_args): Likewise.
4603 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4604 (MIPS_ICACHE_SYNC): Likewise.
4605 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4606 (nios2_trampoline_init): Likewise.
4607 * config/pa/pa.c (hppa_tls_call): Likewise.
4608 (pa_trampoline_init): Likewise.
4609 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4610 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4611 (expand_strn_compare): Likewise.
4612 (rs6000_generate_compare): Likewise.
4613 (rs6000_expand_float128_convert): Likewise.
4614 (output_profile_hook): Likewise.
4615 (rs6000_trampoline_init): Likewise.
4616 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4617 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4618 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4619 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4620 (rs6000_generate_compare): Likewise.
4621 (rs6000_expand_float128_convert): Likewise.
4622 (output_profile_hook): Likewise.
4623 (rs6000_trampoline_init): Likewise.
4624 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4625 * config/sh/sh.c (sh_trampoline_init): Likewise.
4626 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4627 (sparc_emit_float_lib_cmp): Likewise.
4628 (sparc32_initialize_trampoline): Likewise.
4629 (sparc64_initialize_trampoline): Likewise.
4630 (sparc_profile_hook): Likewise.
4631 * config/spu/spu.c (ea_load_store): Likewise.
4632 * config/spu/spu.md (floatunssidf2): Likewise.
4633 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4634 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4635 * config/visium/visium.c (expand_block_move_4): Likewise.
4636 (expand_block_move_2): Likewise.
4637 (expand_block_move_1): Likewise.
4638 (expand_block_set_4): Likewise.
4639 (expand_block_set_2): Likewise.
4640 (expand_block_set_1): Likewise.
4641 (visium_trampoline_init): Likewise.
4642 (visium_profile_hook): Likewise.
4643 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4644 (xtensa_setup_frame_addresses): Likewise.
4645 (xtensa_trampoline_init): Likewise.
4646 * except.c (sjlj_emit_function_enter): Likewise.
4647 (sjlj_emit_function_exit): Likewise.
4648 * explow.c (allocate_dynamic_stack_space): Likewise.
4649 (probe_stack_range): Likewise.
4650 * expr.c (convert_mode_scalar): Likewise.
4651 * optabs.c (expand_binop): Likewise.
4652 (expand_twoval_binop_libfunc): Likewise.
4653 (expand_unop): Likewise.
4654 (prepare_cmp_insn): Likewise.
4655 (prepare_float_lib_cmp): Likewise.
4656 (expand_float): Likewise.
4657 (expand_fix): Likewise.
4658 (expand_fixed_convert): Likewise.
4659 (maybe_emit_sync_lock_test_and_set): Likewise.
4660 (expand_atomic_compare_and_swap): Likewise.
4661 (expand_mem_thread_fence): Likewise.
4662 (expand_atomic_fetch_op): Likewise.
4664 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4666 * doc/generic.texi (OpenACC): Adjust URL.
4667 * doc/invoke.texi (C Dialect Options): Ditto.
4669 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4671 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4672 predicate for operand 1. Add (m,<S>) constraint.
4673 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4674 Prevent memory operand 1 with register operand 2.
4676 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4678 PR rtl-optimization/82024
4679 * combine.c (try_combine): If the combination result is a PARALLEL,
4680 and we only need to retain the SET in there that would be placed
4681 at I2, check that we can place that at I3 instead, before doing so.
4683 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4686 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4688 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4689 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4692 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4693 Jeff Law <law@redhat.com>
4695 * varasm.c (bss_initializer_p): Do not put constants into .bss
4696 (categorize_decl_for_section): Handle bss_initializer_p returning
4697 false when DECL_INITIAL is NULL.
4699 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4702 * config/s390/s390.c (s390_can_inline_p): New function.
4704 2017-09-01 Jeff Law <law@redhat.com>
4706 PR tree-optimization/82052
4707 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4708 Always initialize the returned slot after a hash table miss
4709 when INSERT is true.
4711 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4713 * config/s390/s390.md (mem_signal_fence): Remove.
4714 * doc/md.texi (mem_signal_fence): Remove.
4715 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4717 * target-insns.def (mem_signal_fence): Remove.
4719 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4722 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4725 * asan.c (create_odr_indicator): Strip name encoding from assembler
4726 name before appending it after __odr_asan_.
4728 2017-09-01 Martin Liska <mliska@suse.cz>
4730 PR tree-optimization/82059
4731 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4732 frequency only when an edge is redirected.
4734 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4736 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4737 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4738 (arc_conditional_register_usage): Remove ARC600 lp_count
4740 (arc_file_start): Emit Tag_ARC_CPU_variation.
4741 (arc_can_use_doloop_p): New conditions to use ZOLs.
4742 (hwloop_fail): New function.
4743 (hwloop_optimize): Likewise.
4744 (hwloop_pattern_reg): Likewise.
4745 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4746 (arc_reorg_loops): New function, calls reorg_loops.
4747 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4748 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4750 (arc_loop_hazard): Remove function, functionality moved into
4752 (arc_hazard): Remove arc_loop_hazard call.
4753 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4754 into hwloop_optimize.
4755 (arc_label_align): Remove ZOL handling.
4756 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4757 * config/arc/arc.md (doloop_begin): Remove pattern.
4758 (doloop_begin_i): Likewise.
4759 (doloop_end_i): Likewise.
4760 (doloop_fallback): Likewise.
4761 (doloop_fallback_m): Likewise.
4762 (doloop_end): Reimplement expand.
4763 (arc_lp): New pattern for LP instruction.
4764 (loop_end): New pattern.
4765 (loop_fail): Likewise.
4766 (decrement_and_branch_until_zero): Likewise.
4767 * config/arc/arc.opt (mlpc-width): New option.
4768 * doc/invoke.texi (mlpc-width): Document option.
4770 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4772 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4773 (arc_ccfsm_advance): Fix checking for delay slots.
4774 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4776 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4778 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4780 (movhi_insn): Update store instruction constraint which are saving
4781 6-bit short immediates.
4782 (movsi_insn): Consider also short scaled load operations.
4783 (zero_extendhisi2_i): Use Usd constraint instead of T.
4784 (extendhisi2_i): Add q constraint.
4785 (arc_clzsi2): Add type and length attributes.
4786 (arc_ctzsi2): Likewise.
4787 * config/arc/constraints.md (Usc): Update constraint, the
4788 assembler can parse two relocations for a single instruction.
4790 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4792 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4793 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4795 2017-08-31 Olivier Hainque <hainque@adacore.com>
4797 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4798 match as powerpc-wrs-vxworks*.
4800 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4802 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4803 register constraint for by-element operand.
4804 (aarch64_mls_elt_merge<mode>): Likewise.
4806 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4808 * config/arc/arc.c (arc_can_follow_jump): Check for short
4811 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4813 * config.gcc: Use g.opt for arc.
4814 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4815 functionality moved to ...
4816 (legitimate_scaled_address_p): New function, ...here.
4817 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4818 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4819 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4821 (arc_override_options): Handle G option.
4822 (arc_output_pic_addr_const): Correct function definition.
4823 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4824 (arc_decl_anon_ns_mem_p): Delete.
4825 (arc_in_small_data_p): Overhaul this function to take into
4826 consideration the value given via G option.
4827 (arc_rewrite_small_data_1): Renamed and corrected old
4828 arc_rewrite_small_data function.
4829 (arc_rewrite_small_data): New function.
4830 (small_data_pattern): Don't use pic_offset_table_rtx.
4831 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4832 * config/arc/simdext.md (movmisalignv2hi): Use
4833 prepare_move_operands function.
4835 (movmisalign*): Likewise.
4836 * doc/invoke.texi (ARC options): Document -G option.
4838 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4840 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4842 * config/arc/arc.c (arc_print_operand): Output scalled address for
4843 sdata whenever is possible.
4844 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4845 load/stores are available.
4846 (compact_sda_memory_operand): Check for the alignment required by
4847 code density instructions.
4848 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4850 * config/arc/constraints.md (Usd): Update constraint.
4851 (Us0): New constraint.
4852 (Usc): Update constraint.
4854 2017-08-31 Richard Biener <rguenther@suse.de>
4857 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4860 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4862 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4863 Resize type_signature.
4865 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4866 Alan Hayward <alan.hayward@arm.com>
4867 David Sherwood <david.sherwood@arm.com>
4869 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4870 subregs whose inner modes can be stored in GPRs.
4871 (aarch64_classify_index): Likewise.
4873 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4874 Alan Hayward <alan.hayward@arm.com>
4875 David Sherwood <david.sherwood@arm.com>
4877 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4878 (V_INT_EQUIV): ...this.
4879 (v_cmp_result): Rename to...
4880 (v_int_equiv): ...this.
4881 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4882 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4883 (copysign<mode>3): Likewise.
4884 (aarch64_simd_bsl<mode>_internal): Likewise.
4885 (aarch64_simd_bsl<mode>): Likewise.
4886 (vec_cmp<mode><mode>): Likewise.
4887 (vcond<mode><mode>): Likewise.
4888 (vcond<v_cmp_mixed><mode>): Likewise.
4889 (vcondu<mode><v_cmp_mixed>): Likewise.
4890 (aarch64_cm<optab><mode>): Likewise.
4891 (aarch64_cmtst<mode>): Likewise.
4892 (aarch64_fac<optab><mode>): Likewise.
4893 (vec_perm_const<mode>): Likewise.
4894 (vcond_mask_<mode><v_cmp_result>): Rename to...
4895 (vcond_mask_<mode><v_int_equiv>): ...this.
4896 (vec_cmp<mode><v_cmp_result>): Rename to...
4897 (vec_cmp<mode><v_int_equiv>): ...this.
4899 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4900 Alan Hayward <alan.hayward@arm.com>
4901 David Sherwood <david.sherwood@arm.com>
4903 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4905 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4906 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4907 (UNSPEC_LD4_DREG): New unspecs.
4908 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4909 (aarch64_ld2<mode>_dreg_be): Replace with...
4910 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4912 (aarch64_ld3<mode>_dreg_le)
4913 (aarch64_ld3<mode>_dreg_be): Replace with...
4914 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4916 (aarch64_ld4<mode>_dreg_le)
4917 (aarch64_ld4<mode>_dreg_be): Replace with...
4918 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4921 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4923 PR tree-optimization/81987
4924 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4925 insert an initializer in a location not dominated by the stride
4928 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4930 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4931 on the entire header of the finally block in the fallthru case.
4933 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4935 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4937 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4939 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4940 rs6000_emit_move_from_cr and call renamed function.
4941 (rs6000_emit_prologue): Call renamed functions.
4942 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4943 movesi_from_cr, remove volatile CRs.
4945 2017-08-30 Jon Beniston <jon@beniston.com>
4946 Richard Biener <rguenther@suse.de>
4948 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4949 of VECTOR_MODE_P check.
4950 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4951 element vector types.
4953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4955 * df.h (df_read_modify_subreg_p): Remove in favor of...
4956 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4957 const_rtx instead of an rtx.
4958 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4959 * df-problems.c (df_word_lr_mark_ref): Likewise.
4960 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4961 (mark_pseudo_reg_dead): Likewise.
4962 (mark_ref_dead): Likewise.
4963 * reginfo.c (init_subregs_of_mode): Likewise.
4964 * sched-deps.c (sched_analyze_1): Likewise.
4965 * df-scan.c (df_def_record_1): Likewise.
4966 (df_uses_record): Likewise.
4967 (df_read_modify_subreg_p): Remove in favor of...
4968 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4969 const_rtx instead of an rtx.
4971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4972 Alan Hayward <alan.hayward@arm.com>
4973 David Sherwood <david.sherwood@arm.com>
4975 * rtl.h (partial_subreg_p): New function.
4976 * caller-save.c (save_call_clobbered_regs): Use it.
4977 * calls.c (expand_call): Likewise.
4978 * combine.c (combinable_i3pat): Likewise.
4979 (simplify_set): Likewise.
4980 (make_extraction): Likewise.
4981 (make_compound_operation_int): Likewise.
4982 (gen_lowpart_or_truncate): Likewise.
4983 (force_to_mode): Likewise.
4984 (make_field_assignment): Likewise.
4985 (reg_truncated_to_mode): Likewise.
4986 (record_truncated_value): Likewise.
4987 (move_deaths): Likewise.
4988 * cse.c (record_jump_cond): Likewise.
4989 (cse_insn): Likewise.
4990 * cselib.c (cselib_lookup_1): Likewise.
4991 * expmed.c (extract_bit_field_using_extv): Likewise.
4992 * function.c (assign_parm_setup_reg): Likewise.
4993 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4994 * ira-build.c (create_insn_allocnos): Likewise.
4995 * lra-coalesce.c (merge_pseudos): Likewise.
4996 * lra-constraints.c (match_reload): Likewise.
4997 (simplify_operand_subreg): Likewise.
4998 (curr_insn_transform): Likewise.
4999 * lra-lives.c (process_bb_lives): Likewise.
5000 * lra.c (new_insn_reg): Likewise.
5001 (lra_substitute_pseudo): Likewise.
5002 * regcprop.c (mode_change_ok): Likewise.
5003 (maybe_mode_change): Likewise.
5004 (copyprop_hardreg_forward_1): Likewise.
5005 * reload.c (push_reload): Likewise.
5006 (find_reloads): Likewise.
5007 (find_reloads_subreg_address): Likewise.
5008 * reload1.c (alter_reg): Likewise.
5009 (eliminate_regs_1): Likewise.
5010 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5012 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5014 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5015 back to if statements, including unpack.
5017 2017-08-30 Martin Liska <mliska@suse.cz>
5020 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5022 (func_checker::compare_asm_inputs_outputs): ... this function.
5023 (func_checker::compare_gimple_asm): Use the function to compare
5024 also ASM constrains.
5025 * ipa-icf-gimple.h: Rename the function.
5027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5028 Alan Hayward <alan.hayward@arm.com>
5029 David Sherwood <david.sherwood@arm.com>
5031 * coretypes.h (complex_mode): New type.
5032 * gdbhooks.py (build_pretty_printer): Handle it.
5033 * machmode.h (complex_mode): New class.
5034 (complex_mode::includes_p): New function.
5035 (is_complex_int_mode): Likewise.
5036 (is_complex_float_mode): Likewise.
5037 * genmodes.c (get_mode_class): Handle complex mode classes.
5038 * function.c (expand_function_end): Use is_complex_int_mode.
5040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5041 Alan Hayward <alan.hayward@arm.com>
5042 David Sherwood <david.sherwood@arm.com>
5044 * coretypes.h (scalar_mode_pod): New typedef.
5045 * gdbhooks.py (build_pretty_printer): Handle it.
5046 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5047 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5048 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5049 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5053 Alan Hayward <alan.hayward@arm.com>
5054 David Sherwood <david.sherwood@arm.com>
5056 * machmode.h (mode_for_vector): Take a scalar_mode instead
5058 * stor-layout.c (mode_for_vector): Likewise.
5059 * explow.c (promote_mode): Use as_a <scalar_mode>.
5060 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5063 Alan Hayward <alan.hayward@arm.com>
5064 David Sherwood <david.sherwood@arm.com>
5066 * target.def (preferred_simd_mode): Take a scalar_mode
5067 instead of a machine_mode.
5068 * targhooks.h (default_preferred_simd_mode): Likewise.
5069 * targhooks.c (default_preferred_simd_mode): Likewise.
5070 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5071 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5072 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5073 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5074 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5075 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5076 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5077 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5079 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5080 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5081 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5082 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5083 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5084 * doc/tm.texi: Regenerate.
5085 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5089 Alan Hayward <alan.hayward@arm.com>
5090 David Sherwood <david.sherwood@arm.com>
5092 * target.def (scalar_mode_supported_p): Take a scalar_mode
5093 instead of a machine_mode.
5094 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5095 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5096 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5097 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5098 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5099 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5100 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5101 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5102 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5103 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5104 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5105 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5106 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5107 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5109 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5110 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5111 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5112 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5113 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5115 * doc/tm.texi: Regenerate.
5117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5118 Alan Hayward <alan.hayward@arm.com>
5119 David Sherwood <david.sherwood@arm.com>
5121 * coretypes.h (opt_scalar_mode): New typedef.
5122 * gdbhooks.py (build_pretty_printers): Handle it.
5123 * machmode.h (mode_iterator::get_2xwider): Add overload for
5125 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5127 * expr.c (convert_mode_scalar): Likewise.
5128 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5129 * optabs.c (expand_float): Likewise.
5130 (expand_fix): Likewise.
5131 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5134 Alan Hayward <alan.hayward@arm.com>
5135 David Sherwood <david.sherwood@arm.com>
5137 * optabs.c (expand_float): Explicitly check for scalars before
5138 using a branching expansion.
5139 (expand_fix): Likewise.
5141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5142 Alan Hayward <alan.hayward@arm.com>
5143 David Sherwood <david.sherwood@arm.com>
5145 * expr.c (convert_mode): Split scalar handling out into...
5146 (convert_mode_scalar): ...this new function. Treat the modes
5149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5150 Alan Hayward <alan.hayward@arm.com>
5151 David Sherwood <david.sherwood@arm.com>
5153 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5155 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5158 Alan Hayward <alan.hayward@arm.com>
5159 David Sherwood <david.sherwood@arm.com>
5161 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5162 rather than a machine_mode.
5163 (fixed_from_string): Likewise.
5164 (fixed_convert): Likewise.
5165 (fixed_convert_from_int): Likewise.
5166 (fixed_convert_from_real): Likewise.
5167 (real_convert_from_fixed): Likewise.
5168 * fixed-value.c (fixed_from_double_int): Likewise.
5169 (fixed_from_string): Likewise.
5170 (fixed_convert): Likewise.
5171 (fixed_convert_from_int): Likewise.
5172 (fixed_convert_from_real): Likewise.
5173 (real_convert_from_fixed): Likewise.
5174 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5177 Alan Hayward <alan.hayward@arm.com>
5178 David Sherwood <david.sherwood@arm.com>
5180 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5181 of separate mode class checks. Do not allow vector modes here.
5182 (immed_wide_int_const): Use as_a <scalar_mode>.
5183 * explow.c (trunc_int_for_mode): Likewise.
5184 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5185 (wi::shwi): Likewise.
5186 (wi::min_value): Likewise.
5187 (wi::max_value): Likewise.
5188 * dwarf2out.c (loc_descriptor): Likewise.
5189 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5193 Alan Hayward <alan.hayward@arm.com>
5194 David Sherwood <david.sherwood@arm.com>
5196 * tree.h (SCALAR_TYPE_MODE): New macro.
5197 * expr.c (expand_expr_addr_expr_1): Use it.
5198 (expand_expr_real_2): Likewise.
5199 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5200 (fold_convert_const_fixed_from_int): Likewise.
5201 (fold_convert_const_fixed_from_real): Likewise.
5202 (native_encode_fixed): Likewise
5203 (native_encode_complex): Likewise
5204 (native_encode_vector): Likewise.
5205 (native_interpret_fixed): Likewise.
5206 (native_interpret_real): Likewise.
5207 (native_interpret_complex): Likewise.
5208 (native_interpret_vector): Likewise.
5209 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5210 (simd_clone_adjust_argument_types): Likewise.
5211 (simd_clone_init_simd_arrays): Likewise.
5212 (simd_clone_adjust): Likewise.
5213 * stor-layout.c (layout_type): Likewise.
5214 * tree.c (build_minus_one_cst): Likewise.
5215 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5216 * tree-inline.c (estimate_move_cost): Likewise.
5217 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5218 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5219 (vectorizable_reduction): Likewise.
5220 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5221 (vect_recog_mixed_size_cond_pattern): Likewise.
5222 (check_bool_pattern): Likewise.
5223 (adjust_bool_pattern): Likewise.
5224 (search_type_for_mask_1): Likewise.
5225 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5226 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5227 (vectorizable_load): Likewise.
5228 (vectorizable_store): Likewise.
5229 * ubsan.c (ubsan_encode_value): Likewise.
5230 * varasm.c (output_constant): Likewise.
5232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5233 Alan Hayward <alan.hayward@arm.com>
5234 David Sherwood <david.sherwood@arm.com>
5236 * coretypes.h (scalar_mode): New class.
5237 * machmode.h (scalar_mode): Likewise.
5238 (scalar_mode::includes_p): New function.
5239 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5240 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5241 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5242 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5243 * expmed.c (store_bit_field_1): Likewise.
5244 (extract_bit_field_1): Likewise.
5245 * expr.c (write_complex_part): Likewise.
5246 (read_complex_part): Likewise.
5247 (emit_move_complex_push): Likewise.
5248 (expand_expr_real_2): Likewise.
5249 * function.c (assign_parm_setup_reg): Likewise.
5250 (assign_parms_unsplit_complex): Likewise.
5251 * optabs.c (expand_binop): Likewise.
5252 * rtlanal.c (subreg_get_info): Likewise.
5253 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5254 * varasm.c (output_constant_pool_2): Likewise.
5256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5257 Alan Hayward <alan.hayward@arm.com>
5258 David Sherwood <david.sherwood@arm.com>
5260 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5262 (expmed_mult_highpart_optab): Likewise.
5263 (expmed_mult_highpart): Likewise.
5265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5266 Alan Hayward <alan.hayward@arm.com>
5267 David Sherwood <david.sherwood@arm.com>
5269 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5270 instead of a machine_mode.
5271 (builtin_memset_read_str): Likewise.
5272 * builtins.c (c_readstr): Likewise.
5273 (builtin_memcpy_read_str): Likewise.
5274 (builtin_strncpy_read_str): Likewise.
5275 (builtin_memset_read_str): Likewise.
5276 (builtin_memset_gen_str): Likewise.
5277 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5278 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5279 instead of a machine_mode.
5280 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5282 (make_extraction): Likewise.
5283 (try_widen_shift_mode): Take and return scalar_int_modes instead
5285 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5286 a scalar_int_mode instead of a machine_mode.
5287 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5288 (avr_addr_space_pointer_mode): Likewise.
5289 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5290 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5291 (msp430_unwind_word_mode): Likewise.
5292 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5293 (spu_addr_space_pointer_mode): Likewise.
5294 (spu_addr_space_address_mode): Likewise.
5295 (spu_libgcc_cmp_return_mode): Likewise.
5296 (spu_libgcc_shift_count_mode): Likewise.
5297 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5298 (rl78_addr_space_pointer_mode): Likewise.
5299 (fl78_unwind_word_mode): Likewise.
5300 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5302 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5303 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5304 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5305 (mips_valid_pointer_mode): Likewise.
5306 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5307 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5308 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5310 (ft32_addr_space_address_mode): Likewise.
5311 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5312 scalar_int_mode instead of a machine_mode.
5313 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5315 (m32c_addr_space_address_mode): Likewise.
5316 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5317 (rs6000_eh_return_filter_mode): Likewise.
5318 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5319 (rs6000_eh_return_filter_mode): Likewise.
5320 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5321 (s390_libgcc_shift_count_mode): Likewise.
5322 (s390_unwind_word_mode): Likewise.
5323 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5325 * target.def (mode_rep_extended): Likewise.
5326 (valid_pointer_mode): Likewise.
5327 (addr_space.valid_pointer_mode): Likewise.
5328 (eh_return_filter_mode): Return a scalar_int_mode rather than
5330 (libgcc_cmp_return_mode): Likewise.
5331 (libgcc_shift_count_mode): Likewise.
5332 (unwind_word_mode): Likewise.
5333 (addr_space.pointer_mode): Likewise.
5334 (addr_space.address_mode): Likewise.
5335 * doc/tm.texi: Regenerate.
5336 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5338 (do_jump): Use scalar_int_mode for local variables.
5339 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5340 rather than a machine_mode.
5341 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5342 (scompare_loc_descriptor_wide): Likewise.
5343 (scompare_loc_descriptor_narrow): Likewise.
5344 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5346 * except.c (sjlj_emit_dispatch_table): Likewise.
5347 (expand_builtin_eh_copy_values): Likewise.
5348 * explow.c (convert_memory_address_addr_space_1): Likewise.
5349 Take a scalar_int_mode rather than a machine_mode.
5350 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5351 than a machine_mode.
5352 (memory_address_addr_space): Use scalar_int_mode for local variables.
5353 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5354 rather than a machine_mode.
5355 * expmed.c (mask_rtx): Likewise.
5356 (init_expmed_one_conv): Likewise.
5357 (expand_mult_highpart_adjust): Likewise.
5358 (extract_high_half): Likewise.
5359 (expmed_mult_highpart_optab): Likewise.
5360 (expmed_mult_highpart): Likewise.
5361 (expand_smod_pow2): Likewise.
5362 (expand_sdiv_pow2): Likewise.
5363 (emit_store_flag_int): Likewise.
5364 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5366 (extract_low_bits): Likewise.
5367 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5369 * expr.c (pieces_addr::adjust): Likewise.
5370 (can_store_by_pieces): Likewise.
5371 (store_by_pieces): Likewise.
5372 (clear_by_pieces_1): Likewise.
5373 (expand_expr_addr_expr_1): Likewise.
5374 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5375 (expand_expr_real_1): Likewise.
5376 (try_casesi): Likewise.
5377 * final.c (shorten_branches): Likewise.
5378 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5379 type of "mode" to machine_mode.
5380 * internal-fn.c (expand_arith_overflow_result_store): Take a
5381 scalar_int_mode rather than a machine_mode.
5382 (expand_mul_overflow): Use scalar_int_mode for local variables.
5383 * loop-doloop.c (doloop_modify): Likewise.
5384 (doloop_optimize): Likewise.
5385 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5386 than a machine_mode.
5387 (expand_doubleword_shift_condmove): Likewise.
5388 (expand_doubleword_shift): Likewise.
5389 (expand_doubleword_clz): Likewise.
5390 (expand_doubleword_popcount): Likewise.
5391 (expand_doubleword_parity): Likewise.
5392 (expand_absneg_bit): Use scalar_int_mode for local variables.
5393 (prepare_float_lib_cmp): Likewise.
5394 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5395 rather than a machine_mode.
5396 (convert_memory_address_addr_space): Likewise.
5397 (get_mode_bounds): Likewise.
5398 (get_address_mode): Return a scalar_int_mode rather than a
5400 * rtlanal.c (get_address_mode): Likewise.
5401 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5402 than a machine_mode.
5403 * targhooks.c (default_mode_rep_extended): Likewise.
5404 (default_valid_pointer_mode): Likewise.
5405 (default_addr_space_valid_pointer_mode): Likewise.
5406 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5407 than a machine_mode.
5408 (default_libgcc_cmp_return_mode): Likewise.
5409 (default_libgcc_shift_count_mode): Likewise.
5410 (default_unwind_word_mode): Likewise.
5411 (default_addr_space_pointer_mode): Likewise.
5412 (default_addr_space_address_mode): Likewise.
5413 * targhooks.h (default_eh_return_filter_mode): Likewise.
5414 (default_libgcc_cmp_return_mode): Likewise.
5415 (default_libgcc_shift_count_mode): Likewise.
5416 (default_unwind_word_mode): Likewise.
5417 (default_addr_space_pointer_mode): Likewise.
5418 (default_addr_space_address_mode): Likewise.
5419 (default_mode_rep_extended): Take a scalar_int_mode rather than
5421 (default_valid_pointer_mode): Likewise.
5422 (default_addr_space_valid_pointer_mode): Likewise.
5423 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5425 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5426 rather than a machine_mode.
5427 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5428 for local variables.
5429 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5430 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5431 than a machine_mode.
5433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5434 Alan Hayward <alan.hayward@arm.com>
5435 David Sherwood <david.sherwood@arm.com>
5437 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5438 the mode argument to scalar_int_mode.
5439 (do_jump_by_parts_zero_rtx): Likewise.
5440 (do_jump_by_parts_equality_rtx): Likewise.
5441 (do_jump_by_parts_greater): Take a mode argument.
5442 (do_jump_by_parts_equality): Likewise.
5443 (do_jump_1): Update calls accordingly.
5445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5446 Alan Hayward <alan.hayward@arm.com>
5447 David Sherwood <david.sherwood@arm.com>
5449 * is-a.h (safe_dyn_cast): New function.
5450 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5451 (jump_table_for_label): Likewise.
5452 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5453 instead of an rtx_insn *.
5454 (shorten_branches): Use dyn_cast instead of LABEL_P and
5455 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5456 rtx_jump_table_data::get_data_mode.
5457 (final_scan_insn): Likewise.
5459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5460 Alan Hayward <alan.hayward@arm.com>
5461 David Sherwood <david.sherwood@arm.com>
5463 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5464 trying to combine a full-register integer set with a subreg
5467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5468 Alan Hayward <alan.hayward@arm.com>
5469 David Sherwood <david.sherwood@arm.com>
5471 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5472 that is always either address_mode or pointer_mode.
5474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5475 Alan Hayward <alan.hayward@arm.com>
5476 David Sherwood <david.sherwood@arm.com>
5478 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5479 when the two are known to be equal.
5481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5482 Alan Hayward <alan.hayward@arm.com>
5483 David Sherwood <david.sherwood@arm.com>
5485 * simplify-rtx.c (simplify_const_unary_operation): Use
5486 is_a <scalar_int_mode> instead of checking for a nonzero
5487 precision. Forcibly convert op_mode to a scalar_int_mode
5488 in that case. More clearly differentiate the operand and
5489 result modes and use the former when deciding what the value
5490 of a count-bits operation should be. Use is_int_mode instead
5491 of checking for a MODE_INT. Remove redundant check for whether
5492 this mode has a zero precision.
5494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5495 Alan Hayward <alan.hayward@arm.com>
5496 David Sherwood <david.sherwood@arm.com>
5498 * optabs.c (widen_leading): Change the type of the mode argument
5499 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5500 (widen_bswap): Likewise.
5501 (expand_parity): Likewise.
5502 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5503 (expand_ffs): Likewise.
5504 (epand_unop): Check for scalar integer modes before calling the
5507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5508 Alan Hayward <alan.hayward@arm.com>
5509 David Sherwood <david.sherwood@arm.com>
5511 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5513 (expand_expr_real_1): Update call accordingly.
5515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5516 Alan Hayward <alan.hayward@arm.com>
5517 David Sherwood <david.sherwood@arm.com>
5519 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5520 value_mode arguments. Use scalar_int_mode internally.
5521 (store_bit_field_1): Rename the new integer mode from imode
5522 to op0_mode and use it instead of GET_MODE (op0). Update calls
5523 to store_split_bit_field, store_bit_field_using_insv and
5524 store_fixed_bit_field.
5525 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5526 Use scalar_int_mode internally. Use a bit count rather than a mode
5527 when calculating the largest bit size for get_best_mode.
5528 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5529 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5530 Remove assertion that OP0 has a scalar integer mode.
5531 (store_split_bit_field): Add op0_mode and value_mode arguments.
5532 Update calls to extract_fixed_bit_field.
5533 (extract_bit_field_using_extv): Add an op0_mode argument.
5534 Use scalar_int_mode internally.
5535 (extract_bit_field_1): Rename the new integer mode from imode to
5536 op0_mode and use it instead of GET_MODE (op0). Update calls to
5537 extract_split_bit_field, extract_bit_field_using_extv and
5538 extract_fixed_bit_field.
5539 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5540 to extract_split_bit_field and extract_fixed_bit_field_1.
5541 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5542 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5544 (extract_split_bit_field): Add an op0_mode argument. Update call
5545 to extract_fixed_bit_field.
5547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5548 Alan Hayward <alan.hayward@arm.com>
5549 David Sherwood <david.sherwood@arm.com>
5551 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5552 * explow.c (hard_function_value): Likewise.
5553 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5554 convert_to_mode call outside the loop.
5555 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5556 for the mode iterator. Require the mode specified by max_pieces
5558 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5560 (copy_blkmode_to_reg): Likewise.
5561 (set_storage_via_setmem): Likewise.
5562 * optabs.c (prepare_cmp_insn): Likewise.
5563 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5564 * stor-layout.c (finish_bitfield_representative): Likewise.
5566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5567 Alan Hayward <alan.hayward@arm.com>
5568 David Sherwood <david.sherwood@arm.com>
5570 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5571 * expr.c (convert_move): Use them.
5572 (convert_modes): Likewise.
5573 (store_expr_with_bounds): Likewise.
5575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5576 Alan Hayward <alan.hayward@arm.com>
5577 David Sherwood <david.sherwood@arm.com>
5579 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5580 parameter for the mode of "x". Remove the "known_x", "known_mode"
5581 and "known_ret" arguments. Change the type of the mode argument
5583 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5584 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5585 (reg_num_sign_bit_copies_for_combine): Likewise.
5586 * rtlanal.c (nonzero_bits1): Likewise.
5587 (num_sign_bit_copies1): Likewise.
5588 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5589 (reg_num_sign_bit_copies_general): Likewise.
5590 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5591 (reg_nonzero_bits_general): Likewise.
5593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5594 Alan Hayward <alan.hayward@arm.com>
5595 David Sherwood <david.sherwood@arm.com>
5597 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5598 than in subroutines. Return 1 for non-integer modes.
5599 (cached_num_sign_bit_copies): Change the type of the mode parameter
5601 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5602 classes. Handle CONST_INT_P first and then check whether X also
5603 has a scalar integer mode. Check the same thing for inner registers
5604 of a SUBREG and for values that are being extended or truncated.
5606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5607 Alan Hayward <alan.hayward@arm.com>
5608 David Sherwood <david.sherwood@arm.com>
5610 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5611 in subroutines. Return the mode mask for non-integer modes.
5612 (cached_nonzero_bits): Change the type of the mode parameter
5614 (nonzero_bits1): Likewise. Remove early exit for other mode
5615 classes. Handle CONST_INT_P first and then check whether X
5616 also has a scalar integer mode.
5618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5619 Alan Hayward <alan.hayward@arm.com>
5620 David Sherwood <david.sherwood@arm.com>
5622 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5623 Return a scalar_int_mode and assert that the size is greater than
5625 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5626 (op_by_pieces_d::op_by_pieces_d): Likewise.
5627 (op_by_pieces_d::run): Likewise.
5628 (can_store_by_pieces): Likewise.
5630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5631 Alan Hayward <alan.hayward@arm.com>
5632 David Sherwood <david.sherwood@arm.com>
5634 * combine.c (extract_left_shift): Add a mode argument and update
5636 (make_compound_operation_int): Change the type of the mode parameter
5637 to scalar_int_mode and update the call to extract_left_shift.
5639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5640 Alan Hayward <alan.hayward@arm.com>
5641 David Sherwood <david.sherwood@arm.com>
5643 * combine.c (simplify_and_const_int): Change the type of the mode
5644 parameter to scalar_int_mode.
5645 (simplify_and_const_int_1): Likewise. Update recursive call.
5647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5648 Alan Hayward <alan.hayward@arm.com>
5649 David Sherwood <david.sherwood@arm.com>
5651 * combine.c (simplify_compare_const): Check that the mode is a
5652 scalar_int_mode (rather than VOIDmode) before testing its
5654 (simplify_comparison): Move COMPARISON_P handling out of the
5655 loop and restrict the latter part of the loop to scalar_int_modes.
5656 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5657 and when considering SUBREG_REGs. Use is_int_mode instead of
5658 checking GET_MODE_CLASS against MODE_INT.
5660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5661 Alan Hayward <alan.hayward@arm.com>
5662 David Sherwood <david.sherwood@arm.com>
5664 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5665 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5666 shift_unit_mode and for modes involved in scalar shifts.
5668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5669 Alan Hayward <alan.hayward@arm.com>
5670 David Sherwood <david.sherwood@arm.com>
5672 * combine.c (force_int_to_mode): New function, split out from...
5673 (force_to_mode): ...here. Keep xmode up-to-date and use it
5674 instead of GET_MODE (x).
5676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5677 Alan Hayward <alan.hayward@arm.com>
5678 David Sherwood <david.sherwood@arm.com>
5680 * optabs-query.h (extraction_insn::struct_mode): Change type to
5681 opt_scalar_int_mode and update comment.
5682 (extraction_insn::field_mode): Change type to scalar_int_mode.
5683 (extraction_insn::pos_mode): Likewise.
5684 * combine.c (make_extraction): Update accordingly.
5685 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5686 (get_optab_extraction_insn): Likewise.
5687 * recog.c (simplify_while_replacing): Likewise.
5688 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5689 parameter to opt_scalar_int_mode.
5691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5692 Alan Hayward <alan.hayward@arm.com>
5693 David Sherwood <david.sherwood@arm.com>
5695 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5696 to a scalar_int_mode instead of a machine_mode.
5697 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5698 (get_best_mode): Return a boolean and use a pointer argument to store
5699 the selected mode. Replace the limit mode parameter with a bit limit.
5700 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5701 for the values returned by bit_field_mode_iterator::next_mode.
5702 (store_bit_field): Update call to get_best_mode.
5703 (store_fixed_bit_field): Likewise.
5704 (extract_fixed_bit_field): Likewise.
5705 * expr.c (optimize_bitfield_assignment_op): Likewise.
5706 * fold-const.c (optimize_bit_field_compare): Likewise.
5707 (fold_truth_andor_1): Likewise.
5708 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5709 Update for new type of m_mode.
5710 (get_best_mode): As above.
5712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5713 Alan Hayward <alan.hayward@arm.com>
5714 David Sherwood <david.sherwood@arm.com>
5716 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5717 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5718 (store_bit_field): Check is_a <scalar_int_mode> before calling
5719 strict_volatile_bitfield_p.
5720 (extract_bit_field): Likewise.
5722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5723 Alan Hayward <alan.hayward@arm.com>
5724 David Sherwood <david.sherwood@arm.com>
5726 * target.def (cstore_mode): Return a scalar_int_mode.
5727 * doc/tm.texi: Regenerate.
5728 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5729 * targhooks.h (default_cstore_mode): Likewise.
5730 * targhooks.c (default_cstore_mode): Likewise, using a forced
5732 * expmed.c (emit_cstore): Expect the target of the cstore to be
5735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5736 Alan Hayward <alan.hayward@arm.com>
5737 David Sherwood <david.sherwood@arm.com>
5739 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5741 (niter_desc): Likewise mode.
5742 (iv_analyze): Add a mode parameter.
5744 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5745 and change its type to scalar_int_mode.
5746 * loop-iv.c: Update commentary at head of file.
5747 (iv_constant): Pass the mode paraeter before the rtx it describes
5748 and change its type to scalar_int_mode. Remove VOIDmode handling.
5749 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5750 (iv_extend): Likewise.
5751 (shorten_into_mode): Likewise.
5752 (iv_add): Use scalar_int_mode.
5753 (iv_mult): Likewise.
5754 (iv_shift): Likewise.
5755 (canonicalize_iv_subregs): Likewise.
5756 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5757 it describes and change its mode to scalar_int_mode. Also change
5758 the type of the returned inner_mode to scalar_int_mode.
5759 (get_biv_step): Likewise, turning outer_mode from a pointer
5760 into a direct parameter. Update call to get_biv_step_1.
5761 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5762 iv_constant and get_biv_step.
5763 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5764 and change its type to scalar_int_mode. Don't initialise iv->mode
5765 to VOIDmode and remove later checks for its still being VOIDmode.
5766 Update calls to iv_analyze_op and iv_analyze_expr. Check
5767 is_a <scalar_int_mode> when changing the mode under consideration.
5768 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5769 Update call to iv_analyze_expr.
5770 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5771 inner register is not also a scalar_int_mode. Update call to
5773 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5774 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5775 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5776 separate mode class checks. Update calls to iv_analyze. Remove
5777 fix-up of VOIDmodes after iv_analyze_biv.
5778 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5779 don't have a scalar_int_mode. Update call to biv_p.
5781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5782 Alan Hayward <alan.hayward@arm.com>
5783 David Sherwood <david.sherwood@arm.com>
5785 * cfgexpand.c (convert_debug_memory_address): Use
5786 as_a <scalar_int_mode>.
5787 * combine.c (expand_compound_operation): Likewise.
5788 (make_extraction): Likewise.
5789 (change_zero_ext): Likewise.
5790 (simplify_comparison): Likewise.
5791 * cse.c (cse_insn): Likewise.
5792 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5793 (mem_loc_descriptor): Likewise.
5794 (loc_descriptor): Likewise.
5795 * expmed.c (init_expmed_one_mode): Likewise.
5796 (synth_mult): Likewise.
5797 (emit_store_flag_1): Likewise.
5798 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5799 of a comparison with size.
5800 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5801 (reduce_to_bit_field_precision): Likewise.
5802 * function.c (expand_function_end): Likewise.
5803 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5804 * loop-doloop.c (doloop_modify): Likewise.
5805 * optabs.c (expand_binop): Likewise.
5806 (expand_unop): Likewise.
5807 (expand_copysign_absneg): Likewise.
5808 (prepare_cmp_insn): Likewise.
5809 (maybe_legitimize_operand): Likewise.
5810 * recog.c (const_scalar_int_operand): Likewise.
5811 * rtlanal.c (get_address_mode): Likewise.
5812 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5813 (simplify_cond_clz_ctz): Likewise.
5814 * tree-nested.c (get_nl_goto_field): Likewise.
5815 * tree.c (build_vector_type_for_mode): Likewise.
5816 * var-tracking.c (use_narrower_mode): Likewise.
5818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5819 Alan Hayward <alan.hayward@arm.com>
5820 David Sherwood <david.sherwood@arm.com>
5822 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5823 * builtins.c (expand_builtin_signbit): Use it.
5824 * cfgexpand.c (expand_debug_expr): Likewise.
5825 * dojump.c (do_jump): Likewise.
5826 (do_compare_and_jump): Likewise.
5827 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5828 * expmed.c (make_tree): Likewise.
5829 * expr.c (expand_expr_real_2): Likewise.
5830 (expand_expr_real_1): Likewise.
5831 (try_casesi): Likewise.
5832 * fold-const-call.c (fold_const_call_ss): Likewise.
5833 * fold-const.c (unextend): Likewise.
5834 (extract_muldiv_1): Likewise.
5835 (fold_single_bit_test): Likewise.
5836 (native_encode_int): Likewise.
5837 (native_encode_string): Likewise.
5838 (native_interpret_int): Likewise.
5839 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5840 * internal-fn.c (expand_addsub_overflow): Likewise.
5841 (expand_neg_overflow): Likewise.
5842 (expand_mul_overflow): Likewise.
5843 (expand_arith_overflow): Likewise.
5844 * match.pd: Likewise.
5845 * stor-layout.c (layout_type): Likewise.
5846 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5847 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5848 * tree-ssanames.c (get_range_info): Likewise.
5849 * tree-switch-conversion.c (array_value_type) Likewise.
5850 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5851 (vect_recog_divmod_pattern): Likewise.
5852 (vect_recog_mixed_size_cond_pattern): Likewise.
5853 * tree-vrp.c (extract_range_basic): Likewise.
5854 (simplify_float_conversion_using_ranges): Likewise.
5855 * tree.c (int_fits_type_p): Likewise.
5856 * ubsan.c (instrument_bool_enum_load): Likewise.
5857 * varasm.c (mergeable_string_section): Likewise.
5858 (narrowing_initializer_constant_valid_p): Likewise.
5859 (output_constant): Likewise.
5861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5862 Alan Hayward <alan.hayward@arm.com>
5863 David Sherwood <david.sherwood@arm.com>
5865 * machmode.h (NARROWEST_INT_MODE): New macro.
5866 * expr.c (alignment_for_piecewise_move): Use it instead of
5867 GET_CLASS_NARROWEST_MODE (MODE_INT).
5868 (push_block): Likewise.
5869 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5871 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5874 Alan Hayward <alan.hayward@arm.com>
5875 David Sherwood <david.sherwood@arm.com>
5877 * postreload.c (move2add_valid_value_p): Change the type of the
5878 mode parameter to scalar_int_mode.
5879 (move2add_use_add2_insn): Add a mode parameter and use it instead
5881 (move2add_use_add3_insn): Likewise.
5882 (reload_cse_move2add): Update accordingly.
5884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5885 Alan Hayward <alan.hayward@arm.com>
5886 David Sherwood <david.sherwood@arm.com>
5888 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5890 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5891 * optabs.c (expand_unop): Likewise.
5893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5894 Alan Hayward <alan.hayward@arm.com>
5895 David Sherwood <david.sherwood@arm.com>
5897 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5898 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5899 (popcount_loc_descriptor): Likewise.
5900 (bswap_loc_descriptor): Likewise.
5901 (rotate_loc_descriptor): Likewise.
5902 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5903 calling the functions above.
5905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5906 Alan Hayward <alan.hayward@arm.com>
5907 David Sherwood <david.sherwood@arm.com>
5909 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5911 (try_combine): Likewise.
5912 (simplify_if_then_else): Likewise.
5913 * cse.c (cse_insn): Likewise.
5914 * dwarf2out.c (mem_loc_descriptor): Likewise.
5915 * emit-rtl.c (gen_lowpart_common): Likewise.
5916 * simplify-rtx.c (simplify_truncation): Likewise.
5917 (simplify_binary_operation_1): Likewise.
5918 (simplify_const_relational_operation): Likewise.
5919 (simplify_ternary_operation): Likewise.
5920 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
5922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5923 Alan Hayward <alan.hayward@arm.com>
5924 David Sherwood <david.sherwood@arm.com>
5926 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5927 * reload.c (push_reload): Likewise.
5928 (find_reloads): Likewise.
5930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5931 Alan Hayward <alan.hayward@arm.com>
5932 David Sherwood <david.sherwood@arm.com>
5934 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5935 (make_compound_operation_int): Likewise.
5936 (change_zero_ext): Likewise.
5937 * expr.c (convert_move): Likewise.
5938 (convert_modes): Likewise.
5939 * fwprop.c (forward_propagate_subreg): Likewise.
5940 * loop-iv.c (get_biv_step_1): Likewise.
5941 * optabs.c (widen_operand): Likewise.
5942 * postreload.c (move2add_valid_value_p): Likewise.
5943 * recog.c (simplify_while_replacing): Likewise.
5944 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5945 (simplify_binary_operation_1): Likewise. Remove redundant
5946 mode equality check.
5948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5949 Alan Hayward <alan.hayward@arm.com>
5950 David Sherwood <david.sherwood@arm.com>
5952 * combine.c (combine_simplify_rtx): Add checks for
5953 is_a <scalar_int_mode>.
5954 (simplify_if_then_else): Likewise.
5955 (make_field_assignment): Likewise.
5956 (simplify_comparison): Likewise.
5957 * ifcvt.c (noce_try_bitop): Likewise.
5958 * loop-invariant.c (canonicalize_address_mult): Likewise.
5959 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5962 Alan Hayward <alan.hayward@arm.com>
5963 David Sherwood <david.sherwood@arm.com>
5965 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5966 is_a <scalar_int_mode> instead of != BLKmode.
5968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5969 Alan Hayward <alan.hayward@arm.com>
5970 David Sherwood <david.sherwood@arm.com>
5972 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5973 instead of != VOIDmode.
5974 * combine.c (if_then_else_cond): Likewise.
5975 (change_zero_ext): Likewise.
5976 * dwarf2out.c (mem_loc_descriptor): Likewise.
5977 (loc_descriptor): Likewise.
5978 * rtlanal.c (canonicalize_condition): Likewise.
5979 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5982 Alan Hayward <alan.hayward@arm.com>
5983 David Sherwood <david.sherwood@arm.com>
5985 * simplify-rtx.c (simplify_binary_operation_1): Use
5986 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5989 Alan Hayward <alan.hayward@arm.com>
5990 David Sherwood <david.sherwood@arm.com>
5992 * wide-int.h (int_traits<unsigned char>) New class.
5993 (int_traits<unsigned short>) Likewise.
5994 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5995 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5997 * combine.c (set_nonzero_bits_and_sign_copies): Use
5998 is_a <scalar_int_mode>.
5999 (find_split_point): Likewise.
6000 (combine_simplify_rtx): Likewise.
6001 (simplify_logical): Likewise.
6002 (expand_compound_operation): Likewise.
6003 (expand_field_assignment): Likewise.
6004 (make_compound_operation): Likewise.
6005 (extended_count): Likewise.
6006 (change_zero_ext): Likewise.
6007 (simplify_comparison): Likewise.
6008 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6009 (ucompare_loc_descriptor): Likewise.
6010 (minmax_loc_descriptor): Likewise.
6011 (mem_loc_descriptor): Likewise.
6012 (loc_descriptor): Likewise.
6013 * expmed.c (init_expmed_one_mode): Likewise.
6014 * lra-constraints.c (lra_constraint_offset): Likewise.
6015 * optabs.c (prepare_libcall_arg): Likewise.
6016 * postreload.c (move2add_note_store): Likewise.
6017 * reload.c (operands_match_p): Likewise.
6018 * rtl.h (load_extend_op): Likewise.
6019 * rtlhooks.c (gen_lowpart_general): Likewise.
6020 * simplify-rtx.c (simplify_truncation): Likewise.
6021 (simplify_unary_operation_1): Likewise.
6022 (simplify_binary_operation_1): Likewise.
6023 (simplify_const_binary_operation): Likewise.
6024 (simplify_const_relational_operation): Likewise.
6025 (simplify_subreg): Likewise.
6026 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6027 * var-tracking.c (adjust_mems): Likewise.
6028 (prepare_call_arguments): Likewise.
6030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6031 Alan Hayward <alan.hayward@arm.com>
6032 David Sherwood <david.sherwood@arm.com>
6034 * machmode.h (is_int_mode): New fuction.
6035 * combine.c (find_split_point): Use it.
6036 (combine_simplify_rtx): Likewise.
6037 (simplify_if_then_else): Likewise.
6038 (simplify_set): Likewise.
6039 (simplify_shift_const_1): Likewise.
6040 (simplify_comparison): Likewise.
6041 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6042 * cse.c (notreg_cost): Likewise.
6043 (cse_insn): Likewise.
6044 * cselib.c (cselib_lookup_1): Likewise.
6045 * dojump.c (do_jump_1): Likewise.
6046 (do_compare_rtx_and_jump): Likewise.
6047 * dse.c (get_call_args): Likewise.
6048 * dwarf2out.c (rtl_for_decl_init): Likewise.
6049 (native_encode_initializer): Likewise.
6050 * expmed.c (emit_store_flag_1): Likewise.
6051 (emit_store_flag): Likewise.
6052 * expr.c (convert_modes): Likewise.
6053 (store_field): Likewise.
6054 (expand_expr_real_1): Likewise.
6055 * fold-const.c (fold_read_from_constant_string): Likewise.
6056 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6057 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6058 * optabs.c (expand_binop): Likewise.
6059 (expand_unop): Likewise.
6060 (expand_abs_nojump): Likewise.
6061 (expand_one_cmpl_abs_nojump): Likewise.
6062 * simplify-rtx.c (mode_signbit_p): Likewise.
6063 (val_signbit_p): Likewise.
6064 (val_signbit_known_set_p): Likewise.
6065 (val_signbit_known_clear_p): Likewise.
6066 (simplify_relational_operation_1): Likewise.
6067 * tree.c (vector_type_mode): Likewise.
6069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6070 Alan Hayward <alan.hayward@arm.com>
6071 David Sherwood <david.sherwood@arm.com>
6073 * machmode.h (smallest_mode_for_size): Fix formatting.
6074 (smallest_int_mode_for_size): New function.
6075 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6076 instead of smallest_mode_for_size.
6077 * combine.c (make_extraction): Likewise.
6078 * config/arc/arc.c (arc_expand_movmem): Likewise.
6079 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6080 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6081 * config/s390/s390.c (s390_expand_insv): Likewise.
6082 * config/sparc/sparc.c (assign_int_registers): Likewise.
6083 * config/spu/spu.c (spu_function_value): Likewise.
6084 (spu_function_arg): Likewise.
6085 * coverage.c (get_gcov_type): Likewise.
6086 (get_gcov_unsigned_t): Likewise.
6087 * dse.c (find_shift_sequence): Likewise.
6088 * expmed.c (store_bit_field_1): Likewise.
6089 * expr.c (convert_move): Likewise.
6090 (store_field): Likewise.
6091 * internal-fn.c (expand_arith_overflow): Likewise.
6092 * optabs-query.c (get_best_extraction_insn): Likewise.
6093 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6094 * stor-layout.c (layout_type): Likewise.
6095 (initialize_sizetypes): Likewise.
6096 * targhooks.c (default_get_mask_mode): Likewise.
6097 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6100 Alan Hayward <alan.hayward@arm.com>
6101 David Sherwood <david.sherwood@arm.com>
6103 * machmode.h (opt_mode::else_blk): New function.
6104 (int_mode_for_mode): Declare.
6105 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6106 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6108 * cfgexpand.c (expand_debug_expr): Likewise.
6109 * combine.c (gen_lowpart_or_truncate): Likewise.
6110 (gen_lowpart_for_combine): Likewise.
6111 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6112 * config/avr/avr.c (avr_to_int_mode): Likewise.
6113 (avr_out_plus_1): Likewise.
6114 (avr_out_plus): Likewise.
6115 (avr_out_round): Likewise.
6116 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6117 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6118 (s390_expand_vcond): Likewise.
6119 * config/spu/spu.c (spu_split_immediate): Likewise.
6120 (spu_expand_mov): Likewise.
6121 * dse.c (get_stored_val): Likewise.
6122 * expmed.c (store_bit_field_1): Likewise.
6123 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6125 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6126 (extract_low_bits): Likewise.
6127 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6128 handling rather than repeating the check.
6129 (emit_group_store): Likewise.
6130 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6131 * optabs.c (expand_absneg_bit): Likewise.
6132 (expand_copysign_absneg): Likewise.
6133 (expand_copysign_bit): Likewise.
6134 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6135 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6136 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6137 * var-tracking.c (prepare_call_arguments): Likewise.
6138 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6139 int_mode_for_mode instead of mode_for_size.
6140 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6143 Alan Hayward <alan.hayward@arm.com>
6144 David Sherwood <david.sherwood@arm.com>
6146 * machmode.h (int_mode_for_size): New function.
6147 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6148 instead of mode_for_size.
6149 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6151 * combine.c (expand_field_assignment): Use int_mode_for_size
6152 instead of mode_for_size.
6153 (make_extraction): Likewise.
6154 (simplify_shift_const_1): Likewise.
6155 (simplify_comparison): Likewise.
6156 * dojump.c (do_jump): Likewise.
6157 * dwarf2out.c (mem_loc_descriptor): Likewise.
6158 * emit-rtl.c (init_derived_machine_modes): Likewise.
6159 * expmed.c (flip_storage_order): Likewise.
6160 (convert_extracted_bit_field): Likewise.
6161 * expr.c (copy_blkmode_from_reg): Likewise.
6162 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6163 * internal-fn.c (expand_mul_overflow): Likewise.
6164 * lower-subreg.c (simple_move): Likewise.
6165 * optabs-libfuncs.c (init_optabs): Likewise.
6166 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6167 * tree.c (vector_type_mode): Likewise.
6168 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6169 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6170 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6171 * tree-vect-stmts.c (vectorizable_load): Likewise.
6172 (vectorizable_store): Likewise.
6174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6175 Alan Hayward <alan.hayward@arm.com>
6176 David Sherwood <david.sherwood@arm.com>
6178 * coretypes.h (pod_mode): New type.
6179 (scalar_int_mode_pod): New typedef.
6180 * machmode.h (pod_mode): New class.
6181 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6182 * genmodes.c (emit_mode_int_n): Update accordingly.
6183 * lower-subreg.h (target_lower_subreg): Change type to
6184 scalar_int_mode_pod.
6185 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6186 scalar_int_mode_pod.
6188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6189 Alan Hayward <alan.hayward@arm.com>
6190 David Sherwood <david.sherwood@arm.com>
6192 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6193 machine_mode to scalar_int_mode.
6194 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6195 (rs6000_option_override_internal): Remove cast to int.
6196 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6197 machine_mode to scalar_int_mode.
6198 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6199 (rs6000_option_override_internal): Remove cast to int.
6200 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6201 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6203 * config/s390/s390.c (s390_expand_builtin): Likewise.
6204 * coretypes.h (scalar_int_mode): New type.
6205 (opt_scalar_int_mode): New typedef.
6206 * machmode.h (scalar_int_mode): New class.
6207 (scalar_int_mode::includes_p): New function.
6208 (byte_mode): Change type to scalar_int_mode.
6209 (word_mode): Likewise.
6210 (ptr_mode): Likewise.
6211 * emit-rtl.c (byte_mode): Likewise.
6212 (word_mode): Likewise.
6213 (ptr_mode): Likewise.
6214 (init_derived_machine_modes): Update accordingly.
6215 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6216 and MODE_PARTIAL_INT.
6217 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6218 opt_scalar_int_mode.
6220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6221 Alan Hayward <alan.hayward@arm.com>
6222 David Sherwood <david.sherwood@arm.com>
6224 * target.def (libgcc_floating_mode_supported_p): Take a
6226 * doc/tm.texi: Regenerate.
6227 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6229 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6230 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6234 Alan Hayward <alan.hayward@arm.com>
6235 David Sherwood <david.sherwood@arm.com>
6237 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6238 * doc/tm.texi: Regenerate.
6239 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6240 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6241 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6242 * targhooks.h (default_floatn_mode): Likewise.
6243 * targhooks.c (default_floatn_mode): Likewise.
6244 * tree.c (build_common_tree_nodes): Update accordingly.
6246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6247 Alan Hayward <alan.hayward@arm.com>
6248 David Sherwood <david.sherwood@arm.com>
6250 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6251 (mode_iterator::iterate_p): Likewise.
6252 (mode_iterator::get_wider): Likewise.
6253 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6256 Alan Hayward <alan.hayward@arm.com>
6257 David Sherwood <david.sherwood@arm.com>
6259 * coretypes.h (opt_scalar_float_mode): New typedef.
6260 * machmode.h (float_mode_for_size): New function.
6261 * emit-rtl.c (double_mode): Delete.
6262 (init_emit_once): Use float_mode_for_size.
6263 * stor-layout.c (layout_type): Likewise.
6264 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6267 Alan Hayward <alan.hayward@arm.com>
6268 David Sherwood <david.sherwood@arm.com>
6270 * output.h (assemble_real): Take a scalar_float_mode.
6271 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6272 * config/arm/arm.md (consttable_4): Likewise.
6273 (consttable_8): Likewise.
6274 (consttable_16): Likewise.
6275 * config/mips/mips.md (consttable_float): Likewise.
6276 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6277 * varasm.c (assemble_real): Take a scalar_float_mode.
6278 (output_constant_pool_2): Update accordingly.
6279 (output_constant): Likewise.
6281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6282 Alan Hayward <alan.hayward@arm.com>
6283 David Sherwood <david.sherwood@arm.com>
6285 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6286 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6287 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6288 (native_encode_real): Likewise.
6289 (native_interpret_real): Likewise.
6290 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6291 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6294 Alan Hayward <alan.hayward@arm.com>
6295 David Sherwood <david.sherwood@arm.com>
6297 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6298 <scalar_float_mode>. Simplify.
6299 (gen_extend_conv_libfunc): Likewise.
6301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6302 Alan Hayward <alan.hayward@arm.com>
6303 David Sherwood <david.sherwood@arm.com>
6305 * coretypes.h (scalar_float_mode): New type.
6306 * machmode.h (mode_traits::from_int): Use machine_mode if
6307 USE_ENUM_MODES is defined.
6308 (is_a): New function.
6310 (dyn_cast): Likewise.
6311 (scalar_float_mode): New class.
6312 (scalar_float_mode::includes_p): New function.
6313 (is_float_mode): Likewise.
6314 * gdbhooks.py (MachineModePrinter): New class.
6315 (build_pretty_printer): Use it for scalar_float_mode.
6316 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6317 (format_helper::format_helper): Turn into a template.
6318 * genmodes.c (get_mode_class): New function.
6319 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6320 or machine_mode if none.
6321 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6322 as_a <scalar_float_mode>.
6323 * dwarf2out.c (mem_loc_descriptor): Likewise.
6324 (insert_float): Likewise.
6325 (add_const_value_attribute): Likewise.
6326 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6327 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6328 (expand_unop): Update accordingly.
6329 (expand_abs_nojump): Likewise.
6330 (expand_copysign_absneg): Take a scalar_float_mode.
6331 (expand_copysign_bit): Likewise.
6332 (expand_copysign): Update accordingly.
6334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6335 Alan Hayward <alan.hayward@arm.com>
6336 David Sherwood <david.sherwood@arm.com>
6338 * coretypes.h (opt_mode): New class.
6339 * machmode.h (opt_mode): Likewise.
6340 (opt_mode::else_void): New function.
6341 (opt_mode::require): Likewise.
6342 (opt_mode::exists): Likewise.
6343 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6344 (GET_MODE_2XWIDER_MODE): Likewise.
6345 (mode_iterator::get_wider): Update accordingly.
6346 (mode_iterator::get_2xwider): Likewise.
6347 (mode_iterator::get_known_wider): Likewise, turning into a template.
6348 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6349 forcing a wider mode to exist.
6350 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6351 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6352 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6353 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6354 * lower-subreg.c (init_lower_subreg): Likewise.
6355 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6356 on the final iteration.
6357 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6358 a wider mode exists before asking for a move pattern.
6359 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6360 forcing a wider mode to exist.
6361 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6362 returning false if no such mode exists.
6363 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6364 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6365 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6366 Avoid checking for a MODE_INT if we already know the mode is not a
6368 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6369 forcing a wider mode to exist.
6370 (expmed_mult_highpart_optab): Likewise.
6371 (expmed_mult_highpart): Likewise.
6372 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6374 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6375 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6376 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6377 * internal-fn.c (expand_mul_overflow): Update use of
6378 GET_MODE_2XWIDER_MODE.
6379 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6380 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6381 GET_MODE_WIDER_MODE.
6382 (convert_plusminus_to_widen): Likewise.
6383 * tree-switch-conversion.c (array_value_type): Likewise.
6384 * var-tracking.c (emit_note_insn_var_location): Likewise.
6385 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6386 Return false inside rather than outside the loop if no wider mode
6388 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6389 and GET_MODE_2XWIDER_MODE
6390 (can_compare_p): Use else_void.
6391 * gdbhooks.py (OptMachineModePrinter): New class.
6392 (build_pretty_printer): Use it for opt_mode.
6394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6395 Alan Hayward <alan.hayward@arm.com>
6396 David Sherwood <david.sherwood@arm.com>
6398 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6399 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6402 Alan Hayward <alan.hayward@arm.com>
6403 David Sherwood <david.sherwood@arm.com>
6405 * machmode.h (mode_traits): New structure.
6406 (get_narrowest_mode): New function.
6407 (mode_iterator::start): Likewise.
6408 (mode_iterator::iterate_p): Likewise.
6409 (mode_iterator::get_wider): Likewise.
6410 (mode_iterator::get_known_wider): Likewise.
6411 (mode_iterator::get_2xwider): Likewise.
6412 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6413 (FOR_EACH_MODE): Likewise.
6414 (FOR_EACH_MODE_FROM): Likewise.
6415 (FOR_EACH_MODE_UNTIL): Likewise.
6416 (FOR_EACH_WIDER_MODE): Likewise.
6417 (FOR_EACH_2XWIDER_MODE): Likewise.
6418 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6419 * combine.c (simplify_comparison): Likewise
6420 * config/i386/i386.c (type_natural_mode): Likewise.
6421 * cse.c (cse_insn): Likewise.
6422 * dse.c (find_shift_sequence): Likewise.
6423 * emit-rtl.c (init_derived_machine_modes): Likewise.
6424 (init_emit_once): Likewise.
6425 * explow.c (hard_function_value): Likewise.
6426 * expmed.c (extract_fixed_bit_field_1): Likewise.
6427 (extract_bit_field_1): Likewise.
6428 (expand_divmod): Likewise.
6429 (emit_store_flag_1): Likewise.
6430 * expr.c (init_expr_target): Likewise.
6431 (convert_move): Likewise.
6432 (alignment_for_piecewise_move): Likewise.
6433 (widest_int_mode_for_size): Likewise.
6434 (emit_block_move_via_movmem): Likewise.
6435 (copy_blkmode_to_reg): Likewise.
6436 (set_storage_via_setmem): Likewise.
6437 (compress_float_constant): Likewise.
6438 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6439 * optabs-query.c (get_best_extraction_insn): Likewise.
6440 * optabs.c (expand_binop): Likewise.
6441 (expand_twoval_unop): Likewise.
6442 (expand_twoval_binop): Likewise.
6443 (widen_leading): Likewise.
6444 (widen_bswap): Likewise.
6445 (expand_parity): Likewise.
6446 (expand_unop): Likewise.
6447 (prepare_cmp_insn): Likewise.
6448 (prepare_float_lib_cmp): Likewise.
6449 (expand_float): Likewise.
6450 (expand_fix): Likewise.
6451 (expand_sfix_optab): Likewise.
6452 * postreload.c (move2add_use_add2_insn): Likewise.
6453 * reg-stack.c (reg_to_stack): Likewise.
6454 * reginfo.c (choose_hard_reg_mode): Likewise.
6455 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6456 * stor-layout.c (mode_for_size): Likewise.
6457 (smallest_mode_for_size): Likewise.
6458 (mode_for_vector): Likewise.
6459 (finish_bitfield_representative): Likewise.
6460 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6461 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6462 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6463 * var-tracking.c (prepare_call_arguments): Likewise.
6465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6466 Alan Hayward <alan.hayward@arm.com>
6467 David Sherwood <david.sherwood@arm.com>
6469 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6470 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6471 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6472 * machmode.h (mode_size): Move earlier in file.
6473 (mode_precision): Likewise.
6474 (mode_inner): Likewise.
6475 (mode_nunits): Likewise.
6476 (mode_unit_size): Likewise.
6477 (unit_unit_precision): Likewise.
6478 (mode_wider): Likewise.
6479 (mode_2xwider): Likewise.
6480 (machine_mode): New class.
6481 (mode_to_bytes): New function.
6482 (mode_to_bits): Likewise.
6483 (mode_to_precision): Likewise.
6484 (mode_to_inner): Likewise.
6485 (mode_to_unit_size): Likewise.
6486 (mode_to_unit_precision): Likewise.
6487 (mode_to_nunits): Likewise.
6488 (GET_MODE_SIZE): Use mode_to_bytes.
6489 (GET_MODE_BITSIZE): Use mode_to_bits.
6490 (GET_MODE_PRECISION): Use mode_to_precision.
6491 (GET_MODE_INNER): Use mode_to_inner.
6492 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6493 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6494 (GET_MODE_NUNITS): Use mode_to_nunits.
6495 * system.h (ALWAYS_INLINE): New macro.
6496 * config/powerpcspe/powerpcspe-c.c
6497 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6498 int for arg1_mode and arg2_mode.
6500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6501 Alan Hayward <alan.hayward@arm.com>
6502 David Sherwood <david.sherwood@arm.com>
6504 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6505 Prefix mode names with E_ in case statements.
6506 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6507 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6508 (aarch64_split_simd_move): Likewise.
6509 (aarch64_gen_storewb_pair): Likewise.
6510 (aarch64_gen_loadwb_pair): Likewise.
6511 (aarch64_gen_store_pair): Likewise.
6512 (aarch64_gen_load_pair): Likewise.
6513 (aarch64_get_condition_code_1): Likewise.
6514 (aarch64_constant_pool_reload_icode): Likewise.
6515 (get_rsqrte_type): Likewise.
6516 (get_rsqrts_type): Likewise.
6517 (get_recpe_type): Likewise.
6518 (get_recps_type): Likewise.
6519 (aarch64_gimplify_va_arg_expr): Likewise.
6520 (aarch64_simd_container_mode): Likewise.
6521 (aarch64_emit_load_exclusive): Likewise.
6522 (aarch64_emit_store_exclusive): Likewise.
6523 (aarch64_expand_compare_and_swap): Likewise.
6524 (aarch64_gen_atomic_cas): Likewise.
6525 (aarch64_emit_bic): Likewise.
6526 (aarch64_emit_atomic_swap): Likewise.
6527 (aarch64_emit_atomic_load_op): Likewise.
6528 (aarch64_evpc_trn): Likewise.
6529 (aarch64_evpc_uzp): Likewise.
6530 (aarch64_evpc_zip): Likewise.
6531 (aarch64_evpc_ext): Likewise.
6532 (aarch64_evpc_rev): Likewise.
6533 (aarch64_evpc_dup): Likewise.
6534 (aarch64_gen_ccmp_first): Likewise.
6535 (aarch64_gen_ccmp_next): Likewise.
6536 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6537 (alpha_emit_xfloating_libcall): Likewise.
6538 (emit_insxl): Likewise.
6539 (alpha_arg_type): Likewise.
6540 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6541 (arc_preferred_simd_mode): Likewise.
6542 (arc_secondary_reload): Likewise.
6543 (get_arc_condition_code): Likewise.
6544 (arc_print_operand): Likewise.
6545 (arc_legitimate_constant_p): Likewise.
6546 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6547 * config/arc/arc.md (casesi_load): Likewise.
6548 (casesi_compact_jump): Likewise.
6549 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6550 (cc_use_register): Likewise.
6551 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6552 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6553 (arm_init_iwmmxt_builtins): Likewise.
6554 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6555 (neon_expand_vector_init): Likewise.
6556 (arm_attr_length_move_neon): Likewise.
6557 (maybe_get_arm_condition_code): Likewise.
6558 (arm_emit_vector_const): Likewise.
6559 (arm_preferred_simd_mode): Likewise.
6560 (arm_output_iwmmxt_tinsr): Likewise.
6561 (thumb1_output_casesi): Likewise.
6562 (thumb2_output_casesi): Likewise.
6563 (arm_emit_load_exclusive): Likewise.
6564 (arm_emit_store_exclusive): Likewise.
6565 (arm_expand_compare_and_swap): Likewise.
6566 (arm_evpc_neon_vuzp): Likewise.
6567 (arm_evpc_neon_vzip): Likewise.
6568 (arm_evpc_neon_vrev): Likewise.
6569 (arm_evpc_neon_vtrn): Likewise.
6570 (arm_evpc_neon_vext): Likewise.
6571 (arm_validize_comparison): Likewise.
6572 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6573 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6574 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6575 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6576 (c6x_preferred_simd_mode): Likewise.
6577 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6578 (epiphany_rtx_costs): Likewise.
6579 * config/epiphany/predicates.md (proper_comparison_operator):
6581 * config/frv/frv.c (condexec_memory_operand): Likewise.
6582 (frv_emit_move): Likewise.
6583 (output_move_single): Likewise.
6584 (output_condmove_single): Likewise.
6585 (frv_hard_regno_mode_ok): Likewise.
6586 (frv_matching_accg_mode): Likewise.
6587 * config/h8300/h8300.c (split_adds_subs): Likewise.
6588 (h8300_rtx_costs): Likewise.
6589 (h8300_print_operand): Likewise.
6590 (compute_mov_length): Likewise.
6591 (output_logical_op): Likewise.
6592 (compute_logical_op_length): Likewise.
6593 (compute_logical_op_cc): Likewise.
6594 (h8300_shift_needs_scratch_p): Likewise.
6595 (output_a_shift): Likewise.
6596 (compute_a_shift_length): Likewise.
6597 (compute_a_shift_cc): Likewise.
6598 (expand_a_rotate): Likewise.
6599 (output_a_rotate): Likewise.
6600 * config/i386/i386.c (classify_argument): Likewise.
6601 (function_arg_advance_32): Likewise.
6602 (function_arg_32): Likewise.
6603 (function_arg_64): Likewise.
6604 (function_value_64): Likewise.
6605 (ix86_gimplify_va_arg): Likewise.
6606 (ix86_legitimate_constant_p): Likewise.
6607 (put_condition_code): Likewise.
6608 (split_double_mode): Likewise.
6609 (ix86_avx256_split_vector_move_misalign): Likewise.
6610 (ix86_expand_vector_logical_operator): Likewise.
6611 (ix86_split_idivmod): Likewise.
6612 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6613 (ix86_build_const_vector): Likewise.
6614 (ix86_build_signbit_mask): Likewise.
6615 (ix86_match_ccmode): Likewise.
6616 (ix86_cc_modes_compatible): Likewise.
6617 (ix86_expand_branch): Likewise.
6618 (ix86_expand_sse_cmp): Likewise.
6619 (ix86_expand_sse_movcc): Likewise.
6620 (ix86_expand_int_sse_cmp): Likewise.
6621 (ix86_expand_vec_perm_vpermi2): Likewise.
6622 (ix86_expand_vec_perm): Likewise.
6623 (ix86_expand_sse_unpack): Likewise.
6624 (ix86_expand_int_addcc): Likewise.
6625 (ix86_split_to_parts): Likewise.
6626 (ix86_vectorize_builtin_gather): Likewise.
6627 (ix86_vectorize_builtin_scatter): Likewise.
6628 (avx_vpermilp_parallel): Likewise.
6629 (inline_memory_move_cost): Likewise.
6630 (ix86_tieable_integer_mode_p): Likewise.
6631 (x86_maybe_negate_const_int): Likewise.
6632 (ix86_expand_vector_init_duplicate): Likewise.
6633 (ix86_expand_vector_init_one_nonzero): Likewise.
6634 (ix86_expand_vector_init_one_var): Likewise.
6635 (ix86_expand_vector_init_concat): Likewise.
6636 (ix86_expand_vector_init_interleave): Likewise.
6637 (ix86_expand_vector_init_general): Likewise.
6638 (ix86_expand_vector_set): Likewise.
6639 (ix86_expand_vector_extract): Likewise.
6640 (emit_reduc_half): Likewise.
6641 (ix86_emit_i387_round): Likewise.
6642 (ix86_mangle_type): Likewise.
6643 (ix86_expand_round_sse4): Likewise.
6644 (expand_vec_perm_blend): Likewise.
6645 (canonicalize_vector_int_perm): Likewise.
6646 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6647 (expand_vec_perm_1): Likewise.
6648 (expand_vec_perm_interleave3): Likewise.
6649 (expand_vec_perm_even_odd_pack): Likewise.
6650 (expand_vec_perm_even_odd_1): Likewise.
6651 (expand_vec_perm_broadcast_1): Likewise.
6652 (ix86_vectorize_vec_perm_const_ok): Likewise.
6653 (ix86_expand_vecop_qihi): Likewise.
6654 (ix86_expand_mul_widen_hilo): Likewise.
6655 (ix86_expand_sse2_abs): Likewise.
6656 (ix86_expand_pextr): Likewise.
6657 (ix86_expand_pinsr): Likewise.
6658 (ix86_preferred_simd_mode): Likewise.
6659 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6660 * config/i386/sse.md (*andnot<mode>3): Likewise.
6661 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6662 (*<code><mode>3): Likewise.
6663 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6664 (ia64_expand_atomic_op): Likewise.
6665 (ia64_arg_type): Likewise.
6666 (ia64_mode_to_int): Likewise.
6667 (ia64_scalar_mode_supported_p): Likewise.
6668 (ia64_vector_mode_supported_p): Likewise.
6669 (expand_vec_perm_broadcast): Likewise.
6670 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6671 (iq2000_function_arg_advance): Likewise.
6672 (iq2000_function_arg): Likewise.
6673 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6674 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6675 (m68k_libcall_value): Likewise.
6676 (m68k_function_value): Likewise.
6677 (sched_attr_op_type): Likewise.
6678 * config/mcore/mcore.c (mcore_output_move): Likewise.
6679 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6681 (microblaze_function_arg): Likewise.
6682 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6683 (mips_print_operand): Likewise.
6684 (mips_mode_ok_for_mov_fmt_p): Likewise.
6685 (mips_vector_mode_supported_p): Likewise.
6686 (mips_preferred_simd_mode): Likewise.
6687 (mips_expand_vpc_loongson_even_odd): Likewise.
6688 (mips_expand_vec_unpack): Likewise.
6689 (mips_expand_vi_broadcast): Likewise.
6690 (mips_expand_vector_init): Likewise.
6691 (mips_expand_vec_reduc): Likewise.
6692 (mips_expand_msa_cmp): Likewise.
6693 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6694 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6695 (cc_flags_for_mode): Likewise.
6696 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6697 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6698 (nds32_output_casesi_pc_relative): Likewise.
6699 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6700 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6701 (nvptx_gen_unpack): Likewise.
6702 (nvptx_gen_pack): Likewise.
6703 (nvptx_gen_shuffle): Likewise.
6704 (nvptx_gen_wcast): Likewise.
6705 (nvptx_preferred_simd_mode): Likewise.
6706 * config/pa/pa.c (pa_secondary_reload): Likewise.
6707 * config/pa/predicates.md (base14_operand): Likewise.
6708 * config/powerpcspe/powerpcspe-c.c
6709 (altivec_resolve_overloaded_builtin): Likewise.
6710 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6712 (rs6000_preferred_simd_mode): Likewise.
6713 (output_vec_const_move): Likewise.
6714 (rs6000_expand_vector_extract): Likewise.
6715 (rs6000_split_vec_extract_var): Likewise.
6716 (reg_offset_addressing_ok_p): Likewise.
6717 (rs6000_legitimate_offset_address_p): Likewise.
6718 (rs6000_legitimize_address): Likewise.
6719 (rs6000_emit_set_const): Likewise.
6720 (rs6000_const_vec): Likewise.
6721 (rs6000_emit_move): Likewise.
6722 (spe_build_register_parallel): Likewise.
6723 (rs6000_darwin64_record_arg_recurse): Likewise.
6724 (swap_selector_for_mode): Likewise.
6725 (spe_init_builtins): Likewise.
6726 (paired_init_builtins): Likewise.
6727 (altivec_init_builtins): Likewise.
6728 (do_load_for_compare): Likewise.
6729 (rs6000_generate_compare): Likewise.
6730 (rs6000_expand_float128_convert): Likewise.
6731 (emit_load_locked): Likewise.
6732 (emit_store_conditional): Likewise.
6733 (rs6000_output_function_epilogue): Likewise.
6734 (rs6000_handle_altivec_attribute): Likewise.
6735 (rs6000_function_value): Likewise.
6736 (emit_fusion_gpr_load): Likewise.
6737 (emit_fusion_p9_load): Likewise.
6738 (emit_fusion_p9_store): Likewise.
6739 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6740 (fusion_gpr_mem_load): Likewise.
6741 (fusion_addis_mem_combo_load): Likewise.
6742 (fusion_addis_mem_combo_store): Likewise.
6743 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6744 (fusion_gpr_mem_load): Likewise.
6745 (fusion_addis_mem_combo_load): Likewise.
6746 (fusion_addis_mem_combo_store): Likewise.
6747 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6749 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6750 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6751 (rs6000_preferred_simd_mode): Likewise.
6752 (output_vec_const_move): Likewise.
6753 (rs6000_expand_vector_extract): Likewise.
6754 (rs6000_split_vec_extract_var): Likewise.
6755 (reg_offset_addressing_ok_p): Likewise.
6756 (rs6000_legitimate_offset_address_p): Likewise.
6757 (rs6000_legitimize_address): Likewise.
6758 (rs6000_emit_set_const): Likewise.
6759 (rs6000_const_vec): Likewise.
6760 (rs6000_emit_move): Likewise.
6761 (rs6000_darwin64_record_arg_recurse): Likewise.
6762 (swap_selector_for_mode): Likewise.
6763 (paired_init_builtins): Likewise.
6764 (altivec_init_builtins): Likewise.
6765 (rs6000_expand_float128_convert): Likewise.
6766 (emit_load_locked): Likewise.
6767 (emit_store_conditional): Likewise.
6768 (rs6000_output_function_epilogue): Likewise.
6769 (rs6000_handle_altivec_attribute): Likewise.
6770 (rs6000_function_value): Likewise.
6771 (emit_fusion_gpr_load): Likewise.
6772 (emit_fusion_p9_load): Likewise.
6773 (emit_fusion_p9_store): Likewise.
6774 * config/rx/rx.c (rx_gen_move_template): Likewise.
6775 (flags_from_mode): Likewise.
6776 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6777 (s390_slb_comparison): Likewise.
6778 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6779 (s390_vector_mode_supported_p): Likewise.
6780 (s390_cc_modes_compatible): Likewise.
6781 (s390_match_ccmode_set): Likewise.
6782 (s390_canonicalize_comparison): Likewise.
6783 (s390_emit_compare_and_swap): Likewise.
6784 (s390_branch_condition_mask): Likewise.
6785 (s390_rtx_costs): Likewise.
6786 (s390_secondary_reload): Likewise.
6787 (__SECONDARY_RELOAD_CASE): Likewise.
6788 (s390_expand_cs): Likewise.
6789 (s390_preferred_simd_mode): Likewise.
6790 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6791 * config/sh/sh.c (sh_print_operand): Likewise.
6792 (dump_table): Likewise.
6793 (sh_secondary_reload): Likewise.
6794 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6795 * config/sh/sh.md (casesi_worker_1): Likewise.
6796 (casesi_worker_2): Likewise.
6797 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6798 (fcc_comparison_operator): Likewise.
6799 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6800 (emit_soft_tfmode_cvt): Likewise.
6801 (sparc_preferred_simd_mode): Likewise.
6802 (output_cbranch): Likewise.
6803 (sparc_print_operand): Likewise.
6804 (sparc_expand_vec_perm_bmask): Likewise.
6805 (vector_init_bshuffle): Likewise.
6806 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6807 (spu_vector_mode_supported_p): Likewise.
6808 (spu_expand_insv): Likewise.
6809 (spu_emit_branch_or_set): Likewise.
6810 (spu_handle_vector_attribute): Likewise.
6811 (spu_builtin_splats): Likewise.
6812 (spu_builtin_extract): Likewise.
6813 (spu_builtin_promote): Likewise.
6814 (spu_expand_sign_extend): Likewise.
6815 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6816 (tilegx_simd_int): Likewise.
6817 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6818 (tilepro_simd_int): Likewise.
6819 * config/v850/v850.c (const_double_split): Likewise.
6820 (v850_print_operand): Likewise.
6821 (ep_memory_offset): Likewise.
6822 * config/vax/vax.c (vax_rtx_costs): Likewise.
6823 (vax_output_int_move): Likewise.
6824 (vax_output_int_add): Likewise.
6825 (vax_output_int_subtract): Likewise.
6826 * config/visium/predicates.md (visium_branch_operator): Likewise.
6827 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6828 (visium_print_operand_address): Likewise.
6829 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6830 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6831 (xtensa_expand_conditional_branch): Likewise.
6832 (xtensa_copy_incoming_a7): Likewise.
6833 (xtensa_output_literal): Likewise.
6834 * dfp.c (decimal_real_maxval): Likewise.
6835 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6838 Alan Hayward <alan.hayward@arm.com>
6839 David Sherwood <david.sherwood@arm.com>
6841 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6842 (mode_nunits_inline): Likewise.
6843 (mode_inner_inline): Likewise.
6844 (mode_unit_size_inline): Likewise.
6845 (mode_unit_precision_inline): Likewise.
6846 (emit_insn_modes_h): Likewise. Also emit a #define of the
6848 (emit_mode_wider): Add an E_ prefix to mode names.
6849 (emit_mode_complex): Likewise.
6850 (emit_mode_inner): Likewise.
6851 (emit_mode_adjustments): Likewise.
6852 (emit_mode_int_n): Likewise.
6853 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6854 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6855 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6856 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6857 (CRC32_BUILTIN, ENTRY): Likewise.
6858 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6859 (aarch64_pop_regs): Likewise.
6860 (aarch64_process_components): Likewise.
6861 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6862 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6863 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6864 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6865 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6866 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6868 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6869 (pcmpistr): Likewise.
6870 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6871 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6872 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6874 * config/rl78/rl78.c (mduc_regs): Likewise.
6875 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6876 (htm_expand_builtin): Likewise.
6877 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6878 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6879 * config/xtensa/xtensa.c (print_operand): Likewise.
6880 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6881 (NUM_MODE_VECTOR_INT): Likewise.
6882 * genoutput.c (null_operand): Likewise.
6883 (output_operand_data): Likewise.
6884 * genrecog.c (print_parameter_value): Likewise.
6885 * lra.c (debug_operand_data): Likewise.
6887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6889 * dwarf2out.c (scompare_loc_descriptor_wide)
6890 (scompare_loc_descriptor_narrow): New functions, split out from...
6891 (scompare_loc_descriptor): ...here.
6892 * expmed.c (emit_store_flag_int): New function, split out from...
6893 (emit_store_flag): ...here.
6895 2017-08-30 Richard Biener <rguenther@suse.de>
6897 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6898 (dwarf2out_early_finish): Move setting of AT_pubnames from
6899 early debug output to early finish.
6901 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6903 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6904 and -mdata-region to the assembler.
6906 2017-08-30 Richard Biener <rguenther@suse.de>
6908 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6910 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6911 (dwarf2out_early_global_decl): Only generate a DIE for the
6912 abstract origin if it doesn't already exist or is a declaration DIE.
6913 (resolve_addr): Do not add the linkage name twice when
6914 generating a stub DIE for the DW_TAG_GNU_call_site target.
6916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6918 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6919 Use machine_mode rather than int for arg1_mode.
6921 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6924 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6925 that the second argument of the built-in functions to unpack
6926 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6927 switch statement instead a lot of if statements.
6928 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6929 Allow 64-bit values to be in Altivec registers as well as
6930 traditional floating point registers.
6931 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6933 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6935 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6936 MAX_REGS_PER_ADDRESS == 1.
6938 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6940 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6941 * config/i386/i386.c (ix86_option_override_internal): Simplify
6942 setting of opts->x_flag_entry.
6944 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6945 Jakub Jelinek <jakub@redhat.com>
6946 Richard Biener <rguenther@suse.de>
6948 PR tree-optimization/81503
6949 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6950 folded constant fits in the target type; reorder tests for clarity.
6952 2017-08-29 Martin Liska <mliska@suse.cz>
6954 * passes.def: Include pass_lower_switch.
6955 * stmt.c (dump_case_nodes): Remove and move to
6956 tree-switch-conversion.
6957 (case_values_threshold): Likewise.
6958 (expand_switch_as_decision_tree_p): Likewise.
6959 (emit_case_decision_tree): Likewise.
6960 (expand_case): Likewise.
6961 (balance_case_nodes): Likewise.
6962 (node_has_low_bound): Likewise.
6963 (node_has_high_bound): Likewise.
6964 (node_is_bounded): Likewise.
6965 (emit_case_nodes): Likewise.
6966 (struct simple_case_node): New struct.
6967 (add_case_node): Remove.
6968 (emit_case_dispatch_table): Use vector instead of case_list.
6969 (reset_out_edges_aux): Remove.
6970 (compute_cases_per_edge): Likewise.
6971 (expand_case): Build list of simple_case_node.
6972 (expand_sjlj_dispatch_table): Use it.
6973 * tree-switch-conversion.c (struct case_node): Moved from
6974 stmt.c and adjusted.
6975 (emit_case_nodes): Likewise.
6976 (node_has_low_bound): Likewise.
6977 (node_has_high_bound): Likewise.
6978 (node_is_bounded): Likewise.
6979 (case_values_threshold): Likewise.
6980 (reset_out_edges_aux): Likewise.
6981 (compute_cases_per_edge): Likewise.
6982 (add_case_node): Likewise.
6983 (dump_case_nodes): Likewise.
6984 (balance_case_nodes): Likewise.
6985 (expand_switch_as_decision_tree_p): Likewise.
6986 (emit_jump): Likewise.
6987 (emit_case_decision_tree): Likewise.
6988 (try_switch_expansion): Likewise.
6989 (do_jump_if_equal): Likewise.
6990 (emit_cmp_and_jump_insns): Likewise.
6991 (fix_phi_operands_for_edge): New function.
6992 (record_phi_operand_mapping): Likewise.
6993 (class pass_lower_switch): New pass.
6994 (pass_lower_switch::execute): New function.
6995 (make_pass_lower_switch): Likewise.
6996 (conditional_probability):
6997 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6998 * tree-pass.h: Add make_pass_lower_switch.
7000 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7003 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7006 2017-08-29 Richard Biener <rguenther@suse.de>
7008 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7009 we do not add a DW_AT_inline attribute twice.
7010 (gen_subprogram_die): Remove code setting DW_AT_inline on
7011 DECL_ABSTRACT_P nodes.
7013 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7015 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7016 calls to internal functions.
7017 (gimplify_modify_expr): Likewise.
7018 * tree-call-cdce.c (use_internal_fn): Likewise.
7019 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7020 (convert_to_divmod): Set the nothrow flag.
7021 * tree-if-conv.c (predicate_mem_writes): Likewise.
7022 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7023 (vectorizable_call): Likewise.
7024 (vectorizable_store): Likewise.
7025 (vectorizable_load): Likewise.
7026 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7027 (vect_recog_mask_conversion_pattern): Likewise.
7029 2017-08-29 Martin Liska <mliska@suse.cz>
7032 * gcc.c (driver_handle_option): Add new argument.
7033 * opts-common.c (handle_option): Pass
7034 target_option_override_hook.
7035 * opts-global.c (lang_handle_option): Add new option.
7036 (set_default_handlers): Add new argument.
7037 (decode_options): Likewise.
7038 * opts.c (target_handle_option): Likewise.
7039 (common_handle_option): Call target_option_override_hook.
7040 * opts.h (struct cl_option_handler_func): Add hook for
7041 target option override.
7042 (struct cl_option_handlers): Likewise.
7043 (set_default_handlers): Add new argument.
7044 (decode_options): Likewise.
7045 (common_handle_option): Likewise.
7046 (target_handle_option): Likewise.
7047 * toplev.c (toplev::main): Pass targetm.target_option.override
7050 2017-08-29 Richard Biener <rguenther@suse.de>
7051 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7053 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7054 life to the active subtree.
7056 2017-08-28 Jeff Law <law@redhat.com>
7058 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7059 derive_equivalences.
7060 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7062 (edge_info::derive_equivalences): New private member function
7064 * tree-ssa-dom.c (class edge_info): Changed from a struct
7065 to a class. Add ctor/dtor, methods and data members.
7066 (edge_info::edge_info): Renamed from allocate_edge_info.
7067 Initialize additional members.
7068 (edge_info::~edge_info): New.
7069 (free_dom_edge_info): Delete the edge info.
7070 (record_edge_info): Use new class & associated member functions.
7071 Tighten forms for testing for edge equivalences.
7072 (record_temporary_equivalences): Iterate over the simple
7073 equivalences rather than assuming there's only one per edge.
7074 (cprop_into_successor_phis): Iterate over the simple
7075 equivalences rather than assuming there's only one per edge.
7076 (optimize_stmt): Use operand_equal_p rather than pointer
7077 equality for mini-DSE code.
7079 2017-08-28 Nathan Sidwell <nathan@acm.org>
7081 * gcc.c (execute): Fold SIGPIPE handling into switch
7082 statement. Adjust internal error message.
7084 2017-08-28 Richard Biener <rguenther@suse.de>
7087 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7088 Do nothing for removed DIEs.
7090 2017-08-28 Richard Biener <rguenther@suse.de>
7092 PR tree-optimization/81977
7093 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7096 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7099 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7100 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7101 using targetm.gen_mem_thread_fence.
7103 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7106 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7107 predicate to register_operand. Reorder operands.
7108 (*btr<mode>): Ditto.
7109 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7110 (*btr<mode>_mask): Ditto.
7112 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7114 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7115 * config/rs6000/xmmintrin.h: New file.
7116 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7118 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7121 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7122 parameter and_insn and return it.
7123 (recombine_lvx_pattern): Insert a copy to ensure availability of
7124 the base register of the copied masking operation at the point of
7125 the instruction replacement.
7126 (recombine_stvx_pattern): Likewise.
7128 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7130 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7131 undocumented switches.
7132 (-mpower9-dform-vector): Likewise.
7133 (-mpower9-dform): Likewise.
7134 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7135 comments to delete references to -mpower9-dform* switches.
7136 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7137 Delete reference to -mpower9-dform* switches, test for
7138 -mpower9-vector instead.
7139 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7140 (OTHER_P9_VECTOR_MASKS): Likewise.
7141 (POWERPC_MASKS): Likewise.
7142 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7143 tests against -mpower9-dform* to -mpower9-vector. Delete code
7144 that checked for -mpower9-dform* consistancy with other options.
7145 Add test for -mpower9-misc to enable other power9 switches.
7146 (rs6000_init_hard_regno_mode_ok): Likewise.
7147 (rs6000_option_override_internal): Likewise.
7148 (rs6000_emit_prologue): Likewise.
7149 (rs6000_emit_epilogue): Likewise.
7150 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7151 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7152 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7154 (emit_fusion_p9_store): Likewise.
7155 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7156 resetting these macros if the assembler does not support ISA 3.0
7158 (TARGET_P9_DFORM_VECTOR): Likewise.
7159 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7160 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7162 2017-08-25 Alan Modra <amodra@gmail.com>
7164 PR rtl-optimization/81747
7165 * cse.c (cse_extended_basic_block): Don't attempt to record
7166 equivalences for degenerate conditional branches that branch
7167 to their fall-through.
7169 2017-08-24 Martin Sebor <msebor@redhat.com>
7172 * gimple-fold.c (size_must_be_zero_p): New function.
7173 (gimple_fold_builtin_memory_op): Call it.
7175 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7177 * config/rs6000/mm_malloc.h: New file.
7179 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7181 PR tree-optimization/81913
7182 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7183 analysis when either IVs in condition can wrap.
7185 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7187 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7188 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7190 2017-08-24 Richard Biener <rguenther@suse.de>
7193 * targhooks.c (default_target_can_inline_p): Properly
7194 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7195 is present and always compare.
7196 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7197 infer -mfpmath=sse from TARGET_SSE_P.
7198 (ix86_can_inline_p): Properly use target_option_default_node when
7199 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7201 2017-08-24 Richard Biener <rguenther@suse.de>
7204 * dwarf2out.c (output_die): Handle flag_generate_offload like
7206 (output_comp_unit): Likewise.
7207 (gen_array_type_die): Likewise.
7208 (dwarf2out_early_finish): Likewise.
7209 (note_variable_value_in_expr): Likewise.
7210 (dwarf2out_finish): Likewise. Adjust assert.
7211 * cgraphunit.c (symbol_table::compile): Move setting of
7212 flag_generate_offload earlier ...
7213 (symbol_table::finalize_compilation_unit): ... here, before
7214 early debug finalization.
7216 2017-08-24 Richard Biener <rguenther@suse.de>
7218 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7219 and ipa-fnsummary.h.
7220 (ix86_can_inline_p): When ix86_fpmath flags do not match
7221 check whether the callee uses FP math at all.
7223 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7226 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7227 instead of TYPE_PRECISION.
7229 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7230 Alan Hayward <alan.hayward@arm.com>
7231 David Sherwood <david.sherwood@arm.com>
7233 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7235 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7237 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7239 (ix86_frame::stack_realign_allocate): New field.
7240 (struct machine_frame_state): Modify comments.
7241 (machine_frame_state::sp_realigned_fp_end): New field.
7242 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7244 (sp_valid_at): Add assertion to assure no attempt to access invalid
7245 offset of a realigned stack.
7246 (fp_valid_at): Likewise.
7247 (choose_baseaddr): Modify comments.
7248 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7249 ix86_expand_prologue.
7250 (ix86_expand_prologue): Modify stack realignment and allocation.
7251 (ix86_expand_epilogue): Modify comments.
7252 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7253 avx2_runtime, avx512f, and avx512f_runtime.
7255 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7257 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7258 (mstackrealign): Do not init to -1.
7259 * config/i386/i386.c (ix86_option_override_internal):
7260 Check opts_set, not opts when setting default value of
7261 opts->x_ix86_force_align_arg_pointer.
7263 2017-08-23 Richard Biener <rguenther@suse.de>
7265 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7266 lang_hooks.decl_printable_name.
7267 * print-rtl-function.c (print_rtx_function): Likewise.
7268 * tree-pretty-print.c (dump_function_header): Likewise.
7270 2017-08-23 Richard Biener <rguenther@suse.de>
7273 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7274 -g0 at compile-time.
7276 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7279 * doc/sourcebuild.texi (Other hardware attributes):
7282 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7285 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7286 Fix single-use check.
7288 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7290 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7292 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7294 * doc/install.texi: Modify to add more details on running selected
7297 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7299 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7300 is combined with -mabi=ms.
7301 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7304 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7306 PR tree-optimization/81488
7307 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7308 and cached_basis fields.
7309 (MAX_SPREAD): New constant.
7310 (alloc_cand_and_find_basis): Initialize new fields.
7311 (clear_visited): New function.
7312 (create_phi_basis_1): Rename from create_phi_basis, set visited
7313 and cached_basis fields.
7314 (create_phi_basis): New wrapper function.
7315 (phi_add_costs_1): Rename from phi_add_costs, add spread
7316 parameter, set visited field, short-circuit when limits reached.
7317 (phi_add_costs): New wrapper function.
7318 (record_phi_increments_1): Rename from record_phi_increments, set
7320 (record_phi_increments): New wrapper function.
7321 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7322 (phi_incr_cost): New wrapper function.
7323 (all_phi_incrs_profitable_1): Rename from
7324 all_phi_incrs_profitable, set visited field.
7325 (all_phi_incrs_profitable): New wrapper function.
7327 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7328 Alan Hayward <alan.hayward@arm.com>
7329 David Sherwood <david.sherwood@arm.com>
7331 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7332 that takes the outer and inner modes.
7333 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7334 comparison as the canonical test for a paradoxical subreg.
7335 * combine.c (simplify_set): Use paradoxical_subreg_p.
7336 (make_extraction): Likewise.
7337 (force_to_mode): Likewise.
7338 (rtx_equal_for_field_assignment_p): Likewise.
7339 (gen_lowpart_for_combine): Likewise.
7340 (simplify_comparison): Likewise.
7341 * cse.c (equiv_constant): Likewise.
7342 * expmed.c (store_bit_field_1): Likewise.
7343 * final.c (alter_subreg): Likewise.
7344 * fwprop.c (propagate_rtx): Likewise.
7345 (forward_propagate_subreg): Likewise.
7346 * ira-conflicts.c (ira_build_conflicts): Likewise.
7347 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7348 * lra-constraints.c (curr_insn_transform): Likewise.
7349 (split_reg): Likewise.
7350 * lra-eliminations.c (move_plus_up): Likewise.
7351 (lra_eliminate_regs_1): Likewise.
7352 * recog.c (general_operand): Likewise.
7353 * ree.c (combine_reaching_defs): Likewise.
7354 * reload.c (push_reload): Likewise.
7355 (find_reloads): Likewise.
7356 * reload1.c (elimination_effects): Likewise.
7357 (compute_reload_subreg_offset): Likewise.
7358 (choose_reload_regs): Likewise.
7359 * rtlanal.c (subreg_lsb_1): Likewise.
7360 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7361 (simplify_subreg): Likewise.
7362 * var-tracking.c (track_loc_p): Likewise.
7363 * emit-rtl.c (byte_lowpart_offset): Likewise.
7364 (paradoxical_subreg_p): Delete out-of-line definition.
7366 2017-08-22 Jeff Law <law@redhat.com>
7368 PR tree-optimization/81741
7369 PR tree-optimization/71947
7370 * tree-ssa-dom.c: Include tree-inline.h.
7371 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7372 equivalences if one is more expensive to compute than the other.
7373 * tree-ssa-scopedtables.h (class const_or_copies): Make
7374 record_const_or_copy_raw method private.
7375 (class avail_exprs_stack): New method simplify_binary_operation.
7376 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7377 avail_exprs_stack::simplify_binary_operation as needed.
7378 (avail_exprs_stack::simplify_binary_operation): New function.
7380 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7382 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7383 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7384 (DOT_SYMBOLS): Likewise.
7385 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7386 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7387 (RS6000_ABI_NAME): Likewise.
7388 (TARGET_CMODEL): Likewise.
7389 (TOC_SECTION_ASM_OP): Likewise.
7390 (SET_CMODEL): New macro.
7391 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7393 2017-08-22 Richard Biener <rguenther@suse.de>
7395 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7396 to allow for free-lang-data replacements similar to verify_type_variant.
7398 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7399 Alan Hayward <alan.hayward@arm.com>
7400 David Sherwood <david.sherwood@arm.com>
7402 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7403 VOIDmode for the LABEL_REF.
7405 2017-08-22 Richard Biener <rguenther@suse.de>
7407 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7409 2017-08-22 Richard Biener <rguenther@suse.de>
7411 * common.opt (feliminate-dwarf2-dups): Ignore.
7412 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7413 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7414 same_die_p_wrap, compute_section_prefix,
7415 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7416 (comdat_symbol_id, comdat_symbol_number): Likewise.
7417 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7419 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7420 (output_die): Mark unreachable path unreachable.
7421 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7422 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7423 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7424 (dwarf2out_early_finish): Likewise.
7426 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7428 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7430 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7433 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7434 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7435 (avr_attribute_table) <io, io_low, address>: Initialize
7436 .decl_required with true.
7438 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7440 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7441 undocumented debugging options.
7442 (-mvsx-scalar-double): Likewise.
7443 (-mallow-df-permute): Likewise.
7444 (-mvectorize-builtins): Likewise.
7445 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7446 (rs6000_builtin_vectorized_function): Likewise.
7447 (rs6000_builtin_md_vectorized_function): Likewise.
7448 (rs6000_opt_vars): Likewise.
7450 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7453 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7454 (*btrq_imm): Rename from *btrq.
7455 (*btcq_imm): Rename from *btcq.
7456 (btsc): New code attribute.
7457 (*<btsc><mode>): New insn pattern.
7458 (*btr<mode>): Ditto.
7459 (*<btsc><mode>_mask): New insn_and_split pattern.
7460 (*btr<mode>_mask): Ditto.
7462 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7463 Alan Hayward <alan.hayward@arm.com>
7464 David Sherwood <david.sherwood@arm.com>
7466 * function.c (pad_below): Simplify padding calculation.
7468 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7469 Alan Hayward <alan.hayward@arm.com>
7470 David Sherwood <david.sherwood@arm.com>
7472 * target.def (function_prologue): Remove frame size argument.
7473 (function_epilogue): Likewise.
7474 * doc/tm.texi: Regenerate.
7475 * final.c (final_start_function): Update call to function_prologue.
7476 (final_end_function): Update call to function_epilogue.
7477 (default_function_pro_epilogue): Remove frame size argument.
7478 * output.h (default_function_pro_epilogue): Likewise.
7479 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7480 (arm_output_function_prologue): Likewise.
7481 * config/frv/frv.c (frv_function_prologue): Likewise.
7482 (frv_function_epilogue): Likewise.
7483 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7484 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7485 (ia64_output_function_epilogue): Likewise.
7486 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7487 (m32r_output_function_epilogue): Likewise.
7488 * config/microblaze/microblaze.c (microblaze_function_prologue)
7489 (microblaze_function_epilogue): Likewise.
7490 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7491 (mips_output_function_epilogue): Likewise.
7492 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7493 (mmix_target_asm_function_epilogue): Likewise.
7494 * config/msp430/msp430.c (msp430_start_function): Likewise.
7495 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7496 (nds32_asm_function_epilogue): Likewise.
7497 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7498 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7499 (pa_output_function_epilogue): Likewise.
7500 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7501 (rs6000_output_function_epilogue): Likewise.
7502 * config/rl78/rl78.c (rl78_start_function): Likewise.
7503 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7504 (rs6000_output_function_epilogue): Likewise.
7505 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7506 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7507 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7508 (sparc_asm_function_epilogue): Likewise.
7510 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7512 * tree.h (type_has_mode_precision_p): New function.
7513 * convert.c (convert_to_integer_1): Use it.
7514 * expr.c (expand_expr_real_2): Likewise.
7515 (expand_expr_real_1): Likewise.
7516 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7517 * match.pd: Likewise.
7518 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7519 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7520 * tree-tailcall.c (process_assignment): Likewise.
7521 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7522 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7523 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7524 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7525 (vectorizable_assignment): Likewise.
7526 (vectorizable_shift): Likewise.
7527 (vectorizable_operation): Likewise.
7528 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7530 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7532 * match.pd: Add pow (C, x) simplification.
7534 2017-08-21 Richard Biener <rguenther@suse.de>
7536 PR tree-optimization/81900
7537 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7538 for blocks with abnormal predecessors.
7539 (compute_antic): Do not set visited flag prematurely.
7541 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7544 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7546 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7548 * stor-layout.h (vector_type_mode): Move to...
7549 * tree.h (vector_type_mode): ...here.
7550 * stor-layout.c (vector_type_mode): Move to...
7551 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7553 2017-08-21 Richard Biener <rguenther@suse.de>
7555 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7556 register_external_die hooks.
7557 (debug_false_tree_charstarstar_uhwistar): Declare.
7558 (debug_nothing_tree_charstar_uhwi): Likewise.
7559 * debug.c (do_nothing_debug_hooks): Adjust.
7560 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7561 (debug_nothing_tree_charstar_uhwi): Likewise.
7562 * dbxout.c (dbx_debug_hooks): Adjust.
7563 (xcoff_debug_hooks): Likewise.
7564 * sdbout.c (sdb_debug_hooks): Likewise.
7565 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7566 * dwarf2out.c (macinfo_label_base): New global.
7567 (dwarf2out_register_external_die): New function for the
7568 register_external_die hook.
7569 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7570 (dwarf2_debug_hooks): Use them.
7571 (dwarf2_lineno_debug_hooks): Adjust.
7572 (struct die_struct): Add with_offset flag.
7573 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7574 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7575 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7576 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7577 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7578 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7579 defining section names for the early LTO debug variants.
7580 (reset_indirect_string): New helper.
7581 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7582 (print_dw_val): Add support for offsetted symbol references.
7583 (get_ultimate_context): Split out from is_cxx.
7584 (is_cxx): Use get_ultimate_context.
7585 (is_fortran): Add decl overload.
7586 (compute_comp_unit_symbol): Split out worker from
7587 compute_section_prefix.
7588 (compute_section_prefix): Call compute_comp_unit_symbol and
7589 set comdat_type_p here.
7590 (output_die): Skip DIE symbol output for the LTO added one.
7591 Handle DIE symbol references with offset.
7592 (output_comp_unit): Guard section name mangling properly.
7593 For LTO debug sections emit a symbol at the section beginning
7594 which we use to refer to its DIEs.
7595 (add_abstract_origin_attribute): For DIEs registered via
7596 dwarf2out_register_external_die directly refer to the early
7597 DIE rather than indirectly through the shadow one we created.
7598 Remove obsolete call to dwarf2out_abstract_function for
7599 non-function/block origins.
7600 (gen_array_type_die): When generating early LTO debug do
7601 not emit DW_AT_string_length.
7602 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7603 late when in LTO. As suggested place a gcc_unreachable for
7604 the DECL_ABSTRACT_P case.
7605 (gen_subprogram_die): Avoid another specification DIE
7606 for early built declarations/definitions for the late LTO case.
7607 (gen_variable_die): Add type references for late duplicated VLA dies
7609 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7610 we have the abstract instance already.
7611 (process_scope_var): Adjust decl DIE contexts in LTO which
7612 first puts them in limbo.
7613 (gen_decl_die): Do not generate type DIEs late apart from
7614 types for VLAs or for decls we do not yet have a DIE. Do not
7615 call dwarf2out_abstract_function late.
7616 (dwarf2out_early_global_decl): Make sure to create DIEs
7617 for abstract instances of a decl first.
7618 (dwarf2out_late_global_decl): Adjust comment.
7619 (output_macinfo_op): With multiple macro sections use
7620 macinfo_label_base to distinguish labels.
7621 (output_macinfo): Likewise. Update macinfo_label_base.
7622 Pass in the line info label.
7623 (note_variable_value_in_expr): When generating LTO resolve
7624 all variable values here by generating DIEs as needed.
7625 (init_sections_and_labels): Add early LTO debug flag parameter
7626 and generate different sections and names if set. Add generation
7627 counter for the labels so we can have multiple of them.
7628 (reset_dies): Helper to allow DIEs to be output multiple times.
7629 (dwarf2out_finish): When outputting DIEs to the fat part of an
7630 LTO object first reset DIEs.
7631 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7632 (modified_type_die): Check for decl_ultimate_origin being self
7634 (gen_type_die_with_usage): Likewise.
7635 (gen_typedef_die): Allow decl_ultimate_origin being self.
7636 (set_decl_abstract_flags): Remove.
7637 (set_block_abstract_flags): Likewise.
7638 (dwarf2out_abstract_function): Treat the early generated DIEs
7639 as the abstract copy and only add DW_AT_inline and
7640 DW_AT_artificial here and call set_decl_origin_self.
7641 If the DIE has an abstract origin don't do anything.
7642 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7643 if we have none yet (Go fails to build one, PR78628).
7644 (variably_modified_type_p): Prevent endless recursion for Ada
7645 cyclic pointer types.
7646 * lto-streamer-in.c: Include debug.h.
7647 (dref_queue): New global.
7648 (lto_read_tree_1): Stream in DIE references.
7649 (lto_input_tree): Register DIE references.
7650 (input_function): Stream DECL_DEBUG_ARGS.
7651 * lto-streamer-out.c: Include debug.h.
7652 (lto_write_tree_1): Output DIE references.
7653 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7654 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7655 (output_function): Stream DECL_DEBUG_ARGS.
7656 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7657 Stream DECL_ABSTRACT_ORIGIN.
7658 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7659 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7660 DECL_CONTEXT for file-scope decls.
7661 * lto-streamer.h (struct dref_entry): Declare.
7662 (dref_queue): Likewise.
7663 * cfgexpand.c (pass_expand::execute): Do not call the
7664 outlining_inline_function hook here.
7665 * lto-wrapper.c (debug_obj): New global.
7666 (tool_cleanup): Unlink it if required.
7667 (debug_objcopy): New function.
7668 (run_gcc): Handle early debug sections in the IL files by
7669 extracting them to separate files, partially linkin them and
7670 feeding the result back as result to the linker.
7671 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7672 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7673 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7674 sections into a separate segment.
7675 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7677 (darwin_asm_dwarf_section): Likewise.
7678 (darwin_asm_output_dwarf_offset): Likewise.
7679 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7681 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7682 Alan Hayward <alan.hayward@arm.com>
7683 David Sherwood <david.sherwood@arm.com>
7685 * read-md.h (md_reader::record_potential_iterator_use): Replace
7686 pointer argument with an rtx and an index.
7687 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7688 (apply_mode_iterator): Likewise.
7689 (apply_code_iterator): Likewise.
7690 (apply_int_iterator): Likewise.
7691 (apply_subst_iterator): Likewise.
7692 (record_iterator_use): Likewise.
7693 (record_attribute_use): Likewise.
7694 (md_reader::record_potential_iterator_use): Likewise. Update calls
7695 to record_iterator_use and apply_iterator.
7696 (iterator_use): Replace ptr with x and index.
7697 (attribute_use): Likewise.
7698 (apply_attribute_uses): Update calls to apply_iterator.
7699 (apply_iterators): Likewise. Update initialization of iterator_use.
7700 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7701 and record_potential_iterator_use.
7702 (rtx_reader::read_rtx_operand): Likewise.
7704 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7705 Alan Hayward <alan.hayward@arm.com>
7706 David Sherwood <david.sherwood@arm.com>
7708 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7711 2017-08-21 Richard Biener <rguenther@suse.de>
7714 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7715 at struct end conservatively when comparing common bases.
7717 2017-08-21 Richard Biener <rguenther@suse.de>
7719 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7720 (mem_ref_in_stmt): Remove.
7721 (determine_max_movement): Use ref index to get at the reference.
7722 (invariantness_dom_walker::before_dom_children): Deal with
7723 lim data already initialized.
7724 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7726 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7728 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7729 (TARGET_ISA_ROUND): Ditto.
7730 (TARGET_ROUND): Ditto.
7731 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7732 * config/i386/i386.md: Ditto.
7733 * config/i386/sse.md: Ditto.
7734 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7735 with OPTION_MASK_ISA_SSE4_1.
7737 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7740 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7741 __builtin_ia32_lzcnt_u16.
7743 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7746 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7747 (rs6000_set_current_function): Rewrite function to use it.
7749 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7752 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7753 and TYPE_WARN_IF_NOT_ALIGN.
7754 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7755 (handle_warn_if_not_align): New.
7756 (place_union_field): Call handle_warn_if_not_align.
7757 (place_field): Call handle_warn_if_not_align.
7758 Copy TYPE_WARN_IF_NOT_ALIGN.
7759 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7760 (layout_type): Likewise.
7761 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7763 (tree_decl_common): Add warn_if_not_align.
7764 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7765 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7766 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7767 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7768 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7769 * doc/extend.texi: Document warn_if_not_aligned attribute.
7770 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7772 2017-08-17 Martin Liska <mliska@suse.cz>
7775 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7776 (get_data_dependence): Use it as pointer type.
7777 (distribute_loop): Likewise.
7779 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7781 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7782 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7783 (p8_vmrgow_<mode>_direct): New define_insn.
7784 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7785 handle endianness for vmrgew and vmrgow permute patterns.
7787 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7789 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7790 * config/rs6000/rs6000-cpus.def: Remove comment.
7791 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7792 (POWERPC_MASKS): Likewise.
7793 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7794 use of TARGET_VSX_TIMODE.
7795 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7796 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7797 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7798 (rs6000_option_override_internal): Remove dead code.
7799 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7800 (rs6000_legitimize_reload_address): Likewise.
7801 (rs6000_legitimate_address_p): Likewise.
7802 (rs6000_opt_masks): Delete "vsx-timode".
7803 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7804 from function comment.
7805 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7806 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7807 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7809 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7810 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7811 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7814 (define_peephole2): Likewise.
7816 2017-08-17 Martin Sebor <msebor@redhat.com>
7819 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7820 past the end of an array.
7821 (test_pp_format): Add test cases.
7823 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7825 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7826 missing ECF_NOTHROW flags.
7828 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7831 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7832 operands residing in integer registers.
7833 (*vsx_le_perm_load_<mode>): Likewise.
7834 (*vsx_le_perm_store_<mode>): Likewise.
7835 (define_peephole2): Add peepholes to optimize the above.
7837 2017-08-17 Marek Polacek <polacek@redhat.com>
7840 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7841 to mimic what shorten_compare did. Change the return type to bool.
7842 (fold_cond_expr_with_comparison): Update call to
7843 operand_equal_for_comparison_p.
7844 (fold_ternary_loc): Likewise.
7846 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7848 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7850 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7852 2017-08-17 Richard Biener <rguenther@suse.de>
7854 * tree-ssa-structalias.c (solve_graph): When propagating
7855 to successors update the graphs succ edges and avoid duplicate work.
7857 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7860 * config/i386/i386.c (ix86_option_override_internal): Save target
7861 specific options after ix86_stack_protector_guard_reg was changed.
7863 2017-08-17 Richard Biener <rguenther@suse.de>
7865 PR tree-optimization/81827
7866 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7867 (new_var_info): Initialize it conservatively.
7868 (get_call_vi): Mark register vars.
7869 (new_scalar_tmp_constraint_exp): Likewise.
7870 (handle_rhs_call): Likewise.
7871 (handle_const_call): Likewise.
7872 (create_function_info_for): Likewise.
7873 (solve_constraints): Sort varinfos to separate register from
7874 non-register vars to pack points-to solution bitmaps during
7877 2017-08-17 Marek Polacek <polacek@redhat.com>
7879 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7881 2017-08-17 Richard Biener <rguenther@suse.de>
7883 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7884 to true when overflow is undefined and we saturated the result.
7886 2017-08-17 Alan Modra <amodra@gmail.com>
7889 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7890 Don't use store multiple if only one reg needs saving.
7891 (interesting_frame_related_regno): New function.
7892 (rs6000_frame_related): Don't emit frame info for regs that
7894 (rs6000_emit_epilogue): Likewise.
7896 2017-08-16 Nathan Sidwell <nathan@acm.org>
7898 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7899 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7900 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7901 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7902 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7903 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7904 (verify_type): Adjust for TYPE_BINFO move.
7905 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7906 process TYPE_BINFO directly.
7907 (hash_tree): Likewise.
7908 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7910 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7913 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7915 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7917 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7920 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7921 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7922 Add ix86_binary_operator_ok to insn constraint.
7923 (*iordi_1_bts): Ditto.
7924 (*xordi_1_btc): Ditto.
7925 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7926 Update corresponding peephole2 pattern.
7930 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7932 PR tree-optimization/81832
7933 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7934 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7936 2017-08-16 Marek Polacek <polacek@redhat.com>
7939 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7940 regarding size_zero_node.
7942 2017-08-16 Martin Liska <mliska@suse.cz>
7945 * config.gcc: Respect previously set extra_objs in case
7948 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7950 PR tree-optimization/81835
7951 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7952 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7953 not depend on the phi.
7955 2017-08-16 Alan Modra <amodra@gmail.com>
7957 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7960 2017-08-16 Alan Modra <amodra@gmail.com>
7962 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7963 (save_reg_p): ..into this. Update all callers.
7964 (first_reg_to_save): Simplify.
7966 2017-08-16 Alan Modra <amodra@gmail.com>
7968 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7971 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7975 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7976 constant count if that count is less than 32.
7978 2017-08-15 Nathan Sidwell <nathan@acm.org>
7980 * gcc.c (execute): Emit friendlier message if inferior is killed
7981 by an external cause.
7983 2017-08-15 Richard Biener <rguenther@suse.de>
7985 PR tree-optimization/81790
7986 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7987 CONSTRUCTORs from simplifying and VN.
7989 2017-08-14 Martin Sebor <msebor@redhat.com>
7991 * builtin-attrs.def: Add comments.
7993 2017-08-14 Martin Sebor <msebor@redhat.com>
7996 * doc/extend.texi (attribute nonstring): Document new attribute.
7998 2017-08-14 Martin Sebor <msebor@redhat.com>
8001 * tree-diagnostic.c (default_tree_printer): Handle %G.
8002 * gimple-pretty-print.h (percent_G_format): Declare new function.
8003 * gimple-pretty-print.c (percent_G_format): Define.
8004 * tree-pretty-print.c (percent_K_format): Add argument.
8006 2017-08-14 Martin Sebor <msebor@redhat.com>
8008 PR translation/79998
8009 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8010 Remove a stray space.
8012 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8015 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8016 (*iordi_1_bts): Ditto.
8017 (*xordi_1_btc): Ditto.
8019 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8022 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8023 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8025 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8026 quoted strings, and make more translator-friendly.
8027 (darwin_rs6000_override_options): Likewise.
8028 (rs6000_option_override_internal): Likewise.
8029 (rs6000_return_in_memory): Fix overlong line.
8030 (init_cmulative_args): Use quoted strings, and make more
8031 translator-friendly.
8032 (rs6000_pass_by_reference): Fix overlong line.
8033 (def_builtin): Use quoted strings.
8034 (altivec_expand_predicate_builtin): Use quoted strings, and make
8035 more translator-friendly.
8036 (htm_expand_builtin): Use quoted strings.
8037 (cpu_expand_builtin): Use quoted strings, and make more
8038 translator-friendly.
8039 (altivec_expand_builtin): Likewise.
8040 (paired_expand_predicate_builtin): Likewise.
8041 (rs6000_invalid_builtin): Likewise.
8042 (builtin_function_type): Use quoted strings.
8043 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8044 more translator-friendly.
8045 (rs6000_trampoline_init): Likewise.
8046 (rs6000_handle_altivec_attribute): Likewise.
8047 (rs6000_inner_target_options): Use quoted strings.
8048 (rs6000_disable_incompatible_switches): Likewise.
8049 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8050 strings, and make more translator-friendly.
8051 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8053 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8055 PR tree-optimization/81799
8056 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8057 cond_expr to simple gimple operand.
8059 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8062 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8064 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8068 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8069 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8070 TARGET_GASISR_PROLOGUES.
8071 * config/avr/avr.c (avr_option_override): Same.
8072 (avr_pass_pre_proep::execute): Same.
8074 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8077 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8078 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8080 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8082 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8083 define_insn_and_split. Split to a memory load from 0 in
8084 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8085 using PTR mode iterator.
8086 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8087 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8088 (*add_tp_<mode>): Redefine as define_insn_and_split.
8089 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8090 address space. Merge with *add_tp_x32 using PTR mode iterator.
8091 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8092 Split to an add with a memory load from 0 in
8093 DEFAULT_TLS_SEG_REG address space.
8095 2017-08-12 Andrew Pinski <apinski@cavium.com>
8097 * config/aarch64/aarch64-option-extensions.def (rdma):
8098 Fix feature string to what Linux prints out in /proc/cpuinfo.
8100 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8103 * dwarf2out.c (modified_type_die): For C typedef types that have
8104 an ultimate origin, process the ultimate origin instead of the
8106 (gen_typedef_die): Assert that input DECLs have no ultimate
8108 (gen_type_die_with_usage): For typedef variants that have an
8109 ultimate origin, just call gen_decl_die on the original DECL.
8110 (process_scope_var): Avoid creating DIEs for local typedefs and
8111 concrete static variables.
8113 2017-08-12 Alan Modra <amodra@gmail.com>
8117 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8118 match gnu-user.h startfile.
8119 (ENDFILE_LINUX_SPEC): Similarly.
8121 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8124 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8126 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8128 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8129 * config/aarch64/aarch64.md (mov<mode>): Change.
8130 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8131 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8132 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8134 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8136 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8137 for storage order barriers.
8139 2017-08-11 Martin Liska <mliska@suse.cz>
8142 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8143 variables of void type.
8145 2017-08-11 Martin Liska <mliska@suse.cz>
8147 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8148 TARGET_SUPPORTS_ALIASES.
8149 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8150 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8151 (optimize_weakref): Likewise.
8152 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8153 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8154 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8156 2017-08-11 Martin Liska <mliska@suse.cz>
8159 * config/i386/i386.c (make_resolver_func): Do complete
8160 refactoring of the function.
8162 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8165 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8166 * config/i386/i386.c (ix86_stack_protect_guard): Use
8167 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8168 * doc/invoke.texi (x86 Options): Document
8169 -mstack-protector-guard-symbol= option.
8171 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8173 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8174 * config/i386/i386.c (ix86_split_stack_guard): New function.
8175 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8176 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8177 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8178 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8179 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8180 (split_stack_space_check): Call ix86_split_stack_guard.
8182 2017-08-10 Martin Sebor <msebor@redhat.com>
8184 * print-tree.c (print_node): Print location using the established
8186 Replace spaces with colons.
8187 (debug_raw, debug): Ditto.
8189 2017-08-10 Martin Sebor <msebor@redhat.com>
8192 * pretty-print.c (pp_format): Correct the handling of %s precision.
8194 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8197 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8199 (ix86_finalize_stack_frame_flags): This. Also clear
8200 frame_pointer_needed if -fno-omit-frame-pointer is used without
8202 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8203 with ix86_finalize_stack_frame_flags.
8204 (ix86_expand_epilogue): Likewise.
8205 (ix86_expand_split_stack_prologue): Likewise.
8206 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8208 2017-08-10 Martin Liska <mliska@suse.cz>
8211 * c-attribs.c (handle_target_attribute):
8212 Report warning for an empty string argument of target attribute.
8214 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8217 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8219 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8220 or DECL_NONLOCAL labels.
8221 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8222 or DECL_NONLOCAL labels here.
8224 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8226 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8227 to indicate when early gimple folding has been disabled.
8228 (rs6000_gimple_fold_builtin): Add debug content.
8229 (rs6000_invalid_builtin): Fix whitespace.
8230 (rs6000_expand_builtin): Fix whitespace.
8231 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8233 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8236 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8237 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8239 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8241 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8242 (qdf24xx): Likewise.
8243 * config/aarch64/aarch64.md: Include falkor.md.
8244 * config/aarch64/falkor.md: New.
8246 2017-08-09 Marek Polacek <polacek@redhat.com>
8249 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8250 * diagnostic.c (emit_diagnostic): Add a comment.
8251 (emit_diagnostic_valist): New function.
8253 2017-08-09 Marek Polacek <polacek@redhat.com>
8256 * input.c (make_location): New overload.
8257 * input.h (make_location): Declare.
8259 2017-08-08 Alan Modra <amodra@gmail.com>
8260 H.J. Lu <hongjiu.lu@intel.com>
8263 * gcc.c (NO_PIE_SPEC): Delete.
8264 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8266 (LINK_PIE_SPEC): ..to here.
8267 (LINK_COMMAND_SPEC): Support -no-pie.
8268 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8269 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8271 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8272 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8273 (ENDFILE_CRTEND_SPEC): Similarly.
8275 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8278 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8279 (mstack-protector-guard-offset=): Ditto.
8280 * config/i386/i386.c (ix86_option_override): Handle
8281 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8283 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8284 ix86_stack_protect_guard_offset variables.
8285 (TARGET_STACK_PROTECT_GUARD): Always define.
8286 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8287 and -mstack-protector-guard-offset= options.
8289 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8291 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8292 boundary case for the last candidate.
8294 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8296 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8297 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8299 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8302 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8303 * config/aarch64/aarch64-builtins.c
8304 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8305 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8306 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8308 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8309 Andrew Pinski <pinskia@gmail.com>
8312 * internal-fn.def (XORSIGN): New.
8313 * optabs.def (xorsign_optab): New.
8314 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8315 (convert_expand_mult_copysign): New.
8316 (pass_optimize_widening_mul::execute): Call
8317 convert_expand_mult_copysign.
8319 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8321 PR tree-optimization/81354
8322 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8323 Insert on edges rather than explicitly creating landing pads.
8324 (analyze_candidates_and_replace): Commit edge inserts.
8326 2017-08-08 Richard Biener <rguenther@suse.de>
8329 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8330 (expand_simple_operations): Also look through ADDR_EXPRs with
8331 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8333 2017-08-08 Richard Biener <rguenther@suse.de>
8335 PR tree-optimization/81723
8336 * tree-vect-slp.c (struct bst_traits): New hash traits.
8337 (bst_fail): New global.
8338 (vect_build_slp_tree_2): New worker, split out from ...
8339 (vect_build_slp_tree): ... this now wrapping it with using
8340 bst_fail set to cache SLP tree build fails. Properly handle
8342 (vect_analyze_slp_instance): Allocate and free bst_fail.
8344 2017-08-08 Martin Liska <mliska@suse.cz>
8347 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8348 LABEL_DECLs that can be from a different function.
8350 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8352 PR tree-optimization/81744
8353 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8354 loop's number of iterations.
8356 2017-08-08 Martin Liska <mliska@suse.cz>
8358 * asan.c: Include header files.
8359 * attribs.c (build_decl_attribute_variant): New function moved
8361 (build_type_attribute_qual_variant): Likewise.
8362 (cmp_attrib_identifiers): Likewise.
8363 (simple_cst_list_equal): Likewise.
8364 (omp_declare_simd_clauses_equal): Likewise.
8365 (attribute_value_equal): Likewise.
8366 (comp_type_attributes): Likewise.
8367 (build_type_attribute_variant): Likewise.
8368 (lookup_ident_attribute): Likewise.
8369 (remove_attribute): Likewise.
8370 (merge_attributes): Likewise.
8371 (merge_type_attributes): Likewise.
8372 (merge_decl_attributes): Likewise.
8373 (merge_dllimport_decl_attributes): Likewise.
8374 (handle_dll_attribute): Likewise.
8375 (attribute_list_equal): Likewise.
8376 (attribute_list_contained): Likewise.
8377 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8378 (lookup_attribute_by_prefix): Likewise.
8379 * bb-reorder.c: Include header files.
8380 * builtins.c: Likewise.
8381 * calls.c: Likewise.
8382 * cfgexpand.c: Likewise.
8383 * cgraph.c: Likewise.
8384 * cgraphunit.c: Likewise.
8385 * convert.c: Likewise.
8386 * dwarf2out.c: Likewise.
8387 * final.c: Likewise.
8388 * fold-const.c: Likewise.
8389 * function.c: Likewise.
8390 * gimple-expr.c: Likewise.
8391 * gimple-fold.c: Likewise.
8392 * gimple-pretty-print.c: Likewise.
8393 * gimple.c: Likewise.
8394 * gimplify.c: Likewise.
8395 * hsa-common.c: Likewise.
8396 * hsa-gen.c: Likewise.
8397 * internal-fn.c: Likewise.
8398 * ipa-chkp.c: Likewise.
8399 * ipa-cp.c: Likewise.
8400 * ipa-devirt.c: Likewise.
8401 * ipa-fnsummary.c: Likewise.
8402 * ipa-inline.c: Likewise.
8403 * ipa-visibility.c: Likewise.
8405 * lto-cgraph.c: Likewise.
8406 * omp-expand.c: Likewise.
8407 * omp-general.c: Likewise.
8408 * omp-low.c: Likewise.
8409 * omp-offload.c: Likewise.
8410 * omp-simd-clone.c: Likewise.
8411 * opts-global.c: Likewise.
8412 * passes.c: Likewise.
8413 * predict.c: Likewise.
8414 * sancov.c: Likewise.
8415 * sanopt.c: Likewise.
8416 * symtab.c: Likewise.
8417 * toplev.c: Likewise.
8418 * trans-mem.c: Likewise.
8419 * tree-chkp.c: Likewise.
8420 * tree-eh.c: Likewise.
8421 * tree-into-ssa.c: Likewise.
8422 * tree-object-size.c: Likewise.
8423 * tree-parloops.c: Likewise.
8424 * tree-profile.c: Likewise.
8425 * tree-ssa-ccp.c: Likewise.
8426 * tree-ssa-live.c: Likewise.
8427 * tree-ssa-loop.c: Likewise.
8428 * tree-ssa-sccvn.c: Likewise.
8429 * tree-ssa-structalias.c: Likewise.
8430 * tree-ssa.c: Likewise.
8431 * tree-streamer-in.c: Likewise.
8432 * tree-vectorizer.c: Likewise.
8433 * tree-vrp.c: Likewise.
8435 * ubsan.c: Likewise.
8436 * varasm.c: Likewise.
8437 * varpool.c: Likewise.
8438 * tree.c: Remove functions moved to attribs.[ch].
8440 * config/aarch64/aarch64.c: Add attrs.h header file.
8441 * config/alpha/alpha.c: Likewise.
8442 * config/arc/arc.c: Likewise.
8443 * config/arm/arm.c: Likewise.
8444 * config/avr/avr.c: Likewise.
8445 * config/bfin/bfin.c: Likewise.
8446 * config/c6x/c6x.c: Likewise.
8447 * config/cr16/cr16.c: Likewise.
8448 * config/cris/cris.c: Likewise.
8449 * config/darwin.c: Likewise.
8450 * config/epiphany/epiphany.c: Likewise.
8451 * config/fr30/fr30.c: Likewise.
8452 * config/frv/frv.c: Likewise.
8453 * config/ft32/ft32.c: Likewise.
8454 * config/h8300/h8300.c: Likewise.
8455 * config/i386/winnt.c: Likewise.
8456 * config/ia64/ia64.c: Likewise.
8457 * config/iq2000/iq2000.c: Likewise.
8458 * config/lm32/lm32.c: Likewise.
8459 * config/m32c/m32c.c: Likewise.
8460 * config/m32r/m32r.c: Likewise.
8461 * config/m68k/m68k.c: Likewise.
8462 * config/mcore/mcore.c: Likewise.
8463 * config/microblaze/microblaze.c: Likewise.
8464 * config/mips/mips.c: Likewise.
8465 * config/mmix/mmix.c: Likewise.
8466 * config/mn10300/mn10300.c: Likewise.
8467 * config/moxie/moxie.c: Likewise.
8468 * config/msp430/msp430.c: Likewise.
8469 * config/nds32/nds32-isr.c: Likewise.
8470 * config/nds32/nds32.c: Likewise.
8471 * config/nios2/nios2.c: Likewise.
8472 * config/nvptx/nvptx.c: Likewise.
8473 * config/pa/pa.c: Likewise.
8474 * config/pdp11/pdp11.c: Likewise.
8475 * config/powerpcspe/powerpcspe.c: Likewise.
8476 * config/riscv/riscv.c: Likewise.
8477 * config/rl78/rl78.c: Likewise.
8478 * config/rx/rx.c: Likewise.
8479 * config/s390/s390.c: Likewise.
8480 * config/sh/sh.c: Likewise.
8481 * config/sol2.c: Likewise.
8482 * config/sparc/sparc.c: Likewise.
8483 * config/spu/spu.c: Likewise.
8484 * config/stormy16/stormy16.c: Likewise.
8485 * config/tilegx/tilegx.c: Likewise.
8486 * config/tilepro/tilepro.c: Likewise.
8487 * config/v850/v850.c: Likewise.
8488 * config/vax/vax.c: Likewise.
8489 * config/visium/visium.c: Likewise.
8490 * config/xtensa/xtensa.c: Likewise.
8492 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8495 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8496 constraints since the -mupper-regs-* switches have been
8498 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8499 into a vector from a double word element that was extracted from
8500 another vector, and eliminate extra XXPERMDI instructions.
8501 (vsx_concat_<mode>_2): Likewise.
8502 (vsx_concat_<mode>_3): Likewise.
8503 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8504 concat to allow optimizing inserts from previous extracts.
8506 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8508 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8509 memory reference to a SSP offset in TLS address space.
8510 (ix86_print_operand) <case '@'>: Remove.
8511 (ix86_print_operand_punct_valid_p): Remove '@' code.
8512 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8514 (stack_tls_protect_set_<mode>): Remove.
8515 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8516 (stack_tls_protect_test_<mode>): Remove.
8517 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8519 2017-08-07 Olivier Hainque <hainque@adacore.com>
8522 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8524 2017-08-07 Douglas Rupp <rupp@adacore.com>
8526 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8527 variable was referenced as multidir in command.
8529 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8532 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8535 2017-08-07 Martin Liska <mliska@suse.cz>
8537 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8538 * config/rl78/rl78.c: Add include of attribs.h.
8539 * config/sh/sh.c: Likewise.
8540 * config/v850/v850.c: Likewise.
8542 2017-08-07 Tom de Vries <tom@codesourcery.com>
8545 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8547 2017-08-07 Martin Liska <mliska@suse.cz>
8549 * config/mips/mips.c: Include attribs.h.
8551 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8554 * doc/invoke.texi: Document change in behvaior for -Ofast for
8557 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8559 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8561 (aarch64_pop_regs): Likewise.
8562 (aarch64_gen_load_pair): Likewise.
8563 (aarch64_save_callee_saves): Likewise.
8564 (aarch64_restore_callee_saves): Likewise.
8566 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8568 * config/i386/i386.c: Revert the last change.
8570 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8573 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8575 (ix86_finalize_stack_frame_flags): This. Also clear
8576 frame_pointer_needed if -fno-omit-frame-pointer is used without
8578 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8579 with ix86_finalize_stack_frame_flags.
8580 (ix86_expand_epilogue): Likewise.
8581 (ix86_expand_split_stack_prologue): Likewise.
8583 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8586 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8587 to P_AES for Westmere.
8589 2017-08-07 Jonathan Yong <10walls@gmail.com>
8591 * config/i386/mingw.opt (fset-stack-executable): Removed.
8592 * config/i386/cygming.opt (fset-stack-executable): Moved
8594 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8596 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8598 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8600 2017-08-07 Marek Polacek <polacek@redhat.com>
8603 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8605 2017-08-07 Martin Liska <mliska@suse.cz>
8607 * attribs.h (canonicalize_attr_name): New function.
8608 (cmp_attribs): Move from c-format.c and adjusted.
8609 (is_attribute_p): Moved from tree.h.
8610 * tree-inline.c: Add new includes.
8611 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8612 (private_is_attribute_p): Remove.
8613 (private_lookup_attribute): Likewise.
8614 (private_lookup_attribute_by_prefix): Simplify.
8615 (remove_attribute): Use is_attribute_p.
8616 * tree.h: Remove removed declarations.
8618 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8621 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8622 instead of computing it in the function. Formatting fix.
8623 (expand_case): Don't rely on default_edge being the first edge,
8624 clear it if removing it, pass default_edge to
8625 emit_case_dispatch_table.
8626 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8629 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8631 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8632 insn in the function, emit NOP after the insn.
8634 2017-08-06 Tom de Vries <tom@codesourcery.com>
8636 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8639 2017-08-06 Tom de Vries <tom@codesourcery.com>
8641 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8644 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8646 PR tree-optimization/57371
8647 * match.pd: New pattern.
8649 2017-08-04 Marek Polacek <polacek@redhat.com>
8652 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8653 perform the computation in offset_int.
8655 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8657 PR tree-optimization/81136
8658 * tree-vectorizer.h: Include tree-hash-traits.h.
8659 (vec_base_alignments): New typedef.
8660 (vec_info): Add a base_alignments field.
8661 (vect_record_base_alignments): Declare.
8662 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8664 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8665 (create_data_ref): Add an is_conditional_in_stmt argument.
8666 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8667 the is_conditional_in_stmt field.
8668 (data_ref_loc): Add an is_conditional_in_stmt field.
8669 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8670 (find_data_references_in_stmt): Update call to create_data_ref.
8671 (graphite_find_data_references_in_stmt): Likewise.
8672 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8673 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8674 (vect_record_base_alignment): New function.
8675 (vect_record_base_alignments): Likewise.
8676 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8677 for nested statements even if we fail to compute a misalignment.
8678 Use pooled base alignments for unconditional references.
8679 (vect_find_same_alignment_drs): Compare base addresses instead
8681 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8682 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8684 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8686 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8687 Add an explicit name for the enum. Use auto_vec for slp_instances
8689 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8691 (_bb_vec_info): Add a constructor and destructor.
8692 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8693 (destroy_loop_vec_info): Delete.
8694 (vect_destroy_datarefs): Likewise.
8695 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8696 (vec_info::vec_info): New function.
8697 (vec_info::~vec_info): Likewise.
8698 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8699 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8700 destroy_loop_vec_info.
8701 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8702 (_loop_vec_info::_loop_vec_info): ...this.
8703 (destroy_loop_vec_info): Replace with...
8704 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8705 the stmt_vec_infos. Leave handling of vec_info information to its
8706 destructor. Remove explicit vector releases.
8707 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8708 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8709 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8710 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8711 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8712 (destroy_bb_vec_info): Replace with...
8713 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8714 information to its destructor.
8715 (vect_slp_analyze_bb_1): Use new and delete instead of
8716 new_bb_vec_info and destroy_bb_vec_info.
8717 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8720 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8722 * tree-data-ref.h (subscript): Add access_fn field.
8723 (data_dependence_relation): Add could_be_independent_p.
8724 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8725 (same_access_functions): Move to tree-data-ref.c.
8726 * tree-data-ref.c (ref_contains_union_access_p): New function.
8727 (access_fn_component_p): Likewise.
8728 (access_fn_components_comparable_p): Likewise.
8729 (dr_analyze_indices): Add a reference to access_fn_component_p.
8730 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8732 (constant_access_functions): Likewise.
8733 (add_other_self_distances): Likewise.
8734 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8735 (initialize_data_dependence_relation): Use XCNEW and remove
8736 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8737 of access functions that have the same type. Allow the
8738 subsequence to end with different bases in some circumstances.
8739 Record the chosen access functions in SUB_ACCESS_FN.
8740 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8741 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8742 (subscript_dependence_tester_1): Likewise dra and drb.
8743 (build_classic_dist_vector): Update calls accordingly.
8744 (subscript_dependence_tester): Likewise.
8745 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8746 DDR_COULD_BE_INDEPENDENT_P.
8747 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8748 comp_alias_ddrs instead of may_alias_ddrs.
8749 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8751 (vect_analyze_data_ref_dependence): Use it if
8752 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8753 distance vectors if that fails.
8754 (dependence_distance_ge_vf): New function.
8755 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8756 LOOP_VINFO_MAY_ALIAS_DDRS.
8758 2017-08-04 Richard Biener <rguenther@suse.de>
8761 * fold-const.c (fold_binary_loc): Properly restrict
8762 minus_var0 && minus_var1 case when associating undefined overflow
8765 2017-08-04 Tom de Vries <tom@codesourcery.com>
8767 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8769 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8771 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8772 Don't start diagnostic messages with a capital letter.
8773 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8775 (rs6000_invalid_builtin): Likewise.
8776 (rs6000_trampoline_init): Likewise.
8778 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8781 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8782 after setting changeable df flags.
8784 2017-08-03 Richard Biener <rguenther@suse.de>
8786 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8787 up if the use is in USE - X.
8789 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8791 * toplev.c (dumpfile.h): New include.
8792 (internal_error_reentered): New static function. Use it...
8793 (internal_error_function): ...here to handle reentered internal_error.
8795 2017-08-03 Richard Biener <rguenther@suse.de>
8798 * fold-const.c (split_tree): Add minus_var and minus_con
8799 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8800 here but always use minus_*.
8801 (associate_trees): Assert we never associate with MINUS_EXPR
8802 and NULL first operand. Do not recurse for PLUS_EXPR operands
8803 when associating as MINUS_EXPR either.
8804 (fold_binary_loc): Track minus_var and minus_con.
8806 2017-08-03 Tom de Vries <tom@codesourcery.com>
8809 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8810 ACCEL_COMPILER, apply finish_options on
8811 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8813 2017-08-03 Tom de Vries <tom@codesourcery.com>
8816 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8817 function_entry_patch_area_size > 0.
8819 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8822 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8823 instead of 10??LU, perform unit multiplication in wide_int,
8824 don't change alloc_object_size_limit if the limit is larger
8827 PR tree-optimization/81655
8828 PR tree-optimization/81588
8829 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8830 the case when ranges[i].low and high are 1 for unsigned type with
8834 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8835 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8837 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8839 * tree-vrp.h: Add include guard.
8841 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8844 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8845 (ud2): New insn pattern.
8846 * config/i386/i386.c (ix86_expand_epilogue):
8847 For naked functions, generate ud2 instead of trap insn.
8849 2017-08-02 Marek Polacek <polacek@redhat.com>
8852 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8854 2017-08-02 Tom de Vries <tom@codesourcery.com>
8855 Cesar Philippidis <cesar@codesourcery.com>
8857 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8858 Add missing edge probabilities.
8860 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8862 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8865 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8868 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8869 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8870 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8871 prologue_seq sequences - if any.
8873 2017-08-02 Richard Biener <rguenther@suse.de>
8875 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8876 via vectors if supported, integer extracts via punning if supported
8877 or otherwise vector extracts.
8879 2017-08-02 Richard Biener <rguenther@suse.de>
8881 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8883 (bitmap_insert_into_set): ... this.
8885 2017-08-02 Richard Biener <rguenther@suse.de>
8887 PR tree-optimization/81633
8889 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8891 PR tree-optimization/71752
8892 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8894 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8896 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8897 (machine_function::call_ms2sysv_pad_out): Remove field.
8898 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8899 (ix86_compute_frame_layout): Likewise.
8901 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8904 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8905 attribute with interrupt attribute.
8907 2017-08-01 Andrew Pinski <apinski@cavium.com>
8909 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8910 BIT_INSERT_EXPR's operand 1
8911 to see if the types precision matches.
8913 2017-08-01 Martin Liska <mliska@suse.cz>
8916 * builtins.c (expand_builtin_memcpy_args): Remove.
8917 (expand_builtin_memcpy): Call newly added function
8918 expand_builtin_memory_copy_args.
8919 (expand_builtin_memcpy_with_bounds): Likewise.
8920 (expand_builtin_mempcpy): Remove last argument.
8921 (expand_builtin_mempcpy_with_bounds): Likewise.
8922 (expand_builtin_memory_copy_args): New function created from
8923 expand_builtin_mempcpy_args with small modifications.
8924 (expand_builtin_mempcpy_args): Remove.
8925 (expand_builtin_stpcpy): Remove unused argument.
8926 (expand_builtin): Likewise.
8927 (expand_builtin_with_bounds): Likewise.
8929 2017-08-01 Martin Liska <mliska@suse.cz>
8932 Make mempcpy more optimal (PR middle-end/70140).
8934 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8937 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8938 __builtin_vec_cmpne verify both arguments are compatible vectors
8939 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8940 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8941 move computation of aligned to after checking the argument types.
8945 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8946 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8949 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8950 Doug Rupp <rupp@adacore.com>
8951 Olivier Hainque <hainque@adacore.com>
8953 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8954 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8956 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8957 for TARGET_OS_CPP_BUILTIN.
8958 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8959 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8961 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8962 passing required abi options to the assembler for EABI configurations.
8963 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8964 of .text.hot and .text.unlikely sections for kernel modules when
8965 using ARM style exceptions.
8966 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8967 options. Add EXTRA_CC1_SPEC.
8968 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8970 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8972 (ARM_TARGET2_DWARF_FORMAT): Define.
8973 * config/arm/t-vxworks: Adjust multilib control to removal of the
8974 Diab command line options.
8976 2017-08-01 Martin Liska <mliska@suse.cz>
8978 PR gcov-profile/81561
8979 * gcov.c (unblock): Make unblocking safe as we need to preserve
8980 index correspondence of blocks and block_lists.
8982 2017-08-01 Richard Biener <rguenther@suse.de>
8984 PR tree-optimization/81181
8985 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8986 (compute_antic): ... end of iteration here.
8988 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8990 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8991 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8992 (ftree-slp-vectorize): Likewise.
8993 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8994 can no longer be set independent of flag_tree_loop_vectorize.
8995 * omp-general.c (emp_max_vf): Likewise.
8996 * opts.c (enable_fdo_optimizations): Remove references to
8997 flag_tree_vectorize, these are now implicit.
8998 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8999 and leave it for the options machinery.
9001 2017-08-01 Martin Liska <mliska@suse.cz>
9004 * builtins.c (expand_builtin_memcpy_args): Remove.
9005 (expand_builtin_memcpy): Call newly added function
9006 expand_builtin_memory_copy_args.
9007 (expand_builtin_memcpy_with_bounds): Likewise.
9008 (expand_builtin_mempcpy): Remove last argument.
9009 (expand_builtin_mempcpy_with_bounds): Likewise.
9010 (expand_builtin_memory_copy_args): New function created from
9011 expand_builtin_mempcpy_args with small modifications.
9012 (expand_builtin_mempcpy_args): Remove.
9013 (expand_builtin_stpcpy): Remove unused argument.
9014 (expand_builtin): Likewise.
9015 (expand_builtin_with_bounds): Likewise.
9017 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9020 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9021 print "ds:" only for immediates in generic address space.
9023 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9026 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9027 (ix86_function_ok_for_sibcall): Return false for naked functions.
9029 2017-08-01 Richard Biener <rguenther@suse.de>
9031 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9032 (compute_antic): Seed worklist with exit block predecessors.
9033 * cfganal.c (dfs_find_deadend): For a cycle return the source
9034 of the edge closing it.
9036 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9038 * config/aarch64/aarch64.c
9039 (aarch64_can_const_movi_rtx_p): Move 0 check.
9041 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9043 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9044 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9046 * match.pd: Ditto in address comparison pattern.
9048 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9050 PR tree-optimization/81627
9051 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9052 closed ssa form for store-store chain.
9054 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9056 PR tree-optimization/81620
9057 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9058 for store-store chain.
9060 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9062 PR tree-optimization/81588
9063 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9064 ranges[i].in_p, invert comparison code ccode. For >/>=,
9065 swap rhs1 and rhs2 and comparison code unconditionally,
9066 for </<= don't do that. Don't swap rhs1/rhs2 again if
9067 ranges[i].in_p, instead invert comparison code ccode if
9068 opcode or oe->rank is BIT_IOR_EXPR.
9071 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9072 a direct optab to conversion optab.
9073 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9074 with GET_MODE_INNER as last argument instead of optab_handler.
9075 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9076 vector extraction if possible and optab is available.
9077 * expr.c (store_constructor): Use convert_optab_handler instead
9078 of optab_handler. Use vector initialization from smaller
9079 vectors if possible and optab is available.
9080 * tree-vect-stmts.c (vectorizable_load): Likewise.
9081 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9083 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9084 of vec_init from half-sized vectors with the same element mode.
9085 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9086 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9087 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9088 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9089 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9090 after mode in gen_vec_extract* calls.
9091 (vec_extract<mode>): Renamed to ...
9092 (vec_extract<mode><ssescalarmodelower>): ... this.
9093 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9094 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9095 element mode after mode in gen_vec_init* calls.
9096 (VEC_INIT_HALF_MODE): New mode iterator.
9097 (vec_init<mode>): Renamed to ...
9098 (vec_init<mode><ssescalarmodelower>): ... this.
9099 (vec_init<mode><ssehalfvecmodelower>): New expander.
9100 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9101 (vec_extractv2sfsf): ... this.
9102 (vec_initv2sf): Renamed to ...
9103 (vec_initv2sfsf): ... this.
9104 (vec_extractv2si): Renamed to ...
9105 (vec_extractv2sisi): ... this.
9106 (vec_initv2si): Renamed to ...
9107 (vec_initv2sisi): ... this.
9108 (vec_extractv4hi): Renamed to ...
9109 (vec_extractv4hihi): ... this.
9110 (vec_initv4hi): Renamed to ...
9111 (vec_initv4hihi): ... this.
9112 (vec_extractv8qi): Renamed to ...
9113 (vec_extractv8qiqi): ... this.
9114 (vec_initv8qi): Renamed to ...
9115 (vec_initv8qiqi): ... this.
9116 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9117 (vec_init<mode>): Renamed to ...
9118 (vec_init<mode><VEC_base_l>): ... this.
9119 (vec_extract<mode>): Renamed to ...
9120 (vec_extract<mode><VEC_base_l>): ... this.
9121 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9122 (vec_initv2sfsf): ... this.
9123 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9124 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9125 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9126 element mode after mode in gen_vec_init* calls.
9127 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9128 (vec_init<mode><Vel>): ... this.
9129 (vec_extract<mode>): Renamed to ...
9130 (vec_extract<mode><Vel>): ... this.
9131 * config/aarch64/iterators.md (Vel): New mode attribute.
9132 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9133 Add element mode after mode in gen_vec_extract* calls.
9134 * config/s390/vector.md (non_vec_l): New mode attribute.
9135 (vec_extract<mode>): Renamed to ...
9136 (vec_extract<mode><non_vec_l>): ... this.
9137 (vec_init<mode>): Renamed to ...
9138 (vec_init<mode><non_vec_l>): ... this.
9139 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9140 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9142 * config/arm/iterators.md (V_elem_l): New mode attribute.
9143 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9144 (vec_extract<mode><V_elem_l>): ... this.
9145 (vec_extractv2di): Renamed to ...
9146 (vec_extractv2didi): ... this.
9147 (vec_init<mode>): Renamed to ...
9148 (vec_init<mode><V_elem_l>): ... this.
9149 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9150 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9151 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9152 Add element mode after gen_vec_extract* calls.
9153 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9154 (vec_init<mode><unitmode>): ... this.
9155 (vec_extract<mode>): Renamed to ...
9156 (vec_extract<mode><unitmode>): ... this.
9157 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9158 (vec_init<mode><unitmode>): ... this.
9159 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9160 (vec_initv2sfsf): ... this.
9161 (vec_extractv2sf): Renamed to ...
9162 (vec_extractv2sfsf): ... this.
9163 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9164 Add element mode after gen_vec_extract* calls.
9165 * config/mips/mips.md (unitmode): New mode iterator.
9166 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9167 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9168 * config/spu/spu.md (inner_l): New mode attribute.
9169 (vec_init<mode>): Renamed to ...
9170 (vec_init<mode><inner_l>): ... this.
9171 (vec_extract<mode>): Renamed to ...
9172 (vec_extract<mode><inner_l>): ... this.
9173 * config/sparc/sparc.md (veltmode): New mode iterator.
9174 (vec_init<VMALL:mode>): Renamed to ...
9175 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9176 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9177 (vec_initv2sisi): ... this.
9178 (vec_initv2sf): Renamed to ...
9179 (vec_initv2sfsf): ... this.
9180 (vec_extractv2sf): Renamed to ...
9181 (vec_extractv2sfsf): ... this.
9182 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9183 (vec_init<mode>): Renamed to ...
9184 (vec_init<mode><VEC_base_l>): ... this.
9185 (vec_extract<mode>): Renamed to ...
9186 (vec_extract<mode><VEC_base_l>): ... this.
9187 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9188 (vec_initv2sfsf): ... this.
9189 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9190 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9191 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9192 gen_vec_init* calls.
9194 2017-08-01 Richard Biener <rguenther@suse.de>
9196 PR tree-optimization/81297
9197 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9198 TREE_OVERFLOW from INTEGER_CSTs.
9200 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9202 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9204 2017-07-31 Carl Love <cel@us.ibm.com>
9206 * config/rs6000/rs6000-c: Add support for built-in functions
9207 vector signed char vec_xl_be (signed long long, signed char *);
9208 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9209 vector signed int vec_xl_be (signed long long, signed int *);
9210 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9211 vector signed long long vec_xl_be (signed long long, signed long long *);
9212 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9213 vector signed short vec_xl_be (signed long long, signed short *);
9214 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9215 vector double vec_xl_be (signed long long, double *);
9216 vector float vec_xl_be (signed long long, float *);
9217 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9218 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9219 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9221 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9222 (altivec_expand_builtin): Add switch statement to call
9223 altivec_expand_xl_be for each builtin.
9224 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9225 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9226 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9227 __builtin_vsx_le_be_v16qi.
9228 * doc/extend.texi: Update the built-in documentation file for the
9229 new built-in functions.
9231 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9234 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9236 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9238 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9240 * config.gcc: Add z14.
9241 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9242 CPU model numbers for z13s and z14.
9243 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9245 * config/s390/s390-opts.h (enum processor_type): Rename
9246 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9247 * config/s390/s390.c (processor_table): Add field for CPU name to
9248 be passed to Binutils.
9249 (s390_asm_output_machine_for_arch): Use the new field in
9250 processor_table for Binutils.
9251 (s390_expand_builtin): Replace arch12 with z14.
9252 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9253 (s390_get_sched_attrmask): Likewise.
9254 (s390_get_unit_mask): Likewise.
9255 * config/s390/s390.opt: Add z14 to processor_type enum.
9257 2017-07-31 Martin Jambor <mjambor@suse.cz>
9260 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9261 regardless of optimization level.
9263 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9264 Martin Liska <mliska@suse.cz>
9266 * predict.def: Remove old comment and adjust probability.
9267 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9270 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9273 * config/i386/i386.c (ix86_function_naked): New function.
9274 (ix86_can_use_return_insn_p): Return false for naked functions.
9275 (ix86_expand_prologue): Skip prologue for naked functions.
9276 (ix86_expand_epilogue): Skip epilogue for naked functions
9277 and emit trap instruction.
9278 (ix86_warn_func_return): New function.
9279 (ix86_attribute_table): Add "naked" attribute specification.
9280 (TARGET_WARN_FUNC_RETURN): Define.
9281 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9283 2017-07-31 Martin Liska <mliska@suse.cz>
9285 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9286 (dump_gimple_bb_header): Always dump BB info.
9287 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9289 2017-07-31 Martin Liska <mliska@suse.cz>
9292 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9293 also with current_function_decl non-null equality.
9295 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9298 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9299 change type to the element type, instead add eltype variable and
9300 use it where we are interested in the element type.
9302 PR tree-optimization/81603
9303 * ipa-polymorphic-call.c
9304 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9305 offset arithmetic in offset_int, bail out if the resulting bit offset
9306 doesn't fit into shwi.
9308 2017-07-31 Martin Liska <mliska@suse.cz>
9310 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9311 (gimplify_save_expr): Fix comment.
9313 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9316 * config/i386/i386.c (ix86_function_arg): Update arguments for
9318 (ix86_compute_frame_layout): Set the initial stack offset to
9319 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9320 INCOMING_FRAME_SP_OFFSET.
9321 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9322 stack before exception handler returns.
9323 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9324 the 'ERROR_CODE' for exception handler.
9326 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9328 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9329 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9330 (ASM_OUTPUT_REG_POP): Ditto.
9331 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9332 instead of asm_fprintf to output pure string.
9334 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9336 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9337 to imported_module_or_decl hook.
9338 (debug_nothing_tree_tree_tree_bool): Remove.
9339 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9340 * debug.c (do_nothing_debug_hooks): Use
9341 debug_nothing_tree_tree_tree_bool_bool instead of
9342 debug_nothing_tree_tree_tree_bool.
9343 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9344 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9345 * sdbout.c (sdb_debug_hooks): Likewise.
9346 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9347 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9349 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9350 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9351 attribute, don't add anything.
9353 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9355 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9356 (fold_build2_stat_loc): Likewise.
9357 (fold_build3_stat_loc): Likewise.
9358 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9359 (fold_build1_loc): Remove macro.
9360 (fold_build2_loc): Likewise.
9361 (fold_build3_loc): Likewise.
9363 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9365 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9366 (gimple_build_debug_bind_source_stat): Likewise.
9367 * gimple.h (gimple_build_debug_bind): Remove macro.
9368 (gimple_build_debug_bind_source): Likewise.
9370 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9372 * bitmap.c (bitmap_alloc): Adjust.
9373 (bitmap_gc_alloc): Likewise.
9374 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9376 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9378 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9379 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9380 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9381 (bitmap_gc_alloc_stat): Likewise.
9382 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9384 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9386 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9387 * rtl.h (shallow_copy_rtx): Remove macro.
9389 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9391 * emit-rtl.c (gen_raw_REG): Adjust.
9392 * gengenrtl.c (gendef): Likewise.
9393 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9394 * rtl.h (rtx_alloc): Remove macro.
9396 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9398 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9399 (build_tree_list_stat): Likewise.
9400 * tree.h (build_tree_list): Remove macro.
9401 (build_tree_list_vec): Likewise.
9403 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9405 * tree.c (make_vector_stat): Remove _stat from name.
9406 (build_vector_stat): Likewise.
9407 * tree.h (make_vector_stat): Remove macro.
9408 (build_vector_stat): Likewise.
9410 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9412 * tree.h (build_var_debug_value): Remove prototype.
9414 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9416 * tree.c (tree_cons_stat): Remove _stat from name.
9417 * tree.h (tree_cons): Remove macro.
9419 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9421 * tree.c (build_vl_exp_stat): Remove _stat from name.
9422 * tree.h (build_vl_exp): Remove macro.
9424 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9426 * tree.c (build_decl_stat): Remove _stat from name.
9427 * tree.h (build_decl): Remove macro.
9429 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9431 * gimple.c (gimple_build_with_ops_stat): Adjust.
9432 (gimple_alloc_stat): Remove _stat from name.
9433 * gimple.h (gimple_alloc): Remove macro.
9435 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9437 * tree.c (make_tree_vec_stat): Remove _stat from name.
9438 (grow_tree_vec_stat): Likewise.
9439 * tree.h (make_tree_vec_stat): Adjust prototype.
9440 (grow_tree_vec_stat): Likewise.
9441 (make_tree_vec): Remove macro.
9442 (grow_tree_vec): Likewise.
9444 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9446 * fold-const.c (fold_build1_stat_loc): Adjust.
9447 (fold_build2_stat_loc): Likewise.
9448 (fold_build3_stat_loc): Likewise.
9449 * tree.c (build0_stat): Remove _stat from name.
9450 (build1_stat): Likewise.
9451 (build2_stat): Likewise.
9452 (build3_stat): Likewise.
9453 (build4_stat): Likewise.
9454 (build5_stat): Likewise.
9455 * tree.h (build1_loc): Remove macro, and rename _stat function
9457 (build2_loc): Likewise.
9458 (build3_loc): Likewise.
9459 (build4_loc): Likewise.
9460 (build5_loc): Likewise.
9462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9464 * tree.c (make_int_cst_stat): Remove _stat from name.
9465 * tree.h (make_int_cst_stat): Adjust prototype.
9466 (make_int_cst): Remove macro.
9468 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9470 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9471 * tree.h (make_tree_binfo_stat): Adjust prototype.
9472 (make_tree_binfo): Remove.
9474 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9476 * tree.c (copy_node_stat): Rename to copy_node.
9477 (build_distinct_type_copy): Adjust.
9478 * tree.h (copy_node_stat): Adjust prototype.
9479 (copy_node): Remove macro.
9481 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9483 * tree.c (make_node_stat): rename to make_node.
9484 (build_tree_list_stat): Adjust.
9485 (build0_stat): Likewise.
9486 (build2_stat): Likewise.
9487 (build3_stat): Likewise.
9488 (build4_stat): Likewise.
9489 (build5_stat): Likewise.
9490 (build_decl_stat): Likewise.
9491 * tree.h (make_node_stat): Adjust prototype.
9492 (make_node): remove macro.
9494 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9496 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9497 (PPC_FEATURE2_SCV): Likewise.
9498 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9500 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9502 * config/aarch64/aarch64.c
9503 (aarch64_internal_mov_immediate): Add new special pattern.
9504 * config/aarch64/aarch64.md (*movdi_aarch64):
9505 Add reg/32bit const mov case.
9507 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9508 Richard Sandiford <richard.sandiford@linaro.org>
9510 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9511 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9512 Add integer and movi cases.
9513 (movi-split-hf-df-sf split, fp16): New.
9514 (enabled): Added TARGET_FP_F16INST.
9515 * config/aarch64/iterators.md (GPF_HF): New.
9516 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9518 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9520 * config/aarch64/aarch64.c
9521 (aarch64_simd_container_mode): Add prototype.
9522 (aarch64_expand_mov_immediate): Add HI support.
9523 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9524 (aarch64_can_const_movi_rtx_p): New.
9525 (aarch64_preferred_reload_class):
9526 Remove restrictions of using FP registers for certain SIMD operations.
9527 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9528 (aarch64_valid_floating_const): Add integer move validation.
9529 (aarch64_simd_imm_scalar_p): Remove.
9530 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9531 (aarch64_legitimate_constant_p): Expand list of supported cases.
9532 * config/aarch64/aarch64-protos.h
9533 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9534 (aarch64_reinterpret_float_as_int): New.
9535 (aarch64_simd_imm_scalar_p): Remove.
9536 * config/aarch64/constraints.md (Uvi): New.
9537 (Dd): Split into Ds and new Dd.
9538 * config/aarch64/aarch64.md (*movsi_aarch64):
9540 (*movdi_aarch64): Add SIMD mov case.
9542 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9544 * tree-predcom.c: (struct chain): Handle store-store chain in which
9545 stores for elimination only store loop invariant values.
9546 (execute_pred_commoning_chain): Ditto.
9547 (prepare_initializers_chain_store_elim): Ditto.
9548 (prepare_finalizers): Ditto.
9549 (is_inv_store_elimination_chain): New function.
9550 (initialize_root_vars_store_elim_1): New function.
9552 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9554 * tree-predcom.c: Revise general description of the pass.
9555 (enum chain_type): New enum type for store elimination.
9556 (struct chain): New field supporting store elimination.
9557 (struct component): Ditto.
9558 (dump_chain): Dump store-stores chain.
9559 (release_chain): Release resources.
9560 (split_data_refs_to_components): Compute and create component
9561 contains only stores for elimination.
9562 (get_chain_last_ref_at): New function.
9563 (make_invariant_chain): Initialization.
9564 (make_rooted_chain): Specify chain type in parameter and record it.
9565 (add_looparound_copies): Skip for store-stores chain.
9566 (determine_roots_comp): Compute type of chain and pass it to
9568 (initialize_root_vars_store_elim_2): New function.
9569 (finalize_eliminated_stores): New function.
9570 (remove_stmt): Handle store for elimination.
9571 (execute_pred_commoning_chain): Execute predictive commoning on
9573 (determine_unroll_factor): Skip unroll for store-stores chain.
9574 (prepare_initializers_chain_store_elim): New function.
9575 (prepare_initializers_chain): Hanlde store-store chain.
9576 (prepare_finalizers_chain, prepare_finalizers): New function.
9577 (tree_predictive_commoning_loop): Return integer value indicating
9578 if loop is unrolled or lcssa form is corrupted.
9579 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9581 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9583 * tree-predcom.c (initialize_root): Delete.
9584 (execute_pred_commoning_chain): Initialize root vars and replace
9585 reference of non-combined chain directly, rather than call above
9588 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9590 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9591 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9593 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9595 * tree-predcom.c (struct chain): New field init_seq.
9596 (release_chain): Release init_seq.
9597 (prepare_initializers_chain): Record intialization stmts in above
9599 (insert_init_seqs): New function.
9600 (tree_predictive_commoning_loop): Call insert_init_seqs.
9602 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9604 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9606 2017-07-28 Richard Biener <rguenther@suse.de>
9608 * match.pd: Remove superfluous :c.
9609 * genmatch.c (simplify::id): Add member.
9610 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9612 (current_id): New global.
9613 (dt_node::parent): Move from ...
9614 (dt_operand::parent): ... here. Add for_id member.
9615 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9616 (decision_tree::find_node): Relax order requirement when
9617 merging DT_TRUE nodes to ones inbetween the current simplify
9618 and the one we try to merge with. Add diagnostic whenever
9619 we need to enforce pattern order by not merging.
9620 (decision_tree::insert): Set current_id.
9621 (decision_tree::print_node): Dump parent node and for_id.
9622 (parser::last_id): Add member.
9623 (parser::push_simplify): Assign unique id.
9624 (parser::parser): Initialize last_id.
9626 2017-07-28 Martin Liska <mliska@suse.cz>
9629 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9630 gimple_build_debug_bind.
9632 2017-07-28 Richard Biener <rguenther@suse.de>
9634 PR tree-optimization/81502
9635 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9637 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9639 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9640 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9641 for BIT_FIELD_REF args.
9642 * fold-const.c (make_bit_field_ref): Likewise.
9643 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9645 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9648 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9649 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9650 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9651 Or it into SANITIZER_UNDEFINED.
9652 * ubsan.c: Include gimple-fold.h and varasm.h.
9653 (ubsan_expand_ptr_ifn): New function.
9654 (instrument_pointer_overflow): New function.
9655 (maybe_instrument_pointer_overflow): New function.
9656 (instrument_object_size): Formatting fix.
9657 (pass_ubsan::execute): Call instrument_pointer_overflow
9658 and maybe_instrument_pointer_overflow.
9659 * internal-fn.c (expand_UBSAN_PTR): New function.
9660 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9661 * sanitizer.def (__ubsan_handle_pointer_overflow,
9662 __ubsan_handle_pointer_overflow_abort): New builtins.
9663 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9664 * internal-fn.def (UBSAN_PTR): New internal function.
9665 * opts.c (sanitizer_opts): Add pointer-overflow.
9666 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9667 * fold-const.c (build_range_check): Compute pointer range check in
9668 integral type if pointer arithmetics would be needed. Formatting
9671 2017-07-28 Martin Liska <mliska@suse.cz>
9674 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9675 parameters that are of a variable-length.
9677 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9679 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9681 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9682 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9683 (CRT_CALL_STATIC_FUNCTION): Likewise.
9684 (ASM_DEFAULT_SPEC): New define.
9685 (ASM_SPEC32): Likewise.
9686 (ASM_SPEC64): Likewise.
9687 (ASM_SPEC_COMMON): Likewise.
9688 (ASM_SPEC): Likewise.
9689 (INVALID_64BIT): Likewise.
9690 (LINK_OS_DEFAULT_SPEC): Likewise.
9691 (LINK_OS_SPEC32): Likewise.
9692 (LINK_OS_SPEC64): Likewise.
9693 (POWERPC_LINUX): Likewise.
9694 (PTRDIFF_TYPE): Likewise.
9695 (RESTORE_FP_PREFIX): Likewise.
9696 (RESTORE_FP_SUFFIX): Likewise.
9697 (SAVE_FP_PREFIX): Likewise.
9698 (SAVE_FP_SUFFIX): Likewise.
9699 (SIZE_TYPE): Likewise.
9700 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9701 (TARGET_64BIT): Likewise.
9702 (TARGET_64BIT): Likewise.
9703 (TARGET_AIX): Likewise.
9704 (WCHAR_TYPE_SIZE): Likewise.
9705 (WCHAR_TYPE): Undefine.
9706 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9707 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9708 (CPP_OS_RTEMS_SPEC): Delete.
9709 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9710 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9712 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9714 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9716 PR tree-optimization/81578
9717 * tree-parloops.c (build_new_reduction): Bail out if
9718 reduction_code isn't one of the standard OpenMP reductions.
9719 Move the details printing after that decision.
9721 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9723 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9724 related to reload_in_progress.
9725 (splat_input_operand): Likewise.
9726 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9728 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9730 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9731 (TARGET_INSTANTIATE_DECLS): Likewise.
9732 (legitimate_indexed_address_p): Delete reload_in_progress code.
9733 (rs6000_debug_legitimate_address_p): Likewise.
9734 (rs6000_eliminate_indexed_memrefs): Likewise.
9735 (rs6000_emit_le_vsx_store): Likewise.
9736 (rs6000_emit_move_si_sf_subreg): Likewise.
9737 (rs6000_emit_move): Likewise.
9738 (register_to_reg_type): Likewise.
9739 (rs6000_pre_atomic_barrier): Likewise.
9740 (rs6000_machopic_legitimize_pic_address): Likewise.
9741 (rs6000_allocate_stack_temp): Likewise.
9742 (rs6000_address_for_fpconvert): Likewise.
9743 (rs6000_address_for_altivec): Likewise.
9744 (rs6000_secondary_memory_needed_rtx): Delete function.
9745 (rs6000_check_sdmode): Likewise.
9746 (rs6000_alloc_sdmode_stack_slot): Likewise.
9747 (rs6000_instantiate_decls): Likewise.
9748 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9749 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9750 Delete reload_in_progress.
9751 (*vec_reload_and_plus_<mptrsize>): Likewise.
9752 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9753 (vsx_div_v2di): Likewise.
9754 (vsx_udiv_v2di): Likewise.
9756 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9758 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9759 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9760 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9761 (rs6000_debug_reg_global): Delete print of LRA status.
9762 (rs6000_option_override_internal): Delete dead LRA related code.
9763 (rs6000_lra_p): Delete function.
9764 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9766 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9768 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9769 * config/riscv/rtems.h: New file.
9771 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9772 Sudakshina Das <sudi.das@arm.com>
9774 * config/aarch64/aarch64.md
9775 (define_split for and<mode>3nr_compare): Move
9776 non aarch64_logical_operand to a register.
9777 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9778 register immediate operand to a register.
9779 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9781 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9784 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9786 2017-07-27 Richard Biener <rguenther@suse.de>
9788 PR tree-optimization/81573
9789 PR tree-optimization/81494
9790 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9791 multi defuse cycle case.
9793 2017-07-27 Richard Biener <rguenther@suse.de>
9795 PR tree-optimization/81571
9796 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9799 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9801 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9802 earlier and only if MASK_FPU is set. Adjust formatting.
9804 2017-07-27 Martin Liska <mliska@suse.cz>
9806 * opt-functions.awk: Add validation of value of Init.
9807 * optc-gen.awk: Pass new argument.
9809 2017-07-27 Martin Liska <mliska@suse.cz>
9811 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9812 Fix wrong condition.
9814 2017-07-27 Martin Liska <mliska@suse.cz>
9816 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9817 BBs and edges seen by autoFDO.
9819 2017-07-27 Richard Biener <rguenther@suse.de>
9821 PR tree-optimization/81502
9822 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9823 with incompatible but same sized type.
9824 (execute_update_addresses_taken): Likewise.
9826 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9828 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9829 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9831 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9834 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9835 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9836 Change s_operand to memory_operand.
9838 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9840 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9841 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9842 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9843 Emit instructions rather than returning an expression. Handle TFmode
9844 and KFmode by casting to TImode.
9845 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9846 (rs6000_emit_le_vsx_store): Likewise.
9847 * config/rs6000/vsx.md (VSX_TI): New iterator.
9848 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9849 (*vsx_le_undo_permute_<mode>): Likewise.
9850 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9851 emit the split sequence.
9852 (*vsx_le_perm_store_<mode>): Likewise.
9854 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9856 PR tree-optimization/81555
9857 PR tree-optimization/81556
9858 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9859 if true, force CHANGED for the recursive invocation.
9860 (reassociate_bb): Remember original length of ops array, pass
9861 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9863 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9864 attributes for noipa attribute. For naked attribute use
9865 lookup_attribute first before lookup_attribute_spec.
9866 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9868 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9869 for functions with noipa attribute.
9870 (cgraph_externally_visible_p): Return true for functions with noipa
9872 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9873 for functions with noipa attribute.
9874 * doc/extend.texi: Document noipa function attribute.
9875 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9876 also for functions with noipa attribute.
9877 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9879 2017-07-26 Andrew Pinski <apinski@cavium.com>
9881 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9882 vec_unalign_load_cost and vec_unalign_store_cost.
9884 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9886 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9887 -mvsx-small-integer option.
9888 (ISA_3_0_MASKS_IEEE): Likewise.
9889 (OTHER_VSX_VECTOR_MASKS): Likewise.
9890 (POWERPC_MASKS): Likewise.
9891 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9892 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9893 code, only testing for DImode being allowed in non-VSX floating
9895 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9896 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9898 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9899 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9900 TARGET_P8_VECTOR test.
9901 (rs6000_secondary_reload_simple_move): Likewise.
9902 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9903 since TARGET_P9_VECTOR was already tested.
9904 (rs6000_opt_masks): Remove -mvsx-small-integer.
9905 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9906 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9908 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9909 test for TARGET_VEXTRACTUB was used, and that uses
9911 (p9 extract splitter): Likewise.
9912 (vsx_extract_<mode>_di_p9): Likewise.
9913 (vsx_extract_<mode>_store_p9): Likewise.
9914 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9915 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9916 the elimination of TARGET_VSX_SMALL_INTEGER.
9917 (vsx_extract_<mode>_p8): Likewise.
9918 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9919 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9920 (vsx_set_<mode>_p9): Likewise.
9921 (vsx_set_v4sf_p9): Likewise.
9922 (vsx_set_v4sf_p9_zero): Likewise.
9923 (vsx_insert_extract_v4sf_p9): Likewise.
9924 (vsx_insert_extract_v4sf_p9_2): Likewise.
9925 * config/rs6000/rs6000.md (sign extend splitter): Change
9926 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9927 (floatsi<mode>2_lfiwax_mem): Likewise.
9928 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9929 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9930 since a test for TARGET_P9_VECTOR was used.
9931 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9932 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9933 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9934 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9935 TARGET_P8_VECTOR test.
9936 (fix_trunc<mode>si2_stfiwx): Likewise.
9937 (fix_trunc<mode>si2_internal): Likewise.
9938 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9939 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9941 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9942 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9943 TARGET_P8_VECTOR test.
9944 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9945 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9946 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9948 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9949 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9950 since a test for TARGET_P9_VECTOR was used.
9951 (splitter for loading small constants): Likewise.
9953 2017-07-26 Andrew Pinski <apinski@cavium.com>
9955 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9958 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9961 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9962 (fp_valid_at): Likewise.
9964 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9966 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9967 (qdf24xx_addrcost_table): Likewise.
9968 (cortexa57_tunings): Update to use generic_branch_cost.
9969 (cortexa72_tunings): Likewise.
9970 (cortexa73_tunings): Likewise.
9971 (qdf24xx_tunings): Likewise.
9973 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9975 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9976 (thunderx2t99_branch_cost): Likewise.
9977 (cortexa35_tunings): Update to use generic_branch_cost.
9978 (cortexa53_tunings): Likewise.
9979 (cortexa57_tunings): Likewise.
9980 (cortexa72_tunings): Likewise.
9981 (cortexa73_tunings): Likewise.
9982 (thunderx2t99_tunings): Likewise.
9984 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9986 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9987 (sparc_option_override): Honour MASK_FSMULD.
9988 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9989 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9990 * config/sparc/sparc.opt (mfsmuld): New option.
9991 * doc/invoke.texi (mfsmuld): Document option.
9993 2017-07-26 Marek Polacek <polacek@redhat.com>
9996 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9998 2017-07-26 Richard Biener <rguenther@suse.de>
10000 * gimple-match-head.c (do_valueize): Return OP if valueize
10002 (get_def): New helper to get at the def stmt of a SSA name
10003 if valueize allows.
10004 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10005 do_valueize to get at the def stmt.
10006 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10008 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10010 PR middle-end/46932
10011 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10013 2017-07-26 Martin Liska <mliska@suse.cz>
10016 * function.c (expand_function_start): Make expansion of
10017 nonlocal_goto_save_area after parm_birth_insn.
10019 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10021 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10022 from all CPU target flags enable members.
10024 2017-07-26 Richard Biener <rguenther@suse.de>
10026 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10027 (decision_tree::gen): Make 'type' const.
10028 (write_predicate): Likewise.
10030 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10032 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10033 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10034 (rs6000_option_override_internal): Likewise.
10035 (rs6000_expand_vector_set): Likewise.
10036 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10037 (TARGET_UPPER_REGS_SF): Likewise.
10038 (TARGET_UPPER_REGS_DI): Likewise.
10039 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10040 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10041 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10042 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10043 (Splitters for DI constants in Altivec registers): Likewise.
10044 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10045 (vsx_set_v4sf_p9): Likewise.
10046 (vsx_set_v4sf_p9_zero): Likewise.
10047 (vsx_insert_extract_v4sf_p9): Likewise.
10048 (vsx_insert_extract_v4sf_p9_2): Likewise.
10050 2017-07-25 Carl Love <cel@us.ibm.com>
10052 * doc/extend.texi: Update the built-in documentation file for the
10053 existing built-in functions
10054 vector signed char vec_cnttz (vector signed char);
10055 vector unsigned char vec_cnttz (vector unsigned char);
10056 vector signed short vec_cnttz (vector signed short);
10057 vector unsigned short vec_cnttz (vector unsigned short);
10058 vector signed int vec_cnttz (vector signed int);
10059 vector unsigned int vec_cnttz (vector unsigned int);
10060 vector signed long long vec_cnttz (vector signed long long);
10061 vector unsigned long long vec_cnttz (vector unsigned long long);
10063 2017-07-25 Andrew Pinski <apinski@cavium.com>
10065 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10066 accesses where the use is for the first operand of a BIT_INSERT.
10068 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10071 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10072 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10074 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10076 * config/i386/gstabs.h: Delete.
10077 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10079 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10081 * config/i386/i386.c (ix86_decompose_address): Do not check for
10082 register RTX when looking at index_reg or base_reg.
10083 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10085 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10087 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10088 to update EH info here.
10090 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10092 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10094 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10096 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10098 2017-07-25 Torsten Duwe <duwe@suse.de>
10100 * common.opt: Introduce -fpatchable-function-entry
10101 command line option, and its variables function_entry_patch_area_size
10102 and function_entry_patch_area_start.
10103 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10104 including a two-value parser.
10105 * target.def (print_patchable_function_entry): New target hook.
10106 * targhooks.h (default_print_patchable_function_entry): New function.
10107 * targhooks.c (default_print_patchable_function_entry): Likewise.
10108 * toplev.c (process_options): Switch off IPA-RA if
10109 patchable function entries are being generated.
10110 * varasm.c (assemble_start_function): Look at the
10111 patchable-function-entry command line switch and current
10112 function attributes and maybe generate NOP instructions by
10113 calling the print_patchable_function_entry hook.
10114 * doc/extend.texi: Document patchable_function_entry attribute.
10115 * doc/invoke.texi: Document -fpatchable_function_entry
10116 command line option.
10117 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10119 * doc/tm.texi: Re-generate.
10121 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10124 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10125 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10127 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10129 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10131 2017-07-25 Richard Biener <rguenther@suse.de>
10133 PR tree-optimization/81455
10134 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10135 not walk in cycles when looking for guards.
10137 2017-07-25 Richard Biener <rguenther@suse.de>
10139 PR tree-optimization/81529
10140 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10141 when optimizing backedge uses.
10143 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10145 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10147 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10148 to dl_section_ref. On AIX, append an expression to subtract
10149 the size of the section length to dl_section_ref.
10151 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10153 * configure.ac: If any of the config.* scripts fail, exit 1.
10154 * configure: Regenerate.
10156 2017-07-25 Richard Biener <rguenther@suse.de>
10158 PR middle-end/81546
10159 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10160 of immediate uses, be more verbose on errors.
10162 2017-07-25 Richard Biener <rguenther@suse.de>
10164 PR tree-optimization/81510
10165 * tree-vect-loop.c (vect_is_simple_reduction): When the
10166 reduction stmt is not inside the loop bail out.
10168 2017-07-25 Richard Biener <rguenther@suse.de>
10170 PR tree-optimization/81303
10171 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10172 profitability check against LOOP_VINFO_NITERSM1.
10174 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10176 * domwalk.c (cmp_bb_postorder): Simplify.
10177 (sort_bbs_postorder): New function. Use it...
10178 (dom_walker::walk): ...here to optimize common cases.
10180 2017-07-25 Martin Liska <mliska@suse.cz>
10183 * ipa-visibility.c (function_and_variable_visibility): Make the
10184 redirection just on target that supports aliasing.
10185 Fix GNU coding style.
10187 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10190 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10192 * config/arm/rtems.h: Likewise.
10193 * config/bfin/rtems.h: Likewise.
10194 * config/i386/rtemself.h: Likewise.
10195 * config/lm32/rtems.h: Likewise.
10196 * config/m32c/rtems.h: Likewise.
10197 * config/m68k/rtemself.h: Likewise.
10198 * config/microblaze/rtems.h: Likewise.
10199 * config/mips/rtems.h: Likewise.
10200 * config/moxie/rtems.h: Likewise.
10201 * config/nios2/rtems.h: Likewise.
10202 * config/powerpcspe/rtems.h: Likewise.
10203 * config/rs6000/rtems.h: Likewise.
10204 * config/rtems.h: Likewise.
10205 * config/sh/rtems.h: Likewise.
10206 * config/sh/rtemself.h: Likewise.
10207 * config/sparc/rtemself.h: Likewise.
10209 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10212 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10213 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10214 * tree-ssa-structalias.c (alias_get_name): Same.
10216 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10219 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10220 instructions if no du chain is found.
10222 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10224 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10226 2017-07-25 Richard Biener <rguenther@suse.de>
10228 PR middle-end/81505
10229 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10232 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10234 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10235 upper-regs options.
10236 (ISA_2_7_MASKS_SERVER): Likewise.
10237 (ISA_3_0_MASKS_IEEE): Likewise.
10238 (OTHER_P8_VECTOR_MASKS): Likewise.
10239 (OTHER_VSX_VECTOR_MASKS): Likewise.
10240 (POWERPC_MASKS): Likewise.
10241 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10242 duplicate list of options.
10243 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10244 explicit -mupper-regs options.
10245 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10246 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10247 alias for -mupper-regs-df.
10248 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10249 (rs6000_init_hard_regno_mode_ok): Likewise.
10250 (rs6000_option_override_internal): Likewise.
10251 (rs6000_opt_masks): Likewise.
10252 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10253 options in terms of whether -mvsx or -mpower8-vector was used.
10254 (TARGET_UPPER_REGS_DI): Likewise.
10255 (TARGET_UPPER_REGS_SF): Likewise.
10256 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10257 -mupper-regs-* options.
10259 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10261 * passes.c (emergency_dump_function): Print some empty lines and a
10262 header before the RTL dump.
10264 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10266 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10268 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10271 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10272 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10274 2017-07-24 Carl Love <cel@us.ibm.com>
10276 * config/rs6000/rs6000-c.c: Add support for built-in functions
10277 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10278 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10279 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10280 vec_extract_fp_from_shortl): Add defines for the two builtins.
10281 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10282 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10284 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10285 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10286 * doc/extend.texi: Update the built-in documentation file for the
10287 new built-in function.
10289 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10292 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10294 * doc/generic.texi: Likewise.
10295 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10296 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10298 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10300 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10301 (aarch64_mls_elt_merge<mode>): Likewise.
10303 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10305 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10306 having __cxa_atexit.
10308 2017-07-23 Michael Collison <michael.collison@arm.com>
10310 * config/arm/arm.c (arm_option_override): Deprecate
10311 use of -mstructure-size-boundary.
10312 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10313 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10315 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10318 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10319 Reduce cost estimate for direct moves.
10321 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10324 * config/i386/i386.c (ix86_option_override_internal): Disable
10325 BMI, BMI2 and TBM instructions for -m16.
10327 2017-07-21 Carl Love <cel@us.ibm.com>
10329 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10330 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10331 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10332 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10333 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10334 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10335 VMULOSW): New enum "unspec" values.
10336 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10337 altivec_vmulosw): New patterns.
10338 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10339 VMULOSW): Add definitions.
10341 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10343 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10344 (qdf24xx): Likewise.
10345 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10346 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10347 (AARCH64_FL_V8_1): Renumber.
10348 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10349 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10350 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10351 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10352 rdma to feature modifiers list.
10354 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10356 PR middle-end/56727
10357 * ipa-visibility (function_and_variable_visibility): Convert
10358 recursive PLT call to direct call if appropriate.
10360 2017-07-21 Andrew Pinski <apinski@cavium.com>
10362 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10363 operand 1 to see if the types precision matches.
10364 * fold-const.c (operand_equal_p): Likewise.
10366 2017-07-21 Richard Biener <rguenther@suse.de>
10368 PR tree-optimization/81303
10369 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10370 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10371 (vect_peeling_hash_get_lowest_cost): Adjust.
10372 (vect_enhance_data_refs_alignment): Likewise. Use
10373 vect_get_peeling_costs_all_drs to compute the penalty for no
10374 peeling to match up costs.
10376 2017-07-21 Richard Biener <rguenther@suse.de>
10378 PR tree-optimization/81500
10379 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10380 we didn't identify a reduction path.
10382 2017-07-21 Tom de Vries <tom@codesourcery.com>
10383 Cesar Philippidis <cesar@codesourcery.com>
10385 PR gcov-profile/81442
10386 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10389 2017-07-21 Tom de Vries <tom@codesourcery.com>
10392 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10394 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10395 nvptx_override_options_after_change.
10397 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10399 * dwarf2out.c (output_file_names): Avoid double testing for
10400 dwarf_version >= 5.
10402 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10404 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10406 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10408 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10410 (try_crossjump_to_edge): Do not punt on partitioned functions.
10412 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10414 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10415 Put all BBs reachable only via paths crossing cold region to cold
10417 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10419 2016-07-21 Richard Biener <rguenther@suse.de>
10421 PR tree-optimization/81303
10422 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10423 into account prologue and epilogue iterations when raising
10424 min_profitable_iters to sth at least covering one vector iteration.
10426 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10428 * config/arm/arm.c (arm_test_cpu_arch_dat):
10431 2017-07-20 Nathan Sidwell <nathan@acm.org>
10433 Remove TYPE_METHODS.
10434 * tree.h (TYPE_METHODS): Delete.
10435 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10436 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10437 (dbxout_type_methods): Scan TYPE_FIELDS.
10438 (dbxout_type): Don't check TYPE_METHODS here.
10439 * function.c (use_register_for_decl): Always ignore register for
10440 class types when not optimizing.
10441 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10442 * tree.c (free_lang_data_in_type): Stitch out member functions and
10443 templates from TYPE_FIELDS.
10444 (build_distinct_type_copy, verify_type_variant,
10445 verify_type): Member fns are on TYPE_FIELDS.
10446 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10447 * tree-pretty-print.c (dump_generic_node): Likewise.
10449 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10452 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10453 V2TImode and V4TImode.
10454 (ix86_expand_vector_extract): Likewise.
10455 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10456 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10457 (ssescalarmode): Handle V4TImode and V2TImode.
10458 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10459 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10460 (VEXTRACTI128_MODE): New mode iterator.
10461 (splitter for *vec_extractv?ti first element): New.
10462 (VEC_INIT_MODE): New mode iterator.
10463 (vec_init<mode>): Consolidate 3 expanders into one using
10464 VEC_INIT_MODE mode iterator.
10466 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10468 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10469 non_spilled_static_chain_regno_p.
10471 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10473 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10475 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10477 * bb-reorder.c (connect_traces): Allow copying of blocks within
10480 2017-07-20 Richard Biener <rguenther@suse.de>
10482 * gimple.h (gimple_phi_result): Add gphi * overload.
10483 (gimple_phi_result_ptr): Likewise.
10484 (gimple_phi_arg): Likewise. Adjust index assert to only
10485 allow actual argument accesses rather than all slots available
10487 (gimple_phi_arg_def): Add gphi * overload.
10488 * tree-phinodes.c (make_phi_node): Initialize only actual
10490 (resize_phi_node): Clear memory not covered by old node,
10491 do not initialize excess argument slots.
10492 (reserve_phi_args_for_new_edge): Initialize new argument slot
10495 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10497 PR tree-optimization/81388
10499 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10501 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10502 by removing computation of may_be_zero.
10504 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10505 Tom de Vries <tom@codesourcery.com>
10507 PR middle-end/81030
10508 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10509 when gimple level profile disagrees with what RTL expander did.
10511 2017-07-20 Richard Biener <rguenther@suse.de>
10513 PR tree-optimization/61171
10514 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10515 (vect_analyze_stmt): Add slp instance parameter.
10516 (vectorizable_reduction): Likewise.
10517 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10518 (vect_is_simple_reduction): Deal with chains not detected
10519 as SLP reduction chain, specifically not properly associated
10520 chains containing a mix of plus/minus.
10521 (get_reduction_op): Remove.
10522 (get_initial_defs_for_reduction): Simplify, pass in whether
10523 this is a reduction chain, pass in the SLP node for the PHIs.
10524 (vect_create_epilog_for_reduction): Get the SLP instance as
10526 (vectorizable_reduction): Get the SLP instance as arg.
10527 During analysis remember the SLP node with the PHIs in the
10528 instance. Simplify getting at the vectorized reduction PHIs.
10529 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10530 through SLP instance.
10531 (vect_slp_analyze_operations): Likewise.
10532 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10533 (vect_transform_stmt): Likewise.
10535 2017-07-20 Tom de Vries <tom@codesourcery.com>
10537 PR tree-optimization/81489
10538 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10539 read of phi arg location to before loop that modifies phi.
10541 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10543 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10546 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10548 PR middle-end/81331
10549 * except.c (execute): Fix ordering issue.
10551 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10553 PR rtl-optimization/81423
10554 * combine.c (make_compound_operation_int): Don't try to optimize
10555 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10557 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10559 PR rtl-optimization/81423
10560 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10561 with a constant that is -1 in the truncated to mode.
10563 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10565 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10566 (determine_unlikely_bbs): ... here.
10567 * predict.h (propagate_unlikely_bbs_forward): Declare.
10568 * cfgexpand.c (pass_expand::execute): Use it.
10569 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10571 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10572 propagate_unlikely_bbs_forward.
10574 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10576 PR middle-end/81331
10577 * except.c (maybe_add_nop_after_section_switch): New function.
10580 2017-07-19 Tom de Vries <tom@codesourcery.com>
10582 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10584 2017-07-19 Tom de Vries <tom@codesourcery.com>
10586 * gimple.h (gimple_phi_arg): Make assert more strict.
10588 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10590 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10591 * config/rs6000/mmintrin.h: New file.
10592 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10594 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10596 PR tree-optimization/81346
10597 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10599 2017-07-19 Tom de Vries <tom@codesourcery.com>
10601 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10603 2017-07-19 Tom de Vries <tom@codesourcery.com>
10605 * config/nvptx/nvptx-modes.def: Add V2DImode.
10606 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10607 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10608 (nvptx_output_mov_insn): Handle lack of mov.b128.
10609 (nvptx_print_operand): Handle 'H' and 'L' codes.
10610 (nvptx_vector_mode_supported): Allow V2DImode.
10611 (nvptx_preferred_simd_mode): New function.
10612 (nvptx_data_alignment): New function.
10613 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10614 nvptx_preferred_simd_mode.
10615 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10617 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10619 2017-07-19 Tom de Vries <tom@codesourcery.com>
10621 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10622 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10623 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10624 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10625 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10626 (mov<VECIM>_insn): New define_insn.
10627 (define_expand "mov<VECIM>): New define_expand.
10629 2017-07-19 Tom de Vries <tom@codesourcery.com>
10631 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10633 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10635 PR tree-optimization/81346
10636 * fold-const.h (fold_div_compare, range_check_type): Declare.
10637 * fold-const.c (range_check_type): New function.
10638 (build_range_check): Use range_check_type.
10639 (fold_div_compare): No longer static, rewritten into
10640 a match.pd helper function.
10641 (fold_comparison): Don't call fold_div_compare here.
10642 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10643 as helper function.
10645 2017-07-19 Nathan Sidwell <nathan@acm.org>
10647 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10648 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10649 * tree.c (find_decls_types_r, verify_type): Use
10650 TYPE_{MIN,MAX}_VALUE_RAW.
10651 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10652 (hash_tree): Likewise.
10653 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10655 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10658 2017-07-18 Tom de Vries <tom@codesourcery.com>
10660 PR middle-end/81464
10661 * omp-expand.c (expand_omp_for_static_chunk): Handle
10662 equal-argument loop exit phi.
10664 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10667 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10668 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10669 operand 2 predicate.
10670 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10671 operand 2 predicate.
10672 (ror,rol -> rorx splitters): Use const_int_operand as
10673 operand 2 predicate.
10675 2017-06-18 Richard Biener <rguenther@suse.de>
10677 PR tree-optimization/81410
10678 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10679 the gap in the ! slp_perm SLP case after each group.
10681 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10683 PR middle-end/81463
10684 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10687 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10689 PR middle-end/81462
10690 * predict.c (set_even_probabilities): Cleanup; do not affect
10691 probabilities that are already known.
10692 (combine_predictions_for_bb): Call even when count is set.
10694 2017-07-18 Nathan Sidwell <nathan@acm.org>
10696 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10699 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10702 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10703 optimization for loop niter analysis.
10705 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10708 * config/avr/avr.c (avr_optimize_casesi): Don't use
10709 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10711 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10713 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10714 body_cost_vec from _vect_peel_extended_info.
10715 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10716 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10719 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10721 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10723 2017-07-18 Richard Biener <rguenther@suse.de>
10725 PR tree-optimization/80620
10726 PR tree-optimization/81403
10727 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10728 info when re-using a VN table entry.
10730 2017-07-18 Richard Biener <rguenther@suse.de>
10732 PR tree-optimization/81418
10733 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10734 vectype_in. Verify that with lane-reducing reduction operations
10735 we have a single def-use cycle.
10737 2017-07-17 Carl Love <cel@us.ibm.com>
10739 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10742 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10743 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10744 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10745 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10746 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10747 VMULOSW): New enum "unspec" values.
10748 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10749 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10750 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10751 altivec_vmulosw): New patterns.
10752 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10753 VMULOSW): Add definitions.
10754 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10756 * config/alpha/alpha.c: Include predict.h.
10758 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10760 * tree-vrp.c (compare_assert_loc): Fix comparison function
10761 to return predictable results.
10763 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10765 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10767 (subdi3): Likewise.
10768 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10769 * doc/invoke.texi (mexpand-adddi): Update text.
10771 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10773 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10774 that also clobbers the CC register. The old expand code is moved
10776 (*arc_clzsi2): ... here.
10777 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10778 the CC register. The old expand code is moved to ...
10779 (arc_ctzsi2): ... here.
10781 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10783 * config/arc/arc.opt (mindexed-loads): Use initial value
10784 TARGET_INDEXED_LOADS_DEFAULT.
10785 (mauto-modify-reg): Use initial value
10786 TARGET_AUTO_MODIFY_REG_DEFAULT.
10787 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10788 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10789 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10790 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10792 2017-07-17 Martin Liska <mliska@suse.cz>
10795 * opts.c (finish_options): Do not allow -fgnu-tm
10796 w/ -fsanitize={kernel-,}address. Say sorry.
10798 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10801 * tree-loop-distribution.c (classify_partition): Only assert on
10802 numer of iterations.
10803 (merge_dep_scc_partitions): Delete prameter. Update function call.
10804 (distribute_loop): Remove code handling loop with unknown niters.
10805 (pass_loop_distribution::execute): Skip loop with unknown niters.
10807 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10810 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10811 function sort_partitions_by_post_order.
10813 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10815 PR tree-optimization/81374
10816 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10817 the max index of basic blocks, rather than number of basic blocks.
10819 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10821 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10823 (arc_legitimate_pic_operand_p): Likewise.
10824 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10826 (arc_needs_pcl_p): Likewise.
10827 (arc_legitimate_pc_offset_p): Likewise.
10828 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10829 function is also used in constrains.md.
10830 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10831 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10832 PLUS. Only return true/false in known cases, otherwise assert.
10833 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10834 is already called in arc_legitimate_constant_p.
10835 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10837 (LEGITIMATE_PIC_OPERAND_P): Use
10838 arc_raw_symbolic_reference_mentioned_p function.
10839 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10844 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10845 Andrew Burgess <andrew.burgess@embecosm.com>
10847 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10848 (arc_return_address_register): New function.
10849 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10850 (arc_handle_fndecl_attribute): Add naked attribute.
10851 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10852 (TARGET_WARN_FUNC_RETURN): Likewise.
10853 (arc_allocate_stack_slots_for_args): New function.
10854 (arc_warn_func_return): Likewise.
10855 (machine_function): Change type fn_type.
10856 (arc_compute_function_type): Consider new naked function type,
10857 change function return type.
10858 (arc_must_save_register): Adapt to handle new
10859 arc_compute_function_type's return type.
10860 (arc_expand_prologue): Likewise.
10861 (arc_expand_epilogue): Likewise.
10862 (arc_return_address_regs): Delete.
10863 (arc_return_address_register): New function.
10864 (arc_epilogue_uses): Use above function.
10865 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10866 (arc_function_type): Change encoding, add naked type.
10867 (ARC_INTERRUPT_P): Change to handle the new encoding.
10868 (ARC_FAST_INTERRUPT_P): Likewise.
10869 (ARC_NORMAL_P): Define.
10870 (ARC_NAKED_P): Likewise.
10871 (arc_compute_function_type): Delete prototype.
10872 * config/arc/arc.md (in_ret_delay_slot): Use
10873 arc_return_address_register function.
10874 (simple_return): Likewise.
10875 (p_return_i): Likewise.
10877 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10879 PR tree-optimization/81428
10880 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10881 can't be built for those types.
10883 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10885 Remove stuff dead since r239246.
10887 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10888 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10889 (avr_inform_devices): Remove dead stuff.
10891 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10893 * config/arm/arm_neon.h: Fix softp typo.
10895 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10897 PR tree-optimization/81365
10898 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10899 aggregate moves onto bb predecessor edges, make sure there are no
10900 loads that could alias the lhs in between the start of bb and the
10903 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10906 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10907 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10908 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10910 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10912 PR tree-optimization/81396
10913 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10914 (init_symbolic_number): Initialize it to 1.
10915 (perform_symbolic_merge): Add n_ops from both operands into the new
10917 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10918 without base_addr as useless if they need more than one operation.
10919 (bswap_replace): Handle !bswap case for NULL base_addr.
10921 2017-07-17 Tom de Vries <tom@codesourcery.com>
10924 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10927 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10929 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10930 conditional builtin define __FIX_LEON3FT_B2BST.
10932 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10934 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10935 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10938 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10940 PR rtl-optimization/81424
10941 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10942 to remove potential trapping from operands if -fnon-call-exceptions.
10944 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10946 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10947 profile_proability for scalling.
10948 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10950 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10952 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10954 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10956 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10957 fixpoint arithmetics.
10959 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10961 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10962 fixpoint arithmetics.
10964 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10966 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10967 fixpoint arithmetics.
10969 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10971 * profile-count.h (profile_probability::from_reg_br_prob_note,
10972 profile_probability::to_reg_br_prob_note): New functions.
10973 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10974 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10975 * predict.c (probability_reliable_p): Update.
10976 (edge_probability_reliable_p): Update.
10977 (br_prob_note_reliable_p): Update.
10978 (invert_br_probabilities): Update.
10979 (add_reg_br_prob_note): New function.
10980 (combine_predictions_for_insn): Update.
10981 * asan.c (asan_clear_shadow): Update.
10982 * cfgbuild.c (compute_outgoing_frequencies): Update.
10983 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10984 (update_br_prob_note): Update.
10985 (rtl_verify_edges): Update.
10986 (purge_dead_edges): Update.
10987 (fixup_reorder_chain): Update.
10988 * emit-rtl.c (try_split): Update.
10989 * ifcvt.c (cond_exec_process_insns): Update.
10990 (cond_exec_process_if_block): Update.
10991 (dead_or_predicable): Update.
10992 * internal-fn.c (expand_addsub_overflow): Update.
10993 (expand_neg_overflow): Update.
10994 (expand_mul_overflow): Update.
10995 * loop-doloop.c (doloop_modify): Update.
10996 * loop-unroll.c (compare_and_jump_seq): Update.
10997 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10998 * predict.h: Update.
10999 * reorg.c (mostly_true_jump): Update.
11001 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11002 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11003 * config/arc/arc.c: (emit_unlikely_jump): Update.
11004 * config/arm/arm.c: (emit_unlikely_jump): Update.
11005 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11006 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11007 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11008 (ix86_print_operand): Update.
11009 (ix86_split_fp_branch): Update.
11010 (predict_jump): Update.
11011 * config/ia64/ia64.c (ia64_print_operand): Update.
11012 * config/mmix/mmix.c (mmix_print_operand): Update.
11013 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11014 (rs6000_expand_split_stack_prologue): Update.
11015 * config/rs6000/rs6000.c: Update.
11016 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11017 (s390_expand_vec_movstr): Update.
11018 (s390_expand_cs_tdsi): Update.
11019 (s390_expand_split_stack_prologue): Update.
11020 * config/sh/sh.c (sh_print_operand): Update.
11021 (expand_cbranchsi4): Update.
11022 (expand_cbranchdi4): Update.
11023 * config/sparc/sparc.c (output_v9branch): Update.
11024 * config/spu/spu.c (get_branch_target): Update.
11025 (ea_load_store_inline): Update.
11026 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11027 * config/tilepro/tilepro.c: Update.
11029 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11031 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11032 (gimplify_save_expr): Likewise.
11034 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11036 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11038 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11040 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11042 * timevar.def (TV_IPA_FNSUMMARY): Define.
11044 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11046 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11047 to back store errata sensitive sequence from being generated.
11048 (sqrtdf2_fix): Likewise.
11050 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11052 * tree-ssa-threadupdate.c (compute_path_counts,
11053 update_joiner_offpath_counts): Use profile_probability.
11055 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11058 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11060 * config/arm/arm-c.c (arm_cpu_builtins): Define
11061 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11063 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11065 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11066 array entries to represent __ieee128 versions of the
11067 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11068 scalar_extract_sig, and scalar_insert_exp built-in functions.
11069 (altivec_resolve_overloaded_builtin): Add special case handling
11070 for the __builtin_scalar_insert_exp function, as represented by
11071 the P9V_BUILTIN_VEC_VSIEDP constant.
11072 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11073 exponent support for __ieee128 argument.
11074 (VSESQP): Add scalar extract signature support for __ieee128
11076 (VSTDCNQP): Add scalar test negative support for __ieee128
11078 (VSIEQP): Add scalar insert exponent support for __int128 argument
11079 with __ieee128 result.
11080 (VSIEQPF): Add scalar insert exponent support for __ieee128
11081 argument with __ieee128 result.
11082 (VSTDCQP): Add scalar test data class support for __ieee128
11084 (VSTDCNQP): Add overload support for scalar test negative with
11085 __ieee128 argument.
11086 (VSTDCQP): Add overload support for scalar test data class
11087 __ieee128 argument.
11088 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11089 UNSPEC_VSX_SXSIGDP.
11090 (UNSPEC_VSX_SIEXPQP): New constant.
11091 (xsxexpqp): New insn for VSX scalar extract exponent quad
11093 (xsxsigqp): New insn for VSX scalar extract significand quad
11095 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11096 precision with floating point argument.
11097 (xststdcqp): New expand for VSX scalar test data class quad
11099 (xststdcnegqp): New expand for VSX scalar test negative quad
11101 (xststdcqp): New insn to match expansions for VSX scalar test data
11102 class quad precision and VSX scalar test negative quad precision.
11103 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11104 special case operand checking to enforce that second operand of
11105 VSX scalar test data class with quad precision argument is a 7-bit
11107 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11108 prototypes and descriptions of __ieee128 versions of
11109 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11110 scalar_test_data_class, and scalar_test_neg built-in functions.
11112 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11114 PR tree-optimization/81162
11115 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11116 replace a negate with an add.
11118 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11120 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11122 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11124 * config/arm/arm-c.c (arm_cpu_builtins): Define
11125 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11127 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11129 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11130 (armv8-r): Set ARM Cortex-R52 as default CPU.
11131 * config/arm/arm-tables.opt: Regenerate.
11132 * config/arm/arm-tune.md: Regenerate.
11133 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11135 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11136 extension for -mcpu=cortex-r52.
11138 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11140 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11141 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11142 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11143 (fp-armv8): Define it as FP_ARMv8 only.
11144 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11145 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11147 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11148 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11149 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11150 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11151 than TARGET_FPU_ARMV8.
11152 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11153 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11154 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11156 * config/arm/neon.md (neon_vrint): Likewise.
11157 (neon_vcvt): Likewise.
11158 (neon_<fmaxmin_op><mode>): Likewise.
11159 (<fmaxmin><mode>3): Likewise.
11160 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11161 * config/arm/predicates.md (arm_cond_move_operator): Check against
11162 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11164 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11166 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11167 to top of function.
11169 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11171 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11172 loop in comment with memset.
11174 2017-07-14 Martin Liska <mliska@suse.cz>
11176 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11177 * dwarf2out.c (is_java): Remove the function.
11178 (output_pubname): Remove usage of the function.
11179 (lower_bound_default): Remove usage of DW_LANG_Java.
11180 (gen_compile_unit_die): Likewise.
11181 * gcc.c: Remove compiler defaults for .java and .zip files.
11182 * gimple-expr.c (remove_suffix): Change as there's no longer
11183 extension than 4-letter one.
11184 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11185 (gimplify_save_expr): Likewise.
11186 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11187 as it's possible even for other languages than Java.
11188 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11189 * lto-opts.c (lto_write_options): Remove reference to Java.
11190 * opts.c (strip_off_ending): Update file extension handling.
11191 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11192 * tree-eh.c (lower_resx): Likewise.
11193 * tree.c (free_lang_data_in_type): Remove dead code.
11194 (find_decls_types_r): Likewise.
11195 (build_common_builtin_nodes): Remove Java from a comment.
11196 (verify_type): Remove dead code.
11197 * varasm.c (assemble_external): Remove Java from a comment.
11199 2017-07-14 Martin Liska <mliska@suse.cz>
11201 * opts.c (finish_options): Add quotes.
11202 (common_handle_option): Likewise.
11204 2017-07-14 Martin Liska <mliska@suse.cz>
11206 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11207 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11208 Remove N_SO_PASCAL.
11209 * dwarf2out.c (lower_bound_default): Do not handle
11211 (gen_compile_unit_die): Likewise.
11212 * gcc.c: Remove default extension binding for GNU Pascal.
11213 * stmt.c: Remove Pascal language from a comment.
11214 * xcoffout.c: Likewise.
11216 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11219 * diagnostic-show-locus.c (fixit_cmp): New function.
11220 (layout::layout): Sort m_fixit_hints.
11221 (column_range::column_range): Assert that the values are valid.
11222 (struct char_span): New struct.
11223 (correction::overwrite): New method.
11224 (struct source_line): New struct.
11225 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11226 calls in terms of classes source_line and char_span, and
11227 correction::overwrite.
11228 (selftest::test_overlapped_fixit_printing_2): New function.
11229 (selftest::diagnostic_show_locus_c_tests): Call it.
11231 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11233 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11234 early if there is no lhs.
11236 2017-07-13 Martin Liska <mliska@suse.cz>
11238 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11239 (gen_reference_type_die): Likewise.
11240 * stor-layout.c: Remove Pascal-related comment.
11242 2017-07-13 Martin Liska <mliska@suse.cz>
11244 * opts.c (finish_options): Add quotes to error messages.
11245 (parse_sanitizer_options): Likewise.
11247 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11249 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11251 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11253 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11255 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11257 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11259 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11261 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11264 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11265 provides the hardware capability bits, define the macro
11266 __BUILTIN_CPU_SUPPORTS__.
11267 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11268 if GLIBC does not provide the hardware capability bits. Add a
11269 gcc_unreachable call if the built-in cpu function is neither
11270 __builtin_cpu_is nor __builtin_cpu_supports.
11271 (rs6000_get_function_versions_dispatcher): Change the warning
11272 that an old GLIBC is used which does not export the capability
11273 bits to be an error.
11274 * doc/extend.texi (target_clones attribute): Document the
11275 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11276 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11277 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11278 the macros defined by GCC if the newer GLIBC is available.
11280 2017-07-12 Jeff Law <law@redhat.com>
11282 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11283 remaining includes slightly.
11284 * config/riscv/riscv-builtins.c: Include profile-count.h.
11286 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11289 * config/avr/avr.c (avr_set_current_function): In diagnostic
11290 messages: Quote keywords and (parts of) identifiers.
11291 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11294 2017-07-12 Carl Love <cel@us.ibm.com>
11296 * config/rs6000/rs6000-c.c: Add support for built-in functions
11297 vector bool char vec_revb (vector bool char);
11298 vector bool short vec_revb (vector short char);
11299 vector bool int vec_revb (vector bool int);
11300 vector bool long long vec_revb (vector bool long long);
11301 * doc/extend.texi: Update the built-in documentation file for the
11302 new built-in functions.
11304 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11306 * config/s390/s390.md: Remove movcc splitter.
11308 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11310 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11311 load/store on condition.
11313 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11316 * config/avr/avr.c (avr_encode_section_info)
11317 [progmem && !TREE_READONLY]: Error if progmem object needs
11320 2017-07-11 Michael Collison <michael.collison@arm.com>
11322 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11325 2017-07-11 Carl Love <cel@us.ibm.com>
11327 * config/rs6000/rs6000-c.c: Add support for builtins
11328 vector unsigned int vec_parity_lsbb (vector signed int);
11329 vector unsigned int vec_parity_lsbb (vector unsigned int);
11330 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11331 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11332 vector unsigned long long vec_parity_lsbb (vector signed long long);
11333 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11334 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11335 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11336 * doc/extend.texi: Update the built-in documentation file for the
11337 new built-in functions.
11339 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11341 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11342 (layout::m_primary_loc): New field.
11343 (layout::layout): Initialize new field. Move location filtering
11344 logic from here to...
11345 (layout::maybe_add_location_range): ...this new method. Add
11346 support for filtering to just the lines already specified by other
11348 (layout::will_show_line_p): New method.
11349 (gcc_rich_location::add_location_if_nearby): New method.
11350 (selftest::test_add_location_if_nearby): New test function.
11351 (selftest::diagnostic_show_locus_c_tests): Call it.
11352 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11355 2017-07-11 Tom de Vries <tom@codesourcery.com>
11357 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11358 (bb_first_real_insn): New function.
11359 (nvptx_single): Add extra initialization of broadcasted condition
11362 2017-07-11 Nathan Sidwell <nathan@acm.org>
11364 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11366 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11368 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11369 Binutils doc as TEXI will mess them up.
11370 * doc/invoke.texi (AVR Options): Same here.
11372 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11374 * config/sparc/sparc.opt (mfix-ut700): New option.
11375 (mfix-gr712rc): Likewise.
11376 (sparc_fix_b2bst): New variable.
11377 * doc/invoke.texi (SPARC options): Document them.
11378 (ARM options): Fix warnings.
11379 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11380 instructions to prevent sequences that can trigger the store-store
11381 errata for certain LEON3FT processors.
11382 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11383 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11384 * config/sparc/sparc.md (fix_b2bst): New attribute.
11385 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11387 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11390 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11392 (*rsqrtsf2_sse): Ditto.
11394 (div<mode>3): Macroize insn from divdf3 and divsf3
11395 using MODEF mode iterator.
11397 2017-07-10 Martin Sebor <msebor@redhat.com>
11399 PR tree-optimization/80397
11400 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11401 instead of testing for equality to INTEGER_TYPE.
11403 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11405 * config.gcc: Remove uclibc from arc target spec.
11407 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11409 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11411 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11414 * lto-wrapper.c (remove_option): New function.
11415 (merge_and_complain): Merge PIC/PIE options more realistically.
11417 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11419 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11423 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11424 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11425 * config.in: Regenerate.
11426 * configure: Regenerate.
11427 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11428 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11429 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11430 (TARGET_GASISR_PROLOGUES): ...target mask.
11431 * common/config/avr/avr-common.c
11432 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11433 Set -mgas-isr-prologues.
11434 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11435 INSERT_PASS_BEFORE for it.
11436 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11437 * config/avr/avr.c (avr_option_override)
11438 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11439 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11440 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11441 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11442 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11444 (make_avr_pass_pre_proep): New function.
11445 (emit_push_sfr) <treg>: Add argument to function and use it
11446 instead of TMP_REG.
11447 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11448 and set machine->gasisr.yes.
11449 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11450 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11451 __gcc_isr.n_pushed to .L__stack_usage.
11452 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11453 (avr_asm_final_postscan_insn): ...this new static function.
11454 * config/avr/avr.h (machine_function)
11455 <is_no_gccisr, use_L__stack_usage>: New fields.
11456 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11457 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11458 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11459 (gasisr, *gasisr): New expander and insn.
11460 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11461 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11462 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11464 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11466 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11469 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11471 Move jump-tables out of .text again.
11474 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11475 (ASM_OUTPUT_ADDR_VEC): New function.
11476 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11477 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11478 INSN_ADDRESSes as asm comment.
11479 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11480 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11481 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11482 * config/avr/avr.md (*tablejump): Adjust comment.
11483 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11484 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11486 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11487 (avr_output_addr_vec): New proto.
11488 (avr_log_t) <insn_addresses>: New field.
11490 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11493 * config/i386/i386.c (ix86_function_arg_advance): Set
11494 outgoing_args_on_stack to true if there are outgoing arguments
11496 (ix86_function_arg): Likewise.
11497 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11498 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11499 * config/i386/i386.h (machine_function): Add
11500 outgoing_args_on_stack.
11502 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11504 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11505 supporting pthreds.
11506 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11508 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11510 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11511 (REAL_H): Remove $(MACHMODE_H).
11512 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11514 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11515 $(MACHMODE_H) and double-int.h.
11516 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11518 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11521 2017-07-07 Andrew Pinski <apinski@cavium.com>
11523 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11524 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11526 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11528 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11529 Add warning if GCC was not configured to link against a GLIBC that
11530 exports the hardware capability bits.
11531 (make_resolver_func): Make resolver function private and not a
11532 COMDAT function. Create the name with clone_function_name instead
11533 of make_unique_name.
11536 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11537 correct operand in doing the split.
11539 2017-07-07 Carl Love <cel@us.ibm.com>
11541 * config/rs6000/rs6000-c: Add support for built-in function
11542 vector unsigned short vec_pack_to_short_fp32 (vector float,
11544 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11545 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11546 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11547 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11548 (convert_4f32_8i16): Add define_expand.
11549 * doc/extend.texi: Update the built-in documentation file for the
11550 new built-in function.
11552 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11554 * config/sparc/m8.md: New file.
11555 * config/sparc/sparc.md: Include m8.md.
11557 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11559 * config/sparc/sparc.opt: New option -mvis4b.
11560 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11561 (sparc_option_override): Handle VIS4B.
11562 (enum sparc_builtins): Define
11563 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11564 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11565 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11566 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11567 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11568 (check_constant_argument): New function.
11569 (sparc_vis_init_builtins): Define builtins
11570 __builtin_vis_dictunpack{8,16,32},
11571 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11572 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11573 __builtin_vis_fpcmpde{8,16,32}shl and
11574 __builtin_vis_fpcmpur{8,16,32}shl.
11575 (sparc_expand_builtin): Check that the constant operands to
11576 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11577 constant and in range.
11578 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11580 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11581 (SPARC_IMM5_P): Likewise.
11582 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11583 (enabled): Handle vis4b.
11584 (UNSPEC_DICTUNPACK): New unspec.
11585 (UNSPEC_FPCMPSHL): Likewise.
11586 (UNSPEC_FPUCMPSHL): Likewise.
11587 (UNSPEC_FPCMPDESHL): Likewise.
11588 (UNSPEC_FPCMPURSHL): Likewise.
11589 (cpu_feature): New CPU feature `vis4b'.
11590 (dictunpack{8,16,32}): New insns.
11591 (FPCSMODE): New mode iterator.
11592 (fpcscond): New code iterator.
11593 (fpcsucond): Likewise.
11594 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11595 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11596 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11597 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11598 * config/sparc/constraints.md: Define constraints `q' for unsigned
11599 2-bit integer constants and `t' for unsigned 5-bit integer
11601 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11603 (imm5_operand_dictunpack16): Likewise.
11604 (imm5_operand_dictunpack32): Likewise.
11605 (imm2_operand): Likewise.
11606 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11607 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11608 ditunpack* and fpcmp*shl builtins.
11610 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11612 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11613 * config.in: Add HAVE_AS_SPARC6 define.
11614 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11616 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11618 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11619 (CPP_CPU_SPEC): Handle m8.
11620 (ASM_CPU_SPEC): Likewise.
11621 * config/sparc/sparc-opts.h (enum processor_type): Add
11623 * config/sparc/sparc.c (m8_costs): New struct.
11624 (sparc_option_override): Handle TARGET_CPU_m8.
11625 (sparc32_initialize_trampoline): Likewise.
11626 (sparc64_initialize_trampoline): Likewise.
11627 (sparc_issue_rate): Likewise.
11628 (sparc_register_move_cost): Likewise.
11629 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11630 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11631 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11632 (CPP_CPU_SPEC): Handle M8.
11633 (ASM_CPU_SPEC): Likewise.
11634 (AS_M8_FLAG): Define.
11635 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11636 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11637 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11639 * configure: Regenerate.
11640 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11643 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11645 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11647 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11648 ("*movdi_insn_sp32"): Do not set v3pipe.
11649 ("*movsi_insn"): Likewise.
11650 ("*movdi_insn_sp64"): Likewise.
11651 ("*movsf_insn"): Likewise.
11652 ("*movdf_insn_sp32"): Likewise.
11653 ("*movdf_insn_sp64"): Likewise.
11654 ("*zero_extendsidi2_insn_sp64"): Likewise.
11655 ("*sign_extendsidi2_insn"): Likewise.
11656 ("*mov<VM32:mode>_insn"): Likewise.
11657 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11658 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11659 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11660 ("<vlop:code><VL:mode>3"): Likewise.
11661 ("*not_<vlop:code><VL:mode>3"): Likewise.
11662 ("*nand<VL:mode>_vis"): Likewise.
11663 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11664 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11665 ("one_cmpl<VL:mode>2"): Likewise.
11666 ("faligndata<VM64:mode>_vis"): Likewise.
11667 ("alignaddrsi_vis"): Likewise.
11668 ("alignaddrdi_vis"): Likweise.
11669 ("alignaddrlsi_vis"): Likewise.
11670 ("alignaddrldi_vis"): Likewise.
11671 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11672 ("bmaskdi_vis"): Likewise.
11673 ("bmasksi_vis"): Likewise.
11674 ("bshuffle<VM64:mode>_vis"): Likewise.
11675 ("cmask8<P:mode>_vis"): Likewise.
11676 ("cmask16<P:mode>_vis"): Likewise.
11677 ("cmask32<P:mode>_vis"): Likewise.
11678 ("pdistn<P:mode>_vis"): Likewise.
11679 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11681 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11683 * config/sparc/sparc.md ("subtype"): New insn attribute.
11684 ("*wrgsr_sp64"): Set insn subtype.
11685 ("*rdgsr_sp64"): Likewise.
11686 ("alignaddrsi_vis"): Likewise.
11687 ("alignaddrdi_vis"): Likewise.
11688 ("alignaddrlsi_vis"): Likewise.
11689 ("alignaddrldi_vis"): Likewise.
11690 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11691 ("fexpand_vis"): Likewise.
11692 ("fpmerge_vis"): Likewise.
11693 ("faligndata<VM64:mode>_vis"): Likewise.
11694 ("bshuffle<VM64:mode>_vis"): Likewise.
11695 ("cmask8<P:mode>_vis"): Likewise.
11696 ("cmask16<P:mode>_vis"): Likewise.
11697 ("cmask32<P:mode>_vis"): Likewise.
11698 ("fchksm16_vis"): Likewise.
11699 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11700 ("fmean16_vis"): Likewise.
11701 ("fp<plusminus_insn>64_vis"): Likewise.
11702 ("<plusminus_insn>v8qi3"): Likewise.
11703 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11704 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11705 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11706 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11707 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11708 ("*movqi_insn"): Likewise.
11709 ("*movhi_insn"): Likewise.
11710 ("*movsi_insn"): Likewise.
11711 ("movsi_pic_gotdata_op"): Likewise.
11712 ("*movdi_insn_sp32"): Likewise.
11713 ("*movdi_insn_sp64"): Likewise.
11714 ("movdi_pic_gotdata_op"): Likewise.
11715 ("*movsf_insn"): Likewise.
11716 ("*movdf_insn_sp32"): Likewise.
11717 ("*movdf_insn_sp64"): Likewise.
11718 ("*zero_extendhisi2_insn"): Likewise.
11719 ("*zero_extendqihi2_insn"): Likewise.
11720 ("*zero_extendqisi2_insn"): Likewise.
11721 ("*zero_extendqidi2_insn"): Likewise.
11722 ("*zero_extendhidi2_insn"): Likewise.
11723 ("*zero_extendsidi2_insn_sp64"): Likewise.
11724 ("ldfsr"): Likewise.
11725 ("prefetch_64"): Likewise.
11726 ("prefetch_32"): Likewise.
11727 ("tie_ld32"): Likewise.
11728 ("tie_ld64"): Likewise.
11729 ("*tldo_ldub_sp32"): Likewise.
11730 ("*tldo_ldub1_sp32"): Likewise.
11731 ("*tldo_ldub2_sp32"): Likewise.
11732 ("*tldo_ldub_sp64"): Likewise.
11733 ("*tldo_ldub1_sp64"): Likewise.
11734 ("*tldo_ldub2_sp64"): Likewise.
11735 ("*tldo_ldub3_sp64"): Likewise.
11736 ("*tldo_lduh_sp32"): Likewise.
11737 ("*tldo_lduh1_sp32"): Likewise.
11738 ("*tldo_lduh_sp64"): Likewise.
11739 ("*tldo_lduh1_sp64"): Likewise.
11740 ("*tldo_lduh2_sp64"): Likewise.
11741 ("*tldo_lduw_sp32"): Likewise.
11742 ("*tldo_lduw_sp64"): Likewise.
11743 ("*tldo_lduw1_sp64"): Likewise.
11744 ("*tldo_ldx_sp64"): Likewise.
11745 ("*mov<VM32:mode>_insn"): Likewise.
11746 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11747 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11749 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11751 * config/sparc/sparc.md ("type"): New insn type viscmp.
11752 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11754 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11755 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11756 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11757 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11759 ("n7_vis_logical_11cycle"): Likewise.
11760 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11761 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11762 * config/sparc/niagara.md ("niag_vis"): Likewise.
11763 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11764 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11766 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11768 * config/sparc/sparc.md: New instruction type `bmask'.
11769 (bmaskdi_vis): Use the `bmask' type.
11770 (bmasksi_vis): Likewise.
11771 * config/sparc/ultra3.md (us3_array): Likewise.
11772 * config/sparc/niagara7.md (n7_array): Likewise.
11773 * config/sparc/niagara4.md (n4_array): Likewise.
11774 * config/sparc/niagara2.md (niag2_vis): Likewise.
11775 (niag3_vis): Likewise.
11776 * config/sparc/niagara.md (niag_vis): Likewise.
11778 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11780 * ipa-comdats.c: Remove optimize check from gate.
11781 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11782 for functions not optimized.
11783 (ipa_fn_summary_read): Skip optimize check.
11784 (ipa_fn_summary_write): Likewise.
11785 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11787 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11789 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11790 (check_callers): Give up on caller which is not optimized.
11791 (inline_small_functions): Likewise.
11792 (ipa_inline): Do not give up when not optimizing.
11793 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11794 away unoptimizes cdtors.
11795 (whole_program_function_and_variable_visibility): Do
11796 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11797 * ipa.c (process_references): Do not check optimize.
11798 (symbol_table::remove_unreachable_nodes): Update optimize check.
11799 (set_writeonly_bit): Update optimize check.
11800 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11801 (pass_ipa_single_use::gate): Remove.
11803 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11805 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11806 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11807 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11808 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11809 permute_load, permute_store, adjust_extract, adjust_splat,
11810 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11811 replace_swap_with_copy, dump_swap_insn_table,
11812 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11813 recombine_lvx_pattern, recombine_stvx_pattern,
11814 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11815 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11816 to file rs6000-p8swap.c.
11817 * config/rs6000/rs6000-p8swap.c: New file.
11818 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11819 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11820 and rs6000*-*-* targets.
11822 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11824 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11826 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11828 * lto-wrapper.c (merge_and_complain): Do not merge
11829 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11830 fsigned_zeros, ftrapping_math, fwrapv.
11831 (append_compiler_options): Do not track these options.
11832 (append_linker_options): Likewie
11834 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11836 * cgraphunit.c (cgraph_node::finalize_function): When
11837 !flag_toplevel_reorde set no_reorder flag.
11838 (varpool_node::finalize_decl): Likewise.
11839 (symbol_table::compile): Drop no toplevel reorder path.
11841 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11843 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11844 edges; zero probability is not better than uninitialized.
11846 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11848 * asan.h (asan_sanitize_allocas_p): Declare.
11849 * asan.c (asan_sanitize_allocas_p): New function.
11850 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11851 (handle_builtin_alloca): Likewise.
11852 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11853 if !asan_sanitize_allocas_p.
11854 * params.def (asan-instrument-allocas): Add new option.
11855 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11856 * opts.c (common_handle_option): Disable allocas sanitization for
11859 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11861 * asan.c: Include gimple-fold.h.
11862 (get_last_alloca_addr): New function.
11863 (handle_builtin_stackrestore): Likewise.
11864 (handle_builtin_alloca): Likewise.
11865 (asan_emit_allocas_unpoison): Likewise.
11866 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11867 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11868 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11869 (instrument_builtin_call): Pass gimple iterator to
11870 get_mem_refs_of_builtin_call.
11871 (last_alloca_addr): New global.
11872 * asan.h (asan_emit_allocas_unpoison): Declare.
11873 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11874 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11875 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11876 if function calls alloca.
11877 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11878 * gimple-fold.h (replace_call_with_value): Declare.
11879 * internal-fn.c: Include asan.h.
11880 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11881 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11883 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11885 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11886 (C_SELFTEST_FLAGS): New.
11887 (CPP_SELFTEST_FLAGS): New.
11888 (SELFTEST_DEPS): New, from deps of s-selftest.
11889 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11890 (CPP_SELFTEST_DEPS): New.
11891 (selftest): Add dependency on s-selftest-c++.
11892 (s-selftest): Rename to...
11893 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11894 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11895 than SELFTEST_FLAGS.
11896 (selftest-gdb): Rename to...
11897 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11899 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11900 (selftest-valgrind): Rename to...
11901 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11903 (selftest-valgrind): Reintroduce as an alias for
11904 selftest-c-valgrind.
11905 (s-selftest-c++): New.
11906 (selftest-c++-gdb): New.
11907 (selftest-c++-valgrind): New.
11909 2017-07-06 Olivier Hainque <hainque@adacore.com>
11911 * gcc.c (process_command): When deciding if undefined variables
11912 should be ignored when processing specs, accept "gcc -v" as well.
11914 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11916 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11917 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11919 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11921 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11922 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11923 * config/arm/arm-tables.opt: Regenerate.
11924 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11926 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11928 2017-07-06 Carl Love <cel@us.ibm.com>
11930 * ChangeLog: Clean up from mid air collision
11932 2017-07-06 Carl Love <cel@us.ibm.com>
11934 * config/rs6000/rs6000-c.c: Add support for built-in functions
11935 vector signed int vec_subc (vector signed int, vector signed int);
11936 vector signed __int128 vec_subc (vector signed __int128,
11937 vector signed __int128);
11938 vector unsigned __int128 vec_subc (vector unsigned __int128,
11939 vector unsigned __int128);
11940 vector signed int vec_sube (vector signed int, vector signed int,
11941 vector signed int);
11942 vector unsigned int vec_sube (vector unsigned int,
11943 vector unsigned int,
11944 vector unsigned int);
11945 vector signed __int128 vec_sube (vector signed __int128,
11946 vector signed __int128,
11947 vector signed__int128);
11948 vector unsigned __int128 vec_sube (vector unsigned __int128,
11949 vector unsigned __int128,
11950 vector unsigned __int128);
11951 vector signed int vec_subec (vector signed int, vector signed int,
11952 vector signed int);
11953 vector unsigned int vec_subec (vector unsigned int,
11954 vector unsigned int,
11955 vector unsigned int);
11956 vector signed __int128 vec_subec (vector signed __int128,
11957 vector signed __int128,
11958 vector signed__int128);
11959 vector unsigned __int128 vec_subec (vector unsigned __int128,
11960 vector unsigned __int128,
11961 vector unsigned __int128);
11962 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11963 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11964 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11965 BU_ALTIVEC_OVERLOAD_X definitions.
11966 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11967 * doc/extend.texi: Update the built-in documentation file for the new
11968 built-in functions.
11970 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11973 * diagnostic-show-locus.c (layout::layout): Use start and finish
11974 spelling location for the start and finish of each range.
11975 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11976 Add unused aspect param.
11977 * input.c (expand_location_1): Add "aspect" param, and use it
11978 to access the correct part of the location.
11979 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11981 (expand_location_to_spelling_point): Likewise.
11982 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11983 param, and pass it to expand_location_1.
11985 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11987 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11988 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11989 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11990 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11991 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11992 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11993 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11994 _mm_maskz_getmant_ss): New intrinsics.
11995 (__builtin_ia32_getexpss128_mask): Changed to ...
11996 __builtin_ia32_getexpss128_round ... this.
11997 (__builtin_ia32_getexpsd128_mask): Changed to ...
11998 __builtin_ia32_getexpsd128_round ... this.
11999 * config/i386/i386-builtin-types.def
12000 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12001 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12002 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12003 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12004 __builtin_ia32_getmantss_mask_round): New builtins.
12005 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12006 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12007 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12008 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12009 * config/i386/sse.md
12010 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12011 avx512f_sgetexp<mode><mask_scalar_name>
12012 <round_saeonly_scalar_name> ... this.
12013 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12014 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12015 vgetexp<ssescalarmodesuffix>
12016 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12017 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12018 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12019 avx512f_vgetmant<mode><mask_scalar_name>
12020 <round_saeonly_scalar_name> ... this.
12021 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12022 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12023 vgetmant<ssescalarmodesuffix>
12024 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12025 %0<mask_scalar_operand4>, %1, %2
12026 <round_saeonly_scalar_mask_op4>, %3} ... this.
12027 * config/i386/subst.md (mask_scalar_operand4,
12028 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12029 round_saeonly_scalar_nimm_predicate): New subst attributes.
12031 2017-07-06 Julia Koval <julia.koval@intel.com>
12033 * config/i386/i386.c (ix86_erase_embedded_rounding):
12034 Remove code for old rounding pattern.
12036 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12038 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12040 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12042 * doc/sourcebuild.texi (Test Directives, Variants of
12043 dg-require-support): Add documentation for dg-require-stack-check.
12045 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12047 * config/i386/subst.md (mask_scalar, round_scalar,
12048 round_saeonly_scalar): New meta-templates.
12049 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12050 round_scalar_mask_operand3, round_scalar_mask_op3,
12051 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12052 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12053 round_saeonly_scalar_constraint,
12054 round_saeonly_scalar_prefix): New subst attribute.
12055 * config/i386/sse.md
12056 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12057 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12058 <round_scalar_name> ... this.
12059 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12060 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12061 <round_scalar_name> ... this.
12062 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12063 <sse>_vm<code><mode>3<mask_scalar_name>
12064 <round_saeonly_scalar_name> ... this.
12065 (v<plusminus_mnemonic><ssescalarmodesuffix>
12066 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12067 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12068 v<plusminus_mnemonic><ssescalarmodesuffix>
12069 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12070 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12071 (v<multdiv_mnemonic><ssescalarmodesuffix>
12072 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12073 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12074 v<multdiv_mnemonic><ssescalarmodesuffix>
12075 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12076 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12077 (v<maxmin_float><ssescalarmodesuffix>
12078 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12079 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12080 v<maxmin_float><ssescalarmodesuffix>
12081 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12082 %0<mask_scalar_operand3>, %1, %<iptr>2
12083 <round_saeonly_scalar_mask_op3>} ... this.
12085 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12087 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12088 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12090 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12091 Alan Hayward <alan.hayward@arm.com>
12092 David Sherwood <david.sherwood@arm.com>
12094 * combine.c (simplify_if_then_else): Remove "enum" before
12096 * compare-elim.c (can_eliminate_compare): Likewise.
12097 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12099 (aarch64_lookup_simd_builtin_type): Likewise.
12100 (aarch64_simd_builtin_type): Likewise.
12101 (aarch64_init_simd_builtin_types): Likewise.
12102 (aarch64_simd_expand_args): Likewise.
12103 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12105 (aarch64_reverse_mask): Likewise.
12106 (aarch64_simd_emit_reg_reg_move): Likewise.
12107 (aarch64_gen_adjusted_ldpstp): Likewise.
12108 (aarch64_ccmp_mode_to_code): Likewise.
12109 (aarch64_operands_ok_for_ldpstp): Likewise.
12110 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12111 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12113 (aarch64_min_divisions_for_recip_mul): Likewise.
12114 (aarch64_reassociation_width): Likewise.
12115 (aarch64_get_condition_code_1): Likewise.
12116 (aarch64_simd_emit_reg_reg_move): Likewise.
12117 (aarch64_simd_attr_length_rglist): Likewise.
12118 (aarch64_reverse_mask): Likewise.
12119 (aarch64_operands_ok_for_ldpstp): Likewise.
12120 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12121 (aarch64_gen_adjusted_ldpstp): Likewise.
12122 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12124 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12125 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12126 (arm_lookup_simd_builtin_type): Likewise.
12127 (arm_simd_builtin_type): Likewise.
12128 (arm_init_simd_builtin_types): Likewise.
12129 (arm_expand_builtin_args): Likewise.
12130 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12131 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12132 (ft32_setup_incoming_varargs): Likewise.
12133 (ft32_function_arg): Likewise.
12134 (ft32_function_arg_advance): Likewise.
12135 (ft32_pass_by_reference): Likewise.
12136 (ft32_arg_partial_bytes): Likewise.
12137 (ft32_valid_pointer_mode): Likewise.
12138 (ft32_addr_space_pointer_mode): Likewise.
12139 (ft32_addr_space_legitimate_address_p): Likewise.
12140 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12142 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12143 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12144 (iamcu_alignment): Likewise.
12145 (canonicalize_vector_int_perm): Likewise.
12146 (ix86_noce_conversion_profitable_p): Likewise.
12147 (ix86_mpx_bound_mode): Likewise.
12148 (ix86_operands_ok_for_move_multiple): Likewise.
12149 * config/microblaze/microblaze-protos.h
12150 (microblaze_expand_conditional_branch_reg): Likewise.
12151 * config/microblaze/microblaze.c
12152 (microblaze_expand_conditional_branch_reg): Likewise.
12153 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12155 (rs6000_reassociation_width): Likewise.
12156 (rs6000_invalid_binary_op): Likewise.
12157 (fusion_p9_p): Likewise.
12158 (emit_fusion_p9_load): Likewise.
12159 (emit_fusion_p9_store): Likewise.
12160 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12162 (riscv_hard_regno_mode_ok_p): Likewise.
12163 (riscv_address_insns): Likewise.
12164 (riscv_split_symbol): Likewise.
12165 (riscv_legitimize_move): Likewise.
12166 (riscv_function_value): Likewise.
12167 (riscv_hard_regno_nregs): Likewise.
12168 (riscv_expand_builtin): Likewise.
12169 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12170 (riscv_build_integer): Likewise.
12171 (riscv_split_integer): Likewise.
12172 (riscv_legitimate_constant_p): Likewise.
12173 (riscv_cannot_force_const_mem): Likewise.
12174 (riscv_regno_mode_ok_for_base_p): Likewise.
12175 (riscv_valid_base_register_p): Likewise.
12176 (riscv_valid_offset_p): Likewise.
12177 (riscv_valid_lo_sum_p): Likewise.
12178 (riscv_classify_address): Likewise.
12179 (riscv_legitimate_address_p): Likewise.
12180 (riscv_address_insns): Likewise.
12181 (riscv_load_store_insns): Likewise.
12182 (riscv_force_binary): Likewise.
12183 (riscv_split_symbol): Likewise.
12184 (riscv_force_address): Likewise.
12185 (riscv_legitimize_address): Likewise.
12186 (riscv_move_integer): Likewise.
12187 (riscv_legitimize_const_move): Likewise.
12188 (riscv_legitimize_move): Likewise.
12189 (riscv_address_cost): Likewise.
12190 (riscv_subword): Likewise.
12191 (riscv_output_move): Likewise.
12192 (riscv_canonicalize_int_order_test): Likewise.
12193 (riscv_emit_int_order_test): Likewise.
12194 (riscv_function_arg_boundary): Likewise.
12195 (riscv_pass_mode_in_fpr_p): Likewise.
12196 (riscv_pass_fpr_single): Likewise.
12197 (riscv_pass_fpr_pair): Likewise.
12198 (riscv_get_arg_info): Likewise.
12199 (riscv_function_arg): Likewise.
12200 (riscv_function_arg_advance): Likewise.
12201 (riscv_arg_partial_bytes): Likewise.
12202 (riscv_function_value): Likewise.
12203 (riscv_pass_by_reference): Likewise.
12204 (riscv_setup_incoming_varargs): Likewise.
12205 (riscv_print_operand): Likewise.
12206 (riscv_elf_select_rtx_section): Likewise.
12207 (riscv_save_restore_reg): Likewise.
12208 (riscv_for_each_saved_reg): Likewise.
12209 (riscv_register_move_cost): Likewise.
12210 (riscv_hard_regno_mode_ok_p): Likewise.
12211 (riscv_hard_regno_nregs): Likewise.
12212 (riscv_class_max_nregs): Likewise.
12213 (riscv_memory_move_cost): Likewise.
12214 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12215 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12216 (rl78_addr_space_address_mode): Likewise.
12217 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12219 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12220 (rs6000_reassociation_width): Likewise.
12221 (rs6000_invalid_binary_op): Likewise.
12222 (fusion_p9_p): Likewise.
12223 (emit_fusion_p9_load): Likewise.
12224 (emit_fusion_p9_store): Likewise.
12225 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12226 (ok_for_simple_move_operands): Likewise.
12227 (ok_for_simple_move_strict_operands): Likewise.
12228 (ok_for_simple_arith_logic_operands): Likewise.
12229 (visium_legitimize_reload_address): Likewise.
12230 (visium_select_cc_mode): Likewise.
12231 (output_cbranch): Likewise.
12232 (visium_split_double_move): Likewise.
12233 (visium_expand_copysign): Likewise.
12234 (visium_expand_int_cstore): Likewise.
12235 (visium_expand_fp_cstore): Likewise.
12236 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12237 (visium_function_arg): Likewise.
12238 (visium_function_arg_advance): Likewise.
12239 (visium_libcall_value): Likewise.
12240 (visium_setup_incoming_varargs): Likewise.
12241 (visium_legitimate_constant_p): Likewise.
12242 (visium_legitimate_address_p): Likewise.
12243 (visium_legitimize_address): Likewise.
12244 (visium_secondary_reload): Likewise.
12245 (visium_register_move_cost): Likewise.
12246 (visium_memory_move_cost): Likewise.
12247 (prepare_move_operands): Likewise.
12248 (ok_for_simple_move_operands): Likewise.
12249 (ok_for_simple_move_strict_operands): Likewise.
12250 (ok_for_simple_arith_logic_operands): Likewise.
12251 (visium_function_value_1): Likewise.
12252 (rtx_ok_for_offset_p): Likewise.
12253 (visium_legitimize_reload_address): Likewise.
12254 (visium_split_double_move): Likewise.
12255 (visium_expand_copysign): Likewise.
12256 (visium_expand_int_cstore): Likewise.
12257 (visium_expand_fp_cstore): Likewise.
12258 (visium_split_cstore): Likewise.
12259 (visium_select_cc_mode): Likewise.
12260 (visium_split_cbranch): Likewise.
12261 (output_cbranch): Likewise.
12262 (visium_print_operand_address): Likewise.
12263 * expmed.c (flip_storage_order): Likewise.
12264 * expmed.h (emit_cstore): Likewise.
12265 (flip_storage_order): Likewise.
12266 * genrecog.c (validate_pattern): Likewise.
12267 * hsa-gen.c (gen_hsa_addr): Likewise.
12268 * internal-fn.c (expand_arith_overflow): Likewise.
12269 * ira-color.c (allocno_copy_cost_saving): Likewise.
12270 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12271 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12272 (process_invariant_for_inheritance): Likewise.
12273 * lra-eliminations.c (move_plus_up): Likewise.
12274 * omp-low.c (lower_oacc_reductions): Likewise.
12275 * simplify-rtx.c (simplify_subreg): Likewise.
12276 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12277 (TARGET_CHKP_BOUND_MODE): Likewise..
12278 * targhooks.c (default_chkp_bound_mode): Likewise.
12279 (default_setup_incoming_vararg_bounds): Likewise.
12280 * targhooks.h (default_chkp_bound_mode): Likewise.
12281 (default_setup_incoming_vararg_bounds): Likewise.
12282 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12283 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12284 (have_whole_vector_shift): Likewise.
12285 * tree-vect-stmts.c (vectorizable_load): Likewise.
12286 * doc/tm.texi: Regenerate.
12288 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12290 Graceful degrade if Binutils PR21472 is not available.
12293 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12294 .rodata in flash test fails.
12295 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12296 * confgure: Regenerate.
12297 * config.in: Regenerate.
12298 * config/avr/avr.c (avr_asm_named_section)
12299 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12300 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12301 (avr_asm_init_sections): Same.
12303 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12305 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12306 (fma<VH:mode>4_intrinsic): Likewise.
12307 (*fmsub<VCVTF:mode>4): Likewise.
12308 (*fmsub<VH:mode>4_intrinsic): Likewise.
12310 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12313 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12314 Don't depend on "optimize > 0".
12315 (out_movhi_r_mr, out_movqi_mr_r): Same.
12316 (out_movhi_mr_r, out_movqi_r_mr): Same.
12317 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12318 io_address_operand on "optimize > 0".
12320 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12322 * tree-loop-distribution.c: Add general explanantion on the pass.
12323 (generate_loops_for_partition): Mark distributed loop.
12324 (pg_add_dependence_edges): New parameter. Handle alias data
12325 dependence specially and record it in the parameter if asked.
12326 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12327 (init_partition_graph_vertices, add_partition_graph_edge): New.
12328 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12329 (free_partition_graph_vdata, build_partition_graph): New.
12330 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12331 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12332 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12333 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12334 (version_for_distribution_p, finalize_partitions): New.
12335 (distribute_loop): Handle alias data dependence specially. Factor
12336 out loop fusion code as functions and call these functions.
12338 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12340 * tree-loop-distribution.c (classify_partition): New parameter and
12341 better handle reduction statement.
12342 (rdg_build_partitions): Revise comment.
12343 (distribute_loop): Compute statements in all partitions and pass it
12344 to classify_partition.
12346 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12348 * tree-loop-distribution.c (enum partition_type): New.
12349 (struct partition): New field type.
12350 (partition_merge_into): Add parameter. Update partition type.
12351 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12352 (build_rdg_partition_for_vertex): Compute partition type.
12353 (rdg_build_partitions): Dump partition type.
12354 (distribute_loop): Update calls to partition_merge_into.
12356 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12358 * tree-loop-distribution.c (struct ddr_hasher): New.
12359 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12361 (classify_partition): Call get_data_dependence.
12362 (pg_add_dependence_edges): Ditto.
12363 (distribute_loop): Release data dependence hash table.
12365 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12367 * tree-loop-distribution.c (ref_base_address): Delete.
12368 (similar_memory_accesses): Rename ...
12369 (share_memory_accesses): ... to this. Check if partitions access
12370 the same memory reference.
12371 (distribute_loop): Call share_memory_accesses.
12373 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12375 * tree-loop-distribution.c (struct partition): New field recording
12376 its data reference.
12377 (partition_alloc, partition_free): Init and release data refs.
12378 (partition_merge_into): Merge data refs.
12379 (build_rdg_partition_for_vertex): Collect data refs for partition.
12380 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12382 (distribute_loop): Remve data refs from vertice data of partition
12385 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12387 * tree-loop-distribution.c (params.h): Include header file.
12388 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12389 (datarefs_vec): New global var.
12390 (create_rdg_vertices): Use datarefs_vec directly.
12391 (free_rdg): Don't free data references.
12392 (build_rdg): Update use. Don't free data references.
12393 (distribute_loop): Compute global variable for data references.
12394 Bail out if there are too many data references.
12396 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12398 * tree-loop-distribution.c (loop_nest): New global var.
12399 (build_rdg): Use loop directly, rather than loop nest.
12400 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12402 (distribute_loop): Compute global variable loop nest. Update use.
12404 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12406 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12407 (partition_merge_into): New parameter. Dump reason for fusion.
12408 (distribute_loop): Update use of partition_merge_into.
12410 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12412 * tree-loop-distribution.c (bb_top_order_index): New.
12413 (bb_top_order_index_size, bb_top_order_cmp): New.
12414 (stmts_from_loop): Use topological order.
12415 (pass_loop_distribution::execute): Compute and release topological
12416 order for basic blocks.
12418 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12420 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12423 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12425 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12426 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12427 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12428 * internal-fn.def (LOOP_DIST_ALIAS): New.
12429 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12430 (fold_loop_internal_call): ... this.
12431 (vect_loop_dist_alias_call): New function.
12432 (set_uid_loop_bbs): Call fold_loop_internal_call.
12433 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12436 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12439 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12440 Require dead FLAGS_REG at the beginning of a peephole.
12442 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12445 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12446 arguments in the call to __builtin_ia32_sbb_u32.
12447 (_subborrow_u64): Swap _X and _Y arguments in the call to
12448 __builtin_ia32_sbb_u64.
12450 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12453 * tree-vrp.c (compare_assert_loc): Turn into a function template
12454 with stable template parameter. Only test if a->e is NULL,
12455 !a->e == !b->e has been verified already. Use e == NULL or
12456 e != NULL instead of e or ! e tests. If stable is true, don't use
12457 iterative_hash_expr, on the other side allow a or b or both NULL
12458 and sort the NULLs last.
12459 (process_assert_insertions): Sort using compare_assert_loc<false>
12460 instead of compare_assert_loc, later sort using
12461 compare_assert_loc<true> before calling process_assert_insertions_for
12462 in a loop. Use break instead of continue once seen NULL pointer.
12464 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12466 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12467 Cortex-R7 and Cortex-R8 processors.
12469 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12471 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12472 uninitialized while src is not.
12474 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12476 * common/config/arm/arm-common.c: Adjust include path for
12478 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12479 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12480 (arm-cpu-data.h): Likewise.
12481 (arm-cpu-cdata.h): Likewise.
12482 * config/arm/arm-cpu.h: Delete.
12483 * config/arm/arm-cpu-cdata.h: Delete.
12484 * config/arm/arm-cpu-data.h: Delete.
12486 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12488 * config/arm/arm-cpus.in (cortex-a55): New.
12489 (cortex-a75): Likewise.
12490 (cortex-a75.cortex-a55): Likewise.
12491 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12493 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12494 * config/arm/arm-cpu-cdata.h: Regenerate.
12495 * config/arm/arm-cpu-data.h: Regenerate.
12496 * config/arm/arm-cpu.h: Regenerate.
12497 * config/arm/arm-tables.opt: Regenerate.
12498 * config/arm/arm-tune.md: Regenerate.
12500 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12502 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12504 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12506 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12509 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12511 PR tree-optimization/81292
12512 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12513 full_string_p, also call adjust_related_strinfos if the adjustment
12514 is simple, otherwise invalidate related strinfos.
12516 2017-07-04 Martin Liska <mliska@suse.cz>
12519 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12520 newly created variable as DECL_IGNORED_P.
12522 2017-07-04 Martin Liska <mliska@suse.cz>
12525 * ipa-inline.c (inline_small_functions):
12526 Use xstrdup_for_dump.
12528 2017-07-04 Tom de Vries <tom@codesourcery.com>
12530 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12532 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12535 * config/darwin.c (darwin_function_switched_text_sections):
12538 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12540 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12542 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12544 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12546 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12548 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12549 min_profitable_iters, and th as inclusive lower bounds.
12550 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12551 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12552 for min_profitable_iters and min_profitable_estimate.
12553 (vect_transform_loop): Treat th as an inclusive lower bound.
12554 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12556 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12559 * config/darwin.c (darwin_function_switched_text_sections):
12560 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12561 in two pieces, and suppress the use of buf.
12563 2017-07-03 Nathan Sidwell <nathan@acm.org>
12565 * hash-table.h (hash_table_mod1): Fix indentation.
12567 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12569 PR middle-end/81290
12570 * predict.c (force_edge_cold): Be more careful about propagation
12572 * profile-count.h (profile_probability::guessed,
12573 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12575 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12577 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12579 * doc/invoke.texi (rcpc architecture extension): Document it.
12581 2017-07-03 Richard Biener <rguenther@suse.de>
12583 PR tree-optimization/60510
12584 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12585 the scalar reduction PHI and use it.
12586 (vectorizable_reduction): Properly guard the single_defuse_cycle
12587 path for non-SLP reduction chains where we cannot use it.
12588 Rework reduc_def/index and vector type deduction. Rework
12589 vector operand gathering during reduction op code-gen.
12590 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12591 chains dissolve the chain and leave it to non-SLP reduction
12594 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12596 * tree-data-ref.h (dr_alignment): Declare.
12597 * tree-data-ref.c (dr_alignment): New function.
12598 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12599 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12601 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12603 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12605 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12606 and base_misalignment fields.
12607 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12608 * tree-data-ref.c: Include builtins.h.
12609 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12610 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12611 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12612 * tree-vect-data-refs.c: Include tree-cfg.h.
12613 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12614 fields instead of calculating an alignment here.
12615 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12616 innermost_loop_behavior fields.
12618 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12620 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12622 (DR_STEP_ALIGNMENT): New macro.
12623 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12624 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12625 (create_data_ref): Print it.
12626 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12627 to tell whether the step preserves vector (mis)alignment.
12628 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12629 Move the check for an integer step and generalise to all INTEGER_CST.
12630 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12631 Print the outer step alignment.
12633 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12635 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12636 with offset_alignment.
12637 (DR_ALIGNED_TO): Delete.
12638 (DR_OFFSET_ALIGNMENT): New macro.
12639 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12640 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12641 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12642 (create_data_ref): Likewise.
12643 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12644 (vect_analyze_data_refs): Likewise.
12645 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12646 creating dummy innermost behavior.
12648 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12650 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12651 with a "innermost_loop_behavior *" and refeence tree.
12652 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12653 (create_data_ref): Update call accordingly.
12654 * tree-predcom.c (find_looparound_phi): Likewise.
12656 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12658 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12659 fields with dr_wrt_vec_loop.
12660 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12661 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12662 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12663 (vect_dr_behavior): New function.
12664 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12665 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12666 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12667 track whether the step preserves the misalignment.
12668 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12669 Use vect_dr_behavior.
12670 (vect_setup_realignment): Update call accordingly.
12671 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12672 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12673 call to vect_create_addr_base_for_vector_ref.
12674 (vect_create_cond_for_align_checks): Likewise.
12675 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12676 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12677 (vect_recog_mask_conversion_pattern): Likewise.
12678 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12679 (new_stmt_vec_info): Remove redundant zeroing.
12681 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12683 * common/config/arm/arm-common.c (arm_be8_option): New function.
12684 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12685 (ISA_ARMv6): Add isa_bit_be8.
12686 * config/arm/arm.h (arm_be8_option): Add prototype.
12687 (BE8_SPEC_FUNCTION): New define.
12688 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12689 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12690 (mlittle-endian): Similarly.
12691 (mbe8, mbe32): New options.
12692 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12693 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12695 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12697 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12699 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12701 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12702 (cleanup_tree_cfg_bb): Use it.
12703 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12705 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12707 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12710 * loop-doloop.c (add_test): Update profile.
12712 2017-07-03 Martin Liska <mliska@suse.cz>
12715 * sanopt.c (rewrite_usage_of_param): New function.
12716 (sanitize_rewrite_addressable_params): Likewise.
12717 (pass_sanopt::execute): Call rewrite_usage_of_param.
12719 2017-07-03 Richard Biener <rguenther@suse.de>
12721 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12722 back to using VIEW_CONVERT_EXPR.
12724 2017-07-03 Martin Liska <mliska@suse.cz>
12727 * doc/extend.texi: Document when a resolver function is
12728 generated for target_clones.
12730 2017-07-03 Martin Liska <mliska@suse.cz>
12732 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12733 and shadow memory of auto variables which are subject of
12734 use-after-scope sanitization.
12735 (asan_expand_mark_ifn): Add do set only when is_poison.
12737 2016-07-03 Richard Biener <rguenther@suse.de>
12739 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12741 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12742 (vectorizable_reduction): Perform reduction PHI creation when
12743 visiting a reduction PHI and adjust and simplify code generation
12744 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12745 (vect_transform_loop): Visit reduction PHIs.
12746 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12747 defs into the SLP tree.
12748 (vect_build_slp_tree): Reduction defs terminate the recursion.
12749 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12751 (vect_get_vec_defs_for_stmt_copy): Export.
12752 (vect_get_vec_defs): Likewise.
12753 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12755 (vect_get_vec_defs_for_stmt_copy): Declare.
12756 (vect_get_vec_defs): Likewise.
12758 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12760 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12761 parameter with a "loop" parameter and use it instead of the
12762 loop containing DR_STMT. Don't check simple_iv when doing
12763 BB analysis. Describe the two analysis modes in the comment.
12765 2017-07-03 Tom de Vries <tom@codesourcery.com>
12767 PR tree-optimization/69468
12768 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12769 (find_same_succ_bb): Handle ignore_edge_flags.
12771 2017-07-03 Tom de Vries <tom@codesourcery.com>
12773 PR tree-optimization/81192
12774 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12776 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12778 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12780 2017-07-03 Tom de Vries <tom@codesourcery.com>
12782 PR tree-optimization/81192
12783 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12784 BB_SAME_SUCC (bb) == NULL.
12786 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12788 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12791 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12793 * dumpfile.c: Include profile-count.h
12794 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12796 (insert_cond_bb): Update profile.
12797 * tree-cfg.h (insert_cond_bb): Update prototype.
12798 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12799 * tree-dump.c: Do not include tree-cfg.
12801 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12803 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12805 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12807 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12810 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12812 * tree-complex.c (expand_complex_div_wide): update profile.
12814 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12815 Alan Hayward <alan.hayward@arm.com>
12816 David Sherwood <david.sherwood@arm.com>
12818 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12819 (CORETYPES_H): New define.
12820 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12821 (insn-modes-inline.h, s-modes-inline-h): New rules.
12822 (generated_files): Add insn-modes-inline.h.
12823 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12824 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12825 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12826 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12827 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12828 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12829 (build/gencodes.o, build/genconditions.o): Likewise.
12830 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12831 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12832 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12833 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12834 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12835 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12836 * coretypes.h: Include everything up to real.h for generators.
12837 Include insn-modes.h first. Include wide-int-print.h after
12838 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12839 * machmode.h: Don't include insn-modes.h here.
12840 * function-tests.c: Remove includes of signop.h, machmode.h,
12841 double-int.h and wide-int.h.
12843 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12845 * optc-save-gen.awk: Likewise.
12846 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12847 * godump.c: Remove include of wide-int-print.h.
12848 * pretty-print.h: Likewise.
12849 * wide-int-print.cc: Likewise.
12850 * wide-int.cc: Likewise.
12851 * hash-map-tests.c: Remove include of signop.h.
12852 * hash-set-tests.c: Likewise.
12853 * rtl-tests.c: Likewise.
12854 * mkconfig.sh: Remove include of machmode.h.
12855 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12857 (emit_insn_modes_inline_h): ...this new function. Emit the code
12858 into an insn-modes-inline.h header file, adding appropriate
12859 include guards and end comments.
12860 (emit_insn_modes_c_header): Remove include of machmode.h.
12861 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12863 (main): Handle -i flag and call emit_insn_modes_inline_h when
12866 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12868 * tree-ssa-strlen.c (strinfo): Rename the length field to
12869 nonzero_chars. Add a full_string_p field.
12870 (compare_nonzero_chars, zero_length_string_p): New functions.
12871 (get_addr_stridx): Add an offset_out parameter.
12872 Use compare_nonzero_chars.
12873 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12874 (new_strinfo): Update after above changes to strinfo.
12875 (set_endptr_and_length): Set full_string_p.
12876 (get_string_length): Update after above changes to strinfo.
12877 (unshare_strinfo): Update call to new_strinfo.
12878 (maybe_invalidate): Likewise.
12879 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12880 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12881 as a uhwi instead of an shwi. Update after above changes to
12882 strinfo and new_strinfo.
12883 (zero_length_string): Assert that chainsi contains full strings.
12884 Use zero_length_string_p. Update call to new_strinfo.
12885 (adjust_related_strinfos): Update after above changes to strinfo.
12886 Copy full_string_p from origsi.
12887 (adjust_last_stmt): Use zero_length_string_p.
12888 (handle_builtin_strlen): Update after above changes to strinfo and
12889 new_strinfo. Install the lhs as the string length if the previous
12890 entry didn't describe a full string.
12891 (handle_builtin_strchr): Update after above changes to strinfo
12893 (handle_builtin_strcpy): Likewise.
12894 (handle_builtin_strcat): Likewise.
12895 (handle_builtin_malloc): Likewise.
12896 (handle_pointer_plus): Likewise.
12897 (handle_builtin_memcpy): Likewise. Track nonzero characters
12898 that aren't necessarily followed by a nul terminator.
12899 (handle_char_store): Likewise.
12901 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12903 PR tree-optimization/80769
12904 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12905 for malloc and calloc. Document the new invariant that all related
12906 strinfos have delayed lengths or none do.
12907 (verify_related_strinfos): Move earlier in file.
12908 (set_endptr_and_length): New function, split out from...
12909 (get_string_length): ...here. Also set the lengths of related
12911 (zero_length_string): Assert that chainsi has known (rather than
12913 (adjust_related_strinfos): Likewise.
12915 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12917 PR tree-optimization/81136
12918 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12919 assert that two references with the same misalignment have the same
12920 compile-time misalignment if those compile-time misalignments
12923 2017-07-01 Andi Kleen <ak@linux.intel.com>
12925 * print-tree.c (print_node): Print all attributes.
12927 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12929 * cfg.c (scale_bbs_frequencies): New function.
12930 * cfg.h (scale_bbs_frequencies): Declare it.
12931 * cfgloopanal.c (single_likely_exit): Cleanup.
12932 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12934 (scale_loop_profile): Likewise.
12935 (loop_version): Likewise.
12936 (create_empty_loop_on_edge): Update.
12937 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12938 scale_loop_frequencies, scale_loop_profile, loopify,
12939 loop_version): Update prototypes.
12940 * modulo-sched.c (sms_schedule): Update.
12941 * predict.c (unlikely_executed_edge_p): Also check probability.
12942 (probably_never_executed_edge_p): Fix typo.
12943 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12944 * tree-parloops.c (gen_parallel_loop): Update.
12945 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12946 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12947 * tree-ssa-loop-split.c (split_loop): Update.
12948 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12949 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12950 (vect_loop_versioning): Update.
12951 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12953 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12955 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12957 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12959 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12960 to keep profile consistent.
12962 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12964 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12965 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12966 * profile-count.h (max_safe_multiplier): Make unsigned.
12967 (profile_count::guessed_zero): New.
12969 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12971 * bb-reorder.c (fix_up_crossing_landing_pad,
12972 fix_crossing_conditional_branches): Use make_single_succ_edge
12973 to keep profile consistent.
12975 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12977 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12980 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12983 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12984 the right scopes, make sure cond_jump isn't preserved between multiple
12985 iterations. Search for fallthru edge whenever there are 3+ edges and
12986 use find_fallthru_edge for it.
12988 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12990 Patch by Alexander Monakov <amonakov@ispras.ru>
12991 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12992 probabilities consistently.
12994 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12996 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12997 emit_cmp_and_jump_insns.
12999 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13002 * tree-inline.c (expand_call_inline): Combine profile statuses.
13004 2017-06-30 Andrew Pinski <apinski@cavium.com>
13006 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13007 fold_stmt returned true.
13009 2017-06-30 Nathan Sidwell <nathan@acm.org>
13011 * ggc.h (empty_string): Delete.
13012 * cfgexpand.c (expand_asm_stmt): Use plain "".
13013 * optabs.c (expand_asm_memory_barrier): Likewise.
13014 * stringpool.c (empty_string): Delete.
13015 (digit_vector, digit_string): Delete.
13016 (ggc_alloc_string): Use plain "", don't optimize single digit
13017 strings. Use ggc_alloc_atomic.
13019 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13021 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13022 comparison set and one other set, use the cost of the non-comparison
13025 2017-06-30 Nathan Sidwell <nathan@acm.org>
13027 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13030 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13032 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13033 loops. Remove now unneeded calls to gimple_switch_set_label() that
13034 just set removed labels to NULL_TREE.
13036 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13038 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13039 (set_range_info): ...here. Only call set_range_info_raw if domain
13041 (set_nonzero_bits): Call set_range_info_raw.
13042 * tree-ssanames.h (set_range_info_raw): New.
13044 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13047 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13048 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13049 of nonimmediate_operand and <store_mask_constraint> instead of m
13050 for the input operand. For V8FI iterator, always split if input
13051 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13052 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13053 <store_mask_predicate> instead of register_operand and
13054 <store_mask_constraint> instead of v for the input operand. Make
13055 sure both operands aren't MEMs for if not <mask_applied>.
13057 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13059 * lto-wrapper.c (copy_file) Close both file descriptors before
13062 2017-06-30 Martin Liska <mliska@suse.cz>
13065 * multiple_target.c (create_dispatcher_calls): Make ifunc
13066 also for function that don't have calls or are not referenced.
13068 2017-06-30 Richard Biener <rguenther@suse.de>
13070 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13071 analyze the first scalar stmt. Move vector type computation
13072 for the BB case here from ...
13073 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13074 live operation processing in the SLP case properly.
13076 2017-06-30 Richard Biener <rguenther@suse.de>
13078 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13080 2017-06-30 Martin Liska <mliska@suse.cz>
13083 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13084 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13086 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13088 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13091 2017-06-30 Richard Biener <rguenther@suse.de>
13093 PR tree-optimization/81249
13094 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13095 condition reduction result to original scalar type.
13097 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13099 * profile-count.h (enum profile_quality): Fix typos and whitespace
13102 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13104 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13105 type for branch probabilities.
13107 2017-06-29 Julian Brown <julian@codesourcery.com>
13108 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13110 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13111 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13112 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13113 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13115 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13117 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13118 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13119 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13120 CC usage from generic code to here.
13121 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13122 CC usage into the target macros.
13124 2017-06-29 Maya Rashish <coypu@sdf.org>
13126 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13129 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13131 * arm/arm-builtins.c: Include profile-count.h
13132 * except.c (sjlj_emit_function_enter): Use
13133 profile_probability::unlikely.
13135 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13137 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13138 and tocrel_offset be pointer args rather than implicitly using
13140 (legitimate_constant_pool_address_p, rs6000_emit_move,
13141 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13142 tocrel_offset and use in toc_relative_expr_p call.
13143 (print_operand, print_operand_address): Use static tocrel_base_oac
13144 and tocrel_offset_oac.
13145 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13148 2017-06-29 Maya Rashish <coypu@sdf.org>
13150 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13152 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13154 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13155 objects, take into account only the alignment of 'op0' and 'mode1' if
13158 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13160 * ccmp.c (ccmp_tree_comparison_p): New function.
13161 (ccmp_candidate_p): Update to use above function.
13162 (get_compare_parts): New function.
13163 (expand_ccmp_next): Update to use new functions.
13164 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13166 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13167 take mode as argument.
13168 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13169 * expr.c (expand_expr_real_1): Pass mode as argument.
13171 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13173 * combine.c (combine_instructions): Print insns to dump_file, together
13176 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13178 * asan.c (asan_emit_stack_protection): Update.
13179 (create_cond_insert_point): Update.
13180 * auto-profile.c (afdo_propagate_circuit): Update.
13181 * basic-block.h (struct edge_def): Turn probability to
13182 profile_probability.
13183 (EDGE_FREQUENCY): Update.
13184 * bb-reorder.c (find_traces_1_round): Update.
13185 (better_edge_p): Update.
13186 (sanitize_hot_paths): Update.
13187 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13188 (make_single_succ_edge): Update.
13189 (check_bb_profile): Update.
13190 (dump_edge_info): Update.
13191 (update_bb_profile_for_threading): Update.
13192 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13193 probabilitycount to 0.
13194 * cfgbuild.c (compute_outgoing_frequencies): Update.
13195 * cfgcleanup.c (try_forward_edges): Update.
13196 (outgoing_edges_match): Update.
13197 (try_crossjump_to_edge): Update.
13198 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13199 (expand_gimple_tailcall): Update.
13200 (construct_init_block): Use make_single_succ_edge.
13201 (construct_exit_block): Use make_single_succ_edge.
13202 * cfghooks.c (verify_flow_info): Update.
13203 (redirect_edge_succ_nodup): Update.
13204 (split_edge): Update.
13205 (account_profile_record): Update.
13206 * cfgloopanal.c (single_likely_exit): Update.
13207 * cfgloopmanip.c (scale_loop_profile): Update.
13208 (set_zero_probability): Remove.
13209 (duplicate_loop_to_header_edge): Update.
13210 * cfgloopmanip.h (loop_version): Update prototype.
13211 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13212 (force_nonfallthru_and_redirect): Update.
13213 (update_br_prob_note): Update.
13214 (rtl_verify_edges): Update.
13215 (purge_dead_edges): Update.
13216 (rtl_lv_add_condition_to_bb): Update.
13217 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13218 * cgraphunit.c (init_lowered_empty_function): Update.
13219 (cgraph_node::expand_thunk): Update.
13220 * cilk-common.c: Include profile-count.h
13221 * dojump.c (inv): Remove.
13222 (jumpifnot): Update.
13223 (jumpifnot_1): Update.
13224 (do_jump_1): Update.
13226 (do_jump_by_parts_greater_rtx): Update.
13227 (do_compare_rtx_and_jump): Update.
13228 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13229 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13230 * dwarf2cfi.c: Include profile-count.h
13231 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13232 (sjlj_emit_dispatch_table): Likewise.
13233 * explow.c: Include profile-count.h
13234 * expmed.c (emit_store_flag_force): Update.
13235 (do_cmp_and_jump): Update.
13236 * expr.c (compare_by_pieces_d::generate): Update.
13237 (compare_by_pieces_d::finish_mode): Update.
13238 (emit_block_move_via_loop): Update.
13239 (store_expr_with_bounds): Update.
13240 (store_constructor): Update.
13241 (expand_expr_real_2): Update.
13242 (expand_expr_real_1): Update.
13243 * expr.h (try_casesi, try_tablejump): Update prototypes.
13244 * gimple-pretty-print.c (dump_probability): Update.
13245 (dump_profile): New.
13246 (dump_gimple_label): Update.
13247 (dump_gimple_bb_header): Update.
13248 * graph.c (draw_cfg_node_succ_edges): Update.
13249 * hsa-gen.c (convert_switch_statements): Update.
13250 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13251 (find_if_case_1): Update.
13252 (find_if_case_2): Update.
13253 * internal-fn.c (expand_arith_overflow_result_store): Update.
13254 (expand_addsub_overflow): Update.
13255 (expand_neg_overflow): Update.
13256 (expand_mul_overflow): Update.
13257 (expand_vector_ubsan_overflow): Update.
13258 * ipa-cp.c (good_cloning_opportunity_p): Update.
13259 * ipa-split.c (split_function): Use make_single_succ_edge.
13260 * ipa-utils.c (ipa_merge_profiles): Update.
13261 * loop-doloop.c (add_test): Update.
13262 (doloop_modify): Update.
13263 * loop-unroll.c (compare_and_jump_seq): Update.
13264 (unroll_loop_runtime_iterations): Update.
13265 * lra-constraints.c (lra_inheritance): Update.
13266 * lto-streamer-in.c (input_cfg): Update.
13267 * lto-streamer-out.c (output_cfg): Update.
13268 * mcf.c (adjust_cfg_counts): Update.
13269 * modulo-sched.c (sms_schedule): Update.
13270 * omp-expand.c (expand_omp_for_init_counts): Update.
13271 (extract_omp_for_update_vars): Update.
13272 (expand_omp_ordered_sink): Update.
13273 (expand_omp_for_ordered_loops): Update.
13274 (expand_omp_for_generic): Update.
13275 (expand_omp_for_static_nochunk): Update.
13276 (expand_omp_for_static_chunk): Update.
13277 (expand_cilk_for): Update.
13278 (expand_omp_simd): Update.
13279 (expand_omp_taskloop_for_outer): Update.
13280 (expand_omp_taskloop_for_inner): Update.
13281 * omp-simd-clone.c (simd_clone_adjust): Update.
13282 * optabs.c (expand_doubleword_shift): Update.
13283 (expand_abs): Update.
13284 (emit_cmp_and_jump_insn_1): Update.
13285 (expand_compare_and_swap_loop): Update.
13286 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13287 * predict.c (predictable_edge_p): Update.
13288 (edge_probability_reliable_p): Update.
13289 (set_even_probabilities): Update.
13290 (combine_predictions_for_insn): Update.
13291 (combine_predictions_for_bb): Update.
13292 (propagate_freq): Update.
13293 (estimate_bb_frequencies): Update.
13294 (force_edge_cold): Update.
13295 * profile-count.c (profile_count::dump): Add missing space into dump.
13296 (profile_count::debug): Add newline.
13297 (profile_count::differs_from_p): Explicitly convert to unsigned.
13298 (profile_count::stream_in): Update.
13299 (profile_probability::dump): New member function.
13300 (profile_probability::debug): New member function.
13301 (profile_probability::differs_from_p): New member function.
13302 (profile_probability::differs_lot_from_p): New member function.
13303 (profile_probability::stream_in): New member function.
13304 (profile_probability::stream_out): New member function.
13305 * profile-count.h (profile_count_quality): Rename to ...
13306 (profile_quality): ... this one.
13307 (profile_probability): New.
13308 (profile_count): Update.
13309 * profile.c (compute_branch_probabilities): Update.
13310 * recog.c (peep2_attempt): Update.
13311 * sched-ebb.c (schedule_ebbs): Update.
13312 * sched-rgn.c (find_single_block_region): Update.
13313 (compute_dom_prob_ps): Update.
13314 (schedule_region): Update.
13315 * sel-sched-ir.c (compute_succs_info): Update.
13316 * stmt.c (struct case_node): Update.
13317 (do_jump_if_equal): Update.
13318 (get_outgoing_edge_probs): Update.
13319 (conditional_probability): Update.
13320 (emit_case_dispatch_table): Update.
13321 (expand_case): Update.
13322 (expand_sjlj_dispatch_table): Update.
13323 (emit_case_nodes): Update.
13324 * targhooks.c: Update.
13325 * tracer.c (better_p): Update.
13326 (find_best_successor): Update.
13327 * trans-mem.c (expand_transaction): Update.
13328 * tree-call-cdce.c: Update.
13329 * tree-cfg.c (gimple_split_edge): Upate.
13330 (move_sese_region_to_fn): Upate.
13331 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13332 * tree-eh.c (lower_resx): Upate.
13333 (cleanup_empty_eh_move_lp): Upate.
13334 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13335 * tree-inline.c (copy_edges_for_bb): Update.
13336 (copy_cfg_body): Update.
13337 * tree-parloops.c (gen_parallel_loop): Update.
13338 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13339 (gimple_gen_time_profiler): Update.
13340 * tree-ssa-dce.c (remove_dead_stmt): Update.
13341 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13342 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13343 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13344 (unloop_loops): Update.
13345 (try_peel_loop): Update.
13346 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13347 * tree-ssa-loop-split.c (connect_loops): Update.
13348 (split_loop): Update.
13349 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13350 (hoist_guard): Update.
13351 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13352 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13353 (value_replacement): Update.
13354 * tree-ssa-reassoc.c (branch_fixup): Update.
13355 * tree-ssa-tail-merge.c (replace_block_by): Update.
13356 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13357 (create_edge_and_update_destination_phis): Update.
13358 (compute_path_counts): Update.
13359 (recompute_probabilities): Update.
13360 (update_joiner_offpath_counts): Update.
13361 (freqs_to_counts_path): Update.
13362 (duplicate_thread_path): Update.
13363 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13364 (struct switch_conv_info): Update.
13365 (gen_inbound_check): Update.
13366 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13367 (vect_do_peeling): Update.
13368 (vect_loop_versioning): Update.
13369 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13370 (optimize_mask_stores): Update.
13371 * ubsan.c (ubsan_expand_null_ifn): Update.
13372 * value-prof.c (gimple_divmod_fixed_value): Update.
13373 (gimple_divmod_fixed_value_transform): Update.
13374 (gimple_mod_pow2): Update.
13375 (gimple_mod_pow2_value_transform): Update.
13376 (gimple_mod_subtract): Update.
13377 (gimple_mod_subtract_transform): Update.
13378 (gimple_ic): Update.
13379 (gimple_stringop_fixed_value): Update.
13380 (gimple_stringops_transform): Update.
13381 * value-prof.h: Update.
13383 2017-06-29 Carl Love <cel@us.ibm.com>
13385 * config/rs6000/rs6000-c.c: Add support for built-in functions
13386 vector signed int vec_signed (vector float);
13387 vector signed long long vec_signed (vector double);
13388 vector signed int vec_signed2 (vector double, vector double);
13389 vector signed int vec_signede (vector double);
13390 vector signed int vec_signedo (vector double);
13391 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13392 instruction generator.
13393 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13394 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13395 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13397 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13398 vunsignede_v2df): Add define_expands.
13399 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13400 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13401 VEC_UNSIGNEDO): Add definitions.
13402 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13403 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13404 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13405 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13406 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13407 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13408 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13409 vec_unsignede, vec_unsignedo): Add builtin defines.
13410 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13412 * doc/extend.texi: Update the built-in documentation file for the
13413 new built-in functions.
13415 2017-06-29 Richard Biener <rguenther@suse.de>
13417 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13418 reduction chains to LOOP_VINFO_REDUCTIONS.
13419 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13420 SLP reductions after processing reduction chains.
13422 2017-06-29 Nathan Sidwell <nathan@acm.org>
13424 * builtins.c (fold_builtin_FUNCTION): Use
13425 lang_hooks.decl_printable_name.
13427 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13429 PR middle-end/81194
13430 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13431 with only one label.
13432 * stmt.c (expand_case): Assert NCASES is greater than one.
13434 2017-06-29 Richard Biener <rguenther@suse.de>
13436 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13438 (group_case_labels): Likewise.
13439 (find_taken_edge): Push sanity checking on val to workers...
13440 (find_taken_edge_cond_expr): ... here
13441 (find_taken_edge_switch_expr): ... and here, handle cases
13442 with just a default label.
13443 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13444 (group_case_labels): Likewise.
13445 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13446 group_case_labels does anything cleanup the CFG again.
13448 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13450 PR tree-optimization/81196
13451 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13452 exit condition comparing two IVs.
13454 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13456 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13457 profile to the dummy entry at the end of the list of architectures.
13458 * config/arm/arm-cpu-cdata.h: Regenerated.
13460 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13461 Michael Collison <michael.collison@arm.com>
13464 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13466 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13467 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13468 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13469 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13470 of shift when the shift amount is masked with constant equal to
13471 the size of the mode.
13472 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13475 2017-06-29 Martin Liska <mliska@suse.cz>
13477 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13479 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13482 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13485 (edge_growth_cache_entry::edge_growth_cache_entry): New
13487 (reset_edge_growth_cache): Update to use constructor.
13489 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13491 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13492 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13493 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13495 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13497 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13498 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13500 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13502 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13503 (*-linux-uclibc*): Add t-uclibc tmake_file.
13504 * config/t-musl: New.
13505 * config/t-uclibc: New.
13507 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13509 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13511 (gen_comm_data): Emit architectural setting of arch_prof.
13512 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13514 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13515 (armv8-m.base, armv8-m.main): Likewise.
13516 * arm-protos.h (arm_build_target): Add profile field.
13517 (arch_option): Likewise.
13518 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13520 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13521 arm_active_target.profile.
13523 2017-06-28 Richard Biener <rguenther@suse.de>
13525 PR middle-end/81227
13526 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13527 TYPE_OVERFLOW_WRAPS.
13528 * match.pd (negate_expr_p): Likewise.
13529 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13530 fold_build2, not fold_binary.
13532 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13534 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13535 Convert memory address to Pmode.
13536 (aarch64_print_operand): Assert MEM operands are always Pmode.
13538 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13541 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13542 Remove redundant if.
13543 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13544 * config/arm/aarch-common-protos.h
13545 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13546 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13548 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13551 * multiple_target.c (create_dispatcher_calls): Set the default
13552 clone to be static, not public.
13554 2017-06-28 Richard Biener <rguenther@suse.de>
13556 * tree-vect-loop.c (vectorizable_reduction): Move special
13557 cond reduction IV var creation ...
13558 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13559 parameter. Use STMT_VINFO_VECTYPE.
13560 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13563 2017-06-28 Martin Liska <mliska@suse.cz>
13566 * ipa-visibility.c (non_local_p): Handle visibility.
13568 2017-06-28 Martin Liska <mliska@suse.cz>
13571 * common.opt: Add IntegerRange to various options.
13572 * opt-functions.awk (integer_range_info): New function.
13573 * optc-gen.awk: Add integer_range_info to cl_options struct.
13574 * opts-common.c (decode_cmdline_option): Handle
13575 CL_ERR_INT_RANGE_ARG.
13576 (cmdline_handle_error): Likewise.
13577 * opts.c (print_filtered_help): Show valid interval in
13578 when --help is provided.
13579 * opts.h (struct cl_option): Add range_min and range_max fields.
13580 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13582 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13584 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13585 (x * C EQ/NE y * C): New transformation.
13587 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13589 * genmultilib (combination_space): Accept '+' in option names.
13591 2017-06-28 Martin Liska <mliska@suse.cz>
13594 * asan.c (instrument_derefs): Bail out inner references
13595 that are hard register variables.
13597 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13600 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13601 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13603 2017-06-28 Richard Biener <rguenther@suse.de>
13605 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13606 (vect_get_slp_defs): Adjust.
13607 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13609 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13611 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13612 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13613 (vectorizable_reduction): Adjust.
13614 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13616 (vect_get_slp_defs): Likewise.
13617 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13618 (vectorizable_bswap): Adjust.
13619 (vectorizable_call): Likewise.
13620 (vectorizable_conversion): Likewise.
13621 (vectorizable_assignment): Likewise.
13622 (vectorizable_shift): Likewise.
13623 (vectorizable_operation): Likewise.
13624 (vectorizable_store): Likewise.
13625 (vectorizable_condition): Likewise.
13626 (vectorizable_comparison): Likewise.
13628 2017-06-28 Michael Collison <michael.collison@arm.com>
13631 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13633 (arm_gen_movmemqi): Removed unused variable 'i'.
13634 Convert 'for' loop into 'while' loop.
13635 (arm_expand_prologue): Remove last unnecessary set of insn.
13636 (thumb_pop): Remove unused variable 'pushed_words'.
13637 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13639 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13641 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13642 * config/s390/s390-protos.h: Add prototype of
13643 s390_rel_address_ok_p.
13644 * config/s390/s390.c (s390_got_symbol): New function.
13645 (s390_rel_address_ok_p): New function.
13646 (legitimize_pic_address): Use s390_rel_address_ok_p.
13647 (s390_load_got): Use s390_got_symbol.
13648 (s390_option_override): Issue error if
13649 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13650 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13652 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13654 2017-06-27 Andrew Pinski <apinski@cavium.com>
13656 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13657 (X * copysign (1.0, X)): New pattern.
13658 (X * copysign (1.0, -X)): New pattern.
13659 (copysign (-1.0, CST)): New pattern.
13661 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13663 * genmultilib (combination_space): Remove variable.
13664 Validate reuse rules against regular expression for any sequence
13665 of multilib options in any order.
13667 2017-06-27 Michael Collison <michael.collison@arm.com>
13669 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13670 call aarch64_split_simd_combine.
13671 * (aarch64_combine_internal<mode>): Delete pattern.
13672 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13673 Allow register and subreg operands.
13675 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13677 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13678 specific need, just fallback on defaults.
13679 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13681 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13682 Olivier Hainque <hainque@adacore.com>
13684 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13686 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13687 targets. Pick a default if no particular attempt applied.
13688 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13691 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13693 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13694 (x86_64-wrs-vxworks7): Likewise.
13696 2017-06-27 Marek Polacek <polacek@redhat.com>
13699 * ubsan.c (instrument_null): Check get_base_address's result for null.
13701 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13703 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13705 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13707 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13708 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13709 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13710 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13711 New function types.
13712 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13713 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13714 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13715 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13716 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13717 BUILT_IN_FEUPDATEENV): New builtins.
13718 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13719 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13720 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13721 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13723 (builtin_structptr_types): Adjust size.
13724 * tree.c (builtin_structptr_types): Add four entries.
13726 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13727 Olivier Hainque <hainque@adacore.com>
13729 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13730 (TLS_SYM): New local macro, forcing reference to __tls__ on
13731 link command lines for VxWorks 7 RTPs, triggering initialization
13733 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13734 OS features TLS support, true for RTPs on VxWorks 7.
13735 * config/vxworks.c (vxworks_override_options): Setup emutls
13738 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13740 * predict.c (test_prediction_value_range): Use -1U instead of -1
13741 to avoid narrowing conversion warning.
13742 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13743 to avoid narrowing conversion warning.
13744 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13746 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13748 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13750 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13751 64bit configurations.
13752 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13753 (SIZE_TYPE): Likewise.
13754 * config/vxworks.c (vxworks_emutls_var_fields): Use
13755 long_unsigned_type_node instead of unsigned_type_node as the offset
13756 field type, which is "pointer" mode in emutls.c.
13758 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13761 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13763 PR middle-end/81207
13764 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13765 gimple_vuse copying separately from gimple_vdef copying.
13767 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13769 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13772 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13773 Olivier Hainque <hainque@adacore.com>
13775 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13776 for all vxworks7 targets.
13777 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13778 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13779 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13780 variations for VX6/VX7 and 32/64bits later on in ...
13781 (VXWORKS_LIB_SPEC): Leverage new macros.
13782 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13783 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13785 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13787 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13788 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13790 2017-06-26 Carl Love <cel@us.ibm.com>
13792 * config/rs6000/rs6000-c.c: Add support for built-in functions
13793 vector bool char vec_reve (vector bool char);
13794 vector signed char vec_reve (vector signed char);
13795 vector unsigned char vec_reve (vector unsigned char);
13796 vector bool int vec_reve (vector bool int);
13797 vector signed int vec_reve (vector signed int);
13798 vector unsigned int vec_reve (vector unsigned int);
13799 vector bool long long vec_reve (vector bool long long);
13800 vector signed long long vec_reve (vector signed long long);
13801 vector unsigned long long vec_reve (vector unsigned long long);
13802 vector bool short vec_reve (vector bool short);
13803 vector signed short vec_reve (vector signed short);
13804 vector double vec_reve (vector double);
13805 vector float vec_reve (vector float);
13806 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13807 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13808 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13809 (altivec_vreve): New pattern.
13810 * config/rs6000/altivec.h (vec_reve): New define.
13811 * doc/extend.texi (vec_rev): Update the built-in documentation file
13812 for the new built-in functions.
13814 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13816 PR tree-optimization/71815
13817 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13819 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13821 (slsr_process_phi): Likewise.
13822 (replace_uncond_cands_and_profitable_phis): Don't replace a
13823 multiply candidate with a stride of 1 (copy or cast).
13824 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13826 (lowest_cost_path): Likewise.
13827 (total_savings): Likewise.
13829 2017-06-26 Richard Biener <rguenther@suse.de>
13832 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13833 Use def_builtin_pure for all gather builtins.
13835 2017-06-26 Richard Biener <rguenther@suse.de>
13837 PR tree-optimization/81203
13838 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13839 non-dominating BBs.
13841 2017-06-26 Marek Polacek <polacek@redhat.com>
13844 * doc/invoke.texi: Document -Wmultistatement-macros.
13846 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13848 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13849 arm_neon_ok_no_float_abi effective target.
13851 2017-06-26 Richard Biener <rguenther@suse.de>
13853 PR tree-optimization/80928
13854 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13855 (copy_bbs): Set BB_DUPLICATED flag early.
13856 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13858 (execute_on_shrinking_pred): Likewise.
13859 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13860 BB_DUPLICATED blocks.
13861 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13862 iterate over all PHIs considering removal of *gsi.
13864 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13866 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13869 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13871 * config/rs6000/rs6000-string.c: (expand_block_clear,
13872 do_load_for_compare, select_block_compare_mode,
13873 compute_current_alignment, expand_block_compare,
13874 expand_strncmp_align_check, expand_strn_compare,
13875 expand_block_move, rs6000_output_load_multiple)
13876 Move functions related to string/block move/compare
13877 to a separate file.
13878 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13879 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13880 for this function which is now used in two files.
13881 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13882 * config.gcc: Add rs6000-string.o to extra_objs for
13883 targets powerpc*-*-* and rs6000*-*-*.
13885 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13888 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13889 32-bit, since indexed is not valid for DImode.
13890 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13891 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13892 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13893 (define_peephole2 for Altivec d-form store): Likewise.
13896 * multiple_target.c (create_dispatcher_calls): Only create the
13897 dispatcher call if the function is the default clone of a
13898 versioned function.
13900 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13902 PR middle-end/80902
13903 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13904 a call, force the call to not be a tail call.
13906 2017-06-23 Jeff Law <law@redhat.com>
13908 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13911 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13913 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13914 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13915 handling for early expansion of vector shifts (sl,sr,sra,rl).
13916 (builtin_function_type): Add vector shift right instructions
13917 to the unsigned argument list.
13919 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13921 rtl-optimizatoin/79286
13922 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13923 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13924 trap. PIC register plus a const unspec without offset can never trap.
13926 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13928 * tree.h (builtin_structptr_type): New type.
13929 (builtin_structptr_types): Declare new array.
13930 * tree.c (builtin_structptr_types): New array.
13931 (free_lang_data, build_common_tree_nodes): Use it.
13933 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13936 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13939 2017-06-22 Matt Turner <mattst88@gmail.com>
13941 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13942 Lake models to skylake case. Assume skylake for unknown
13943 models with clflushopt.
13945 2017-06-22 Jeff Law <law@redhat.com>
13947 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13948 frame sizes that do not satisfy aarch64_uimm12_shift.
13950 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13952 * profile-count.h (apply_probability,
13953 apply_scale, probability_in): Fix checks for zero.
13955 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13957 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13958 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13960 2016-06-22 Richard Biener <rguenther@suse.de>
13962 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13963 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13964 REDUC_MAX_EXPR support.
13965 (vectorizable_reduction): Likewise.
13966 (vect_create_epilog_for_reduction): Likewise.
13968 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13970 * match.pd (A / (1 << B) -> A >> B): New.
13971 * generic-match-head.c: Include optabs-tree.h.
13972 * gimple-match-head.c: Likewise.
13973 * optabs-tree.h (target_supports_op_p): New.
13974 * optabs-tree.c (target_supports_op_p): New.
13976 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13978 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13979 $gcc_cv_ld --help output.
13980 (gcc_cv_ld_demangle): Likewise.
13981 (gcc_cv_ld_eh_frame_hdr): Likewise.
13982 (gcc_cv_ld_pie): Likewise.
13983 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13984 (gcc_cv_ld_buildid): Likewise.
13985 (gcc_cv_ld_sysroot): Likewise.
13986 (ld_bndplt_support): Likewise.
13987 (ld_pushpopstate_support): Likewise.
13988 * configure: Regenerate.
13989 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13991 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13994 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13995 operands indexes to avoid gap between operands and match_dups.
13997 2017-06-21 Andrew Pinski <apinski@cavium.com>
13999 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14000 Increment Arith_shift and Arith_shift_reg by 1.
14001 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14003 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14004 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14005 (aarch64_strip_extend): Add new argument and test for it.
14006 (aarch64_cheap_mult_shift_p): New function.
14007 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14008 add a cost if it is true.
14009 Update calls to aarch64_strip_extend.
14010 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14012 2017-06-21 Andrew Pinski <apinski@cavium.com>
14014 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14016 (thunderxt88): Likewise.
14017 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14018 (thunderx_prefetch_tune): New variable.
14019 (thunderx2t99_prefetch_tune): Update for the correct values.
14020 (thunderxt88_tunings): New variable.
14021 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14022 generic_prefetch_tune.
14023 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14025 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14027 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14028 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14029 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14030 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14031 (aarch64_atomic_cas<mode>, GPI): Likewise.
14033 2017-06-21 Martin Liska <mliska@suse.cz>
14035 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14036 statements on cold and hot labels.
14037 * predict.c (tree_estimate_probability_bb): Remove the
14038 prediction from this place.
14040 2017-06-21 Martin Liska <mliska@suse.cz>
14042 PR tree-optimization/79489
14043 * gimplify.c (maybe_add_early_return_predict_stmt): New
14045 (gimplify_return_expr): Call the function.
14046 * predict.c (tree_estimate_probability_bb): Remove handling
14048 * predict.def: Update comment about early return predictor.
14049 * gimple-predict.h (is_gimple_predict): New function.
14050 * predict.def: Change default value of early return to 66.
14051 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14053 * passes.def: Put pass_strip_predict_hints to the beginning of
14056 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14058 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14059 FUNCTION_DECL declarations.
14060 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14062 (dwaf2out_decl): Likewise.
14063 * godump.c (go_early_global_decl): Skip call to the real debug hook
14064 for FUNCTION_DECL declarations.
14065 * passes.c (rest_of_decl_compilation): Skip call to the
14066 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14067 -fdump-go-spec is passed.
14069 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14071 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14072 Reorder for compactness.
14073 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14074 (def_builtin_pure, def_builtin_pure2): New functions.
14075 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14077 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14079 * match.pd (nop_convert): New predicate.
14080 ((A +- CST1) +- CST2): Allow some NOP conversions.
14082 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14085 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14086 with ctors/dtors if GOVD_SHARED is set.
14088 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14090 * config/aarch64/aarch64.md (movti_aarch64):
14091 Emit mov rather than orr.
14092 (movtf_aarch64): Likewise.
14093 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14094 Emit mov rather than orr.
14096 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14098 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14099 Swap alternatives, make integer dup more expensive.
14101 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14103 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14104 Return true for non-tls symbols.
14106 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14108 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14109 (cortex-a75): Likewise.
14110 (cortex-a75.cortex-a55): Likewise.
14111 * config/aarch64/aarch64-tune.md: Regenerate.
14112 * doc/invoke.texi (-mtune): Document new values for -mtune.
14114 2017-06-21 Tom de Vries <tom@codesourcery.com>
14116 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14117 stack_size feature.
14118 (Effective-Target Keywords, Other attributes): Suggest using
14119 dg-add-options stack_size feature to get stack limit in stack_size
14120 effective target documentation.
14122 2017-06-21 Julian Brown <julian@codesourcery.com>
14123 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14125 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14126 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14127 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14129 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14130 attribute type list for neon_multiply.
14131 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14132 attribute type list for neon_multiply.
14133 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14134 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14135 attribute type list for neon_multiply.
14136 * config/arm/types.md (crypto_pmull): Add.
14137 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14138 attribute type list.
14140 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14142 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14145 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14147 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14148 to make sure not to dereference a NULL cost_classes_ptr pointer.
14150 2017-06-20 Carl Love <cel@us.ibm.com>
14152 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14153 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14154 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14155 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14156 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14157 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14158 VMULOSW): New enum "unspec" values.
14159 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14160 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14161 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14162 altivec_vmulosw): New patterns.
14163 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14164 VMULOSW): Add definitions.
14166 2017-06-20 Julia Koval <julia.koval@intel.com>
14168 * config/i386/i386.c: Fix rounding expand for new pattern.
14169 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14171 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14173 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14174 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14176 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14178 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14181 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14183 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14184 architecture, then by implementer ID.
14185 * config/aarch64/aarch64-tune.md: Regenerate.
14187 2017-06-20 Richard Biener <rguenther@suse.de>
14189 PR middle-end/81097
14190 * fold-const.c (split_tree): Fold to type before negating.
14192 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14194 * diagnostic-show-locus.c
14195 (selftest::test_fixit_deletion_affecting_newline): New function.
14196 (selftest::diagnostic_show_locus_c_tests): Call it.
14198 2017-06-20 Andreas Schwab <schwab@suse.de>
14201 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14204 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14206 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14207 __ARM_FEATURE_COPROC according to support.
14209 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14211 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14212 Rewritten to avoid overflow for > 32-bit pointers.
14215 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14216 by removing enum keyword.
14217 (ubsan_type_descriptor): Likewise. Formatting fix.
14220 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14221 splitter): Require TARGET_SSE2 in the condition.
14223 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14226 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14227 for doing vector set of SFmode on ISA 3.0.
14228 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14229 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14231 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14232 SFmode value into a V4SF variable that was extracted from another
14233 V4SF variable without converting the element to double precision
14234 and back to single precision vector format.
14235 (vsx_insert_extract_v4sf_p9_2): Likewise.
14237 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14239 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14240 in UWHI to avoid undefined overflow.
14243 * ubsan.h (enum ubsan_encode_value_phase): New.
14244 (ubsan_encode_value): Change second argument to
14245 enum ubsan_encode_value_phase with default value of
14246 UBSAN_ENCODE_VALUE_GENERIC.
14247 * ubsan.c (ubsan_encode_value): Change second argument to
14248 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14249 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14250 create_tmp_var_raw instead of create_tmp_var and use a
14252 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14253 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14254 ubsan_encode_value callers.
14257 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14258 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14259 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14261 2017-06-19 Richard Biener <rguenther@suse.de>
14263 PR middle-end/81118
14264 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14265 estimates if we changed anything.
14267 2017-06-19 Richard Biener <rguenther@suse.de>
14269 PR tree-optimization/80887
14270 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14271 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14272 simplified lookups, then reset mprts_hook.
14273 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14275 (try_to_simplify): Likewise.
14277 2017-06-19 Martin Liska <mliska@suse.cz>
14280 * gimplify.c (gimplify_switch_expr):
14281 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14283 2017-06-19 Martin Liska <mliska@suse.cz>
14285 * doc/install.texi: Document that PGO runs in 4 stages.
14287 2017-06-19 Martin Liska <mliska@suse.cz>
14290 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14291 to dispatcher function name.
14292 * multiple_target.c (replace_function_decl): New function.
14293 (create_dispatcher_calls): Redirect both edges and references.
14295 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14297 * profile-count.c (profile_count::dump): Dump quality.
14298 (profile_count::differs_from_p): Update for unsigned val.
14299 * profile-count.h (profile_count_quality): New enum.
14300 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14302 2017-06-19 Richard Biener <rguenther@suse.de>
14304 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14305 struct function as arg.
14306 (estimate_numbers_of_iterations): Export overload with loop arg.
14307 (free_numbers_of_iterations_estimates_loop): Use an overload of
14308 free_numbers_of_iterations_estimates instead.
14309 * tree-cfg.c (remove_bb): Adjust.
14310 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14311 * tree-parloops.c (gen_parallel_loop): Likewise.
14312 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14314 (tree_unroll_loops_completely): Likewise.
14315 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14316 Use an overload instead and export.
14317 (estimated_loop_iterations): Adjust.
14318 (max_loop_iterations): Likewise.
14319 (likely_max_loop_iterations): Likewise.
14320 (estimate_numbers_of_iterations): Take struct function as arg
14322 (loop_exits_before_overflow): Adjust.
14323 (free_numbers_of_iterations_estimates_loop): Use an overload.
14324 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14325 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14327 2017-06-19 Richard Biener <rguenther@suse.de>
14330 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14331 RANGE_EXPR conservatively.
14333 2017-06-16 Carl Love <cel@us.ibm.com>
14335 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14336 definitions for vec_float, vec_float2, vec_floato,
14337 vec_floate built-ins.
14338 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14339 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14341 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14342 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14343 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14344 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14345 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14346 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14347 vec_floato): Add builtin defines.
14348 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14349 Update the built-in documentation file for the new built-in
14352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14354 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14355 (mthumb): Mark as the negative of -marm.
14357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14359 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14361 (ARM Options, -mtune): Document that this accepts the same
14362 extension options as -mcpu.
14363 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14365 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14367 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14368 permitted extensions.
14370 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14372 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14373 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14374 (armv8-m.main): Add option +nodsp.
14375 * config/arm/arm-cpu-cdata.h: Regenerated.
14377 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14379 * config/arm/t-fuchsia: New file.
14380 * config.gcc (arm*-*-fuchsia*): Use it.
14382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14384 * config/arm/t-symbian: Rewrite for new option infrastructure.
14386 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14388 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14389 (MULTILIB_REQUIRED): Likewise.
14391 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14393 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14394 (MULTILIB_RESUE): Likewise.
14395 (MULTILIB_MATCHES): Likewise.
14396 (MULTLIB_REQUIRED): Likewise.
14398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14400 * config/arm/t-rtems: Rewrite for new option framework.
14402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14404 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14405 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14406 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14407 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14408 * config/arm/t-multilib: ... here.
14409 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14410 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14411 armv7-a and armv8*-a when A-profile libraries have not been built.
14412 * config/arm/t-rmprofile: Rewrite.
14414 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14416 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14417 with a backslash. Remove the backslash after substituting unescaped
14419 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14423 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14424 the driver to pass through the default setting of -mfloat-abi.
14425 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14427 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14428 (all_feat_combs): New rule.
14429 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14431 * config/arm/t-aprofile: Rewrite.
14433 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14435 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14436 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14437 fpu is not specified by the user/command-line.
14438 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14439 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14440 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14441 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14442 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14443 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14447 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14448 * config/arm/t-arm-elf: Rewritten.
14450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14452 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14453 have some floating-point instructions.
14454 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14455 (TARGET_MAYBE_HARD_FLOAT): New macro.
14456 * config/arm/arm-builtins.c (arm_init_builtins): Use
14457 TARGET_MAYBE_HARD_FLOAT.
14458 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14460 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14462 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14463 (configargs.h): Include it.
14464 (arm_print_hint_for_fpu_option): New function.
14465 (arm_parse_fpu_option): New function.
14466 (candidate_extension): New class.
14467 (arm_canon_for_multilib): New function.
14468 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14469 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14470 (ARCH_CANONICAL_SPECS): New macro.
14471 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14473 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14475 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14476 are set after handling multilib fragments. Set target_cpu_default2
14479 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14481 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14483 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14484 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14486 (check_arch): Likewise.
14487 * config/arm/arm.c (arm_configure_build_target): Handle
14488 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14489 options in the default.
14491 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14493 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14494 when an option is an alias of another.
14495 * config/arm/parsecpu.awk (optalias): New parser token.
14496 (gen_comm_data): Mark non-alias options as such. Emit entries
14497 for extension aliases.
14498 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14499 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14500 (armv6kz, armv6zk, armv6t2): Likewise.
14501 (armv7): Make vfpv3-d16 an alias.
14502 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14504 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14505 Sort in canonical order.
14506 (armv8-a): Sort in canonical order.
14507 (armv8.1-a, armv8.2-a): Likewise.
14508 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14510 (cortex-a9): Sort in canonical order.
14511 * config/arm/arm.c (selftests.h): Include it.
14512 (arm_test_cpu_arch_data): New function.
14513 (arm_run_self_tests): New function.
14514 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14515 (targetm): Move declaration to the end of the file.
14516 * arm-cpu-cdata.h: Regenerated.
14518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14520 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14521 call to target_mode_check describing the type of option passed.
14522 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14523 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14524 arm_parse_cpu_option_name to match parameters against list of
14526 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14527 arm_arch_core_flags data structure.
14528 * config/arm/arm-cpu_cdata.h: Regenerated.
14530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14532 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14534 (arm_print_hint_for_cpu_option): Likewise.
14535 (arm_print_hint_for_arch_option): Likewise.
14536 (arm_parse_cpu_option_name): Likewise.
14537 (arm_parse_arch_option_name): Likewise.
14538 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14539 of entries in the all_fpus list.
14540 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14541 (arm_parse_cpu_option_name): Declare.
14542 (arm_parse_arch_option_name): Declare.
14543 (arm_parse_option_features): Declare.
14544 (arm_intialize_isa): Declare.
14545 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14547 (gen_comm_data): ... here. Make definitions non-static.
14548 * config/arm/arm-cpu-data.h: Regenerated.
14549 * config/arm/arm-cpu-cdata.h: Regenerated.
14551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14553 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14554 (cpu_arch_extension): New structure.
14555 (cpu_arch_option, arch_option, cpu_option): New structures.
14556 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14557 architecture types.
14558 (gen_data): Generate new format data tables.
14559 * config/arm/arm.c (cpu_tune): New structure.
14560 (cpu_option, processors): Delete.
14561 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14562 (arm_print_hint_for_cpu_option): ... this and ...
14563 (arm_print_hint_for_arch_option): ... this.
14564 (arm_parse_arch_cpu_name): Delete. Replace with ...
14565 (arm_parse_cpu_option_name): ... this and ...
14566 (arm_parse_arch_option_name): ... this.
14567 (arm_unrecognized_feature): Change type of target parameter to
14569 (arm_parse_arch_cpu_features): Delete. Replace with ...
14570 (arm_parse_option_features): ... this.
14571 (arm_configure_build_target): Rework to use new configuration data
14573 (arm_print_tune_info): Rework for new configuration data tables.
14574 * config/arm/arm-cpu-data.h: Regenerated.
14575 * config/arm/arm-cpu.h: Regenerated.
14577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14579 * Makefile.in (OBJS): Move sbitmap.o from here ...
14580 (OBJS-libcommon): ... to here.
14582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14584 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14585 (ISA_ALL_CRYPTO): New macro.
14586 (ISA_ALL_SIMD): New macro
14587 (ISA_ALL_FP): New macro.
14588 * config/arm/arm.c (fpu_bitlist): Update initializer.
14589 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14591 (arm9e): Add fpu. Add option for nofp
14592 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14593 (arm926ej-s, arm1026ej-s): Likewise.
14594 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14595 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14596 neon-fp16, neon-vfpv4, nofp and nosimd.
14597 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14598 (cortex-a8): Add fpu. Add option for nofp.
14599 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14600 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14601 (cortex-r4f): Add fpu.
14602 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14603 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14605 (cortex-r8): Likewise.
14606 (cortex-m4): Add fpu. Add option for nofp.
14607 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14608 (cortex-a17.cortex-a7): Likewise.
14609 (cortex-a32): Add fpu. Add options for crypto and nofp.
14610 (cortex-a35, cortex-a53): Likewise.
14611 (cortex-a57): Add fpu. Add option for crypto.
14612 (cortex-a72, cortex-a73): Likewise.
14613 (exynos-m1): Likewise.
14614 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14615 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14616 (cortex-m33): Add fpu. Add option for nofp.
14617 * config/arm/arm-cpu-cdata.h: Regenerated
14618 * config/arm/arm-cpu-data.h: Regenerated.
14620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14622 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14623 (armv5te, armv5tej): Likewise.
14624 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14625 (armv7): Add options fp and vfpv3-d16.
14626 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14627 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14629 (armv7ve): Likewise.
14630 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14631 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14632 (armv8-a): Add nocrypto option.
14633 (armv8.1-a, armv8.2-a): Likewise.
14634 (armv8-m.main): add options fp, fp.dp and nofp.
14636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14638 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14640 (armv8-a+crc): Delete.
14641 (armv8.1-a): Add options simd, crypto and nofp.
14642 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14643 (armv8.2-a+fp16): Delete.
14644 (armv8-m.main): Add option dsp.
14645 (armv8-m.main+dsp): Delete.
14646 (cortex-a8): Add fpu. Add nofp option.
14647 (cortex-a9): Add fpu. Add nofp and nosimd options.
14648 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14649 link to main cpu and architecture data structures.
14650 (gen_comm_data): Only put isa attributes from the main architecture
14652 (option): New statement for architecture and CPU entries.
14653 * arm.c (struct cpu_option): New structure.
14654 (struct processors): Add entry for options.
14655 (arm_unrecognized_feature): New function.
14656 (arm_parse_arch_cpu_name): Ignore any characters after the first
14658 (arm_parse_arch_cpu_feature): New function.
14659 (arm_configure_build_target): Separate out any CPU and architecture
14660 features and parse separately. Don't error out if -mfpu=auto is
14661 used with only an architecture string.
14662 (arm_print_asm_arch_directives): New function.
14663 (arm_file_start): Call it.
14664 * config/arm/arm-cpu-cdata.h: Regenerated.
14665 * config/arm/arm-cpu-data.h: Likewise.
14666 * config/arm/arm-tables.opt: Likewise.
14668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14670 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14671 assembler when it is not -mfpu=auto.
14673 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14675 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14676 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14677 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14678 (ASM_CPU_SPEC): Rewrite.
14679 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14680 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14681 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14682 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14683 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14684 copied string is NUL-terminated. Also strip any characters prefixed
14686 (arm_rewrite_selected_arch): New function.
14687 (arm_rewrite_march): New function.
14689 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14691 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14692 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14693 (march, mcpu, mtune): Convert to string-based options.
14694 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14695 (arm_parse_arch_cpu_name): New function.
14696 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14697 identify selected architecture or CPU.
14698 (arm_option_save): New function.
14699 (TARGET_OPTION_SAVE): Redefine.
14700 (arm_option_restore): Restore string options.
14701 (arm_option_print): Print string options.
14703 2017-06-16 Martin Sebor <msebor@redhat.com>
14705 PR tree-optimization/80933
14706 PR tree-optimization/80934
14707 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14708 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14709 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14710 (gimple_fold_builtin): Call them.
14712 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14714 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14715 as unlikely; update profile.
14717 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14719 * predict.c (force_edge_cold): Handle declaring edges impossible
14722 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14724 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14726 (try_unroll_loop_completely): Fix reporting.
14728 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14730 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14732 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14735 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14736 if given a non-constant argument for an intrinsic which requires a
14739 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14741 * profile.c (compare_freqs): New function.
14742 (branch_prob): Sort edge list.
14743 (find_spanning_tree): Assume that the list is priority sorted.
14745 2017-06-16 Richard Biener <rguenther@suse.de>
14747 PR tree-optimization/81090
14748 * passes.def (pass_record_bounds): Remove.
14749 * tree-pass.h (make_pass_record_bounds): Likewise.
14750 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14751 make_pass_record_bounds): Likewise.
14752 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14753 not free niter estimates at the beginning but at the end.
14754 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14756 2017-06-16 Richard Biener <rguenther@suse.de>
14758 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14759 initializer to workaround ICE in host GCC 4.8.
14761 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14763 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14765 (clone_inlined_nodes): Update.
14767 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14769 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14770 prefetch settings, and enable prefetching by default at -O3.
14772 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14774 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14775 Set flag_prefetch_loop_arrays according to tuning data.
14777 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14779 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14780 New tune structure.
14781 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14782 [Unrelated to main purpose of the patch] Place the pointer field last
14783 to enable type checking errors when tune structure are wrongly merged.
14784 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14785 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14786 (thunderx2t99_prefetch_tune): New tune constants.
14787 (tune_params *_tunings): Update all tunings (no functional change).
14788 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14789 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14790 from tunings structures.
14792 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14795 * ubsan.c (instrument_null): Add T argument, use it instead
14796 of computing it based on IS_LHS.
14797 (instrument_object_size): Likewise.
14798 (pass_ubsan::execute): Adjust instrument_null and
14799 instrument_object_size callers to pass gimple_get_lhs or
14800 gimple_assign_rhs1 result to it. Use instrument_null instead of
14801 calling get_base_address and instrument_mem_ref. Handle
14802 aggregate call arguments for object-size sanitization.
14804 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14806 PR tree-optimization/81089
14807 * tree-vrp.c (is_masked_range_test): Validate operands of
14810 2017-06-15 Martin Sebor <msebor@redhat.com>
14813 * dumpfile.c (dump_register): Avoid calling memset to initialize
14814 a class with a default ctor.
14815 * gcc.c (struct compiler): Remove const qualification.
14816 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14817 * hash-table.h: Ditto.
14818 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14820 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14821 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14823 * params.h (struct param_info): Make struct members non-const.
14824 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14825 with default initialization.
14826 * vec.h (vec_copy_construct, vec_default_construct): New helper
14828 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14829 with vec_copy_construct.
14830 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14831 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14832 * doc/invoke.texi (-Wclass-memaccess): Document.
14834 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14836 * emit-rtl.h (is_leaf): Update comment about local
14837 register allocator.
14839 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14842 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14843 for a variable to have a section before checking if the section has a
14845 Set section to.persistent if persistent attribute is set.
14846 Warn if .persistent attribute is used on an automatic variable.
14848 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14850 PR rtl-optimization/80474
14851 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14853 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14855 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14858 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14860 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14861 (casesi): Emit bounds checking as RTL.
14862 (casesi_internal_mips16_<mode>): Remove bounds checking.
14864 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14866 * config/xtensa/xtensa.c (xtensa_option_override): Append
14867 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14868 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14869 xtensa_doloop_hooks): Define unconditionally.
14870 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14872 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14873 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14874 for it in xtensa_option_override.
14875 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14876 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14878 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14880 * doc/cppopts.texi: Document '-' special value to -MF.
14882 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14884 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14885 (cortex_a53_fconst): Likewise.
14886 (cortex_a53_fpmul): Likewise.
14887 (cortex_a53_f_load_64): Likewise.
14888 (cortex_a53_f_load_many): Likewise.
14889 (cortex_a53_advsimd_alu): Likewise.
14890 (cortex_a53_advsimd_alu_q): Likewise.
14891 (cortex_a53_advsimd_mul): Likewise.
14892 (cortex_a53_advsimd_mul_q): Likewise.
14893 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14894 Add missing fmul, r2f_cvt and fconst cases.
14896 2017-06-14 Richard Biener <rguenther@suse.de>
14898 PR middle-end/81088
14899 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14901 (fold_binary_loc): When associating do not treat pre-existing
14902 TREE_OVERFLOW on literal constants as a reason to allow
14903 TREE_OVERFLOW on associated literal constants.
14905 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14907 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14908 (MASK_FEATURES): New macro.
14909 * config/sparc/sparc.c (sparc_option_override): Remove the special
14910 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14912 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14914 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14915 a division of 0 if non-call exceptions are enabled.
14917 2017-06-14 Andrew Pinski <apinski@cavium.com>
14918 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14921 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14922 Improve vector initialization code gen for only variable case.
14924 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14926 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14928 2017-06-14 Richard Biener <rguenther@suse.de>
14930 PR tree-optimization/81083
14931 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14934 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14936 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14937 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14938 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14939 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14940 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14941 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14943 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14945 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14946 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14948 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14950 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14952 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14954 * config/rs6000/t-rtems: Don't handle SPE.
14956 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14958 * config/rs6000/t-linux: Don't handle SPE.
14960 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14962 * config/rs6000/eabispe.h: Delete file.
14964 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14966 * config/rs6000/t-spe: Delete file.
14968 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14970 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14971 (rs6000_legitimate_offset_address_p): Return false for anything in
14972 V2SImode or V2SFmode.
14974 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14976 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14977 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14978 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14980 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14981 (rs6000_legitimate_offset_address_p): Ditto.
14982 (rs6000_emit_move): Ditto.
14983 (rs6000_init_builtins): Remove V4HI_type_node.
14985 2017-06-13 Martin Liska <mliska@suse.cz>
14988 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14989 (gate_asan): Likewise.
14990 * asan.h (asan_no_sanitize_address_p): Remove the function.
14991 (sanitize_flags_p): New function.
14992 * builtins.def: Fix coding style.
14993 * common.opt: Use renamed enum value.
14994 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14995 * doc/extend.texi: Document no_sanitize attribute.
14996 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14997 to SANITIZE_UNDEFINED_NONDEFAULT.
14998 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14999 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15000 Use sanitize_flags_p.
15001 * gimplify.c (gimplify_function_tree): Likewise.
15002 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15003 * opts.c (parse_no_sanitize_attribute): New function.
15004 (common_handle_option): Use renamed enum value.
15005 * opts.h (parse_no_sanitize_attribute): Declare.
15006 * tree.c (sanitize_flags_p): New function.
15007 * tree.h: Declared here.
15008 * tsan.c: Use sanitize_flags_p.
15009 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15010 (instrument_mem_ref): Likewise.
15011 (instrument_bool_enum_load): Likewise.
15012 (do_ubsan_in_current_function): Remove the function.
15013 (pass_ubsan::execute): Use sanitize_flags_p.
15014 * ubsan.h: Remove do_ubsan_in_current_function
15015 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15016 (dump_function_to_file): Use it here.
15018 2017-06-13 Martin Jambor <mjambor@suse.cz>
15020 PR tree-optimization/80803
15021 PR tree-optimization/81063
15022 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15023 (propagate_subaccesses_across_link): Enqueue subtree whenever
15024 necessary instead of relying on the caller.
15026 2017-06-13 Martin Jambor <mjambor@suse.cz>
15028 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15029 that have a first_link.
15030 (sort_and_splice_var_accesses): Do not check first_link before
15032 (subtree_mark_written_and_enqueue): Likewise.
15033 (propagate_all_subaccesses): Likewise and do not stop at first
15034 parent with a first_link.
15036 2017-06-13 Martin Jambor <mjambor@suse.cz>
15038 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15041 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15043 * match.pd: New pattern.
15045 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15047 * tree-vrp.c (is_masked_range_test): New function.
15048 (register_edge_assert_for): Determine ranges for
15051 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15053 PR tree-optimization/67328
15054 * fold-const.c (maskable_range_p): New function.
15055 (build_range_check): Generate bittests if possible.
15057 2017-06-13 Martin Liska <mliska@suse.cz>
15059 * gimple-pretty-print.c (dump_probability): Add new argument.
15060 (dump_edge_probability): Dump both probability and count.
15061 (dump_gimple_label): Likewise.
15062 (dump_gimple_bb_header): Likewise.
15064 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15067 * config/avr/avr-devices.c: Fix indentation.
15068 * config/avr/gen-avr-mmcu-specs.c: Dito.
15070 2017-06-13 Richard Biener <rguenther@suse.de>
15072 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15073 instead get vector type from stmt_info.
15074 (vectorizable_reduction): Adjust. Remove dead code.
15076 2017-06-13 Richard Biener <rguenther@suse.de>
15078 PR middle-end/81065
15079 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15080 case of C * (x * C2 + C3).
15081 (fold_addr_of_array_ref_difference): Properly fold index difference.
15083 2017-06-12 David S. Miller <davem@davemloft.net>
15086 * config/sparc/sparc.md (return expander): Emit frame blockage if
15087 function uses alloca.
15089 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15091 * combine.c (make_field_assignment): Check len rather than the mode
15092 precision when calling force_to_mode.
15094 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15096 Support multilibs and devices that see flash in RAM address range.
15099 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15100 (avr_mcu_t) <flash_pm_offset>: New field.
15101 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15102 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15103 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15104 (AVR_TINY_PM_OFFSET): Remove macro.
15105 * config/avr/avr.opt (-mshort-calls): New option.
15106 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15107 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15108 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15109 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15110 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15111 instead of avr_arch->have_jmp_call.
15112 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15113 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15114 avr_arch->flash_pm_offset to define.
15115 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15116 new field flash_pm_offset. Add entry for avrxmega3.
15117 (avr_texinfo): Add entry for avrxmega3.
15118 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15119 attiny212, attiny214,
15120 attiny412, attiny414, attiny416, attiny417,
15121 attiny814, attiny816, attiny817,
15122 attiny1614, attiny1616, attiny1617,
15123 attiny3214, attiny3216, attiny3217.
15124 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15125 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15126 (avr_print_operand_address) [AVR_TINY]: Same.
15127 (avr_asm_init_sections) <readonly_data_section>: Only patch
15128 callback if avr_arch->flash_pm_offset = 0.
15129 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15130 for rodata if avr_arch->flash_pm_offset != 0.
15131 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15132 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15133 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15134 (m_options): Append opt_rcall.
15135 (m_dirnames): Append dir_rcall.
15136 * config/avr/t-multilib: Regenerate.
15138 * configure.ac [target=avr]: Check whether avrxmega3 default
15139 linker description file works as needed.
15140 * configure: Regenerate.
15141 * doc/avr-mmcu.texi: Regenerate.
15142 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15143 <__AVR_ARCH__>: Document avrxmega3 and 103.
15144 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15145 <__AVR_SHORT_CALLS__>: Document it.
15146 <__AVR_PM_BASE_ADDRESS__>: Document it.
15147 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15148 (AVR Variable Attributes) <progmem>: Document this is
15149 not needed for avrxmega3.
15150 (AVR Named Address Spaces) <__flash>: Dito.
15152 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15154 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15156 2017-06-12 Doug Rupp <rupp@adacore.com>
15158 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15159 Append vxworks-stdint.h to the tm_file list.
15160 * config/vxworks-stdint.h: New file.
15162 2017-06-12 Martin Liska <mliska@suse.cz>
15164 PR tree-optimization/81041
15165 * tree-profile.c (gimple_gen_ic_func_profiler):
15166 Create an extra BB in profile-generate
15167 (gimple_gen_time_profiler): Likewise.
15169 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15171 PR tree-optimization/81003
15172 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15173 (update_range_test): Use it instead of force_gimple_operand_gsi.
15175 2017-06-12 Richard Biener <rguenther@suse.de>
15177 PR tree-optimization/81053
15178 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15179 with backedge value not defined in loop. Simplify def stmt
15182 2017-06-11 Tom de Vries <tom@codesourcery.com>
15185 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15187 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15188 nvptx_cannot_force_const_mem.
15190 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15192 * opts.c (finish_options): Move test for flag_split_stack after
15193 it has been initialized.
15195 2017-06-11 Jason Merrill <jason@redhat.com>
15197 * tree.h (id_equal): New.
15198 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15199 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15200 instead of strcmp of IDENTIFIER_POINTER.
15202 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15204 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15205 (mark_all_inlined_calls_cdtor): Fix formating.
15206 (inline_transform): Rescale profile before inlining.
15208 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15210 * cgraph.h (cgraph_edge::clone): Update prototype.
15211 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15212 (cgraph_node::create_clone): Update.
15213 (cgraph_node::create_version_clone): Update.
15214 * tree-inline.c (copy_bb): Update.
15215 (expand_call_inline): Update.
15217 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15219 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15220 factored out from ...
15221 (rs6000_emit_prologue): ... here.
15223 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15225 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15226 factored out from ...
15227 (rs6000_emit_prologue): ... here.
15229 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15231 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15233 (handle_missing_profiles): Fix computation of tp_first_run.
15234 (counts_to_freqs): Do not touch freqs when count is 0.
15236 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15238 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15241 2017-06-10 Tom de Vries <tom@codesourcery.com>
15243 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15244 attributes): Document signal effective target.
15246 2017-06-10 Tom de Vries <tom@codesourcery.com>
15248 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15249 Document effective target stack_size.
15251 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15253 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15254 to the edit_context if they can be auto-applied.
15256 2017-06-9 Ian Lance Taylor <iant@golang.org>
15258 * opts.c (finish_options): If -fsplit-stack, disable implicit
15259 -forder-blocks-and-partition.
15260 * doc/invoke.texi (Optimize Options): Document that when using
15261 -fsplit-stack -forder-blocks-and-partition is not implicitly
15264 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15266 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15267 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15268 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15269 * builtins.def (abort, trap, unreachable): Declare cold.
15270 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15271 * tree-core.h (ECF_COLD): New.
15272 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15273 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15275 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15277 * predict.c (unlikely_executed_stmt_p): Cleanup.
15279 2017-06-09 Richard Biener <rguenther@suse.de>
15281 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15282 model if the ref is always written to.
15284 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15286 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15288 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15290 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15293 2017-06-09 Tom de Vries <tom@codesourcery.com>
15296 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15297 "target cannot support label values" when encountering LABEL_REF.
15299 2017-06-09 Martin Liska <mliska@suse.cz>
15301 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15302 (gimple_gen_ic_func_profiler): Emit direct comparison
15303 of __gcov_indirect_call_callee with NULL.
15304 (gimple_gen_time_profiler): Change probability from
15305 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15307 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15309 * profile.c (edge_gcov_counts): Turn to pointer.
15310 (compute_branch_probabilities, compute_branch_probabilities): Update.
15311 (branch_prob): Do not clear edge_gcov_count.
15312 * profile.h (edge_gcov_counts): Turn to pointer.
15313 (edge_gcov_count): Update.
15315 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15317 * gimple.h (gimple_check_failed): Mark cold.
15319 2017-06-09 Richard Biener <rguenther@suse.de>
15321 PR tree-optimization/66623
15322 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15323 refactor check_reduction into two parts, properly computing
15324 whether we have to check reduction validity for outer loop
15327 2017-06-09 Richard Biener <rguenther@suse.de>
15329 PR tree-optimization/79483
15330 * graphite-scop-detection.c (order): New global.
15331 (get_order): Compute bb to order mapping that satisfies code
15332 generation constraints.
15333 (cmp_pbbs): New helper.
15334 (build_scops): Start domwalk at entry block, sort generated
15337 2017-06-09 Richard Biener <rguenther@suse.de>
15339 PR middle-end/81007
15340 * ipa-polymorphic-call.c
15341 (ipa_polymorphic_call_context::restrict_to_inner_class):
15342 Skip FIELD_DECLs with error_mark_node type.
15343 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15346 2017-06-09 Martin Liska <mliska@suse.cz>
15348 * predict.c (struct branch_predictor): New struct.
15349 (test_prediction_value_range): New test.
15350 (predict_c_tests): New function.
15351 * selftest-run-tests.c (selftest::run_tests): Run the function.
15352 * selftest.h: Declare new tests.
15354 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15357 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15358 gen_add3_insn did not fail.
15359 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15360 r0, construct that number in a temporary reg and add that reg to r0.
15361 If asked to put the result in r0 as well, fail.
15363 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15365 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15366 for early expansion of vec_eqv.
15368 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15370 PR middle-end/81005
15371 * ubsan.c (instrument_null): Avoid pointless code temporary.
15372 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15374 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15378 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15381 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15384 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15386 * predict.c (unlikely_executed_edge_p): Move ahead.
15387 (probably_never_executed_edge_p): Use it.
15389 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15391 PR middle-end/79988
15392 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15393 gimple_call_builtin_p call.
15395 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15397 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15398 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15399 rtl_check_failed_type2, rtl_check_failed_code1,
15400 rtl_check_failed_code2, rtl_check_failed_code_mode,
15401 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15402 rtvec_check_failed_bounds, rtl_check_failed_flag,
15403 _fatal_insn_not_found, _fatal_insn): Likewise.
15404 * tree.h (tree_contains_struct_check_failed,
15405 tree_check_failed, tree_not_check_failed,
15406 tree_class_check_failed, tree_range_check_failed,
15407 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15408 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15409 tree_operand_check_failed, omp_clause_check_failed,
15410 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15413 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15415 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15416 flag_branch_probabilities.
15417 * ipa-inline.c (edge_badness): Likewise.
15418 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15419 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15420 * predict.c (maybe_hot_frequency_p): Likewise.
15421 (probably_never_executed): Likewise.
15422 * sched-ebb.c (schedule_ebbs): Likewise.
15423 * sched-rgn.c (find_single_block_region): Likewise.
15424 * tracer.c (tail_duplicate): Likewise.
15426 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15428 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15429 longer requires x_flag_profile_use.
15431 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15433 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15434 instead of flag_reorder_blocks_and_partition.
15435 * dbxout.c (dbxout_function_end): Likewise.
15436 * dwarf2out.c (gen_subprogram_die): Likewise.
15437 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15438 * hw-doloop.c (reorg_loops): Likewise.
15439 * varasm.c (assemble_start_function,
15440 assemble_end_function): Likewise.
15441 (decide_function_section): Do not check for
15442 flag_reorder_blocks_and_partition.
15444 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15446 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15448 (chkp_get_hard_register_fake_addr_expr): Ditto.
15449 (chkp_build_addr_expr): Add check for hard reg case.
15450 (chkp_parse_array_and_component_ref): Ditto.
15451 (chkp_find_bounds_1): Ditto.
15452 (chkp_process_stmt): Don't generate bounds store for
15455 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15457 * predict.c (maybe_hot_bb_p): Do not check profile status.
15458 (maybe_hot_edge_p): Likewise.
15459 (probably_never_executed): Check for zero counts even if profile
15461 (unlikely_executed_edge_p): New function.
15462 (unlikely_executed_stmt_p): New function.
15463 (unlikely_executed_bb_p): New function.
15464 (set_even_probabilities): Use unlikely predicates.
15465 (combine_predictions_for_bb): Likewise.
15466 (predict_paths_for_bb): Likewise.
15467 (predict_paths_leading_to_edge): Likewise.
15468 (determine_unlikely_bbs): New function.
15469 (estimate_bb_frequencies): Use it.
15470 (compute_function_frequency): Use zero counts even if profile is
15472 * profile-count.h: Fix typo.
15474 2017-08-08 Julia Koval <julia.koval@intel.com>
15476 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15477 _mm512_mask_cvtsepi16_storeu_epi8,
15478 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15479 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15480 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15481 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15482 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15483 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15484 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15485 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15486 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15487 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15488 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15489 __builtin_ia32_pmovuswb256mem_mask,
15490 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15491 __builtin_ia32_pmovwb512mem_mask): New builtins.
15493 2017-08-08 Julia Koval <julia.koval@intel.com>
15495 PR target/73350,80862
15496 * config/i386/subst.md (round): Fix round pattern.
15497 * config/i386/i386.c (ix86_erase_embedded_rounding):
15498 Fix erasing rounding for the fixed pattern.
15500 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15502 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15504 2017-06-08 Martin Liska <mliska@suse.cz>
15506 PR gcov-profile/80911
15507 * gcov.c (block_info::block_info): New constructor.
15509 2017-06-07 Carl Love <cel@us.ibm.com>
15511 * config/rs6000/rs6000-c: The return type of the following
15512 built-in functions was implemented as int not long long. Fix sign
15513 of return value for the unsigned version of vec_mulo and vec_mule.
15514 vector unsigned long long vec_bperm (vector unsigned long long,
15515 vector unsigned char)
15516 vector signed long long vec_mule (vector signed int,
15518 vector unsigned long long vec_mule (vector unsigned int,
15519 vector unsigned int)
15520 vector signed long long vec_mulo (vector signed int,
15522 vector unsigned long long vec_mulo (vector unsigned int,
15523 vector unsigned int)
15524 * doc/extend.texi: Fix the documentation for the built-in
15527 2017-06-07 Carl Love <cel@us.ibm.com>
15530 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15533 2017-06-07 Carl Love <cel@us.ibm.com>
15535 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15536 support, Generate doublehv for signed int/float for BE case only.
15538 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15540 * doc/invoke.texi (mcx16): Rewrite.
15542 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15544 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15545 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15546 *mov<mode>_softfloat, and an anonymous splitter): Use
15547 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15549 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15551 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15553 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15554 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15555 (rs6000_debug_reg_global): Adjust.
15556 (rs6000_init_hard_regno_mode_ok): Adjust.
15557 (rs6000_dbx_register_number): Adjust.
15558 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15559 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15560 Remove SPE_ACC and SPEFSCR.
15561 (REG_ALLOC_ORDER): Ditto.
15562 (FRAME_POINTER_REGNUM): Change to 111.
15563 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15564 (REG_CLASS_NAMES): Ditto.
15565 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15566 (REGISTER_NAMES): Ditto.
15567 (ADDITIONAL_REG_NAMES): Ditto.
15568 (rs6000_reg_names): Ditto.
15569 * config/rs6000/rs6000.md: Renumber some register number
15572 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15574 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15576 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15577 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15579 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15580 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15581 Delete the SPE high registers.
15582 (REG_ALLOC_ORDER): Ditto.
15583 (enum reg_class): Remove SPE_HIGH_REGS.
15584 (REG_CLASS_NAMES): Ditto.
15585 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15586 (REGISTER_NAMES): Ditto.
15587 (rs6000_reg_names): Ditto.
15588 * doc/tm.texi.in: Remove SPE as example.
15589 * doc/tm.texi: Regenerate.
15591 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15593 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15594 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15595 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15596 * config/rs6000/rs6000.md (type): Remove "brinc".
15598 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15600 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15601 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15602 * config/rs6000/linuxspe.h: Delete file.
15603 * config/rs6000/rs6000.md: Don't include spe.md.
15604 * config/rs6000/spe.h: Delete file.
15605 * config/rs6000/spe.md: Delete file.
15606 * config/rs6000/t-rs6000: Remove spe.md.
15608 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15610 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15611 (reg_or_none500mem_operand): Delete.
15612 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15613 instead of reg_or_none500mem_operand.
15615 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15617 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15618 handling of SPE flags.
15619 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15621 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15623 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15625 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15626 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15627 paired_divv2sf3): Similar.
15628 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15629 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15630 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15632 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15633 Rename the paired_* instruction patterns.
15634 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15636 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15637 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15638 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15639 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15640 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15641 PAIRED_VECTOR_MODE.
15642 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15643 (spe_func_has_64bit_regs_p): Delete.
15644 (spe_expand_predicate_builtin): Delete.
15645 (spe_expand_evsel_builtin): Delete.
15646 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15647 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15648 (invalid_e500_subreg): Delete.
15649 (rs6000_legitimize_address): Always force_reg op2 as well, for
15650 paired single memory accesses.
15651 (rs6000_member_type_forces_blk): Delete.
15652 (rs6000_spe_function_arg): Delete.
15653 (rs6000_expand_unop_builtin): Delete SPE handling.
15654 (rs6000_expand_binop_builtin): Ditto.
15655 (spe_expand_stv_builtin): Delete.
15656 (bdesc_2arg_spe): Delete.
15657 (spe_expand_builtin): Delete.
15658 (spe_expand_predicate_builtin): Delete.
15659 (spe_expand_evsel_builtin): Delete.
15660 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15661 (spe_init_builtins): Delete.
15662 (spe_func_has_64bit_regs_p): Delete.
15663 (savres_routine_name): Delete "info" parameter. Adjust callers.
15664 (rs6000_emit_stack_reset): Ditto.
15665 (rs6000_dwarf_register_span): Delete.
15666 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15667 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15668 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15670 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15672 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15673 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15674 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15675 mulv2sf3, divv2sf3): Delete expanders.
15677 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15679 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15681 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15683 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15684 * config/rs6000/rs6000.c: Ditto.
15686 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15688 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15689 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15690 comparison_operator.
15692 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15694 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15695 * config/rs6000/rs6000.opt: Ditto.
15696 * config/rs6000/t-rtems: Ditto.
15698 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15700 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15701 TARGET_E500_SINGLE by 0, simplify.
15702 * config/rs6000/rs6000.c: Ditto.
15703 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15704 (spe_build_register_parallel): Delete.
15705 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15706 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15707 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15708 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15709 (E500_CONVERT): Delete.
15710 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15712 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15714 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15715 * config/rs6000/dfp.md: Ditto.
15716 (negdd2, *negdd2_fpr): Merge.
15717 (absdd2, *absdd2_fpr): Merge.
15718 (negtd2, *negtd2_fpr): Merge.
15719 (abstd2, *abstd2_fpr): Merge.
15720 * config/rs6000/e500.h: Delete file.
15721 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15722 TARGET_FPRS by 1 and simplify.
15723 * config/rs6000/rs6000-c.c: Ditto.
15724 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15725 TARGET_DF_SPE by 0.
15726 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15728 * config/rs6000/rs6000.md: Ditto.
15729 (floatdidf2, *floatdidf2_fpr): Merge.
15730 (move_from_CR_gt_bit): Delete.
15731 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15732 (E500_CR_IOR_COMPARE): Delete.
15733 (All patterns that require !TARGET_FPRS): Delete.
15734 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15736 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15738 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15740 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15742 * graphds.c (add_edge): Intitialize edge's attached data.
15743 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15744 pointer parameter. Call pointed function on each edge during
15745 graph traversing. Skip traversing the edge when the function
15747 (graphds_dfs, graphds_scc): Ditto.
15748 (for_each_edge): New parameter. Pass the new parameter to callback
15750 * graphds.h (skip_edge_callback): New function pointer type.
15751 (graphds_dfs, graphds_scc): New function pointer parameter.
15752 (graphds_edge_callback, for_each_edge): New parameter.
15754 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15756 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15757 out code checking if runtime alias check is possible to below ...
15758 Call the new function.
15759 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15760 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15762 2017-06-07 Marek Polacek <polacek@redhat.com>
15765 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15766 TYPE_OVERFLOW_WRAPS check.
15768 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15770 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15771 if versioning is required.
15772 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15773 peeling with the check for versioning.
15775 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15777 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15778 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15779 Set true to new parameter if new ssa variable is defined.
15780 (vect_gen_vector_loop_niters): Refactor. Set range information
15781 for the new vector loop bound variable.
15782 (vect_do_peeling): Ditto.
15784 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15786 * tree-affine.c (ssa.h): Include header file.
15787 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15788 has wrapping overflow behavior.
15790 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15792 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15794 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15796 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15797 (tree_to_aff_combination): ... here.
15799 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15801 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15802 reg_pressure model function.
15803 (ivopts_global_cost_for_size): Delete.
15804 (determine_set_costs, iv_ca_recount_cost): Call new model function
15805 ivopts_estimate_reg_pressure.
15807 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15809 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15810 expensive than udiv. Remove floating point cases from mod.
15812 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15814 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15815 Increase idiv cost.
15817 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15819 * config/aarch64/aarch64.md
15820 (copysignsf3): Fix mask generation.
15822 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15824 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15826 (class dump_manager): Add register_dumps method.
15827 * dumpfile.c: Include langhooks.h.
15828 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15829 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15830 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15831 (dump_manager::dump_register): Start with 512 entries instead of 32.
15832 (dump_manager::register_dumps): New method.
15833 * toplev.c (general_init): Instead of invoking register_dumps
15834 langhook, invoke register_dumps method on the dump manager.
15835 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15838 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15840 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15841 Say that # requires an associated define_split to exist, and that
15842 the define_split must be suitable for use after register allocation.
15844 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15846 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15847 (compute_outgoing_frequencies): Also initialize zero counts.
15848 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15849 around loops; preserve more of profile when nothing changes.
15851 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15853 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15855 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15856 * config/arm/arm-cpu-cdata.h: Regenerate.
15857 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15858 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15859 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15860 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15861 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15863 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15865 * config/arm/t-rmprofile: Likewise.
15866 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15868 2017-06-06 David S. Miller <davem@davemloft.net>
15871 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15872 blockage if function uses alloca.
15874 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15876 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15877 New "uid" fields to hold pretty-print IDs of group and ref.
15878 Memory references are now identified as <group_id>:<ref_id>
15879 instead of using [random] addresses.
15880 (dump_mem_details): Simplify, no functional change.
15881 (dump_mem_ref): Simplify and make output more concise.
15882 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15883 (find_or_create_group): Initialize group uid.
15884 (record_ref): Initialize ref uid. Improve debug output.
15885 (prune_group_by_reuse, should_issue_prefetch_p,)
15886 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15887 (mark_nontemporal_store, determine_loop_nest_reuse):
15888 Improve debug output.
15890 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15892 * dbgcnt.def (prefetch): New debug counter.
15893 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15894 (schedule_prefetches): Stop issueing prefetches if debug counter
15897 2017-06-06 Tom de Vries <tom@codesourcery.com>
15899 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15900 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15902 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15904 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15905 Use aarch64_reg_or_zero predicate for operand 4.
15906 (aarch64_compare_and_swap<mode> define_insn_and_split):
15907 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15908 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15910 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15912 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15913 (arm_compute_save_core_reg_mask): This.
15914 (thumb1_compute_save_reg_mask): Rename into ...
15915 (thumb1_compute_save_core_reg_mask): This.
15916 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15917 (arm_compute_frame_layout): Likewise.
15919 2017-06-06 Richard Biener <rguenther@suse.de>
15921 PR tree-optimization/80974
15922 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15923 keep or clear leaders SSA info.
15925 2017-06-06 Tom de Vries <tom@codesourcery.com>
15927 * config/nvptx/nvptx.c (split_mode_p): New function.
15928 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15930 2017-06-06 Tom de Vries <tom@codesourcery.com>
15932 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15934 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15937 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15940 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15942 * shrink-wrap.c (handle_simple_exit): Update profile.
15943 (try_shrink_wrapping): Upate profile.
15945 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15947 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15948 (tree_guess_outgoing_edge_probabilities): New.
15949 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15950 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15952 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15954 * ipa-split.c (split_function): Initialize return bb profile.
15956 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15958 * profile.c (compute_branch_probabilities): Also initialize
15959 EXIT_BLOCK profile.
15961 2017-06-06 Richard Biener <rguenther@suse.de>
15963 PR tree-optimization/80928
15964 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15965 (vect_analyze_loop_operations): Properly guard analysis for
15967 (vect_transform_loop): Likewise.
15968 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15969 (vect_model_induction_cost): Do not cost for pure SLP.
15970 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15971 of induction in inner loop vectorization.
15972 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15973 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15974 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15976 (vect_analyze_slp_cost_1): Cost induction.
15977 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15978 (vect_get_slp_vect_defs): Likewise.
15979 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15980 (vect_transform_stmt): Handle SLP reductions.
15981 * tree-vectorizer.h (vectorizable_induction): Adjust.
15983 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15985 * config/rs6000/rs6000.c (make_resolver_func): Update
15986 init_lowered_empty_function call.
15988 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15990 * doc/invoke.texi: Document the -fprofile-abs-path option.
15991 * common.opt (fprofile-abs-path): New option.
15992 * gcov-io.h (gcov_write_filename): Declare.
15993 * gcov-io.c (gcov_write_filename): New function.
15994 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15995 * profile.c (output_location): Likewise.
15997 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15999 * shring-wrap.c: Revert accidental commit.
16001 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16003 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16005 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16007 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16009 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16010 profile in callgraph edge.
16011 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16012 (apply_scale): Likewise.
16013 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16014 Also scale profile when inlining function with zero profile.
16015 (initialize_cfun): Update exit block profile even when it is zero.
16016 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16017 when profile is read.
16019 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16021 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16022 (CLONE_*): New constants to define the processors we can generate
16023 code for with the target_clone attribute.
16024 (rs6000_clone_map): New array to identify which clone processors
16025 the current program is running on.
16026 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16027 target_clone attribute.
16028 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16029 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16030 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16031 (cpu_expand_builtin): Add support for target_clone attribute.
16032 (rs6000_valid_attribute_p): Allow "default" attribute.
16033 (get_decl_name): New debug function to simplify printing the
16034 current function name in debugging statements.
16035 (rs6000_clone_priority): New functions to support the target_clone
16036 attribute, and be able to generate code to switch between ISA 2.05
16037 through ISA 3.0 (power6 through power9).
16038 (rs6000_compare_version_priority): Likewise.
16039 (rs6000_get_function_versions_dispatcher): Likewise.
16040 (make_resolver_func): Likewise.
16041 (add_condition_to_bb): Likewise.
16042 (dispatch_function_versions): Likewise.
16043 (rs6000_generate_version_dispatcher_body): Likewise.
16044 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16045 (fusion_gpr_load_p): Fix a spacing issue.
16046 * doc/extend.texi (Common Function Attributes): Document that the
16047 PowerPC supports the target_clone attribute.
16049 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16051 * config/arm/arm.h: explain F symbol found in description of ARM
16052 register allocation in its legend.
16054 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16056 * config/mips/frame-header-opt.c: Include profile-count.h.
16057 * config/riscv/riscv.c: Include profile-count.h
16059 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16061 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16063 (sm_set_flag_if_changed): Add bbs field.
16064 (execute_sm_if_changed_flag_set): Pass BBS.
16065 (execute_sm): Update.
16067 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16069 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16072 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16074 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16075 (peephole2): New peephole2 to emit the above.
16076 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16078 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16080 * config/aarch64/aarch64.c (define_peephole2 above
16081 *sub_<shift>_<mode>): New peephole.
16083 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16085 * config/i386/i386.c (make_resolver_func): Update.
16086 * Makefile.in: Add profile-count.h and profile-count.o
16087 * auto-profile.c (afdo_indirect_call): Update to new API.
16088 (afdo_set_bb_count): Update.
16089 (afdo_propagate_edge): Update.
16090 (afdo_propagate_circuit): Update.
16091 (afdo_calculate_branch_prob): Update.
16092 (afdo_annotate_cfg): Update.
16093 * basic-block.h: Include profile-count.h
16094 (struct edge_def): Turn count to profile_count.
16095 (struct basic_block_def): Likewie.
16096 (REG_BR_PROB_BASE): Move to profile-count.h
16097 (RDIV): Move to profile-count.h
16098 * bb-reorder.c (max_entry_count): Turn to profile_count.
16099 (find_traces): Update.
16100 (rotate_loop):Update.
16101 (connect_traces):Update.
16102 (sanitize_hot_paths):Update.
16103 * bt-load.c (migrate_btr_defs): Update.
16104 * cfg.c (RDIV): Remove.
16105 (init_flow): Use alloc_block.
16106 (alloc_block): Uninitialize count.
16107 (unchecked_make_edge): Uninitialize count.
16108 (check_bb_profile): Update.
16109 (dump_edge_info): Update.
16110 (dump_bb_info): Update.
16111 (update_bb_profile_for_threading): Update.
16112 (scale_bbs_frequencies_int): Update.
16113 (scale_bbs_frequencies_gcov_type): Update.
16114 (scale_bbs_frequencies_profile_count): New.
16115 * cfg.h (update_bb_profile_for_threading): Update.
16116 (scale_bbs_frequencies_profile_count): Declare.
16117 * cfgbuild.c (compute_outgoing_frequencies): Update.
16118 (find_many_sub_basic_blocks): Update.
16119 * cfgcleanup.c (try_forward_edges): Update.
16120 (try_crossjump_to_edge): Update.
16121 * cfgexpand.c (expand_gimple_tailcall): Update.
16122 (construct_exit_block): Update.
16123 * cfghooks.c (verify_flow_info): Update.
16124 (dump_bb_for_graph): Update.
16125 (split_edge): Update.
16126 (make_forwarder_block): Update.
16127 (duplicate_block): Update.
16128 (account_profile_record): Update.
16129 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16130 (get_estimated_loop_iterations): Update.
16131 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16132 (single_likely_exit): Update.
16133 * cfgloopmanip.c (scale_loop_profile): Update.
16135 (set_zero_probability): Update.
16136 (lv_adjust_loop_entry_edge): Update.
16137 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16138 (purge_dead_edges): Update.
16139 (rtl_account_profile_record): Update.
16140 * cgraph.c (cgraph_node::create): Uninitialize count.
16141 (symbol_table::create_edge): Uninitialize count.
16142 (cgraph_update_edges_for_call_stmt_node): Update.
16143 (cgraph_edge::dump_edge_flags): Update.
16144 (cgraph_node::dump): Update.
16145 (cgraph_edge::maybe_hot_p): Update.
16146 * cgraph.h: Include profile-count.h
16147 (create_clone), create_edge, create_indirect_edge): Update.
16148 (cgraph_node): Turn count to profile_count.
16149 (cgraph_edge0: Likewise.
16150 (make_speculative, clone): Update.
16151 (create_edge): Update.
16152 (init_lowered_empty_function): Update.
16153 * cgraphclones.c (cgraph_edge::clone): Update.
16154 (duplicate_thunk_for_node): Update.
16155 (cgraph_node::create_clone): Update.
16156 * cgraphunit.c (cgraph_node::analyze): Update.
16157 (cgraph_node::expand_thunk): Update.
16158 * final.c (dump_basic_block_info): Update.
16159 * gimple-streamer-in.c (input_bb): Update.
16160 * gimple-streamer-out.c (output_bb): Update.
16161 * graphite.c (print_global_statistics): Update.
16162 (print_graphite_scop_statistics): Update.
16163 * hsa-brig.c: Include basic-block.h.
16164 * hsa-dump.c: Include basic-block.h.
16165 * hsa-gen.c (T sum_slice): Update.
16166 (convert_switch_statements):Update.
16167 * hsa-regalloc.c: Include basic-block.h.
16168 * ipa-chkp.c (chkp_produce_thunks): Update.
16169 * ipa-cp.c (struct caller_statistics): Update.
16170 (init_caller_stats): Update.
16171 (gather_caller_stats): Update.
16172 (ipcp_cloning_candidate_p): Update.
16173 (good_cloning_opportunity_p): Update.
16174 (get_info_about_necessary_edges): Update.
16175 (dump_profile_updates): Update.
16176 (update_profiling_info): Update.
16177 (update_specialized_profile): Update.
16178 (perhaps_add_new_callers): Update.
16179 (decide_about_value): Update.
16180 (ipa_cp_c_finalize): Update.
16181 * ipa-devirt.c (struct odr_type_warn_count): Update.
16182 (struct decl_warn_count): Update.
16183 (struct final_warning_record): Update.
16184 (possible_polymorphic_call_targets): Update.
16185 (ipa_devirt): Update.
16186 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16187 * ipa-icf.c (sem_function::merge): Update.
16188 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16189 * ipa-inline.c (compute_uninlined_call_time): Update.
16190 (compute_inlined_call_time): Update.
16191 (want_inline_small_function_p): Update.
16192 (want_inline_self_recursive_call_p): Update.
16193 (edge_badness): Update.
16194 (lookup_recursive_calls): Update.
16195 (recursive_inlining): Update.
16196 (inline_small_functions): Update.
16197 (dump_overall_stats): Update.
16198 (dump_inline_stats): Update.
16199 * ipa-profile.c (ipa_profile_generate_summary): Update.
16200 (ipa_propagate_frequency): Update.
16201 (ipa_profile): Update.
16202 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16203 * ipa-utils.c (ipa_merge_profiles): Update.
16204 * loop-doloop.c (doloop_modify): Update.
16205 * loop-unroll.c (report_unroll): Update.
16206 (unroll_loop_runtime_iterations): Update.
16207 * lto-cgraph.c (lto_output_edge): Update.
16208 (lto_output_node): Update.
16209 (input_node): Update.
16210 (input_edge): Update.
16211 (merge_profile_summaries): Update.
16212 * lto-streamer-in.c (input_cfg): Update.
16213 * lto-streamer-out.c (output_cfg): Update.
16214 * mcf.c (create_fixup_graph): Update.
16215 (adjust_cfg_counts): Update.
16216 (sum_edge_counts): Update.
16217 * modulo-sched.c (sms_schedule): Update.
16218 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16219 * predict.c (maybe_hot_count_p): Update.
16220 (probably_never_executed): Update.
16221 (dump_prediction): Update.
16222 (combine_predictions_for_bb): Update.
16223 (propagate_freq): Update.
16224 (handle_missing_profiles): Update.
16225 (counts_to_freqs): Update.
16226 (rebuild_frequencies): Update.
16227 (force_edge_cold): Update.
16228 * predict.h: Include profile-count.h
16229 (maybe_hot_count_p, counts_to_freqs): UPdate.
16230 * print-rtl-function.c: Do not include cfg.h
16231 * print-rtl.c: Include basic-block.h
16232 * profile-count.c: New file.
16233 * profile-count.h: New file.
16234 * profile.c (is_edge_inconsistent): Update.
16235 (correct_negative_edge_counts): Update.
16236 (is_inconsistent): Update.
16237 (set_bb_counts): Update.
16238 (read_profile_edge_counts): Update.
16239 (compute_frequency_overlap): Update.
16240 (compute_branch_probabilities): Update; Initialize and deinitialize
16242 (branch_prob): Update.
16243 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16244 (edge_gcov_count): New.
16245 (bb_gcov_count): New.
16246 * shrink-wrap.c (try_shrink_wrapping): Update.
16247 * tracer.c (better_p): Update.
16248 * trans-mem.c (expand_transaction): Update.
16249 (ipa_tm_insert_irr_call): Update.
16250 (ipa_tm_insert_gettmclone_call): Update.
16251 * tree-call-cdce.c: Update.
16252 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16253 (gimple_duplicate_sese_tail): Update.
16254 (gimple_account_profile_record): Update.
16255 (execute_fixup_cfg): Update.
16256 * tree-inline.c (copy_bb): Update.
16257 (copy_edges_for_bb): Update.
16258 (initialize_cfun): Update.
16259 (freqs_to_counts): Update.
16260 (copy_cfg_body): Update.
16261 (expand_call_inline): Update.
16262 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16263 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16264 (try_unroll_loop_completely): Update.
16265 (try_peel_loop): Update.
16266 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16267 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16268 * tree-ssa-loop-split.c (connect_loops): Update.
16269 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16270 * tree-ssa-reassoc.c (branch_fixup): Update.
16271 * tree-ssa-tail-merge.c (replace_block_by): Update.
16272 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16273 (compute_path_counts): Update.
16274 (update_profile): Update.
16275 (recompute_probabilities): Update.
16276 (update_joiner_offpath_counts): Update.
16277 (estimated_freqs_path): Update.
16278 (freqs_to_counts_path): Update.
16279 (clear_counts_path): Update.
16280 (ssa_fix_duplicate_block_edges): Update.
16281 (duplicate_thread_path): Update.
16282 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16283 (struct switch_conv_info): Update.
16284 * tree-tailcall.c (decrease_profile): Update.
16285 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16286 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16287 * value-prof.c (check_counter): Update.
16288 (gimple_divmod_fixed_value): Update.
16289 (gimple_mod_pow2): Update.
16290 (gimple_mod_subtract): Update.
16291 (gimple_ic_transform): Update.
16292 (gimple_stringop_fixed_value): Update.
16293 * value-prof.h (gimple_ic): Update.
16295 2017-06-02 Carl Love <cel@us.ibm.com>
16297 * config/rs6000/rs6000-c: Add support for built-in functions
16298 vector double vec_doublee (vector signed int);
16299 vector double vec_doublee (vector unsigned int);
16300 vector double vec_doublee (vector float);
16301 vector double vec_doubleh (vector signed int);
16302 vector double vec_doubleh (vector unsigned int);
16303 vector double vec_doubleh (vector float);
16304 vector double vec_doublel (vector signed int);
16305 vector double vec_doublel (vector unsigned int);
16306 vector double vec_doublel (vector float);
16307 vector double vec_doubleo (vector signed int);
16308 vector double vec_doubleo (vector unsigned int);
16309 vector double vec_doubleo (vector float);.
16310 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16311 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16313 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16314 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16315 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16317 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16318 vec_doublel, vec_doubleh.
16319 * doc/extend.texi: Update the built-in documentation file for the
16320 new built-in functions.
16322 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16325 * ipa-inline-analysis.c (free_growth_caches): Set
16326 edge_removal_hook_holder to NULL after removing it.
16328 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16330 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16331 comparision with zero.
16333 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16334 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16335 for early expansion of vec_min and vec_max builtins.
16336 (builtin_function_type): Add min/max unsigned variants to those
16337 identified as having unsigned arguments.
16339 2017-06-02 Olivier Hainque <hainque@adacore.com>
16341 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16343 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16345 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16346 Use VALL_F16 iterator rather than VALL.
16348 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16350 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16351 Emit CBNZ inside loop when doing a strong exchange and comparing
16352 against zero. Generate the CC flags after the loop.
16354 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16356 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16357 (dl_section_ref): New.
16358 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16359 On AIX, append an expression to subtract the size of the
16360 section length to dl_section_ref.
16362 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16364 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16365 for early expansion of vector absolute builtins.
16367 2017-06-02 Richard Biener <rguenther@suse.de>
16369 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16370 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16372 2017-06-02 Richard Biener <rguenther@suse.de>
16374 PR tree-optimization/80948
16375 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16376 stmt order as well.
16378 2017-06-02 Richard Biener <rguenther@suse.de>
16380 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16382 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16383 for inductions as relevant.
16385 2017-06-02 Richard Biener <rguenther@suse.de>
16387 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16388 (vectorizable_induction): ... this. Remove dead code.
16390 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16392 * builtins. (expand_builtin_alloca): Remove second parameter and
16393 infer its value from the first parameter instead.
16394 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16396 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16398 PR rtl-optimization/80903
16399 * loop-doloop.c (add_test): Unshare sequence.
16401 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16403 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16405 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16407 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16409 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16410 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16411 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16412 (xlogue_layout::get_stub_rtx): Make static.
16413 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16414 (xlogue_layout::compute_stub_managed_regs): Rename to...
16415 (xlogue_layout::count_stub_managed_regs): ...this.
16416 (xlogue_layout::is_stub_managed_reg): New function.
16417 (xlogue_layout::m_stub_names): Rename to...
16418 (xlogue_layout::s_stub_names): ...this, make static.
16419 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16420 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16421 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16422 xlogue_layout::s_stub_names): Instantiate statics.
16423 (stub_managed_regs): Remove.
16424 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16425 (disable_call_ms2sysv_xlogues): Rename to...
16426 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16427 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16429 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16430 change after reload_completed.
16431 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16433 (ix86_expand_prologue): Likewise.
16434 (ix86_expand_epilogue): Likewise.
16435 (ix86_expand_split_stack_prologue): Likewise.
16436 (ix86_compute_frame_layout): Remove frame parameter ...
16437 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16438 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16440 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16441 (ix86_frame): Move from here ...
16442 * config/i386/i386.h (ix86_frame): ... to here.
16443 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16444 complete ix86_frame data structure instead. Remove some_ld_name.
16446 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16448 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16449 symbols that hold a DECL_VALUE_EXPR.
16451 2017-06-01 Martin Jambor <mjambor@suse.cz>
16453 PR tree-optimization/80898
16454 * tree-sra.c (process_subtree_disqualification): Removed.
16455 (disqualify_candidate): Do not acll
16456 process_subtree_disqualification.
16457 (subtree_mark_written_and_enqueue): New function.
16458 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16459 RHS has been disqualified and re-queue LHS if necessary. Apart
16460 from that, ignore disqualified RHS.
16462 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16464 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16465 address fetch for z10 or later.
16467 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16469 * config/arc/arc.md (tst_movb): Add guard when splitting.
16471 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16473 * config/arc/arc.c (arc_can_eliminate): Test against
16474 arc_frame_pointer_needed.
16476 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16478 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16479 to prevent store reordering.
16480 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16481 (type): Add block type.
16482 (stack_tie): Define special instruction to be used in
16485 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16487 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16488 constraint. It is not valid for the pattern.
16489 (noncommutative_binary_comparison): Likewise.
16491 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16493 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16496 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16498 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16499 be used by the reg-alloc.
16501 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16503 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16504 reg-alloc when having mul64 or mul32x16 instructions.
16505 (mulsidi3): Likewise.
16506 (umulsidi3): Likewise.
16507 (mulsi32x16): New pattern.
16508 (mulsi64): Likewise.
16509 (mulsidi64): Likewise.
16510 (umulsidi64): Likewise.
16511 (MUL32x16_REG): Define.
16512 (mul64_600): Use MUL32x16_REG.
16513 (mac64_600): Likewise.
16514 (umul64_600): Likewise.
16515 (umac64_600): Likewise.
16517 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16519 * config/arc/arc.md (mulsi3_700): Make it commutative.
16521 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16523 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16525 (*sign_extendsidi2_insn): Likewise for movstosw.
16527 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16529 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16530 the type of the input discriminant value. Convert the
16531 discriminant value of signedness vary.
16533 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16535 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16536 Add to -Wall section.
16538 2017-06-01 Richard Biener <rguenther@suse.de>
16540 PR middle-end/66313
16541 * fold-const.c (fold_plusminus_mult_expr): If the factored
16542 factor may be zero use a wrapping type for the inner operation.
16543 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16544 and handle moved defs.
16545 (process_assignment): Properly guard the unary op case. Return a
16546 tri-state indicating that moving the stmt before the call may allow
16547 to continue. Pass through to_move.
16548 (find_tail_calls): Handle moving unrelated defs before
16551 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16554 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16555 splitter result in the canonical way.
16557 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16559 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16560 also for 32bit target. Update insn attributes.
16561 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16563 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16565 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16566 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16567 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16568 (_mm_maskz_min_ss): New intrinsics.
16570 2017-05-31 Martin Liska <mliska@suse.cz>
16572 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16573 Change comment style to one we normally use.
16574 (vectorizable_reduction): Likewise.
16575 (vectorizable_induction): Likewise.
16576 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16577 (vectorizable_call): Likewise.
16578 (vectorizable_simd_clone_call): Likewise.
16579 (vectorizable_conversion): Likewise.
16580 (vectorizable_assignment): Likewise.
16581 (vectorizable_shift): Likewise.
16582 (vectorizable_operation): Likewise.
16583 (vectorizable_store): Likewise.
16584 (vectorizable_load): Likewise.
16585 * tree-vectorizer.h: Likewise.
16587 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16589 * passes.c (emergency_dump_function): New.
16590 * tree-pass.h (emergency_dump_function): Declare.
16591 * plugin.c (plugins_internal_error_function): Remove.
16592 * plugin.h (plugins_internal_error_function): Remove declaration.
16593 * toplev.c (internal_error_function): New static function. Use it...
16594 (general_init): ...here.
16596 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16598 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16599 (arc_rtx_costs): Add costs for new patterns.
16600 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16601 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16603 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16605 * tree-ssa-strlen.c (get_next_strinfo): New function.
16606 (get_stridx_plus_constant): Use it.
16607 (zero_length_string): Likewise.
16608 (adjust_related_strinfos): Likewise.
16609 (adjust_last_stmt): Likewise.
16611 2017-05-31 Richard Biener <rguenther@suse.de>
16614 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16615 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16617 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16619 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16620 loop_vinfo argument and use of dependence distance vectors.
16621 Check instead whether the two references differ only in their
16622 initial value and assume that they have the same alignment if the
16623 difference is a multiple of the vector alignment.
16624 (vect_analyze_data_refs_alignment): Update call accordingly.
16626 2017-05-31 Martin Liska <mliska@suse.cz>
16629 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16631 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16633 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16634 (create_intersect_range_checks): Move from ...
16635 * tree-data-ref.c (create_intersect_range_checks_index)
16636 (create_intersect_range_checks): ... to here.
16637 (create_runtime_alias_checks): New function factored from ...
16638 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16639 here. Call above function.
16640 * tree-data-ref.h (create_runtime_alias_checks): New function.
16642 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16644 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16645 segment length for dr_b and compute it in wide_int.
16647 2017-05-31 Richard Biener <rguenther@suse.de>
16649 PR tree-optimization/80906
16650 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16651 and pass through iv_map.
16652 (copy_bb_and_scalar_dependences): Adjust.
16653 (translate_pending_phi_nodes): Likewise.
16654 (copy_loop_close_phi_args): Handle code-generating IVs instead
16657 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16659 * diagnostic-color.c (color_dict): Add "type-diff".
16660 (parse_gcc_colors): Update comment.
16661 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16662 -fdiagnostics-show-template-tree and -fno-elide-type.
16663 (GCC_COLORS): Add type-diff to example.
16665 (-fdiagnostics-show-template-tree): New.
16666 (-fno-elide-type): New.
16667 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16668 the pp_format_decoder callback. Call any m_format_postprocessor's
16670 (pretty_printer::pretty_printer): Initialize
16671 m_format_postprocessor.
16672 (pretty_printer::~pretty_printer): Delete any
16673 m_format_postprocessor.
16674 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16675 (class format_postprocessor): New class.
16676 (struct pretty_printer::format_decoder): Document the new parameters.
16677 (struct pretty_printer::m_format_postprocessor): New field.
16678 * tree-diagnostic.c (default_tree_printer): Update for new
16679 bool and const char ** params.
16680 * tree-diagnostic.h (default_tree_printer): Likewise.
16682 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16684 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16685 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16686 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16687 rs6000_gen_cell_microcode code.
16688 (rs6000_final_prescan_insn): Delete.
16689 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16690 "warn-cell-microcode" entries.
16691 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16692 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16693 throughout. Change cc_reg_not_micro_cr0_operand to
16694 cc_reg_not_cr0_operand throughout.
16695 (*extendhi<mode>2_noload): Delete.
16696 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16697 (mwarn-cell-microcode): Delete.
16698 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16699 -mgen-cell-microcode and -mwarn-cell-microcode.
16701 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16704 * config/i386/constraints.md (Yd): New constraint.
16706 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16707 and (?Yd, r) alternatives. Update insn attributes.
16708 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16709 and (?*Yd, r) alternatives. Update insn attributes.
16710 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16712 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16714 * gimplify.c (gimplify_modify_expr): Don't create a
16715 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16718 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16720 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16722 2017-05-30 Richard Biener <rguenther@suse.de>
16724 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16725 and reduc_def fields.
16726 (STMT_VINFO_REDUC_TYPE): New define.
16727 (STMT_VINFO_REDUC_DEF): Likewise.
16728 (vect_force_simple_reduction): Adjust prototype.
16729 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16730 (vect_is_simple_reduction): Remove check_reduction argument.
16731 (vect_force_simple_reduction): Adjust and set
16732 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16733 (vectorizable_reduction): Do not re-do reduction analysis
16734 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16735 * tree-parloops.c (gather_scalar_reductions): Adjust.
16737 2017-05-30 Richard Biener <rguenther@suse.de>
16739 PR middle-end/80901
16740 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16743 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16745 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16746 Introduce unknown_misalignment parameter and remove vf.
16747 (vect_peeling_hash_get_lowest_cost):
16748 Pass unknown_misalignment parameter.
16749 (vect_enhance_data_refs_alignment):
16750 Fix unsupportable data ref treatment.
16752 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16754 * tree-vect-data-refs.c (vect_get_data_access_cost):
16755 Workaround for SLP handling.
16756 (vect_enhance_data_refs_alignment):
16757 Compute costs for doing no peeling at all, compare to the best
16758 peeling costs so far and avoid peeling if cheaper.
16760 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16762 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16763 Return peeling info and set costs to zero for unlimited cost
16765 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16766 with unknown misalignment. Compute and costs for unknown
16767 misalignment, compare them to the costs for known misalignment
16768 and choose the cheapest for peeling.
16770 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16772 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16773 (vect_get_peeling_costs_all_drs): Create function.
16774 (vect_peeling_hash_get_lowest_cost):
16775 Use vect_get_peeling_costs_all_drs.
16776 (vect_peeling_supportable): Create function.
16777 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16779 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16781 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16782 DR_HAS_NEGATIVE_STEP.
16783 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16784 (vect_enhance_data_refs_alignment): Use.
16785 (vect_duplicate_ssa_name_ptr_info): Use.
16786 * tree-vectorizer.h (dr_misalignment): Use.
16787 (known_alignment_for_access_p): Use.
16789 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16792 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16794 (has_section_name): New function.
16796 2017-05-30 Martin Liska <mliska@suse.cz>
16799 * auto-profile.c (get_function_decl_from_block): Fix
16802 2017-05-30 Richard Biener <rguenther@suse.de>
16804 PR middle-end/80876
16805 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16807 2017-05-30 Martin Liska <mliska@suse.cz>
16809 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16810 * dumpfile.h (struct dump_file_info): Remove ctors.
16812 2017-05-30 Martin Liska <mliska@suse.cz>
16814 * predict.def: Fix GNU coding style.
16816 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16818 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16819 Mark 'to' argument with ATTRIBUTE_UNUSED.
16821 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16823 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16824 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16825 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16828 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16830 * doc/install.texi (Options specification): Restore entry of
16831 --enable-sjlj-exceptions.
16833 2017-05-27 Michael Eager <eager@eagercon.com>
16836 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16838 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16840 * config/microblaze/microblaze.h
16841 (FIXED_REGISTERS): Update in macro.
16842 (CALL_USED_REGISTERS): Update in macro.
16844 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16846 * doc/install.texi: Add links to macOS binary distributions.
16848 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16852 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16854 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16856 2017-05-26 Martin Liska <mliska@suse.cz>
16858 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16860 2017-05-26 Martin Liska <mliska@suse.cz>
16862 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16863 always leading ';; '.
16864 (dump_bb_info): Likewise.
16865 (brief_dump_cfg): Likewise.
16866 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16867 * dumpfile.c: Remove usage of TDF_VERBOSE.
16868 * dumpfile.h (enum dump_kind): Likewise.
16869 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16870 * print-tree.c (debug_verbose): Remove.
16871 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16872 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16873 * tree-diagnostic.c (default_tree_printer): Replace
16874 TDF_DIAGNOSTIC with TDF_SLIM.
16876 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16878 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16879 in parameter loop, rather than loop_vinfo.
16880 (create_intersect_range_checks): Ditto.
16881 (vect_create_cond_for_alias_checks): Update call to above functions.
16883 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16885 PR tree-optimization/80815
16886 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16887 for merging runtime alias checks. Handle negative DR_STEPs.
16889 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16891 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16893 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16894 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16895 out code pruning runtime alias checks.
16896 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16897 factored out from above.
16898 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16900 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16902 (prune_runtime_alias_test_list): New decalaration.
16904 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16906 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16907 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16908 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16909 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16910 (operator==, comp_dr_with_seg_len_pair): Ditto.
16911 (vect_prune_runtime_alias_test_list): Ditto.
16913 2017-05-26 Martin Liska <mliska@suse.cz>
16916 * params.def: Bound partial-inlining-entry-probability param.
16918 2017-05-26 Marek Polacek <polacek@redhat.com>
16921 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16924 2017-05-26 Richard Biener <rguenther@suse.de>
16926 PR tree-optimization/80842
16927 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16930 2017-05-26 Richard Biener <rguenther@suse.de>
16932 PR tree-optimization/80844
16933 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16935 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16937 * doc/md.texi (Machine Constraints): Update x86 family
16938 machine constraints section to match 'config/i386/constraints.md'.
16940 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16942 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16944 2017-05-25 Nathan Sidwell <nathan@acm.org>
16946 * doc/invoke.texi (--enable-languages): Update documentation.
16948 2017-05-25 Martin Liska <mliska@suse.cz>
16950 * dumpfile.c: Add TDF_FOLDING.
16951 * dumpfile.h (enum dump_kind): Likewise.
16952 * genmatch.c (dt_simplify::gen_1): Use it.
16954 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16956 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16958 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16960 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16961 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16963 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16965 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16966 * match.pd (X == C): Rewrite it here.
16967 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16968 with_certain_nonzero_bits2): New predicates.
16969 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16971 2017-05-24 Nathan Sidwell <nathan@acm.org>
16973 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16976 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16979 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16981 * config/powerpcspe: New port. Files are copied from the rs6000
16982 port, with "rs6000" in filenames replaced by "powerpcspe".
16984 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16986 PR rtl-optimization/80754
16987 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16989 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16991 * config/sparc/sparc.md (length): Return the correct value for -mflat
16992 sibcalls to match output_sibcall.
16994 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16998 * config/rs6000/rs6000.c (struct machine_function): Add new field
17000 (rs6000_get_separate_components): Init that field, use it.
17001 (rs6000_components_for_bb): Use the field.
17003 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17005 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17007 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17009 PR middle-end/80823
17010 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17012 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17015 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17016 address against address_operand predicate.
17017 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17019 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17021 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17022 parameters passed indirectly.
17024 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17026 * config/i386/i386.md (*movdi_internal): Remove SSE4
17027 alternative 18 (?r, *v). Update insn attributes.
17028 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17029 Update insn attributes.
17030 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17031 Update insn attributes.
17032 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17033 alternative 1 (r, v). Remove isa attribute.
17034 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17035 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17036 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17038 2017-05-23 Tom de Vries <tom@codesourcery.com>
17040 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17043 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17045 * cgraphunit.c (symbol_table::process_new_functions): Update.
17046 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17047 (inline_generate_summary): Rename to ...
17048 (ipa_fn_summary_generate): ... this one.
17049 (inline_read_summary): Rename to ...
17050 (ipa_fn_summary_read): ... this one.
17051 (inline_write_summary): Rename to ...
17052 (ipa_fn_summary_write): ... this one.
17053 (inline_free_summary): Rename to ...
17054 (ipa_free_fn_summary): ... this one.
17055 (pass_data_local_fn_summary, pass_local_fn_summary,
17056 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17057 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17058 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17059 make_pass_ipa_fn_summary): New.
17060 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17061 inline_write_summary, inline_free_summary): Remove.
17062 (ipa_free_fn_summary) : New.
17063 * ipa-inline.c (ipa_inline): Update.
17064 (pass_ipa_inline): Do not generate summaries.
17065 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17067 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17068 and add pass_ipa_fn_summary.
17069 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17071 (make_pass_inline_parameters): Remove.
17073 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17075 * omp-low.c (struct omp_context): Remove "default_kind" member.
17078 * omp-offload.c (execute_oacc_device_lower): Remove the
17079 parallelism dimensions function attributes for unparallelized
17080 OpenACC kernels constructs.
17082 2017-05-23 Martin Liska <mliska@suse.cz>
17084 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17086 (cgraph_edge::make_speculative): Likewise.
17087 (cgraph_edge::resolve_speculation): Likewise.
17088 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17089 (cgraph_node::dump): Likewise.
17090 * cgraph.h: Likewise.
17091 * cgraphunit.c (analyze_functions): Likewise.
17092 (symbol_table::compile): Likewise.
17093 * ipa-cp.c (print_all_lattices): Likewise.
17094 (determine_versionability): Likewise.
17095 (initialize_node_lattices): Likewise.
17096 (ipcp_verify_propagated_values): Likewise.
17097 (estimate_local_effects): Likewise.
17098 (update_profiling_info): Likewise.
17099 (create_specialized_node): Likewise.
17100 (perhaps_add_new_callers): Likewise.
17101 (decide_about_value): Likewise.
17102 (decide_whether_version_node): Likewise.
17103 (identify_dead_nodes): Likewise.
17104 (ipcp_store_bits_results): Likewise.
17105 * ipa-devirt.c (dump_targets): Likewise.
17106 (ipa_devirt): Likewise.
17107 * ipa-icf.c (sem_item::dump): Likewise.
17108 (sem_function::equals): Likewise.
17109 (sem_variable::equals): Likewise.
17110 (sem_item_optimizer::read_section): Likewise.
17111 (sem_item_optimizer::execute): Likewise.
17112 (congruence_class::dump): Likewise.
17113 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17114 (dump_inline_summary): Likewise.
17115 (estimate_node_size_and_time): Likewise.
17116 (inline_analyze_function): Likewise.
17117 * ipa-inline-transform.c (inline_call): Likewise.
17118 * ipa-inline.c (report_inline_failed_reason): Likewise.
17119 (want_early_inline_function_p): Likewise.
17120 (edge_badness): Likewise.
17121 (update_edge_key): Likewise.
17122 (inline_small_functions): Likewise.
17123 * ipa-profile.c (ipa_profile): Likewise.
17124 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17125 (ipa_make_edge_direct_to_target): Likewise.
17126 (remove_described_reference): Likewise.
17127 (ipa_impossible_devirt_target): Likewise.
17128 (propagate_controlled_uses): Likewise.
17129 (ipa_print_node_params): Likewise.
17130 (ipcp_transform_function): Likewise.
17131 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17132 (propagate_pure_const): Likewise.
17133 * ipa-reference.c (generate_summary): Likewise.
17134 (read_write_all_from_decl): Likewise.
17135 (propagate): Likewise.
17136 (ipa_reference_read_optimization_summary): Likewise.
17137 * ipa-utils.c (ipa_merge_profiles): Likewise.
17138 * ipa.c (walk_polymorphic_call_targets): Likewise.
17139 (symbol_table::remove_unreachable_nodes): Likewise.
17140 (ipa_single_use): Likewise.
17141 * passes.c (execute_todo): Likewise.
17142 * predict.c (drop_profile): Likewise.
17143 * symtab.c (symtab_node::get_dump_name): New function.
17144 (symtab_node::dump_name): Likewise.
17145 (symtab_node::dump_asm_name): Likewise.
17146 (symtab_node::dump_references): Likewise.
17147 (symtab_node::dump_referring): Likewise.
17148 (symtab_node::dump_base): Likewise.
17149 (symtab_node::debug_symtab): Likewise.
17150 * tree-sra.c (convert_callers_for_node): Likewise.
17151 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17152 * value-prof.c (init_node_map): Likewise.
17154 2017-05-23 Martin Liska <mliska@suse.cz>
17156 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17157 and symtab_node::debug_symtab to symbol_table::debug.
17158 * cgraphunit.c (analyze_functions): Use the renamed function.
17159 (symbol_table::compile): Likewise.
17160 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17161 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17162 * passes.c (execute_todo): Likewise.
17163 * symtab.c (symbol_table::dump): New function.
17164 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17166 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17168 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17169 that nonconst implies exec.
17171 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17173 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17174 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17175 (inline_edge_summary_vec): Turn into ...
17176 (ipa_call_summaries): ... this one.
17177 (redirect_to_unreachable, edge_set_predicate,
17178 evaluate_properties_for_edge, inline_summary_alloc,
17179 reset_ipa_call_summary, reset_inline_summary,
17180 inline_summary_t::duplicate): Update.
17181 (inline_edge_duplication_hook): Turn to ...
17182 (ipa_call_summary_t::duplicate): ... this one.
17183 (inline_edge_removal_hook): Turn to ...
17184 (ipa_call_summary_t::remove): ... this one.
17185 (dump_inline_edge_summary): Turn to ...
17186 (dump_ipa_call_summary): ... this one.
17187 (estimate_function_body_sizes): Update.
17188 (inline_update_callee_summaries): Update.
17189 (remap_edge_change_prob): Update.
17190 (remap_edge_summaries): Update.
17191 (inline_merge_summary): Update.
17192 (do_estimate_edge_time): Update.
17193 (inline_generate_summary): Update.
17194 (inline_read_section): Update.
17195 (inline_read_summary): Update.
17196 (inline_free_summary): Update.
17197 * ipa-inline.c (can_inline_edge_p): Update.
17198 (compute_inlined_call_time): Update.
17199 (want_inline_small_function_p): Update.
17200 (edge_badness): Update.
17201 (early_inliner): Update.
17202 * ipa-inline.h (inline_edge_summary): Turn to ...
17203 (ipa_call_summary): ... this one.
17204 (ipa_call_summary_t): New class.
17205 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17206 (ipa_call_summaries): New.
17207 (inline_edge_summary): Remove.
17208 (estimate_edge_growth): Update.
17209 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17210 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17211 * ipa-split.c (execute_split_functions): Update.
17212 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17214 2017-05-23 Tom de Vries <tom@codesourcery.com>
17216 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17217 attributes): Document rdrand effective target.
17219 2017-05-23 Tom de Vries <tom@codesourcery.com>
17221 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17222 attributes): Sort alphabetically.
17224 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17226 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17228 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17231 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17232 V2DF/V2DI splat into two separate patterns, one that handles
17233 registers, and the other that only handles memory. Drop support
17234 for splatting from a GPR on ISA 2.07 and then splitting the
17235 splat into direct move and splat.
17236 (vsx_splat_<mode>_reg): Likewise.
17237 (vsx_splat_<mode>_mem): Likewise.
17239 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17241 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17243 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17245 PR middle-end/80809
17246 * omp-low.c (finish_taskreg_remap): New function.
17247 (finish_taskreg_scan): If unit size of ctx->record_type
17248 is non-constant, unshare the size expression and replace
17249 decls in it with possible outer var refs.
17251 PR middle-end/80809
17252 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17253 GOVD_SHARED rather than GOVD_PRIVATE with it.
17254 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17255 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17257 PR middle-end/80853
17258 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17259 as last argument to build_outer_var_ref for pointer bases of array
17260 section reductions.
17262 2017-05-19 Martin Sebor <msebor@redhat.com>
17264 * print-tree.c (print_node): Print DECL_READ_P flag.
17266 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17268 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17269 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17270 * cgraph.c: Likewise.
17271 * cgraphunit.c: Likewise.
17272 * gengtype.c: Likewise.
17273 * ipa-cp.c: Likewise.
17274 * ipa-devirt.c: Likewise.
17275 * ipa-icf.c: Likewise.
17276 * ipa-predicate.c: Likewise.
17277 * ipa-profile.c: Likewise.
17278 * ipa-prop.c: Likewise.
17279 * ipa-split.c: Likewise.
17281 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17282 edge_predicate_pool, dump_inline_hints,
17283 inline_summary::account_size_time, redirect_to_unreachable,
17284 edge_set_predicate, set_hint_predicate,
17285 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17286 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17287 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17288 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17289 ipa_call_summary_t::remove, initialize_growth_caches,
17290 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17291 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17292 mark_modified, unmodified_parm_1, unmodified_parm,
17293 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17294 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17295 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17296 will_be_nonconstant_predicate, record_modified_bb_info,
17297 get_minimal_bb, record_modified, param_change_prob,
17298 phi_result_unknown_predicate, predicate_for_phi_result,
17299 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17300 estimate_function_body_sizes, compute_inline_parameters,
17301 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17302 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17303 inline_update_callee_summaries, remap_edge_change_prob,
17304 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17305 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17306 inline_analyze_function, inline_summary_t::insert,
17307 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17308 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17309 inline_free_summary): Move to ipa-fnsummary.h
17310 (predicate_t): Remove.
17311 * ipa-fnsummary.c: New file.
17312 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17313 (enum inline_hints_vals, inline_hints, agg_position_info,
17314 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17315 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17316 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17317 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17318 inline_read_summary, inline_write_summary, inline_free_summary,
17319 inline_analyze_function, initialize_inline_failed,
17320 inline_merge_summary, inline_update_overall_summary,
17321 compute_inline_parameters): Move to ipa-fnsummary.h
17322 * ipa-fnsummary.h: New file.
17323 * ipa-inline-transform.h: Include ipa-inline.h.
17324 * ipa-inline.c: LIkewise.
17326 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17328 * ipa-inline.c (edge_badness): Use inlined_time instead of
17329 inline_summaries->get.
17331 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17333 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17335 2017-05-22 Nathan Sidwell <nathan@acm.org>
17337 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17338 (fdump-lang): Document 'raw' option.
17339 * dumpfile.h (TDI_tu): Delete.
17340 * dumpfile.c (dump_files): Remove translation-unit.
17341 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17343 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17345 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17346 command option from $(AWK) call.
17347 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17348 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17349 [FORMAT]: Remove handling of variable.
17350 * config/avr/t-multilib: Regenerate.
17352 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17354 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17356 (dump_inline_summary): Do not print self_time.
17357 (estimate_function_body_sizes): Do not set self_time.
17358 (compute_inline_parameters): Likewise.
17359 (inline_read_section, inline_write_summary): Do not stream self_time.
17360 * ipa-inline.h (inline_summary): Drop self_time.
17362 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17364 * ipa-inline-analysis.c (account_size_time): Rename to ...
17365 (inline_summary::account_size_time): ... this one.
17366 (reset_ipa_call_summary): Turn to ...
17367 (ipa_call_summary::reset): ... this one.
17368 (reset_inline_summary): Turn to ...
17369 (inline_summary::reset): ... this one.
17370 (inline_summary_t::remove): Update.
17371 (inline_summary_t::duplicate): Update.
17372 (ipa_call_summary_t::remove): Update.
17373 (dump_inline_summary): Update.
17374 (estimate_function_body_sizes): Update.
17375 (compute_inline_parameters): Update.
17376 (estimate_node_size_and_time): Update.
17377 (inline_merge_summary): Update.
17378 (inline_update_overall_summary): Update.
17379 (inline_read_section): Update.
17380 (inline_write_summary): Update.
17381 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17382 add account_size_time and reset member functions.
17383 (ipa_call_summary): Add reset function.
17384 * ipa-predicate.h (predicate::operator &): Constify.
17386 2017-05-22 Richard Biener <rguenther@suse.de>
17388 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17390 2017-05-19 Jason Merrill <jason@redhat.com>
17392 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17394 2017-05-19 Marek Polacek <polacek@redhat.com>
17397 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17398 TYPE_OVERFLOW_WRAPS checks.
17400 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17402 * tree-core.h (enum omp_clause_default_kind): Add
17403 "OMP_CLAUSE_DEFAULT_PRESENT".
17404 * tree-pretty-print.c (dump_omp_clause): Handle it.
17405 * gimplify.c (enum gimplify_omp_var_data): Add
17406 "GOVD_MAP_FORCE_PRESENT".
17407 (gimplify_adjust_omp_clauses_1): Map it to
17408 "GOMP_MAP_FORCE_PRESENT".
17409 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17411 * gimplify.c (oacc_default_clause): Clarify.
17413 2017-05-19 Nathan Sidwell <nathan@acm.org>
17415 LANG_HOOK_REGISTER_DUMPS
17416 * toplev.c (general_init): Call register dump lang hook.
17417 * doc/invoke.texi: Document -fdump-lang option family.
17418 * dumpfile.c (dump_files): Remove class dump here.
17419 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17420 * dumpfile.h (tree_dump_index): Remove TDI_class.
17421 * langhooks-def.h (lhd_register_dumps): Declare.
17422 (LANG_HOOKS_REGISTER_DUMPS): Define.
17423 (LANG_HOOKS_INITIALIZER): Add it.
17424 * langhooks.c (lhd_register_dumps): Define.
17425 * langhooks.h (struct lang_hooks): Add register_dumps.
17427 2017-05-19 Nathan Sidwell <nathan@acm.org>
17429 * context.h (context::set_passes): New.
17430 * context.c (context::context): Do not create pass manager.
17431 * toplev.c (general_init): Create pass manager here.
17433 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17435 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17436 use this splitter if two add or or instructions would also work for
17437 the constant we want to generate.
17439 2017-05-19 Richard Biener <rguenther@suse.de>
17442 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17443 predicate evaluation.
17445 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17447 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17449 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17452 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17454 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17455 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17456 (inline_edge_summary_vec): Turn into ...
17457 (ipa_call_summaries): ... this one.
17458 (redirect_to_unreachable, edge_set_predicate,
17459 evaluate_properties_for_edge, inline_summary_alloc,
17460 reset_ipa_call_summary, reset_inline_summary,
17461 inline_summary_t::duplicate): Update.
17462 (inline_edge_duplication_hook): Turn to ...
17463 (ipa_call_summary_t::duplicate): ... this one.
17464 (inline_edge_removal_hook): Turn to ...
17465 (ipa_call_summary_t::remove): ... this one.
17466 (dump_inline_edge_summary): Turn to ...
17467 (dump_ipa_call_summary): ... this one.
17468 (estimate_function_body_sizes): Update.
17469 (inline_update_callee_summaries): Update.
17470 (remap_edge_change_prob): Update.
17471 (remap_edge_summaries): Update.
17472 (inline_merge_summary): Update.
17473 (do_estimate_edge_time): Update.
17474 (inline_generate_summary): Update.
17475 (inline_read_section): Update.
17476 (inline_read_summary): Update.
17477 (inline_free_summary): Update.
17478 * ipa-inline.c (can_inline_edge_p): Update.
17479 (compute_inlined_call_time): Update.
17480 (want_inline_small_function_p): Update.
17481 (edge_badness): Update.
17482 (early_inliner): Update.
17483 * ipa-inline.h (inline_edge_summary): Turn to ...
17484 (ipa_call_summary): ... this one.
17485 (ipa_call_summary_t): New class.
17486 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17487 (ipa_call_summaries): New.
17488 (inline_edge_summary): Remove.
17489 (estimate_edge_growth): Update.
17490 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17491 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17492 * ipa-split.c (execute_split_functions): Update.
17493 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17495 2017-05-19 Richard Biener <rguenther@suse.de>
17497 PR middle-end/80764
17498 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17500 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17502 * config/rs6000/rs6000.c (struct machine_function): Add field
17503 fpr_is_wrapped_separately.
17504 (rs6000_get_separate_components): Use 64 components. Handle the
17505 new FPR components.
17506 (rs6000_components_for_bb): Handle the FPR components.
17507 (rs6000_emit_prologue_components): Handle the FPR components.
17508 (rs6000_emit_epilogue_components): Handle the FPR components.
17509 (rs6000_set_handled_components): Handle the FPR components.
17510 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17511 that are already separately shrink-wrapped.
17512 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17513 that are already separately shrink-wrapped.
17515 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17518 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17521 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17522 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17523 cases where the register allocator uses a move and an offsettable
17524 memory operation to/from a FPR register on ISA 2.06/2.07.
17525 (define_peephole2 for Altivec d-form store): Likewise.
17527 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17530 * config/i386/mmx.md (*mov<mode>_internal): Enable
17531 alternatives 11, 12, 13 and 14 also for 32bit targets.
17532 Remove alternatives 15, 16, 17 and 18.
17533 * config/i386/sse.md (vec_concatv2di): Change
17534 alternative (!x, *y) to (x, ?!*Yn).
17536 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17538 * dumpfile.h (enum dump_kind): Remove stray comma.
17540 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17542 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17543 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17544 predicate::num_conditions
17545 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17546 (CHANGED): turn into predicate::changed.
17547 (agg_position_info): Move to ipa-predicate.h
17548 (add_condition, predicate::add_clause, predicate::operator &=,
17549 predicate::or_with, predicate::evaluate, predicate::probability,
17550 dump_condition, dump_clause, predicate::dump,
17551 predicate::remap_after_duplication, predicate::remap_after_inlining,
17552 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17553 (evaluate_conditions_for_known_args): Update.
17554 (set_cond_stmt_execution_predicate): Update.
17555 * ipa-inline.h: Include ipa-predicate.h
17556 (condition, inline_param_summary, conditions, agg_position_info,
17557 predicate): Move to ipa-predicate.h
17558 * ipa-predicate.c: New file.
17559 * ipa-predicate.h: New file.
17561 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17563 * final.c (leaf_function_p): Check we are not in a sequence.
17565 2017-05-18 Martin Liska <mliska@suse.cz>
17567 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17568 * dumpfile.c (dump_register): Use new enum dump_kind.
17569 (get_dump_file_name): Likewise.
17570 (dump_enable_all): Likewise.
17571 (dump_switch_p_1): Likewise.
17572 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17573 * dumpfile.h (enum dump_kind): New enum type.
17574 (struct dump_file_info): Create constructor and
17575 format fields and comments.
17576 * passes.c (pass_manager::register_one_dump_file):
17578 * statistics.c (statistics_early_init): Likewise.
17579 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17580 TDF_TREE with TDF_SLIM.
17581 (gather_memory_references_ref): Likewise.
17583 2017-05-18 Martin Liska <mliska@suse.cz>
17585 * vec.h (struct vnull): Use it.
17587 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17589 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17590 (true_predicate, false_predicate, true_predicate_p,
17591 false_predicate_p): Remove.
17592 (single_cond_predicate, not_inlined_predicate): Turn to member function
17594 (add_condition): Update.
17595 (add_clause): Turn to...
17596 (predicate::add_clause): ... this one; update; allow passing NULL
17598 (and_predicates): Turn to ...
17599 (predicate::operator &=): ... this one.
17600 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17601 (or_predicates): Turn to ...
17602 (predicate::or_with): ... this one.
17603 (evaluate_predicate): Turn to ...
17604 (predicate::evaluate): ... this one.
17605 (predicate_probability): Turn to ...
17606 (predicate::probability): ... this one.
17607 (dump_condition): Update.
17608 (dump_predicate): Turn to ...
17609 (predicate::dump): ... this one.
17610 (account_size_time): Update.
17611 (edge_set_predicate): Update.
17612 (set_hint_predicate): UPdate.
17613 (evaluate_conditions_for_known_args): Update.
17614 (evaluate_properties_for_edge): Update.
17615 (remap_predicate_after_duplication): Turn to...
17616 (predicate::remap_after_duplication): ... this one.
17617 (remap_hint_predicate_after_duplication): Update.
17618 (inline_summary_t::duplicate): UPdate.
17619 (dump_inline_edge_summary): Update.
17620 (dump_inline_summary): Update.
17621 (set_cond_stmt_execution_predicate): Update.
17622 (set_switch_stmt_execution_predicate): Update.
17623 (compute_bb_predicates): Update.
17624 (will_be_nonconstant_expr_predicate): Update.
17625 (will_be_nonconstant_predicate): Update.
17626 (phi_result_unknown_predicate): Update.
17627 (predicate_for_phi_result): Update.
17628 (array_index_predicate): Update.
17629 (estimate_function_body_sizes): Update.
17630 (estimate_node_size_and_time): Update.
17631 (estimate_ipcp_clone_size_and_time): Update.
17632 (remap_predicate): Rename to ...
17633 (predicate::remap_after_inlining): ... this one.
17634 (remap_hint_predicate): Update.
17635 (inline_merge_summary): Update.
17636 (inline_update_overall_summary): Update.
17637 (estimate_size_after_inlining): Update.
17638 (read_predicate): Rename to ...
17639 (predicate::stream_in): ... this one.
17640 (read_inline_edge_summary): Update.
17641 (write_predicate): Rename to ...
17642 (predicate::stream_out): ... this one.
17643 (write_inline_edge_summary): Update.
17644 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17645 (clause_t): Turn to uint32_t
17646 (predicate): Turn to class; implement constructor and operators
17648 (size_time_entry): Update.
17649 (inline_summary): Update.
17650 (inline_edge_summary): Update.
17652 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17654 * fold-const.c (fold_binary_loc): Move transformation...
17655 * match.pd (C - X CMP X): ... here.
17657 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17659 * config/sparc/sparc.c (sparc_option_override): Set function
17660 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17661 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17663 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17665 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17667 2017-05-18 Marek Polacek <polacek@redhat.com>
17670 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17671 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17672 gimple_assign_load_p.
17674 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17676 PR middle-end/80692
17677 * real.c (do_compare): Give decimal_do_compare preference over
17678 comparing just the signs.
17680 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17682 * doc/md.texi (Canonicalization of Instructions): Describe the
17683 canonical form of instructions that inherently set a condition
17686 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17688 PR middle-end/80775
17689 * tree-cfg.c: Move deletion of unreachable case statements to after
17690 the merging of consecutive case labels.
17692 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17694 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17695 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17696 restoring of callee-saved registers.
17698 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17700 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17701 * config/visium/visium.c (single_set_and_flags): Likewise.
17702 * config/visium/visium.md (Substitutions): Likewise.
17704 2017-05-17 Martin Liska <mliska@suse.cz>
17706 * cfg.c: Introduce dump_flags_t type and
17707 use it instead of int type.
17709 * cfghooks.c: Likewise.
17710 * cfghooks.h (struct cfg_hooks): Likewise.
17711 * cfgrtl.c: Likewise.
17712 * cfgrtl.h: Likewise.
17713 * cgraph.c (cgraph_node::get_body): Likewise.
17714 * coretypes.h: Likewise.
17715 * domwalk.c: Likewise.
17716 * domwalk.h: Likewise.
17717 * dumpfile.c (struct dump_option_value_info): Likewise.
17718 (dump_enable_all): Likewise.
17719 (dump_switch_p_1): Likewise.
17720 (opt_info_switch_p): Likewise.
17721 * dumpfile.h (enum tree_dump_index): Likewise.
17722 (struct dump_file_info): Likewise.
17723 * genemit.c: Likewise.
17724 * generic-match-head.c: Likewise.
17725 * gengtype.c (open_base_files): Likewise.
17726 * gimple-pretty-print.c: Likewise.
17727 * gimple-pretty-print.h: Likewise.
17728 * graph.c (print_graph_cfg): Likewise.
17729 * graphite-scop-detection.c (dot_all_sese): Likewise.
17730 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17731 * loop-unroll.c (report_unroll): Likewise.
17732 * passes.c (pass_manager::register_one_dump_file): Likewise.
17733 * print-tree.c: Likewise.
17734 * statistics.c: Likewise.
17735 * tree-cfg.c: Likewise.
17736 * tree-cfg.h: Likewise.
17737 * tree-dfa.c: Likewise.
17738 * tree-dfa.h: Likewise.
17739 * tree-dump.c (dump_function): Likewise.
17740 * tree-dump.h (struct dump_info): Likewise.
17741 * tree-pretty-print.c: Likewise.
17742 * tree-pretty-print.h: Likewise.
17743 * tree-ssa-live.c: Likewise.
17744 * tree-ssa-live.h: Likewise.
17745 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17746 * tree-vect-loop.c: Likewise.
17747 * tree-vect-slp.c: Likewise.
17749 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17750 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17752 PR tree-optimization/80457
17753 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17754 of all arguments to a statement as scalar_to_vec operations.
17755 (vectorizable_call): Adjust call to vect_model_simple_cost for
17757 (vectorizable_conversion): Likewise.
17758 (vectorizable_assignment): Likewise.
17759 (vectorizable_shift): Likewise.
17760 (vectorizable_operation): Likewise.
17761 (vectorizable_comparison): Likewise.
17762 (vect_is_simple_cond): Record the def types for operands.
17763 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17764 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17765 for statement argument count.
17767 2017-05-16 Carl Love <cel@us.ibm.com>
17769 * config/rs6000/rs6000-c: Add support for built-in functions
17770 vector unsigned long long vec_bperm (vector unsigned long long,
17771 vector unsigned char)
17772 vector signed long long vec_mule (vector signed int,
17774 vector unsigned long long vec_mule (vector unsigned int,
17775 vector unsigned int)
17776 vector signed long long vec_mulo (vector signed int,
17778 vector unsigned long long vec_mulo (vector unsigned int,
17779 vector unsigned int)
17780 vector signed char vec_sldw (vector signed char,
17781 vector signed char,
17783 vector unsigned char vec_sldw (vector unsigned char,
17784 vector unsigned char,
17786 vector signed short vec_sldw (vector signed short,
17787 vector signed short,
17789 vector unsigned short vec_sldw (vector unsigned short,
17790 vector unsigned short,
17792 vector signed int vec_sldw (vector signed int,
17795 vector unsigned int vec_sldw (vector unsigned int,
17796 vector unsigned int,
17798 vector signed long long vec_sldw (vector signed long long,
17799 vector signed long long,
17801 vector unsigned long long vec_sldw (vector unsigned long long,
17802 vector unsigned long long,
17804 * config/rs6000/rs6000-c: Add support for built-in functions
17805 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17806 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17807 * doc/extend.texi: Update the built-in documentation for the
17808 new built-in functions.
17810 2017-05-16 Marek Polacek <polacek@redhat.com>
17814 * tree.c (save_expr): Don't fold the expression.
17816 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17818 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17819 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17820 and (?*y,m). Update insn attributes.
17822 2017-05-16 Martin Liska <mliska@suse.cz>
17824 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17825 flags argument of print_gimple_stmt, print_gimple_expr,
17826 print_generic_stmt and print_generic_expr.
17827 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17828 * coretypes.h: Likewise.
17829 * except.c (dump_eh_tree): Likewise.
17830 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17831 * gimple-pretty-print.h: Likewise.
17832 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17833 (backprop::push_to_worklist): Likewise.
17834 (backprop::pop_from_worklist): Likewise.
17835 (backprop::process_use): Likewise.
17836 (backprop::intersect_uses): Likewise.
17837 (note_replacement): Likewise.
17838 * gimple-ssa-store-merging.c
17839 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17840 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17841 (pass_store_merging::execute): Likewise.
17842 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17843 (ssa_base_cand_dump_callback): Likewise.
17844 (dump_incr_vec): Likewise.
17845 (replace_refs): Likewise.
17846 (replace_mult_candidate): Likewise.
17847 (create_add_on_incoming_edge): Likewise.
17848 (create_phi_basis): Likewise.
17849 (insert_initializers): Likewise.
17850 (all_phi_incrs_profitable): Likewise.
17851 (introduce_cast_before_cand): Likewise.
17852 (replace_one_candidate): Likewise.
17853 * gimplify.c (gimplify_expr): Likewise.
17854 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17855 (set_rename): Likewise.
17856 (rename_uses): Likewise.
17857 (copy_loop_phi_nodes): Likewise.
17858 (add_close_phis_to_merge_points): Likewise.
17859 (copy_loop_close_phi_args): Likewise.
17860 (copy_cond_phi_args): Likewise.
17861 (graphite_copy_stmts_from_block): Likewise.
17862 (translate_pending_phi_nodes): Likewise.
17863 * graphite-poly.c (print_pdr): Likewise.
17864 (dump_gbb_cases): Likewise.
17865 (dump_gbb_conditions): Likewise.
17866 (print_scop_params): Likewise.
17867 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17868 (build_cross_bb_scalars_use): Likewise.
17869 (gather_bbs::before_dom_children): Likewise.
17870 * hsa-dump.c (dump_hsa_immed): Likewise.
17871 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17872 (get_replacement_map): Likewise.
17873 * ipa-inline-analysis.c (dump_condition): Likewise.
17874 (estimate_function_body_sizes): Likewise.
17875 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17876 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17877 * ipa-prop.c (ipa_dump_param): Likewise.
17878 (ipa_print_node_jump_functions_for_edge): Likewise.
17879 (ipa_modify_call_arguments): Likewise.
17880 (ipa_modify_expr): Likewise.
17881 (ipa_dump_param_adjustments): Likewise.
17882 (ipa_dump_agg_replacement_values): Likewise.
17883 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17884 * ipa-pure-const.c (check_stmt): Likewise.
17885 (pass_nothrow::execute): Likewise.
17886 * ipa-split.c (execute_split_functions): Likewise.
17887 * omp-offload.c (dump_oacc_loop_part): Likewise.
17888 (dump_oacc_loop): Likewise.
17889 * trans-mem.c (tm_log_emit): Likewise.
17890 (tm_memopt_accumulate_memops): Likewise.
17891 (dump_tm_memopt_set): Likewise.
17892 (dump_tm_memopt_transform): Likewise.
17893 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17894 (print_loop): Likewise.
17895 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17896 (chkp_gather_checks_info): Likewise.
17897 (chkp_get_check_result): Likewise.
17898 (chkp_remove_check_if_pass): Likewise.
17899 (chkp_use_outer_bounds_if_possible): Likewise.
17900 (chkp_reduce_bounds_lifetime): Likewise.
17901 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17902 (chkp_mark_completed_bounds): Likewise.
17903 (chkp_register_incomplete_bounds): Likewise.
17904 (chkp_mark_invalid_bounds): Likewise.
17905 (chkp_maybe_copy_and_register_bounds): Likewise.
17906 (chkp_build_returned_bound): Likewise.
17907 (chkp_get_bound_for_parm): Likewise.
17908 (chkp_build_bndldx): Likewise.
17909 (chkp_get_bounds_by_definition): Likewise.
17910 (chkp_generate_extern_var_bounds): Likewise.
17911 (chkp_get_bounds_for_decl_addr): Likewise.
17912 * tree-chrec.c (chrec_apply): Likewise.
17913 * tree-data-ref.c (dump_data_reference): Likewise.
17914 (dump_subscript): Likewise.
17915 (dump_data_dependence_relation): Likewise.
17916 (analyze_overlapping_iterations): Likewise.
17917 * tree-inline.c (expand_call_inline): Likewise.
17918 (tree_function_versioning): Likewise.
17919 * tree-into-ssa.c (dump_defs_stack): Likewise.
17920 (dump_currdefs): Likewise.
17921 (dump_names_replaced_by): Likewise.
17922 (dump_update_ssa): Likewise.
17923 (update_ssa): Likewise.
17924 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17925 * tree-parloops.c (build_new_reduction): Likewise.
17926 (try_create_reduction_list): Likewise.
17927 (ref_conflicts_with_region): Likewise.
17928 (oacc_entry_exit_ok_1): Likewise.
17929 (oacc_entry_exit_single_gang): Likewise.
17930 * tree-pretty-print.h: Likewise.
17931 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17932 (get_scalar_evolution): Likewise.
17933 (add_to_evolution): Likewise.
17934 (get_loop_exit_condition): Likewise.
17935 (analyze_evolution_in_loop): Likewise.
17936 (analyze_initial_condition): Likewise.
17937 (analyze_scalar_evolution): Likewise.
17938 (instantiate_scev): Likewise.
17939 (number_of_latch_executions): Likewise.
17940 (gather_chrec_stats): Likewise.
17941 (final_value_replacement_loop): Likewise.
17942 (scev_const_prop): Likewise.
17943 * tree-sra.c (dump_access): Likewise.
17944 (disqualify_candidate): Likewise.
17945 (create_access): Likewise.
17946 (reject): Likewise.
17947 (maybe_add_sra_candidate): Likewise.
17948 (create_access_replacement): Likewise.
17949 (analyze_access_subtree): Likewise.
17950 (analyze_all_variable_accesses): Likewise.
17951 (sra_modify_assign): Likewise.
17952 (initialize_constant_pool_replacements): Likewise.
17953 (find_param_candidates): Likewise.
17954 (decide_one_param_reduction): Likewise.
17955 (replace_removed_params_ssa_names): Likewise.
17956 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17957 * tree-ssa-copy.c (dump_copy_of): Likewise.
17958 (copy_prop_visit_cond_stmt): Likewise.
17959 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17960 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17961 (record_equivalences_from_stmt): Likewise.
17962 * tree-ssa-dse.c (compute_trims): Likewise.
17963 (delete_dead_call): Likewise.
17964 (delete_dead_assignment): Likewise.
17965 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17966 (forward_propagate_into_cond): Likewise.
17967 (pass_forwprop::execute): Likewise.
17968 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17969 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17971 (move_computations_worker): Likewise.
17972 (execute_sm): Likewise.
17973 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17974 (remove_exits_and_undefined_stmts): Likewise.
17975 (remove_redundant_iv_tests): Likewise.
17976 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17977 (adjust_iv_update_pos): Likewise.
17978 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17979 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17980 (value_replacement): Likewise.
17981 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17982 * tree-ssa-pre.c (print_pre_expr): Likewise.
17983 (get_representative_for): Likewise.
17984 (create_expression_by_pieces): Likewise.
17985 (insert_into_preds_of_block): Likewise.
17986 (eliminate_insert): Likewise.
17987 (eliminate_dom_walker::before_dom_children): Likewise.
17988 (eliminate): Likewise.
17989 (remove_dead_inserted_code): Likewise.
17990 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17991 * tree-ssa-reassoc.c (get_rank): Likewise.
17992 (eliminate_duplicate_pair): Likewise.
17993 (eliminate_plus_minus_pair): Likewise.
17994 (eliminate_not_pairs): Likewise.
17995 (undistribute_ops_list): Likewise.
17996 (eliminate_redundant_comparison): Likewise.
17997 (update_range_test): Likewise.
17998 (optimize_range_tests_var_bound): Likewise.
17999 (optimize_vec_cond_expr): Likewise.
18000 (rewrite_expr_tree): Likewise.
18001 (rewrite_expr_tree_parallel): Likewise.
18002 (linearize_expr): Likewise.
18003 (break_up_subtract): Likewise.
18004 (linearize_expr_tree): Likewise.
18005 (attempt_builtin_powi): Likewise.
18006 (attempt_builtin_copysign): Likewise.
18007 (transform_stmt_to_copy): Likewise.
18008 (transform_stmt_to_multiply): Likewise.
18009 (dump_ops_vector): Likewise.
18010 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18011 (print_scc): Likewise.
18012 (set_ssa_val_to): Likewise.
18013 (visit_reference_op_store): Likewise.
18014 (visit_use): Likewise.
18015 (sccvn_dom_walker::before_dom_children): Likewise.
18016 (run_scc_vn): Likewise.
18017 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18019 (expr_hash_elt::print): Likewise.
18020 (const_and_copies::pop_to_marker): Likewise.
18021 (const_and_copies::record_const_or_copy_raw): Likewise.
18022 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18023 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18024 (dump_predicates): Likewise.
18025 (find_uninit_use): Likewise.
18026 (warn_uninitialized_phi): Likewise.
18027 (pass_late_warn_uninitialized::execute): Likewise.
18028 * tree-ssa.c (verify_vssa): Likewise.
18029 (verify_ssa): Likewise.
18030 (maybe_optimize_var): Likewise.
18031 * tree-vrp.c (dump_value_range): Likewise.
18032 (dump_all_value_ranges): Likewise.
18033 (dump_asserts_for): Likewise.
18034 (register_edge_assert_for_2): Likewise.
18035 (vrp_visit_cond_stmt): Likewise.
18036 (vrp_visit_switch_stmt): Likewise.
18037 (vrp_visit_stmt): Likewise.
18038 (vrp_visit_phi_node): Likewise.
18039 (simplify_cond_using_ranges_1): Likewise.
18040 (fold_predicate_in): Likewise.
18041 (evrp_dom_walker::before_dom_children): Likewise.
18042 (evrp_dom_walker::push_value_range): Likewise.
18043 (evrp_dom_walker::pop_value_range): Likewise.
18044 (execute_early_vrp): Likewise.
18046 2017-05-16 Richard Biener <rguenther@suse.de>
18048 * dwarf2out.c (loc_list_from_tree_1): Do not create
18049 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18051 2017-05-16 Richard Biener <rguenther@suse.de>
18053 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18055 (note_variable_value_in_expr): If we resolved the decl ref
18056 do not push to the stack.
18058 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18060 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18061 operations in fast-math mode.
18062 (vaddq_f16): Likewise.
18063 (vmul_f16): Likewise.
18064 (vmulq_f16): Likewise.
18065 (vsub_f16): Likewise.
18066 (vsubq_f16): Likewise.
18067 * config/arm/neon.md (add<mode>3): New.
18069 (fma:<VH:mode>3): New. Also remove outdated comment.
18072 2017-05-16 Martin Liska <mliska@suse.cz>
18076 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18077 (odr_types_equivalent_p): Likewise.
18079 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18081 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18083 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18086 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18087 non-interunit SSE move alternatives with '?'.
18088 (zero-extendsidi peephole2): New peephole to skip intermediate
18089 general register in SSE zero-extend sequence.
18091 2017-05-15 Jeff Law <law@redhat.com>
18093 * reorg.c (relax_delay_slots): Create a new variable to hold
18094 the temporary target rather than clobbering TARGET_LABEL.
18096 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18097 missing argument to extract_bit_field call.
18098 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18100 2017-05-15 Martin Liska <mliska@suse.cz>
18103 * gcc.c (process_command): Do not allow empty argument of -o option.
18105 2017-05-15 Renlin Li <renlin.li@arm.com>
18107 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18108 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18109 * config/aarch64/constraints.md (Usf): Add long call check.
18110 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18111 (call_value): Likewise.
18112 (sibcall): Likewise.
18113 (sibcall_value): Likewise.
18115 (call_value_insn): New.
18116 (sibcall_insn): Update rtx pattern.
18117 (sibcall_value_insn): Likewise.
18118 (call_internal): Remove.
18119 (call_value_internal): Likewise.
18120 (sibcall_internal): Likewise.
18121 (sibcall_value_internal): Likewise.
18122 (call_reg): Likewise.
18123 (call_symbol): Likewise.
18124 (call_value_reg): Likewise.
18125 (call_value_symbol): Likewise.
18127 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18130 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18132 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18134 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18135 compatible with CCGOCmode and with CCZmode.
18137 2017-05-14 Martin Sebor <msebor@redhat.com>
18139 PR middle-end/77671
18140 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18141 (gimple_fold_builtin_snprintf): Same.
18142 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18143 (gimple_fold_builtin_snprintf): Same.
18144 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18145 of character types.
18146 (is_call_safe): New function.
18147 (try_substitute_return_value): Call it.
18148 (try_simplify_call): New function.
18149 (pass_sprintf_length::handle_gimple_call): Call it.
18151 2017-05-14 Martin Sebor <msebor@redhat.com>
18153 PR middle-end/80669
18154 * builtins.c (expand_builtin_stpncpy): Simplify.
18156 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18158 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18159 * config/i386/i386.h
18160 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18161 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18162 (struct machine_function): Add new members call_ms2sysv,
18163 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18164 (struct machine_frame_state): New fields sp_realigned and
18165 sp_realigned_offset.
18166 * config/i386/i386.c
18167 (enum xlogue_stub): New enum.
18168 (enum xlogue_stub_sets): New enum.
18169 (class xlogue_layout): New class.
18170 (struct ix86_frame): New fields stack_realign_allocate_offset,
18171 stack_realign_offset and outlined_save_offset. Modify comments to
18172 detail stack layout when using out-of-line stubs.
18173 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18174 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18175 -mcall-ms2sysv-xlogues.
18176 (stub_managed_regs): New static variable.
18177 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18178 registers managed by out-of-line stub.
18179 (disable_call_ms2sysv_xlogues): New function.
18180 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18181 m->call_ms2sysv when appropriate and compute frame layout for
18183 (sp_valid_at, fp_valid_at): New inline functions.
18184 (choose_basereg): New function.
18185 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18187 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18188 Use align parameter of choose_baseaddr to generated aligned SSE movs
18190 (pro_epilogue_adjust_stack): Modify to track
18191 machine_frame_state::sp_realigned.
18192 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18193 (ix86_nsaved_sseregs): Likewise.
18194 (ix86_emit_save_regs): Likewise.
18195 (ix86_emit_save_regs_using_mov): Likewise.
18196 (ix86_emit_save_sse_regs_using_mov): Likewise.
18197 (get_scratch_register_on_entry): Likewise.
18198 (gen_frame_set): New function.
18199 (gen_frame_load): Likewise.
18200 (gen_frame_store): Likewise.
18201 (emit_outlined_ms2sysv_save): Likewise.
18202 (emit_outlined_ms2sysv_restore): Likewise.
18203 (ix86_expand_prologue): Modify stack re-alignment code and call
18204 emit_outlined_ms2sysv_save when appropriate.
18205 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18206 parameter rtx_insn *insn, which allows the function to be used to only
18207 generate the notes.
18208 (ix86_expand_epilogue): Modify validity checks of frame and stack
18209 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18210 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18211 * config/i386/predicates.md
18212 (save_multiple): New predicate.
18213 (restore_multiple): Likewise.
18214 * config/i386/sse.md
18215 (save_multiple<mode>): New pattern.
18216 (save_multiple_realign<mode>): Likewise.
18217 (restore_multiple<mode>): Likewise.
18218 (restore_multiple_and_return<mode>): Likewise.
18219 (restore_multiple_leave_return<mode>): Likewise.
18220 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18222 2017-05-14 Julia Koval <julia.koval@intel.com>
18224 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18225 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18226 (__builtin_ia32_xsetbv): New builtins.
18227 * config/i386/i386.c (ix86_expand_special_args_builtin):
18229 (ix86_expand_builtin): Special expand for new intrinsics.
18230 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18231 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18232 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18234 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18236 * cfganal.c (inverted_post_order_compute): Change argument type
18238 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18239 * df-core.c (rest_of_handle_df_initialize): Adjust.
18240 (rest_of_handle_df_finish): Likewise.
18241 (df_analyze_1): Likewise.
18242 (df_analyze): Likewise.
18243 (loop_inverted_post_order_compute): Change argument to be a vec *.
18244 (df_analyze_loop): Adjust.
18245 (df_get_n_blocks): Likewise.
18246 (df_get_postorder): Likewise.
18247 * df.h (struct df_d): Change field to be a vec.
18248 * lcm.c (compute_laterin): Adjust.
18249 (compute_available): Likewise.
18250 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18251 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18252 * tree-ssa-pre.c (compute_antic): Likewise.
18254 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18256 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18257 (depth_first_search::depth_first_search): Change structure init
18258 function to this constructor.
18259 (depth_first_search::add_bb): Rename function to this member.
18260 (depth_first_search::execute): Likewise.
18261 (flow_dfs_compute_reverse_finish): Adjust.
18263 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18265 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18266 (longest_simple_path): Likewise.
18267 * shrink-wrap.c (spread_components): Likewise.
18268 (disqualify_problematic_components): Likewise.
18269 (emit_common_heads_for_components): Likewise.
18270 (emit_common_tails_for_components): Likewise.
18271 (insert_prologue_epilogue_for_components): Likewise.
18273 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18275 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18278 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18280 * df-core.c (df_set_blocks): Start using auto_bitmap.
18281 (df_compact_blocks): Likewise.
18282 * df-problems.c (df_rd_confluence_n): Likewise.
18283 * df-scan.c (df_insn_rescan_all): Likewise.
18284 (df_process_deferred_rescans): Likewise.
18285 (df_update_entry_block_defs): Likewise.
18286 (df_update_exit_block_uses): Likewise.
18287 (df_entry_block_bitmap_verify): Likewise.
18288 (df_exit_block_bitmap_verify): Likewise.
18289 (df_scan_verify): Likewise.
18290 * lra-constraints.c (lra_constraints): Likewise.
18291 (undo_optional_reloads): Likewise.
18292 (lra_undo_inheritance): Likewise.
18293 * lra-remat.c (calculate_gen_cands): Likewise.
18294 (do_remat): Likewise.
18295 * lra-spills.c (assign_spill_hard_regs): Likewise.
18296 (spill_pseudos): Likewise.
18297 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18298 (bitmap_set_subtract_values): Likewise.
18300 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18302 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18303 management with auto_bitmap.
18304 (fix_inter_tick): Likewise.
18305 (fix_recovery_deps): Likewise.
18306 * ira.c (add_store_equivs): Likewise.
18307 (find_moveable_pseudos): Likewise.
18308 (split_live_ranges_for_shrink_wrap): Likewise.
18309 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18310 (rtx_reuse_manager::seen_def_p): Likewise.
18311 (rtx_reuse_manager::set_seen_def): Likewise.
18312 * print-rtl.h (class rtx_reuse_manager): Likewise.
18314 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18316 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18318 (migrate_btr_def): Likewise.
18319 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18320 * df-core.c (loop_post_order_compute): Likewise.
18321 (loop_inverted_post_order_compute): Likewise.
18322 * hsa-common.h: Likewise.
18323 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18324 * init-regs.c (initialize_uninitialized_regs): Likewise.
18325 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18326 (inline_small_functions): Likewise.
18327 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18328 * ira.c (combine_and_move_insns): Likewise.
18329 (build_insn_chain): Likewise.
18330 * loop-invariant.c (find_invariants): Likewise.
18331 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18332 * predict.c (tree_predict_by_opcode): Likewise.
18333 (predict_paths_leading_to): Likewise.
18334 (predict_paths_leading_to_edge): Likewise.
18335 (estimate_loops_at_level): Likewise.
18336 (estimate_loops): Likewise.
18337 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18338 (spread_components): Likewise.
18339 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18340 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18341 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18342 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18343 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18344 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18345 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18346 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18347 (mark_threaded_blocks): Likewise.
18348 (thread_through_all_blocks): Likewise.
18349 * tree-ssa.c (verify_ssa): Likewise.
18350 (execute_update_addresses_taken): Likewise.
18351 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18353 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18355 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18357 (post_order_compute): Likewise.
18358 (inverted_post_order_compute): Likewise.
18359 (pre_and_rev_post_order_compute_fn): Likewise.
18361 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18363 * genrecog.c (int_set::int_set): Explicitly construct our
18364 auto_vec base class.
18365 * vec.h (auto_vec::auto_vec): New constructor.
18367 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18369 * bitmap.h (class auto_bitmap): New constructor taking
18370 bitmap_obstack * argument.
18372 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18374 * bitmap.h (class auto_bitmap): Change type of m_bits to
18375 bitmap_head, and adjust ctor / dtor and member operators.
18377 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18379 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18380 when returned register mode doesn't match original mode.
18382 2017-05-12 Jeff Law <law@redhat.com>
18383 Jakub Jelinek <jakub@redhat.com>
18385 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18386 we look for cc setter after the compare-elim changes.
18387 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18388 within the vector to match what compare-elim now expects.
18389 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18390 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18392 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18393 after the compare-elim changes.
18394 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18395 the vector to match what compare-elim now expects.
18396 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18397 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18398 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18399 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18400 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18402 * config/visium/visium.c (single_set_and_flags): Fix where
18403 we look for cc setter after the compare-elim changes.
18404 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18405 with the vector to match what compare-elim now expects.
18406 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18407 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18408 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18409 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18410 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18411 (neg<mode>2_insn_set_overflow): Likewise.
18413 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18415 PR middle-end/79794
18416 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18417 maybe_expand_insn call, set ops[0].target. If still set after call,
18418 set alt_rtl. Add extra arg to recursive calls.
18419 (extract_bit_field): Add alt_rtl argument. Pass to
18421 * expmed.h (extract_bit_field): Fix prototype.
18422 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18423 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18424 to extract_bit_field_calls.
18425 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18426 Pass alt_rtl to extract_bit_field calls.
18427 * calls.c (store_unaligned_arguments_into_psuedos)
18428 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18429 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18431 * optabs.h (struct expand_operand): Add target bitfield.
18433 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18435 * compare-elim.c (try_eliminate_compare): Canonicalize
18436 operation with embedded compare to
18437 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18438 (set (reg) (operation)].
18440 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18442 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18445 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18446 cost of adding a carry flag for ADC instruction.
18447 [case MINUS]: Ignore the cost of subtracting a carry flag
18448 for SBB instruction.
18450 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18452 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18454 * config/rs6000/bmiintrin.h: New file.
18455 * config/rs6000/bmi2intrin.h: New file.
18456 * config/rs6000/x86intrin.h: New file.
18458 2017-05-12 Jeff Law <law@redhat.com>
18460 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18463 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18465 PR middle-end/80707
18466 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18468 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18470 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18471 early expansion of vector divide builtins.
18472 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18473 builtins identified as having unsigned arguments.
18475 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18477 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18478 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18479 expansion of vector logical operations (and, andc, or, xor,
18482 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18484 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18485 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18487 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18490 early GIMPLE expansion of vector multiplies.
18492 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18494 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18495 TARGET_HAVE_MOVT conditional.
18496 (movt splitter): Likewise.
18498 2017-05-12 Richard Biener <rguenther@suse.de>
18500 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18501 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18502 Fold all stmts not inplace.
18504 2017-05-12 Richard Biener <rguenther@suse.de>
18506 PR tree-optimization/80713
18507 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18508 inserted_exprs bit for not removed stmts.
18510 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18512 PR middle-end/69921
18513 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18514 parallelized" attribute for parallelized OpenACC kernels.
18515 * omp-offload.c (execute_oacc_device_lower): Use it.
18517 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18518 Set "oacc kernels" attribute.
18519 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18520 parameter. Adjust all users.
18521 (oacc_fn_attrib_kernels_p): Remove function.
18522 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18523 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18524 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18525 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18526 assert "oacc kernels" attribute is set.
18528 2017-05-11 Carl Love <cel@us.ibm.com>
18530 * config/rs6000/rs6000-c: Add support for built-in functions
18531 vector unsigned char vec_popcnt (vector signed char)
18532 vector unsigned char vec_popcnt (vector unsigned char)
18533 vector unsigned short vec_popcnt (vector signed short)
18534 vector unsigned short vec_popcnt (vector unsigned short)
18535 vector unsigned int vec_popcnt (vector signed int)
18536 vector unsigned int vec_popcnt (vector unsigned int)
18537 vector unsigned long long vec_popcnt (vector signed long long)
18538 vector unsigned long long vec_popcnt (vector unsigned long long)
18539 vector signed long long vec_slo (vector signed long long,
18540 vector signed char)
18541 vector signed long long vec_slo (vector signed long long,
18542 vector unsigned char)
18543 vector unsigned long long vec_slo (vector unsigned long long,
18544 vector signed char)
18545 vector unsigned long long vec_slo (vector unsigned long long,
18546 vector unsigned char)
18547 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18548 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18549 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18550 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18551 * doc/extend.texi: Update the built-in documentation file for the
18552 new built-in functions.
18554 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18556 * attribs.h (sorted_attr_string): Move machine independent
18557 functions for target clone support from the i386 port to common
18558 code. Rename ix86_function_versions to common_function_versions.
18559 Rename make_name to make_unique_name.
18560 (common_function_versions): Likewise.
18561 (make_unique_name): Likewise.
18562 (make_dispatcher_decl): Likewise.
18563 (is_function_default_version): Likewise.
18564 * attribs.c (attr_strcmp): Likewise.
18565 (sorted_attr_string): Likewise.
18566 (common_function_versions): Likewise.
18567 (make_unique_name): Likewise.
18568 (make_dispatcher_decl): Likewise.
18569 (is_function_default_version): Likewise.
18570 * config/i386/i386.c (attr_strcmp): Likewise.
18571 (sorted_attr_string): Likewise.
18572 (ix86_function_versions): Likewise.
18573 (make_name): Likewise.
18574 (make_dispatcher_decl): Likewise.
18575 (is_function_default_version): Likewise.
18576 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18578 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18581 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18582 Account for direct move costs for vec_construct of integer
18585 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18588 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18589 (UNSPEC_STX_ATOMIC): Ditto.
18590 (loaddi_via_sse): New insn.
18591 (storedi_via_sse): Ditto.
18592 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18593 Update corresponding peephole2 patterns.
18594 (atomic_storedi_fpu): Ditto.
18596 2017-05-11 Julia Koval <julia.koval@intel.com>
18598 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18599 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18601 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18602 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18603 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18605 2017-05-11 Nathan Sidwell <nathan@acm.org>
18607 * graphite-poly.c: Include dumpfile.h.
18609 * dumpfle.h (dump_function): Declare here ...
18610 * tree-dump.h (dump_function): ... not here.
18611 * dumpfile.c: #include tree-cfg.h.
18612 (dump_function): Move here from ...
18613 * tree-dump.c (dump_function): ... here.
18614 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18615 * graphite-poly.c: Don't include tree-dump.h.
18616 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18617 * print-tree.c: Likewise.
18618 * stor-layout.c: Likewise.
18619 * tree-nested.c: Likewise.
18621 * dumpfile.c (dump_start): Use TDF_FLAGS.
18622 (dump_enable_all): Fix TDF_KIND check thinko.
18624 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18626 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18627 array entries to represent two legal parameterizations of the
18628 overloaded __builtin_cmpb function, as represented by the
18629 P6_OV_BUILTIN_CMPB constant.
18630 (altivec_resolve_overloaded_builtin): Add special case handling
18631 for the __builtin_cmpb function, as represented by the
18632 P6_OV_BUILTIN_CMPB constant.
18633 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18634 (BU_P6_64BIT_2): New macro.
18635 (BU_P6_OVERLOAD_2): New macro
18636 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18637 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18638 (CMPB): Add overload support to represent both 32-bit and 64-bit
18639 compare-bytes function.
18640 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18641 support for TARGET_CMPB.
18642 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18643 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18644 documentation of the __builtin_cmpb overloaded built-in function.
18646 2017-05-11 Richard Biener <rguenther@suse.de>
18648 PR tree-optimization/80705
18649 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18650 bases are not vectorizable.
18652 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18654 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18655 when counting register pressure.
18657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18659 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18660 (iv_ca_more_deps): Renamed to ...
18661 (iv_ca_compare_deps): ... this.
18662 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18664 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18666 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18668 (determine_group_iv_costs): ... here.
18669 (find_inv_vars_cb): Record inv var if it's not recorded before.
18671 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18673 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18674 (get_shiftadd_cost): Ditto.
18676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18678 * tree-ssa-address.c: Include header file.
18679 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18681 (add_to_parts): Refactor.
18682 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18683 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18688 PR tree-optimization/53090
18689 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18691 (extract_cond_operands): Detect condition with IV on both sides
18692 and return COMP_IV_EXPR_2.
18693 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18694 (rewrite_use_compare): Simplify by removing call to function
18695 extract_cond_operands.
18697 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18699 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18700 (extract_cond_operands): Detect condition comparing against non-
18701 invariant bound and return appropriate enum value.
18702 (find_interesting_uses_cond): Update use of extract_cond_operands.
18703 Handle its return value accordingly.
18704 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18706 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18708 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18709 nonlinear iv_use computation in loop invariant sensitive way.
18711 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18713 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18714 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18716 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18718 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18719 (dump_cand): Support iv_cand.inv_exprs.
18720 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18722 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18725 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18727 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18729 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18730 as local function. Include necessary header files.
18731 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18733 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18735 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18737 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18739 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18740 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18741 RSHIFT_EXPR and BIT_NOT_EXPR.
18743 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18745 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18746 (adjust_setup_cost): New parameter supporting round up adjustment.
18747 (struct address_cost_data): Delete.
18748 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18749 (split_address_cost, ptr_difference_cost): Delete.
18750 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18751 (struct ainc_cost_data): New struct.
18752 (get_address_cost_ainc): New function.
18753 (get_address_cost, get_computation_cost): Reimplement.
18754 (determine_group_iv_cost_address): Record inv_expr for all uses of
18756 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18757 (iv_ca_has_deps): Reimplemented to ...
18758 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18760 (iv_ca_extend): Call iv_ca_more_deps.
18762 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18764 * tree-ssa-address.c (struct mem_address): Move to header file.
18765 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18766 * tree-ssa-address.h (struct mem_address): Move from C file.
18767 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18769 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18771 * tree-affine.h (aff_combination_type): New interface.
18772 (aff_combination_zero_p): Remove static.
18773 (aff_combination_const_p): New interface.
18774 (aff_combination_singleton_var_p): New interfaces.
18776 2017-05-11 Richard Biener <rguenther@suse.de>
18778 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18779 Skip unreachable blocks and destinations.
18780 (eliminate): Move stmt removal and fixup ...
18781 (fini_eliminate): ... here. Skip inserted exprs.
18782 (pass_pre::execute): Move fini_pre after fini_eliminate.
18783 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18784 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18785 PRE to get rid of dead code that has invalid SSA form and
18786 split critical edges again.
18788 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18790 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18792 2017-05-11 Richard Biener <rguenther@suse.de>
18794 * passes.c (execute_function_todo): Verify loops if they are
18795 said to be up-to-date.
18796 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18797 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18799 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18802 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18803 handle calling assemble_external ourself.
18806 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18807 modes with zero size. Enhance comment.
18809 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18811 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18812 built-ins for vec_xl and vec_xst with short and char pointer
18815 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18817 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18818 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18819 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18820 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18821 (_mm_maskz_min_round_ss): New intrinsics.
18822 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18823 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18824 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18825 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18826 (__builtin_ia32_minss_mask_round): New builtins.
18827 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18828 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18829 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18831 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18832 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18834 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18837 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18839 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18840 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18841 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18842 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18843 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18844 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18845 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18846 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18847 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18848 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18849 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18850 (__builtin_ia32_mulss_mask_round): New builtins.
18851 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18852 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18853 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18855 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18856 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18858 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18861 2017-05-10 Julia Koval <julia.koval@intel.com>
18863 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18864 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18865 (_mm256_setr_m128i): New intrinsics.
18867 2017-05-10 Julia Koval <julia.koval@intel.com>
18869 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18870 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18871 (_mm_maskz_rcp14_ss): New intrinsics.
18872 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18873 (__builtin_ia32_rcp14ss_mask): New builtins.
18874 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18876 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18878 PR tree-optimization/51513
18879 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18880 (assert_unreachable_fallthru_edge_p): Use it.
18881 (group_case_labels_stmt): Likewise.
18882 * tree-cfg.h: Prototype it.
18883 * stmt.c: Include cfghooks.h and tree-cfg.h.
18884 (emit_case_dispatch_table) <gap_label>: New local variable.
18885 Use it to fill dispatch table gaps.
18886 Test for default_label before updating probabilities.
18887 (expand_case) <default_label>: Remove unneeded initialization.
18888 Test for unreachable default case statement and remove its edge.
18889 Set default_label accordingly.
18890 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18892 2017-05-10 Carl Love <cel@us.ibm.com>
18894 * config/rs6000/rs6000-c: Add support for built-in functions
18895 vector signed char vec_neg (vector signed char)
18896 vector signed short int vec_neg (vector short int)
18897 vector signed int vec_neg (vector signed int)
18898 vector signed long long vec_neg (vector signed long long)
18899 vector float vec_neg (vector float)
18900 vector double vec_neg (vector double)
18901 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18903 * config/rs6000/altivec.h: Add define for vec_neg
18904 * doc/extend.texi: Update the built-in documentation for the
18905 new built-in functions.
18907 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18909 PR tree-optimization/77644
18910 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18912 2017-05-10 Nathan Sidwell <nathan@acm.org>
18914 * dumpfile.h (TDI_lang_all): New.
18915 (TDF_KIND): New. Renumber others
18916 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18918 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18920 (get_dump_file_name): Adjust suffix generation.
18921 (dump_enable_all): Use TDF_KIND.
18922 * doc/invoke.texi (-fdump-lang-all): Document.
18924 * dumpfile.h: Tabify.
18926 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18929 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18930 Move member access before delete.
18932 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18934 * tree-inline.c (expand_call_inline): Split block at stmt
18937 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18940 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18941 are now unused after splitting mov{sf,sd}_hardfloat.
18942 (f32_lr2): Likewise.
18943 (f32_lm): Likewise.
18944 (f32_lm2): Likewise.
18945 (f32_li): Likewise.
18946 (f32_li2): Likewise.
18947 (f32_lv): Likewise.
18948 (f32_sr): Likewise.
18949 (f32_sr2): Likewise.
18950 (f32_sm): Likewise.
18951 (f32_sm2): Likewise.
18952 (f32_si): Likewise.
18953 (f32_si2): Likewise.
18954 (f32_sv): Likewise.
18955 (f32_dm): Likewise.
18956 (f32_vsx): Likewise.
18957 (f32_av): Likewise.
18958 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18959 For movsf, order stores so the VSX stores occur before the GPR
18960 store which encourages the register allocator to use a traditional
18961 FPR instead of a GPR. For movsd, order the stores so that the GPR
18962 store comes before the VSX stores to allow the power6 to work.
18963 This is due to the power6 not having a 32-bit integer store
18964 instruction from a FPR.
18965 (movsf_hardfloat): Likewise.
18966 (movsd_hardfloat): Likewise.
18968 2017-05-09 Martin Sebor <msebor@redhat.com>
18970 PR translation/80280
18971 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18974 PR translation/80280
18975 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18976 data member added in r247778.
18977 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18979 2017-05-09 Nathan Sidwell <nathan@acm.org>
18981 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18983 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18986 2017-05-09 Marek Polacek <polacek@redhat.com>
18988 * doc/invoke.texi: Fix typo.
18990 2017-05-09 Richard Biener <rguenther@suse.de>
18992 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18993 (vrp_val_is_min): Likewise.
18994 (set_value_range_to_value): Likewise.
18995 (set_value_range_to_nonnegative): Likewise.
18996 (gimple_assign_nonzero_p): Likewise.
18997 (gimple_stmt_nonzero_p): Likewise.
18998 (vrp_int_const_binop): Likewise. Remove unreachable case.
18999 (adjust_range_with_scev): Adjust comments.
19000 (compare_range_with_value): Likewise.
19001 (extract_range_from_phi_node): Likewise.
19002 (test_for_singularity): Likewise.
19004 2017-05-09 Richard Biener <rguenther@suse.de>
19006 * tree-vrp.c (get_single_symbol): Add assert that we don't
19007 get overflowed constants as invariant part.
19008 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19009 checks. Use wi::cmp instead of recursing for integer constants.
19010 (compare_values): Just ignore whether we assumed undefined
19011 overflow instead of failing the compare.
19012 (extract_range_for_var_from_comparison_expr): Add comment before the
19013 TREE_NO_WARNING sets.
19014 (test_for_singularity): Likewise.
19015 (extract_range_from_comparison): Do not disable optimization
19016 when we assumed undefined overflow.
19017 (extract_range_basic): Remove init of unused var.
19019 2017-05-09 Richard Biener <rguenther@suse.de>
19021 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19022 (extract_range_from_multiplicative_op_1): Adjust.
19023 (extract_range_from_binary_expr_1): Use int_const_binop.
19025 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19028 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19029 rs6000_store_data_bypass_p in seven define_bypass directives and
19030 in several comments.
19031 * config/rs6000/rs6000-protos.h: Add prototype for
19032 rs6000_store_data_bypass_p function.
19033 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19034 function implements slightly different (rs6000-specific) semantics
19035 than store_data_bypass_p, returning false rather than aborting
19036 with assertion error when arguments do not satisfy the
19037 requirements of store data bypass.
19038 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19039 rs6000_store_data_bypass_p.
19041 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19043 * config/xtensa/xtensa-protos.h
19044 (xtensa_initial_elimination_offset): New declaration.
19045 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19046 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19047 macro definition, add case for FRAME_POINTER_REGNUM when
19048 FRAME_GROWS_DOWNWARD.
19049 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19050 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19051 xtensa_initial_elimination_offset.
19053 2017-05-08 Nathan Sidwell <nathan@acm.org>
19055 * doc/invoke.texi: Alphabetize -fdump options.
19057 2017-05-08 Martin Sebor <msebor@redhat.com>
19059 PR translation/80280
19060 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19062 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19064 * target.def (compute_frame_layout): New optional target hook.
19065 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19066 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19067 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19069 * reload1.c (verify_initial_elim_offsets): Likewise.
19070 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19071 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19072 (arm_get_frame_offsets): Split up into this ...
19073 (arm_compute_frame_layout): ... and this function.
19075 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19077 * config/aarch64/constraints.md (Usa): New constraint.
19078 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19080 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19082 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19083 with_multilib_list after it has been checked.
19085 2017-05-08 Richard Biener <rguenther@suse.de>
19087 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19088 (bitmap_set_subtract_values): Likewise.
19090 2017-05-08 Richard Biener <rguenther@suse.de>
19092 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19093 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19095 (gimple_stmt_nonzero_warnv_p): Rename to ...
19096 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19098 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19099 (extract_range_basic): Adjust, do not disable propagation on
19100 strict overflow sensitive simplification.
19101 (vrp_visit_cond_stmt): Likewise.
19103 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19105 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19106 body size unconditionally.
19108 2017-05-07 Jeff Law <law@redhat.com>
19111 2017-05-06 Jeff Law <law@redhat.com>
19112 PR tree-optimization/78496
19113 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19116 PR tree-optimization/78496
19117 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19118 (simplify_stmt_using_ranges): Call it.
19119 (vrp_dom_walker::before_dom_children): Extract equivalences
19120 from an ASSERT_EXPR with an equality comparison against a
19123 2017-05-06 Jeff Law <law@redhat.com>
19125 PR tree-optimization/78496
19126 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19129 PR tree-optimization/78496
19130 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19131 (simplify_stmt_using_ranges): Call it.
19132 (vrp_dom_walker::before_dom_children): Extract equivalences
19133 from an ASSERT_EXPR with an equality comparison against a
19136 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19138 * lra-constraints.c (lra_copy_reg_equiv): New function.
19139 (split_reg): Use it to copy equivalence information from the
19140 original register to the spill register.
19142 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19144 PR rtl-optimization/75964
19145 * simplify-rtx.c (simplify_const_relational_operation): Remove
19146 invalid handling of comparisons of integer ABS.
19148 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19150 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19151 initialize to zero.
19152 (init_regs): Remove declaration.
19153 (function_arg_advance_32): Initialize error_p as boolean variable.
19155 2017-05-05 Nathan Sidwell <nathan@acm.org>
19157 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19158 lines. Use for (;;).
19160 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19162 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19163 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19164 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19165 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19166 VF=2 that require versioning.
19168 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19170 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19173 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19175 * diagnostic.h (diagnostic_override_option_index): Convert from
19176 macro to inline function.
19178 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19180 * diagnostic.c (last_module_changed_p): New function.
19181 (set_last_module): New function.
19182 (diagnostic_report_current_module): Convert macro usage to
19183 the above functions.
19184 * diagnostic.h (diagnostic_context::last_module): Strengthen
19185 from const line_map * to const line_map_ordinary *.
19186 (diagnostic_last_module_changed): Delete macro.
19187 (diagnostic_set_last_module): Delete macro.
19189 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19191 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19192 with diagnostic_report_diagnostic.
19193 (diagnostic_n_impl_richloc): Likewise.
19194 * diagnostic.h (report_diagnostic): Delete macro.
19195 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19196 with diagnostic_report_diagnostic.
19197 * substring-locations.c (format_warning_va): Likewise.
19199 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19201 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19202 save/restor of format_spec. Move option-printing code to...
19203 (print_option_information): ...this new function, and
19204 reimplement by simply printing to the pretty_printer,
19205 rather than appending to the format string.
19207 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19209 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19210 handling logic into...
19211 (update_effective_level_from_pragmas): ...this new function.
19213 2017-05-04 Andrew Waterman <andrew@sifive.com>
19215 * config/riscv/riscv.opt (mstrict-align): New option.
19216 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19217 (SLOW_UNALIGNED_ACCESS): Define.
19218 (riscv_slow_unaligned_access): Declare.
19219 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19221 (riscv_slow_unaligned_access): New variable.
19222 (rocket_tune_info): Set slow_unaligned_access to true.
19223 (optimize_size_tune_info): Set slow_unaligned_access to false.
19224 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19225 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19226 (riscv_option_override): Set riscv_slow_unaligned_access.
19227 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19229 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19231 * config/riscv/riscv.md: Unify indentation.
19233 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19238 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19240 (extendsi<mode>2): Add support for doing sign extension via
19241 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19242 don't have ISA 3.0 instructions.
19243 (extendsi<mode>2 splitter): Likewise.
19244 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19245 generate the normal insns since SImode can now go in vector
19246 registers. Disallow the special UNSPECs needed for previous
19247 machines to hide SImode being used. Add new insns
19248 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19249 (fix_trunc<mode>si2_stfiwx): Likewise.
19250 (fix_trunc<mode>si2_internal): Likewise.
19251 (fixuns_trunc<mode>si2): Likewise.
19252 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19253 (fctiw<u>z_<mode>_smallint): Likewise.
19254 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19255 of floating point to 32-bit integer from doing a direct move to
19256 the GPR registers to do a store.
19257 (fctiwz_<mode>): Break long line.
19259 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19261 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19262 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19263 (addr_list, addr_offset_valid_p): New.
19264 (split_address_groups): Check offset validity with above function.
19265 (gt-tree-ssa-loop-ivopts.h): Include header file.
19267 2017-05-05 Nathan Sidwell <nathan@acm.org>
19269 * config.gcc (arm*-*-*): Add missing 'fi'.
19271 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19273 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19274 included in -fopt-info does not matter.
19275 * doc/optinfo.texi (-fopt-info): Fix description of default
19276 behavour. Explicitly say order of options included in -fopt-info
19279 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19281 * config.gcc: Allow combinations of aprofile and rmprofile values for
19282 --with-multilib-list.
19283 * config/arm/t-multilib: New file.
19284 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19285 variables. Remove setting of ISA and floating-point ABI in
19286 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19287 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19288 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19289 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19291 * config/arm/t-rmprofile: Likewise except for the matches changes.
19292 * doc/install.texi (--with-multilib-list): Document the combination of
19293 aprofile and rmprofile values and warn about pitfalls in doing that.
19295 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19297 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19298 (movdi_aarch64): Likewise.
19300 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19302 PR tree-optimization/80632
19303 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19305 (build_arrays): Initialize it for virtual phis.
19306 (fix_phi_nodes): Use it for virtual phis.
19308 PR tree-optimization/80558
19309 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19310 [x, y] op z into [x op, y op z] for op & or | if conditions
19313 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19314 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19317 * config/arm/arm.md (use_literal_pool): Remove.
19318 (64-bit immediate split): No longer takes cost into consideration
19319 if arm_disable_literal_pool is enabled.
19320 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19321 used when arm_disable_literal_pool is enabled.
19322 (arm_max_const_double_inline_cost): Remove use of
19323 arm_disable_literal_pool.
19324 (push_minipool_fix): Add assert.
19325 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19326 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19327 (no_literal_pool_sf_immediate): New.
19329 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19331 PR tree-optimization/80613
19332 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19333 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19335 2017-05-05 Richard Biener <rguenther@suse.de>
19337 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19339 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19341 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19342 of this flag from insn conditions due to removal from r247495.
19344 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19346 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19348 (arm_early_store_addr_dep_ptr): Likewise.
19349 * config/arm/aarch-common-protos.h
19350 (arm_early_load_addr_dep_ptr): Add prototype.
19351 (arm_early_store_addr_dep_ptr): Likewise.
19352 * config/arm/cortex-a53.md: Add new bypasses.
19354 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19356 * tree.c (next_type_uid): Change type to unsigned.
19357 (type_hash_canon): Decrement back next_type_uid if
19358 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19359 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19362 2017-05-04 Martin Sebor <msebor@redhat.com>
19364 * builtins.c: Fix a trivial typo in a comment.
19366 PR middle-end/79234
19367 * builtins.c (check_sizes): Adjust to handle reading past the end.
19368 Avoid printing excessive upper bound of ranges. Use %E to print
19369 tree nodes instead of converting them to %wu.
19370 (expand_builtin_memchr): New function.
19371 (compute_dest_size): Rename...
19372 (compute_objsize): ...to this.
19373 (expand_builtin_memcpy): Adjust.
19374 (expand_builtin_mempcpy): Adjust.
19375 (expand_builtin_strcat): Adjust.
19376 (expand_builtin_strcpy): Adjust.
19377 (check_strncat_sizes): Adjust.
19378 (expand_builtin_strncat): Adjust.
19379 (expand_builtin_strncpy): Adjust and simplify.
19380 (expand_builtin_memset): Adjust.
19381 (expand_builtin_bzero): Adjust.
19382 (expand_builtin_memcmp): Adjust.
19383 (expand_builtin): Handle memcmp.
19384 (maybe_emit_chk_warning): Check strncat just once.
19386 2017-05-04 Martin Sebor <msebor@redhat.com>
19388 PR preprocessor/79214
19389 PR middle-end/79222
19390 PR middle-end/79223
19391 * builtins.c (check_sizes): Add inlining context and issue
19392 warnings even when -Wno-system-headers is set.
19393 (check_strncat_sizes): Same.
19394 (expand_builtin_strncat): Same.
19395 (expand_builtin_memmove): New function.
19396 (expand_builtin_stpncpy): Same.
19397 (expand_builtin): Handle memmove and stpncpy.
19399 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19401 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19402 which is not used any more.
19404 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19406 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19408 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19410 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19411 (cortexa53_tunings): Likewise.
19412 (cortexa57_tunings): Likewise.
19413 (cortexa72_tunings): Likewise.
19414 (cortexa73_tunings): Likewise.
19416 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19418 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19419 Set loop alignment to 8.
19421 2017-05-04 Martin Sebor <msebor@redhat.com>
19423 PR translation/80280
19424 * builtins.c (expand_builtin_object_size): Add missing quoting to
19425 %D and like directives.
19426 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19427 (hsa_type_for_tree_type): Same.
19428 (verify_function_arguments): Same.
19429 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19430 * varasm.c (get_section): Same.
19433 2017-05-04 Martin Sebor <msebor@redhat.com>
19435 PR translation/80280
19436 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19438 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19440 * config/aarch64/aarch64.c (generic_addrcost_table):
19441 Change HI/TI mode setting.
19443 2017-05-04 Martin Jambor <mjambor@suse.cz>
19445 PR tree-optimization/80622
19446 * tree-sra.c (comes_initialized_p): New function.
19447 (build_accesses_from_assign): Only set write lazily when
19448 comes_initialized_p is false.
19449 (analyze_access_subtree): Use comes_initialized_p.
19450 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19451 instead of testing for PARM_DECL.
19453 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19455 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19456 constraint on operand 0 to allow more general addressing modes.
19457 Adjust output template.
19458 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19460 * config/aarch64/aarch64-protos.h
19461 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19462 * config/aarch64/constraints.md (Dp): New address constraint.
19463 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19466 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19468 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19469 update use of estimate_ipcp_clone_size_and_time.
19470 (estimate_local_effects): Update use of
19471 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19472 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19473 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19474 Return nonspecialized time.
19476 2017-05-04 Richard Biener <rguenther@suse.de>
19478 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19479 for the last VUSE which def dominates the PHI. Directly call
19481 (get_continuation_for_phi_1): Remove.
19483 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19485 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19486 to explain the use of truncating division. Cap the number of
19487 iterations to the maximum given by nb_iterations_upper_bound,
19490 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19492 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19493 * configure: Regenerate.
19494 * config.in: Regenerate.
19495 * config/i386/driver-mingw32.c: new file.
19496 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19497 * config.host: Link driver-mingw32.o on MinGW host.
19498 * doc/install.texi: Document new --enable-mingw-wildcard configure
19501 2017-05-04 Marek Polacek <polacek@redhat.com>
19503 PR tree-optimization/80612
19504 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19506 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19507 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19509 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19510 (movt splitter): Likewise.
19511 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19512 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19513 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19514 block for Thumb-1 with MOVT.
19515 (thumb2_legitimate_address_p): Move code block ...
19516 (can_avoid_literal_pool_for_label_p): ... into this new function.
19517 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19519 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19520 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19521 "M-profile targets with the MOVT instruction".
19523 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19525 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19526 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19527 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19529 2017-05-04 Martin Liska <mliska@suse.cz>
19531 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19532 variable cond_code.
19534 2017-05-04 Richard Biener <rguenther@suse.de>
19536 * tree.c (array_at_struct_end_p): Handle arrays at struct
19537 end with flexarrays more conservatively. Refactor and treat
19538 arrays of arrays or aggregates more strict. Fix
19539 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19540 * tree.c (array_at_struct_end_p): Adjust prototype.
19541 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19542 * gimple-fold.c (get_range_strlen): Likewise.
19543 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19545 2017-05-04 Richard Biener <rguenther@suse.de>
19547 PR tree-optimization/31130
19548 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19550 (supports_overflow_infinity): Likewise.
19551 (is_negative_overflow_infinity): Likewise.
19552 (is_positive_overflow_infinity): Likewise.
19553 (is_overflow_infinity): Likewise.
19554 (stmt_overflow_infinity): Likewise.
19555 (overflow_infinity_range_p): Likewise.
19556 (usable_range_p): Remove as always returning true.
19557 (make_overflow_infinity): Remove.
19558 (negative_overflow_infinity): Likewise.
19559 (positive_overflow_infinity): Likewise.
19560 (avoid_overflow_infinity): Likewise.
19561 (set_value_range): Adjust accordingly.
19562 (set_value_range_to_nonnegative): Likewise, remove now unused
19563 overflow_infinity arg.
19564 (vrp_operand_equal_p): Adjust.
19565 (update_value_range): Likewise.
19566 (range_int_cst_singleton_p): Likewise.
19567 (operand_less_p): Likewise.
19568 (compare_values_warnv): Likewise.
19569 (extract_range_for_var_from_comparison_expr): Likewise.
19570 (vrp_int_const_binop): Likewise.
19571 (zero_nonzero_bits_from_vr): Likewise.
19572 (extract_range_from_multiplicative_op_1): Likewise.
19573 (extract_range_from_binary_expr_1): Likewise.
19574 (extract_range_from_unary_expr): Likewise.
19575 (extract_range_from_comparison): Likewise.
19576 (extract_range_basic): Likewise.
19577 (adjust_range_with_scev): Likewise.
19578 (compare_ranges): Likewise.
19579 (compare_range_with_value): Likewise.
19580 (dump_value_range): Likewise.
19581 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19583 (simplify_cond_using_ranges): Adjust.
19585 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19587 * brig-builtins.def: Added a builtin for class_f64.
19588 * builtin-types.def: Added a builtin type needed by class_f64.
19590 2017-05-03 Jason Merrill <jason@redhat.com>
19592 * timevar.def: Add TV_CONSTEXPR.
19594 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19596 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19598 2017-05-03 Martin Jambor <mjambor@suse.cz>
19600 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19601 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19602 * ipa-cp.c (ipcp_propagate_stage): Do not call
19603 ipa_update_after_lto_read.
19604 * ipa-inline.c (ipa_inline): Likewise.
19606 2017-05-03 Martin Jambor <mjambor@suse.cz>
19608 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19609 tag. Added a default constructor and a destructor.
19610 (ipa_edge_args_sum_t): New class;
19611 (ipa_edge_args_sum): Declare.
19612 (ipa_edge_args_vector): Remove declaration.
19613 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19614 (ipa_free_edge_args_substructures): Remove declaration.
19615 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19616 (ipa_edge_args_info_available_for_edge_p): Likewise.
19617 * ipa-prop.c (ipa_edge_args_vector): Removed.
19618 (edge_removal_hook_holder): Likewise.
19619 (edge_duplication_hook_holder): Likewise.
19620 (ipa_edge_args_sum): New variable.
19621 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19622 ipa_edge_args_vector.
19623 (ipa_free_edge_args_substructures): Likewise.
19624 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19625 ipa_edge_args_vector.
19626 (ipa_edge_removal_hook): Turned into method
19627 ipa_edge_args_sum_t::remove.
19628 (ipa_edge_duplication_hook): Turned into method
19629 ipa_edge_args_sum_t::duplicate.
19630 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19631 registering edge hooks.
19632 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19633 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19634 ipa_edge_args_sum instead of ipa_edge_args_vector.
19635 * ipa-profile.c (ipa_profile): Likewise.
19637 2017-05-03 Martin Jambor <mjambor@suse.cz>
19639 * symbol-summary.h (function_summary): New method exists.
19640 (function_summary::symtab_removal): Deallocate through release.
19641 (call_summary): New class.
19642 (gt_ggc_mx): New overload.
19643 (gt_pch_nx): Likewise.
19644 (gt_pch_nx): Likewise.
19646 2017-05-03 Jeff Law <law@redhat.com>
19648 PR tree-optimization/78496
19649 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19650 from simplify_cond_using_ranges. Split off code to walk
19651 backwards through casts into ...
19652 (simplify_cond_using_ranges_2): New function.
19653 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19654 (execute_vrp): After identifying jump threads, call
19655 simplify_cond_using_ranges_2.
19657 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19660 * ipa-inline.h (inline_summary): Add ctor.
19661 (create_ggc): Do not use ggc_cleared_alloc.
19663 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19664 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19666 * gcc.c (handle_braces): Support escaping in switch matching
19668 * doc/invoke.texi (Spec Files): Document it.
19669 Remove superfluous @code markup in items.
19671 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19673 * diagnostic-show-locus.c (struct column_range): New struct.
19674 (get_affected_columns): New function.
19675 (get_printed_columns): New function.
19676 (struct correction): New struct.
19677 (correction::ensure_capacity): New function.
19678 (correction::ensure_terminated): New function.
19679 (struct line_corrections): New struct.
19680 (line_corrections::~line_corrections): New dtor.
19681 (line_corrections::add_hint): New function.
19682 (layout::print_trailing_fixits): Reimplement in terms of the new
19684 (selftest::test_overlapped_fixit_printing): New function.
19685 (selftest::diagnostic_show_locus_c_tests): Call it.
19687 2017-05-03 Nathan Sidwell <nathan@acm.org>
19689 Canonicalize canonical type hashing
19690 * tree.h (type_hash_canon_hash): Declare.
19691 * tree.c (type_hash_list, attribute_hash_list): Move into
19692 type_hash_canon_hash.
19693 (build_type_attribute_qual_variant): Break out hash code calc into
19694 type_hash_canon_hash.
19695 (type_hash_canon_hash): New. Generic type hash computation.
19696 (build_range_type_1, build_array_type_1, build_function_type,
19697 build_method_type_directly, build_offset_type, build_complex_type,
19698 make_vector_type): Call it.
19700 2017-05-03 Richard Biener <rguenther@suse.de>
19702 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19703 When all DRs have unknown misaligned do not always peel
19704 when there is a store but apply the same costing model as if
19705 there were only loads.
19707 2017-05-03 Richard Biener <rguenther@suse.de>
19710 PR tree-optimization/80492
19711 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19712 compare_base_decls returning dont-know properly.
19714 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19716 * config/arm/iterators.md (CCSI): New mode iterator.
19717 (arch): New mode attribute.
19718 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19719 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19720 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19721 code iterator for success result mode.
19722 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19723 the corresponding new insn generators.
19725 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19728 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19729 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19731 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19733 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19734 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19735 (DDR_A): Wrap DDR argument in brackets.
19736 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19737 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19738 (DDR_REVERSED_P): Likewise.
19740 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19742 PR tree-optimization/79472
19743 * tree-switch-conversion.c (struct switch_conv_info): Add
19744 contiguous_range and default_case_nonstandard fields.
19745 (collect_switch_conv_info): Compute contiguous_range and
19746 default_case_nonstandard fields, don't clear final_bb if
19747 contiguous_range and only the default case doesn't have the required
19749 (check_all_empty_except_final): Set default_case_nonstandard instead
19750 of failing if contiguous_range and the default case doesn't have empty
19752 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19753 and only the default case doesn't have the required constants. Skip
19755 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19756 if default_case_nonstandard.
19757 (build_constructors): Build constant 1 just once. Assert that default
19758 values aren't inserted in between cases if contiguous_range. Skip
19760 (build_arrays): Skip virtual phis.
19761 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19762 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19763 Handle virtual phis.
19764 (gen_inbound_check): Handle default_case_nonstandard case.
19765 (process_switch): Adjust check_final_bb caller. Call
19766 gather_default_values with the first non-default case instead of
19767 default case if default_case_nonstandard.
19769 2017-05-02 Nathan Sidwell <nathan@acm.org>
19771 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19772 check. Fix formatting.
19774 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19776 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19777 errors when comparing specialized and unspecialized times.
19779 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19781 * diagnostic-show-locus.c
19782 (layout::should_print_annotation_line_p): Make private.
19783 (layout::print_annotation_line): Make private.
19784 (layout::annotation_line_showed_range_p): Make private.
19785 (layout::show_ruler): Make private.
19786 (layout::print_source_line): Make private. Pass in line and
19787 line_width, rather than calling location_get_source_line. Drop
19789 (layout::print_leading_fixits): New method.
19790 (layout::print_any_fixits): Rename to...
19791 (layout::print_trailing_fixits): ...this, and make private.
19792 Don't print newline fixits.
19793 (diagnostic_show_locus): Move logic for printing one row into...
19794 (layout::print_line): ...this new function. Move the
19795 location_get_source_line call and error-handling from
19796 print_source_line to here. Call print_leading_fixits, and rename
19797 print_any_fixits to print_trailing_fixits.
19798 (selftest::test_fixit_insert_containing_newline): Update now that
19799 newlines are partially supported.
19800 (selftest::test_fixit_insert_containing_newline_2): New test.
19801 (selftest::test_fixit_replace_containing_newline): Update comments.
19802 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19803 * edit-context.c (class added_line): New class.
19804 (class edited_line): Describe newline handling in comment.
19805 (edited_line::actually_edited_p): New method.
19806 (edited_line::print_content): Delete redundant decl.
19807 (edited_line::m_predecessors): New field.
19808 (edited_file::print_content): Call edited_line::print_content.
19809 (edited_file::print_diff): Update to support newlines.
19810 (edited_file::print_diff_hunk): Likewise.
19811 (edited_file::print_run_of_changed_lines): New function.
19812 (edited_file::print_diff_line): Convert to...
19813 (print_diff_line): ...this.
19814 (edited_file::get_effective_line_count): New function.
19815 (edited_line::edited_line): Initialize new field m_predecessors.
19816 (edited_line::~edited_line): Clean up m_predecessors.
19817 (edited_line::apply_fixit): Handle newlines.
19818 (edited_line::get_effective_line_count): New function.
19819 (edited_line::print_content): New function.
19820 (edited_line::print_diff_lines): New function.
19821 (selftest::test_applying_fixits_insert_containing_newline): New
19823 (selftest::test_applying_fixits_replace_containing_newline): New
19825 (selftest::insert_line): New function.
19826 (selftest::test_applying_fixits_multiple_lines): Add example of
19828 (selftest::edit_context_c_tests): Call the new tests.
19830 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19832 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19833 parameter cand. Update dump information.
19834 (get_computation_cost): Update uses.
19836 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19838 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19839 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19840 (get_computation_at, rewrite_use_address): Update use of
19841 get_computation_aff.
19843 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19845 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19846 (get_computation): Delete.
19847 (get_computation_cost): Implement like get_computation_cost_at.
19848 Use get_computation_at.
19849 (get_computation_cost_at): Delete.
19850 (rewrite_use_nonlinear_expr): Use get_computation_at.
19851 (rewrite_use_compare, remove_unused_ivs): Ditto.
19853 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19855 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19857 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19859 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19860 (ivopts_global_cost_for_size): Rename parameter and update uses.
19861 (iv_ca_recount_cost): Update uses.
19862 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19863 candidates seperately in n_invs and n_cands.
19864 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19866 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19868 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19869 (find_inv_vars_cb): New.
19870 (find_depends): Renamed to ...
19871 (find_inv_vars): ... this.
19872 (add_candidate_1, force_var_cost): Call find_inv_vars.
19873 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19875 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19877 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19878 inv_vars. Add inv_exprs.
19879 (struct iv_cand): Rename depends_on to inv_vars.
19880 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19881 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19882 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19883 (dump_cand): Dump inv_vars.
19884 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19885 (record_invariant, find_depends, add_candidate_1): Ditto.
19886 (set_group_iv_cost, force_var_cost): Ditto.
19887 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19888 (get_computation_cost_at, get_computation_cost): Ditto.
19889 (determine_group_iv_cost_generic): Ditto.
19890 (determine_group_iv_cost_address): Ditto.
19891 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19892 (determine_group_iv_costs): Ditto.
19893 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19894 (iv_ca_set_remove_invariants): Renamed to ...
19895 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19896 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19897 (iv_ca_set_add_invariants): Renamed to ...
19898 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19899 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19900 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19901 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19902 (create_new_ivs): Remove useless dump.
19904 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19906 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19908 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19909 (iv_ca_set_no_cp, create_new_iv): Ditto.
19911 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19913 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19915 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19917 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19918 function tree_check2.
19920 2017-05-02 Martin Liska <mliska@suse.cz>
19922 * doc/gcov.texi: Add missing preposition.
19923 * gcov.c (function_info::function_info): Properly fill up
19924 all member variables.
19926 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19928 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19930 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19932 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19934 2017-05-02 Martin Liska <mliska@suse.cz>
19937 * lto-streamer-in.c (lto_read_tree_1): Remove
19938 LTO_STREAMER_DEBUG.
19939 * lto-streamer.c (struct tree_hash_entry): Likewise.
19940 (struct tree_entry_hasher): Likewise.
19941 (tree_entry_hasher::hash): Likewise.
19942 (tree_entry_hasher::equal): Likewise.
19943 (lto_streamer_init): Likewise.
19944 (lto_orig_address_map): Likewise.
19945 (lto_orig_address_get): Likewise.
19946 (lto_orig_address_remove): Likewise.
19947 * lto-streamer.h: Likewise.
19948 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19949 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19951 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19953 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19954 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19955 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19956 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19957 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19958 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19959 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19960 (mm_maskz_sub_ss): New intrinsics.
19961 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19962 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19963 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19964 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19965 (__builtin_ia32_subss_mask_round): New builtins.
19966 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19967 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19968 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19970 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19971 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19973 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19976 2017-05-02 Martin Jambor <mjambor@suse.cz>
19978 PR tree-optimization/78687
19979 * tree-sra.c (access): New field parent.
19980 (process_subtree_disqualification): New function.
19981 (disqualify_candidate): Call it.
19982 (build_accesses_from_assign): Reset write flag if creating an
19984 (build_access_subtree): Fill in parent field and also prpagate
19985 down grp_write flag.
19986 (create_artificial_child_access): New parameter set_grp_write, set
19987 grp_write to its value.
19988 (propagate_subaccesses_across_link): Also propagate grp_write flag
19990 (propagate_all_subaccesses): Push the closest parent back to work
19991 queue if add_access_to_work_queue returned true.
19993 2017-05-02 Richard Biener <rguenther@suse.de>
19995 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19996 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19997 -fstrict-overflow documentation.
19998 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19999 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20000 flag_strict_overflow.
20001 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20002 * lto-opts.c (lto_write_options): Do not stream it.
20003 * lto-wrapper.c (merge_and_complain): Do not handle it.
20004 * opts.c (default_options_table): Do not set -fstrict-overflow.
20005 (finish_options): Likewise do not clear it when sanitizing.
20006 * simplify-rtx.c (simplify_const_relational_operation): Do not
20007 test flag_strict_overflow.
20009 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20011 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20012 using enabled attribute.
20013 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20014 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20015 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20016 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20017 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20018 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20019 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20020 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20021 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20022 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20024 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20026 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20028 2017-05-02 Richard Biener <rguenther@suse.de>
20030 PR tree-optimization/80591
20032 2017-04-10 Richard Biener <rguenther@suse.de>
20034 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20037 2017-05-02 Richard Biener <rguenther@suse.de>
20039 PR tree-optimization/80549
20040 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20041 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20042 headers if they do not have a preheader.
20044 2017-05-02 Martin Liska <mliska@suse.cz>
20047 * common.opt: Fix typo.
20048 * doc/invoke.texi: Likewise.
20050 2017-05-01 Jan Beulich <jbeulich@suse.com>
20052 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20053 swapping, add (x,x,m,x,n) alternative.
20055 2017-05-01 Nathan Sidwell <nathan@acm.org>
20057 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20058 unnecessary unadjusted_alignment check.
20060 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20063 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20065 * gimplify.c (gimplify_cilk_detach): New function.
20066 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20067 * tree-core.h: Document EXPR_CILK_SPAWN.
20068 * tree.h (EXPR_CILK_SPAWN): Define.
20070 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20072 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20073 to use new fixit_hint representation, using the "replace" logic.
20074 (get_line_span_for_fixit_hint): Likewise.
20075 (layout::print_any_fixits): Likewise.
20076 (selftest::test_one_liner_many_fixits): Rename to...
20077 (selftest::test_one_liner_many_fixits_1): ...this, and update
20078 comment and expected output to reflect that the multiple fix-it
20079 hints are now consolidated into one insertion.
20080 (selftest::test_one_liner_many_fixits_2): New test.
20081 (selftest::test_diagnostic_show_locus_one_liner): Update for
20083 (selftest::test_fixit_consolidation): Update for fix-it API
20085 * diagnostic.c (print_parseable_fixits): Likewise.
20086 * edit-context.c (edited_line::m_line_events): Convert from
20087 auto_vec <line_event *> to auto_vec <line_event>.
20088 (class line_event): Convert from abstract base class to a concrete
20089 class, taking over the role of replace_event.
20090 (class insert_event): Delete.
20091 (class replace_event): Rename to class line_event. Convert to
20093 (edit_context::add_fixits): Reimplement.
20094 (edit_context::apply_insert): Delete.
20095 (edit_context::apply_replace): Rename to...
20096 (edit_context::apply_fixit): ...this. Convert to half-open range.
20097 (edited_file::apply_insert): Delete.
20098 (edited_file::apply_replace): Rename to...
20099 (edited_file::apply_fixit): ...this.
20100 (edited_line::~edited_line): Drop deletion of events.
20101 (edited_line::apply_insert): Delete.
20102 (edited_line::apply_replace): Rename to...
20103 (edited_line::apply_fixit): ...this. Convert to half-open range.
20104 Update for change to type of m_line_events.
20105 * edit-context.h (edit_context::apply_insert): Delete.
20106 (edit_context::apply_replace): Rename to...
20107 (edit_context::apply_fixit): ...this.
20109 2017-05-01 Martin Sebor <msebor@redhat.com>
20111 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20114 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20117 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20118 __get_cpuid_max returns 0.
20119 (__get_cpuid_count): Ditto.
20121 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20123 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20124 replacement expression is another instance of one of its arguments.
20126 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20129 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20130 check for stack push/pop autoinc.
20131 * config/i386/i386.c (ix86_agi_dependent): Return false
20132 if the only reason why modified_in_p returned true is that
20133 addr is SP based and set_insn is a push or pop.
20135 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20137 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20140 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20143 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20144 (account_size_time): Use two predicates - exec_pred and
20146 (evaluate_conditions_for_known_args): Compute both clause and
20148 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20149 (inline_summary_t::duplicate): Update.
20150 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20152 (compute_inline_parameters): Likewise.
20153 (estimate_edge_size_and_time): Update caluclation of time.
20154 (estimate_node_size_and_time): Compute both time and nonspecialized
20156 (estimate_ipcp_clone_size_and_time): Update.
20157 (inline_merge_summary): Update.
20158 (do_estimate_edge_time): Update.
20159 (do_estimate_edge_size): Update.
20160 (do_estimate_edge_hints): Update.
20161 (inline_read_section, inline_write_summary): Stream both new predicates.
20162 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20164 (compute_inlined_call_time): Cleanup.
20165 (big_speedup_p): Update.
20166 (edge_badness): Update.
20167 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20168 (size_time_entry): Replace predicate by exec_predicate and
20169 nonconst_predicate.
20170 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20171 (estimate_edge_time): Return also nonspec_time.
20172 (reset_edge_growth_cache): Update.
20174 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20176 PR rtl-optimization/80491
20177 * ifcvt.c (noce_process_if_block): When looking for x setter
20178 with missing else_bb, don't check only the insn right before
20179 cond_earliest, but look for the last insn that x is modified in
20180 within the same bb.
20182 PR rtl-optimization/80491
20183 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20185 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20187 PR tree-optimization/80487
20188 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20190 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20192 PR tree-optimization/79697
20193 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20194 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20195 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20197 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20198 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20200 2017-04-28 Martin Sebor <msebor@redhat.com>
20202 PR tree-optimization/80523
20203 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20204 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20206 (maybe_warn, format_directive, parse_directive): Use new functions.
20207 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20209 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20211 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20213 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20215 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20216 target_header_dir): Set correctly.
20217 * configure: Regenerated.
20218 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20219 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20220 instead of SYSTEM_HEADER_DIR.
20222 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20224 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20225 (estimate_local_effects): Likewise.
20226 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20227 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20228 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20229 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20230 do_estimate_edge_time, estimate_edge_time): Likewise.
20231 * ipa-inline-analysis.c (estimate_node_size_and_time,
20232 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20233 (estimate_time_after_inlining): Remove.
20235 2017-04-28 Martin Liska <mliska@suse.cz>
20237 * doc/gcov.texi: Enhance documentation of gcov.
20239 2017-04-28 Martin Liska <mliska@suse.cz>
20241 * doc/gcov.texi: Sort options in alphabetic order.
20242 * doc/gcov-dump.texi: Likewise.
20243 * doc/gcov-tool.texi: Likewise.
20244 * gcov.c (print_usage): Likewise.
20245 * gcov-dump.c (print_usage): Likewise.
20246 * gcov-tool.c (print_merge_usage_message): Likewise.
20247 (print_rewrite_usage_message): Likewise.
20248 (print_overlap_usage_message): Likewise.
20250 2017-04-28 Martin Liska <mliska@suse.cz>
20252 PR gcov-profile/53915
20253 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20255 2017-04-28 Martin Liska <mliska@suse.cz>
20257 PR gcov-profile/79891
20258 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20259 is marked by compiler as living on a line.
20260 (get_cycles_count): Remove usage of the union.
20261 (output_intermediate_file): Likewise.
20262 (find_source): Fix GNU coding style.
20263 (accumulate_line_counts): Remove old non-all block mode.
20264 (output_lines): Remove usage of the union.
20265 * profile.c (output_location): Include all BBs, even if
20266 belonging to a same line (and file) as a previous BB.
20268 2017-04-28 Martin Liska <mliska@suse.cz>
20270 * gcov.c (process_args): Handle new argument 'w'.
20271 (read_graph_file): Assign ID to BBs.
20272 (output_branch_count): Display BB # if verbose flag is set.
20273 (output_lines): Likewise for arcs.
20274 (print_usage): Add '--verbose' option help.
20275 * doc/gcov.texi: Document --verbose (-w) option.
20277 2017-04-28 Martin Liska <mliska@suse.cz>
20279 * gcov.c (struct block_location_info): New struct.
20280 (process_file): Fill up the new structure.
20281 (read_graph_file): Replace usage of encoding by the newly added
20283 (add_line_counts): Likewise.
20284 (accumulate_line_counts): Remove usage of the union.
20285 (function_info::function_info): New function.
20286 (function_info::~function_info): Likewise.
20287 (process_file): Call delete instead of release_function.
20288 (release_function): Release the function.
20289 (release_structures): Call delete instead of release_function.
20290 (solve_flow_graph): Replace usage of num_blocks.
20291 (find_exception_blocks): Likewise.
20292 (output_lines): Fix GNU coding style.
20294 2017-04-28 Martin Liska <mliska@suse.cz>
20297 * coverage.c (coverage_remove_note_file): New function.
20298 * coverage.h: Declare the function.
20299 * toplev.c (finalize): Clean if an error has been seen.
20301 2017-04-28 Martin Liska <mliska@suse.cz>
20303 PR gcov-profile/80031
20304 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20305 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20306 * gcov.c (read_graph_file): Read just number of blocks.
20307 * profile.c (branch_prob): Do not stream 0 flags per a basic
20310 2017-04-28 Martin Liska <mliska@suse.cz>
20312 * gcov-dump.c (tag_*): Add new argument to declarations.
20313 (dump_gcov_file): Likewise.
20314 (tag_blocks): Add and use new argument depth.
20315 (tag_arcs): Likewise.
20316 (tag_lines): Likewise.
20317 (tag_counters): Likewise.
20318 (tag_summary): Likewise.
20319 (dump_working_sets): Use depth to do a proper indentation.
20321 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20324 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20325 * symtab.c (symtab_node::debug_symtab): Move definition here.
20327 2017-04-28 Richard Biener <rguenther@suse.de>
20329 * lto-streamer.h (LTO_major_version): Bump to 7.
20331 2017-04-28 Richard Biener <rguenther@suse.de>
20333 * tree-vrp.c (assert_info): New struct.
20334 (add_assert_info): New helper.
20335 (register_edge_assert_for_2): Refactor to add asserts to a vector
20337 (register_edge_assert_for_1): Likewise.
20338 (register_edge_assert_for): Likewise.
20339 (finish_register_edge_assert_for): New helper actually registering
20340 asserts where live on edge.
20341 (find_conditional_asserts): Adjust.
20342 (find_switch_asserts): Likewise.
20343 (evrp_dom_walker::try_find_new_range): Generalize.
20344 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20346 2017-04-27 Marek Polacek <polacek@redhat.com>
20349 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20350 arg10 and arg11 to itype.
20352 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20354 * doc/extend.texi (Object Size Checking): Improve grammar.
20356 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20359 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20360 that the logic for permitting reciprocal estimates matches that
20363 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20366 * tree.c (type_cache_hasher::equal): Only compare
20367 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20368 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20369 non-aggregate element types.
20370 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20371 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20373 2017-04-27 Richard Biener <rguenther@suse.de>
20375 PR middle-end/80533
20376 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20377 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20378 keeping a reference to a trailing array.
20380 2017-04-27 Richard Biener <rguenther@suse.de>
20382 PR middle-end/80539
20383 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20384 being in loop-closed SSA form conservatively.
20385 (chrec_fold_multiply_poly_poly): Likewise.
20387 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20389 PR middle-end/79665
20390 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20391 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20393 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20396 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20397 (aarch64_function_arg_alignment): Return unsigned int again, but still
20398 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20399 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20400 Don't emit -Wpsabi note.
20401 (aarch64_function_arg_boundary): Likewise.
20402 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20405 2017-04-26 Nathan Sidwell <nathan@acm.org>
20407 * tree.h (crc32_unsigned_n): Declare.
20408 (crc32_unsigned, crc32_unsigned): Make inline.
20409 * tree.c (crc32_unsigned_bits): Replace with ...
20410 (crc32_unsigned_n): ... this.
20411 (crc32_unsigned, crc32_byte): Remove.
20412 (crc32_string): Remove unnecessary braces.
20414 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20416 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20417 * ipa-inline-analysis.c (MAX_TIME): Remove.
20418 (account_size_time): Use sreal for time.
20419 (dump_inline_summary): Update.
20420 (estimate_function_body_sizes): Update.
20421 (estimate_edge_size_and_time): Update.
20422 (estimate_calls_size_and_time): Update.
20423 (estimate_node_size_and_time): Update.
20424 (inline_merge_summary): Update.
20425 (inline_update_overall_summary): Update.
20426 (estimate_time_after_inlining): Update.
20427 (inline_read_section): Update.
20428 (inline_write_summary): Update.
20429 * ipa-inline.c (compute_uninlined_call_time): Update.
20430 (compute_inlined_call_time): Update.
20431 (recursive_inlining): Update.
20432 (inline_small_functions): Update.
20433 (dump_overall_stats): Update.
20434 * ipa-inline.h: Include sreal.h.
20435 (size_time_entry): Turn time to sreal.
20436 (inline_summary): Turn self_time nad time to sreal.
20438 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20440 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20442 (sreal::stream_out, sreal::stream_in): New.
20443 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20445 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20447 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20450 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20453 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20455 (dimode_scalar_chain::compute_convert_gain): Ditto.
20456 (dimode_scalar_chain::make_vector_copies): Ditto.
20457 (dimode_scalar_chain::convert_reg): Ditto.
20458 (dimode_scalar_chain::convert_insn): Ditto.
20459 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20460 (VI248_AVX512BW_1): New mode iterator.
20461 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20462 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20465 2017-04-25 Martin Sebor <msebor@redhat.com>
20467 PR tree-optimization/80497
20468 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20469 constants are representable in HOST_WIDE_INT.
20470 (parse_directive): Ditto.
20472 2017-04-25 Martin Sebor <msebor@redhat.com>
20475 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20476 (new_zero_array): Adjust signature.
20477 (dom_info::dom_init): Used unsigned rather that size_t.
20478 (dom_info::dom_info): Same.
20480 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20481 Jakub Jelinek <jakub@redhat.com>
20484 * config/arm/arm.c: Include gimple.h.
20485 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20486 returns negative, increment ncrn only if it returned positive.
20487 (arm_needs_doubleword_align): Return int instead of bool,
20488 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20489 members, but if there is any such non-FIELD_DECL
20490 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20491 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20492 returns negative, increment nregs only if it returned positive.
20493 (arm_setup_incoming_varargs): Likewise.
20494 (arm_function_arg_boundary): Emit -Wpsabi note if
20495 arm_needs_doubleword_align returns negative, return
20496 DOUBLEWORD_ALIGNMENT only if it returned positive.
20498 2017-04-25 Marek Polacek <polacek@redhat.com>
20501 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20502 first argument to type.
20504 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20507 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20508 type checks to test for compatibility instead of equality.
20510 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20511 Jakub Jelinek <jakub@redhat.com>
20514 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20516 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20517 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20518 the alignment computation, but return their maximum in warn_alignment.
20519 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20520 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20522 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20523 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20526 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20528 * config/arc/simdext.md (dmpyh): Fix typo.
20530 2017-04-25 Richard Biener <rguenther@suse.de>
20532 PR tree-optimization/80492
20533 * alias.c (compare_base_decls): Handle registers with asm
20534 specification conservatively.
20535 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20536 compare_base_decls returning dont-know properly.
20538 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20540 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20541 (legitimate_offset_address_p): New function.
20542 (arc_legitimate_address_p): Use above function.
20544 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20546 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20548 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20550 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20551 ACCH registers whenever they are available.
20553 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20555 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20556 double regs fix when not used.
20558 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20560 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20562 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20563 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20565 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20567 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20568 when emitting short ADD instructions.
20570 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20572 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20574 (cmpsi_cc_c_insn): Likewise.
20575 (cbranchsi4_scratch): Compute proper instruction length using
20576 compact_hreg_operand.
20577 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20579 2017-04-25 Richard Biener <rguenther@suse.de>
20581 PR middle-end/80509
20582 * passes.c (pass_manager::pass_manager): Initialize
20583 m_name_to_pass_map.
20585 2017-04-25 Richard Biener <rguenther@suse.de>
20587 PR tree-optimization/79201
20588 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20590 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20593 * config/s390/vector.md: Split MEM->GPR vector moves for
20594 non-s_operand addresses.
20596 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20599 * config/s390/predicates.md (reload_const_wide_int_operand): New
20601 * config/s390/s390.md ("movti"): Remove d/P alternative.
20602 ("movti_bigconst"): New pattern definition.
20604 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20607 * s390-protos.h (s390_expand_cs_hqi): Removed.
20608 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20609 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20610 modes as well as CCZ1mode and CCZmode.
20611 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20612 signature of s390_emit_compare_and_swap.
20613 (s390_expand_cs_hqi): Likewise, make static.
20614 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20615 compare-and-swap, in some cases.
20616 (s390_expand_cs): Wrapper function.
20617 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20619 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20620 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20621 patterns for small and large integers. Forbid symref memory operands.
20622 Move expander to s390.c. Require cc register.
20623 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20624 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20625 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20626 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20627 symref memory operands. Remove CC mode and call s390_match_ccmode
20629 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20631 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20633 * config/s390/s390.md (define_peephole2): New peephole to help
20634 combining the load-and-test pattern with volatile memory.
20636 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20638 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20639 with CCZmode for TARGET_Z196.
20641 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20643 PR rtl-optimization/80501
20644 * combine.c (make_compound_operation_int): Set subreg_code to SET
20645 even for AND with mask of the sign bit of mode.
20647 PR rtl-optimization/80500
20648 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20649 sum's initial value.
20651 2017-04-25 Julian Brown <julian@codesourcery.com>
20652 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20654 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20656 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20658 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20660 2017-04-25 Julian Brown <julian@codesourcery.com>
20661 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20663 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20664 (thunderx2t99_sha): New Reservation.
20666 2017-04-25 Julian Brown <julian@codesourcery.com>
20667 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20669 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20670 type for 1-element load.
20672 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20674 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20676 2017-04-24 Martin Jambor <mjambor@suse.cz>
20678 PR tree-optimization/80293
20679 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20680 char arrays not totally scalarizable if it is false.
20681 (analyze_all_variable_accesses): Pass correct value in the new
20682 parameter. Add a statistics counter.
20684 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20686 PR middle-end/79931
20687 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20689 2017-04-24 Richard Biener <rguenther@suse.de>
20691 PR tree-optimization/80494
20692 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20693 out for complex types.
20695 2017-04-24 Richard Biener <rguenther@suse.de>
20697 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20698 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20699 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20700 (DFS): Adjust and never fail.
20701 (sccvn_dom_walker::fail): Remove.
20702 (sccvn_dom_walker::before_dom_children): Adjust.
20703 (run_scc_vn): Likewise and never fail.
20704 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20705 (pass_fre::execute): Likewise.
20707 2017-04-24 Richard Biener <rguenther@suse.de>
20709 PR tree-optimization/79725
20710 * tree-ssa-sink.c (statement_sink_location): Return whether
20711 failure reason was zero uses. Move that check later.
20712 (sink_code_in_bb): Deal with zero uses by removing the stmt
20715 2017-04-24 Richard Biener <rguenther@suse.de>
20718 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20719 pointer-based references.
20721 2017-04-24 Richard Biener <rguenther@suse.de>
20724 * pass_manager.h (pass_manager::operator new): Remove.
20725 (pass_manager::operator delete): Likewise.
20726 * passes.c (pass_manager::operator new): Remove.
20727 (pass_manager::operator delete): Likewise.
20728 (pass_manager::pass_manager): Zero individual pass members.
20730 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20733 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20734 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20735 Check "XEXP (src, 1)" operand here.
20736 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20737 Check "XEXP (src, 1)" operand here.
20738 (dimode_scalar_chain::make_vector_copies): Detect count register
20739 of a shift instruction. Zero extend count register from QImode
20740 to DImode to satisfy vector shift pattern count operand predicate.
20741 Substitute vector shift count operand with a DImode copy.
20742 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20745 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20747 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20748 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20749 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20750 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20751 (UNSPEC_NOREX_MEM): Remove definition.
20753 2017-04-21 Richard Biener <rguenther@suse.de>
20755 PR tree-optimization/79547
20756 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20757 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20758 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20759 without any constraints.
20761 2017-04-21 Richard Biener <rguenther@suse.de>
20763 PR tree-optimization/78847
20764 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20766 2017-04-21 Richard Biener <rguenther@suse.de>
20768 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20769 (build_distinct_type_copy): Likewise.
20770 (build_variant_type_copy): Likewise.
20771 * tree.c (build_qualified_type): Pass down mem-stat info.
20772 (build_distinct_type_copy): Likewise.
20773 (build_variant_type_copy): Likewise.
20775 2017-04-21 Richard Biener <rguenther@suse.de>
20777 PR tree-optimization/80237
20778 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20780 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20781 for a simplified result.
20783 2016-04-21 Richard Biener <rguenther@suse.de>
20785 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20786 sth as strict as a simple_iv but a chrec without symbols and an
20787 operand defined in the loop we are peeling (and not some subloop).
20788 (propagate_constants_for_unrolling): Propagate all constants.
20790 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20793 * config/i386/i386.c (print_reg): Remove assert for disalowed
20794 regno values, call output_operand_lossage instead.
20796 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20799 * config/i386/constraints.md (Yc): New register constraint.
20800 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20801 Use Yc constraint for alternative 2 of operand 0. Remove
20802 preferred_for_speed attribute.
20804 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20806 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20807 lastprivate clauses in SIMT case.
20809 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20811 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20813 2017-04-20 Richard Biener <rguenther@suse.de>
20815 PR tree-optimization/57796
20816 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20817 as N scalar stores.
20818 (vect_model_load_cost): Cost gathers as N scalar loads.
20820 2017-04-20 Richard Biener <rguenther@suse.de>
20822 * ggc-page.c (ggc_allocated_p): Rename to ...
20823 (safe_lookup_page_table_entry): ... this and return the lookup
20825 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20827 2017-04-20 Richard Biener <rguenther@suse.de>
20829 PR tree-optimization/80453
20830 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20831 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20832 from the conditions.
20833 (vn_phi_eq): Pass them down.
20834 (vn_phi_lookup): Record them.
20835 (vn_phi_insert): Likewise.
20837 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20839 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20840 uninitialized variable warning to avoid buffer overrun.
20842 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20845 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20846 is suppressed for '{ 0 }' in C.
20848 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20850 * BASE-VER: Set to 8.0.0.
20852 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20854 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20855 priority .init_array and .fini_array section with SECTION_NOTYPE
20858 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20860 PR middle-end/80423
20861 * tree.h (build_array_type): Add typeless_storage default argument.
20862 * tree.c (type_cache_hasher::equal): Also compare
20863 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20864 (build_array_type): Add typeless_storage argument, set
20865 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20867 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20868 (build_array_type): Likewise. Add typeless_storage argument.
20870 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20871 Jakub Jelinek <jakub@redhat.com>
20873 PR tree-optimization/80426
20874 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20875 operation on symbolic operands, also compute the overflow for the
20876 invariant part when the operation degenerates into a negation.
20878 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20881 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20882 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20885 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20887 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20890 * config/avr/avr.c (tree.h): Include it.
20891 (cgraph.h): Include it.
20892 (avr_encode_section_info): Don't warn for uninitialized progmem
20893 variable if it's just an alias.
20895 2017-04-19 Richard Biener <rguenther@suse.de>
20898 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20899 when needed by AutoPGO.
20901 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20904 * doc/lto.texi: Remove an extra 'that'.
20906 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20908 PR rtl-optimization/80429
20909 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20910 are only used in debug insns.
20912 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20913 Vladimir Makarov <vmakarov@redhat.com>
20915 * config/sparc/predicates.md (input_operand): Add comment. Return
20916 true for any memory operand when LRA is in progress.
20917 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20919 2017-04-18 Jeff Law <law@redhat.com>
20922 * mips.md ({return,simple_return}_internal): Do not overwrite
20925 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20927 PR tree-optimization/80443
20928 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20929 instead of adding 1, subtract -1 and similarly instead of subtracting
20932 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20934 PR rtl-optimization/80357
20935 * haifa-sched.c (tmp_bitmap): New variable.
20936 (model_recompute): Handle duplicate use records.
20937 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20938 (free_global_sched_pressure_data): Free it.
20940 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20943 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20944 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20945 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20946 instead of SYSTEM_HEADER_DIR.
20948 2017-04-18 Jeff Law <law@redhat.com>
20950 PR middle-end/80422
20951 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20952 predecessors after walking up the insn chain.
20954 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20957 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20958 sizetype type into debug info.
20960 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20963 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20964 unneeded test for TARGET_UPPER_REGS_SF.
20965 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20967 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20970 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20971 instead of gsi_after_labels.
20973 2017-04-18 Jeff Law <law@redhat.com>
20975 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20979 2017-04-13 Jeff Law <law@redhat.com>
20980 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20981 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20983 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20986 * config/avr/avr.c (intl.h): Include it.
20987 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20989 2017-04-18 Martin Liska <mliska@suse.cz>
20991 PR gcov-profile/78783
20992 * gcov-tool.c (gcov_output_files): Validate that destination
20993 file is either removed by the tool or by a user.
20995 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20996 Guy Benyei <guybe@mellanox.com>
20998 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20999 block, and do not negate it, the stored id is already negative.
21001 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21003 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21005 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21008 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21009 masks of options that should be turned off if the VSX vector
21010 options are turned off.
21011 (OTHER_P8_VECTOR_MASKS): Likewise.
21012 (OTHER_VSX_VECTOR_MASKS): Likewise.
21013 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21014 rs6000_disable_incompatible_switches to validate no type switches
21016 (rs6000_incompatible_switch): New function to disallow turning on
21017 other vector options if -mno-vsx, -mno-power8-vector, or
21018 -mno-power9-vector are specified.
21020 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21022 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21024 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21026 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21027 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21028 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21029 (ARG_POINTER_CFA_OFFSET): Likewise.
21031 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21033 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21034 conditions to take advantage of various optimizations.
21036 2017-04-13 Jeff Law <law@redhat.com>
21038 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21039 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21040 (zero_extendsidi2_dext): Likewise.
21042 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21045 * fold-const.c (fold_ternary_loc): Revert
21046 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21049 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21051 PR rtl-optimization/80343
21052 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21053 new scratch pseudo.
21055 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21058 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21059 to ubsan_encode_value.
21061 2017-04-13 Jeff Law <law@redhat.com>
21063 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21064 appearing in DEBUG_INSNs.
21066 2017-04-13 Martin Liska <mliska@suse.cz>
21068 PR gcov-profile/80413
21069 * gcov-io.c (gcov_write_string): Copy to buffer just when
21070 allocated size is greater than zero.
21072 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21075 * dwarf2out.c (decls_for_scope): Ignore declarations of
21076 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21078 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21081 * ipa-visibility.c (non_local_p): Fix typos.
21082 (localize_node): When localizing symbol in same comdat group,
21083 dissolve the group only when we know external symbols are going
21085 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21087 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21089 PR tree-optimization/79390
21090 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21091 order does not result in usable sequence, retry with reversed operand
21097 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21098 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21099 op0 instead of fold_convert_loc (loc, type, arg0).
21101 2017-04-12 Jeff Law <law@redhat.com>
21103 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21104 has a delay slot in the generated code.
21106 * config/cris/cris.md (cris_preferred_reload_class): Return
21107 GENNONACR_REGS rather than GENERAL_REGS.
21109 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21112 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21113 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21114 signedness of the result type.
21116 2017-04-12 Richard Biener <rguenther@suse.de>
21117 Jeff Law <law@redhat.com>
21119 PR tree-optimization/80359
21120 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21121 trim stores to TARGET_MEM_REFs.
21123 2017-04-12 Richard Biener <rguenther@suse.de>
21125 PR tree-optimization/79390
21126 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21127 threading case even more.
21129 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21132 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21133 for quad_address_p for TImode, instead of just not indexed_address.
21135 2017-04-12 Richard Biener <rguenther@suse.de>
21136 Bernd Edlinger <bernd.edlinger@hotmail.de>
21138 PR middle-end/79671
21139 * alias.c (component_uses_parent_alias_set_from): Handle
21140 TYPE_TYPELESS_STORAGE.
21141 (get_alias_set): Likewise.
21142 * tree-core.h (tree_type_common): Add typeless_storage flag.
21143 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21144 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21145 for types containing members with TYPE_TYPELESS_STORAGE.
21146 (place_field): Likewise.
21147 (layout_type): Likewise for ARRAY_TYPE.
21148 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21149 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21150 TYPE_TYPELESS_STORAGE.
21151 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21153 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21156 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21157 first argument to type.
21159 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21163 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21164 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21165 (rs6000_expand_binop_builtin): Likewise.
21166 (rs6000_expand_ternop_builtin): Likewise; also add missing
21167 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21168 vshasigma built-ins.
21169 * doc/extend.texi: Document that vec_xxpermdi's third argument
21170 must be a constant.
21172 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21174 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21175 Use shift_const cost parameter when calculating gain of STV shifts.
21177 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21179 PR rtl-optimization/70478
21180 * lra-constraints.c (process_alt_operands): Check memory for
21181 disfavoring memory insn operand.
21183 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21185 PR middle-end/80100
21186 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21187 left shift in unsigned HOST_WIDE_INT type.
21189 PR rtl-optimization/80385
21190 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21191 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21194 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21195 if they have any depend clauses.
21197 2017-04-11 Martin Liska <mliska@suse.cz>
21200 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21201 * ipa-split.c (split_function): Create a local comdat symbol
21202 if caller is in a comdat group.
21204 2017-04-11 Martin Liska <mliska@suse.cz>
21207 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21210 2017-04-11 Martin Sebor <msebor@redhat.com>
21212 PR middle-end/80364
21213 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21214 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21216 (directive::set_width, directive::set_precision, format_character):
21218 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21221 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21224 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21225 conflict, set target->arch_name instead of target->cpu_name.
21227 2017-04-11 Richard Biener <rguenther@suse.de>
21229 PR tree-optimization/80374
21230 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21231 build_zero_cst, remove fold_convertible_p check again.
21233 2017-04-11 Martin Liska <mliska@suse.cz>
21236 * ubsan.c (instrument_object_size): Do not instrument register
21239 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21242 * config/i386/i386-builtin-types.def
21243 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21244 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21245 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21246 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21247 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21248 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21249 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21250 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21251 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21252 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21253 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21254 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21255 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21256 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21257 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21258 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21259 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21260 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21261 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21262 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21263 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21264 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21265 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21266 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21267 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21268 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21269 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21270 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21271 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21272 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21273 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21274 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21275 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21276 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21277 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21278 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21279 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21280 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21281 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21282 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21283 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21284 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21285 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21286 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21287 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21288 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21290 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21291 flag to second_arg_count, handle 4 argument function type _COUNT
21292 aliases, handle second_arg_count on second argument rather than last.
21294 2017-04-10 Jeff Law <law@redhat.com>
21296 PR tree-optimization/80374
21297 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21298 record anything if we can not convert integer_zero_node to the
21301 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21304 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21305 Enhance special handling given to the TARGET_P9_MINMAX option in
21306 relation to certain other options.
21308 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21310 PR tree-optimization/80153
21311 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21312 remove POINTER_PLUS_EXPR's base part directly, rather than through
21315 2017-04-10 Richard Biener <rguenther@suse.de>
21316 Bin Cheng <bin.cheng@arm.com>
21318 PR tree-optimization/80153
21319 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21320 the first element of pointer type aff_tree. Build result expr in
21322 (add_elt_to_tree): Convert to type unconditionally. Remove other
21323 fold_convert calls.
21324 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21325 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21327 2017-04-10 Richard Biener <rguenther@suse.de>
21329 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21332 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21334 PR rtl-optimization/70478
21335 * lra-constraints.c (curr_small_class_check): New.
21336 (update_and_check_small_class_inputs): New.
21337 (process_alt_operands): Update curr_small_class_check. Disfavor
21338 alternative insn memory operands. Check available regs for small
21341 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21344 * config/mips/mips.opt (-mvirt): Update description.
21345 * doc/invoke.texi (-mvirt): Likewise.
21347 2017-04-10 Richard Biener <rguenther@suse.de>
21349 PR middle-end/80362
21350 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21351 looking for NEGATE_EXPR in -A / -B to A / B folding.
21353 2017-04-10 Martin Liska <mliska@suse.cz>
21355 PR gcov-profile/80224
21356 * gcov.c (print_usage): Fix usage string.
21357 (get_gcov_intermediate_filename): Remove.
21358 (output_gcov_file): Use both for normal and intermediate format.
21359 (generate_results): Do not initialize special file for
21360 intermediate format.
21362 2017-04-10 Richard Biener <rguenther@suse.de>
21364 PR tree-optimization/80304
21365 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21368 2017-04-10 Nathan Sidwell <nathan@acm.org>
21371 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21372 (rs6000_init_builtins): Use it.
21374 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21376 * config/arm/arm.md (<mrc>): Add mode to SET source.
21377 (<mrrc>): Likewise.
21379 2017-04-10 Richard Biener <rguenther@suse.de>
21381 PR middle-end/80344
21382 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21384 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21387 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21388 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21389 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21390 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21391 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21392 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21393 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21394 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21395 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21396 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21397 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21398 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21399 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21400 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21401 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21402 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21403 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21404 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21405 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21406 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21407 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21408 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21409 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21411 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21413 PR rtl-optimization/70478
21414 * lra-constraints.c: Reverse the last patch.
21416 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21418 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21419 Add comment for WCHAR_T.
21421 2017-04-08 Martin Liska <mliska@suse.cz>
21424 2017-04-07 Martin Liska <mliska@suse.cz>
21427 * ipa-split.c (split_function): Add function part to a same comdat
21430 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21433 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21435 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21437 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21438 (vec_pairing): ...to this.
21439 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21440 (rs6000_sched_init): Adjust for name change.
21441 (struct rs6000_sched_context): Likewise.
21442 (rs6000_init_sched_context): Likewise.
21443 (rs6000_set_sched_context): Likewise.
21445 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21451 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21453 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21454 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21455 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21457 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21459 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21461 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21463 PR rtl-optimization/70703
21464 * ira-color.c (update_conflict_hard_regno_costs): Use
21465 int64_t instead of HOST_WIDE_INT.
21467 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21469 PR rtl-optimization/70478
21470 * lra-constraints.c (process_alt_operands): Disfavor alternative
21471 insn memory operands.
21473 2017-04-07 Jeff Law <law@redhat.com>
21475 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21476 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21478 2017-04-07 Martin Liska <mliska@suse.cz>
21481 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21482 Show error message instead of an ICE.
21484 2017-04-07 Martin Liska <mliska@suse.cz>
21487 * ipa-split.c (split_function): Add function part to a same comdat
21490 2017-04-07 Richard Biener <rguenther@suse.de>
21492 PR middle-end/80341
21493 * tree.c (get_unwidened): Also handle ! for_type case for
21495 * convert.c (do_narrow): Split out from ...
21496 (convert_to_integer_1): ... here. Do not pass final truncation
21497 type to get_unwidened for TRUNC_DIV_EXPR.
21499 2017-04-07 Richard Biener <rguenther@suse.de>
21501 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21503 (aff_combination_const): Adjust.
21504 (aff_combination_scale): Likewise.
21505 (aff_combination_add_elt): Likewise.
21506 (aff_combination_add_cst): Likewise.
21507 (aff_combination_convert): Likewise.
21508 (add_elt_to_tree): Likewise. Remove unused argument.
21509 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21511 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21513 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21515 * config/arm/arm.c (arm_default_short_enums): Use
21516 ARM_DEFAULT_SHORT_ENUMS.
21517 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21519 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21522 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21523 members with redundant out-of-class redeclaration.
21525 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21528 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21529 * config/i386/i386.md (*zero_extendsidi2):
21530 Add (?*x,*x) and (?*v,*v) alternatives.
21532 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21535 * config/i386/i386.c (ix86_expand_builtin)
21536 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21537 mode from insn data. Convert operands to insn operand mode.
21538 Copy operands that don't satisfy insn predicate to a register.
21540 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21542 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21545 2017-04-06 Richard Biener <rguenther@suse.de>
21547 PR tree-optimization/80334
21548 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21549 preserve alignment of accesses.
21551 2017-04-06 Richard Biener <rguenther@suse.de>
21553 PR tree-optimization/80262
21554 * tree-sra.c (build_ref_for_offset): Preserve address-space
21556 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21557 Drop useless address-space information on MEM_REF offsets.
21559 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21561 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21563 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21565 PR rtl-optimization/70703
21566 * ira-color.c (update_conflict_hard_regno_costs): Use
21567 HOST_WIDE_INT instead of long.
21569 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21572 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21573 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21575 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21576 for x86_64 target. Handle -m3dnowa option.
21578 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21580 PR rtl-optimization/70703
21581 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21582 (update_conflict_hard_regno_costs): Use long instead of unsigned
21583 arithmetic for cost calculation.
21585 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21586 Bernd Edlinger <bernd.edlinger@hotmail.de>
21589 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21592 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21595 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21596 ptr_mode with Pmode throughout.
21597 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21598 into probe_stack_range and use DImode.
21600 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21603 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21604 call_eh_return is true.
21606 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21608 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21609 Initialize last_match_fntype_index.
21611 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21614 * tree-nvr.c: Include internal-fn.h.
21615 (pass_return_slot::execute): Ignore internal calls without
21618 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21619 Richard Biener <rguenther@suse.de>
21622 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21623 captures used multiple times, except for the last use.
21624 * generic-match-head.c: Include gimplify.h.
21626 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21628 PR tree-optimization/79390
21629 * target.h (struct noce_if_info): Declare.
21630 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21631 * target.def (noce_conversion_profitable_p): New target hook.
21632 * ifcvt.h (struct noce_if_info): New type, moved from ...
21633 * ifcvt.c (struct noce_if_info): ... here.
21634 (noce_conversion_profitable_p): Renamed to ...
21635 (default_noce_conversion_profitable_p): ... this. No longer
21637 (noce_try_store_flag_constants, noce_try_addcc,
21638 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21639 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21640 instead of noce_conversion_profitable_p.
21641 * config/i386/i386.c: Include ifcvt.h.
21642 (ix86_option_override_internal): Don't override
21643 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21644 (ix86_noce_conversion_profitable_p): New function.
21645 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21646 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21647 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21648 * doc/tm.texi: Regenerated.
21650 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21652 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21655 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21658 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21659 instructions for small multiply cores.
21661 2017-04-04 Jeff Law <law@redhat.com>
21663 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21665 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21666 that are not set by the loop over the elements.
21668 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21671 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21672 int mode, convert_modes it to mode as unsigned, otherwise use
21673 lowpart_subreg to mode rather than SImode.
21674 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21675 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21676 Use DImode instead of SImode for the shift count operand.
21677 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21680 2017-04-04 Richard Biener <rguenther@suse.de>
21682 PR middle-end/80281
21683 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21684 arithmetic done for the negate or the plus. Simplify.
21685 (A - (-B) -> A + B): Likewise.
21686 * fold-const.c (split_tree): Make sure to not negate pointers.
21688 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21690 PR rtl-optimization/60818
21691 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21692 a compare of comparisons with the thing compared if this results
21693 in a different machine mode.
21695 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21697 * alias.c (base_alias_check): Fix typo in comment.
21698 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21699 * cgraphunit.c (symbol_table::compile): Likewise.
21700 * collect2.c (maybe_run_lto_and_relink): Likewise.
21701 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21702 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21703 * config/avr/avr.c (avr_map_op_t): Likewise.
21704 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21705 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21706 * config/epiphany/epiphany.md (movcc): Likewise.
21707 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21708 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21710 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21711 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21712 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21713 * config/sh/sh.c (sh_rtx_costs): Likewise.
21714 * fold-const.c (fold_truth_andor): Likewise.
21715 * genautomata.c (collapse_flag): Likewise.
21716 * gengtype.h (struct type::u::s): Likewise.
21717 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21718 * input.c (FORMAT_AMOUNT): Likewise.
21719 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21720 (known_aggs_to_agg_replacement_list): Likewise.
21721 * ipa-inline-analysis.c: Likewise.
21722 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21723 * ipa-polymorphic-call.c
21724 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21725 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21726 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21728 * modulo-sched.c (apply_reg_moves): Likewise.
21729 * omp-expand.c (build_omp_regions_1): Likewise.
21730 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21731 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21732 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21733 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21734 * value-prof.c: Likewise.
21735 * var-tracking.c (val_reset): Likewise.
21737 2017-04-03 Richard Biener <rguenther@suse.de>
21739 PR tree-optimization/80275
21740 * fold-const.c (split_address_to_core_and_offset): Handle
21743 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21745 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21746 descriptors is at least equal to that of functions.
21748 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21750 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21752 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21755 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21756 (mov<IMOD4:mode>): New expander.
21757 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21759 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21761 PR rtl-optimization/79405
21762 * fwprop.c (propagations_left): New variable.
21763 (forward_propagate_into): Decrement it.
21764 (fwprop_init): Initialize it.
21765 (fw_prop): If the variable has reached zero, stop propagating.
21766 (fwprop_addr): Ditto.
21768 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21771 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21772 a FUNCTION_DECL, pass it as decl instead of origin to
21775 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21777 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21780 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21783 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21784 TARGET_VSX_SMALL_INTEGER.
21786 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21788 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21789 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21791 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21793 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21794 extraction from odd-numbered MSA register.
21796 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21798 PR middle-end/80173
21799 * expmed.c (store_bit_field_1): Don't attempt to create
21800 a word subreg out of hard registers wider than word if they
21801 have HARD_REGNO_NREGS of 1 for their mode.
21803 PR middle-end/80163
21804 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21805 conversions to integer types wider than word and pointer.
21808 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21809 (rtx_equal_for_cselib_p): Pass 0 to it.
21810 * cselib.c (cselib_hasher::equal): Likewise.
21811 (rtx_equal_for_cselib_1): Add depth argument. If depth
21812 is 128, don't look up VALUE locs and punt. Increment
21813 depth in recursive calls when walking VALUE locs.
21815 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21817 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21818 (make_gcov_file_name): Use the canonical path name for generating
21820 (read_line): Fix handling of files with ascii null bytes.
21822 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21824 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21825 to initialise a vector register instead
21826 of using a const_int.
21828 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21830 PR translation/80189
21831 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21832 diagnostic messages.
21834 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21837 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21838 (dfp_diex_<mode>): Update mode of operand 1.
21839 * doc/extend.texi (dxex, dxexq): Document change to return type.
21840 (diex, diexq): Document change to argument type.
21842 2017-03-30 Martin Jambor <mjambor@suse.cz>
21845 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21846 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21847 it reflects the signature changes performed at the callee side.
21848 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21849 to cgraph_build_function_type_skip_args.
21850 (build_function_decl_skip_args): Adjust call to the above function.
21852 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21855 * config/i386/sse.md
21856 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21857 register as dest whenever it is a MEM not rtx_equal_p to the
21858 corresponding dup operand, and when forcing into reg move the
21859 reg into the memory afterwards.
21860 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21861 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21862 for the force_reg mode.
21863 (avx512vl_vextractf128<mode>): Use register as dest either
21864 always when a MEM, or when it is a MEM not rtx_equal_p to the
21865 corresponding dup operand, or even not when it is a CONST_VECTOR
21866 depending on the mode and lo vs. hi.
21867 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21869 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21870 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21871 Likewise. Require that operands[2] is even.
21872 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21873 Remove extraneous parens. Require that operands[2] is a multiple
21875 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21876 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21877 disallow memory then.
21879 2017-03-30 Richard Biener <rguenther@suse.de>
21881 PR tree-optimization/77498
21882 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21883 to non-constants over backedges.
21885 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21887 PR rtl-optimization/80233
21888 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21889 as last_combined_insn. Do not test for BARRIER_P separately.
21891 2017-03-29 Andreas Schwab <schwab@suse.de>
21894 * calls.c (prepare_call_address): Convert funexp to Pmode before
21895 copying to temp reg.
21897 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21899 PR tree-optimization/80158
21900 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21901 Handle possible future case of more than one alternate
21903 (replace_rhs_if_not_dup): Likewise.
21904 (replace_one_candidate): Likewise.
21906 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21908 PR rtl-optimization/80193
21909 * ira.c (ira): Do not check allocation for LRA.
21911 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21913 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21914 (nvptx_output_simt_exit): Declare.
21915 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21916 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21917 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21918 (nvptx_declare_function_name): ...here. Emit declaration of local
21919 memory space buffer for omp_simt_enter insn.
21920 (nvptx_output_unisimt_switch): New.
21921 (nvptx_output_softstack_switch): New.
21922 (nvptx_output_simt_enter): New.
21923 (nvptx_output_simt_exit): New.
21924 * config/nvptx/nvptx.h (struct machine_function): New fields
21925 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21926 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21927 (UNSPECV_SIMT_EXIT): Ditto.
21928 (omp_simt_enter_insn): New insn.
21929 (omp_simt_enter): New expansion.
21930 (omp_simt_exit): New insn.
21931 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21933 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21934 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21935 (expand_GOMP_SIMT_EXIT): New.
21936 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21937 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21938 (GOMP_SIMT_EXIT): Ditto.
21939 * target-insns.def (omp_simt_enter): New insn.
21940 (omp_simt_exit): Ditto.
21941 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21943 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21944 (lower_rec_input_clauses): Likewise.
21945 (lower_lastprivate_clauses): Handle SIMT privatization.
21947 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21948 (ompdevlow_adjust_simt_enter): New.
21949 (find_simtpriv_var_op): New.
21950 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21951 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21953 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21954 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21955 (copy_decl_for_dup_finish): Ditto.
21957 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21959 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21962 * config/i386/i386.c (ix86_option_override_internal): Always
21963 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21965 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21967 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21969 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21971 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21972 mark new edge's irreducible flag accordign to it.
21973 (vect_do_peeling): Check loop preheader edge's irreducible flag
21974 and pass it to function slpeel_add_loop_guard.
21976 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21978 PR tree-optimization/80218
21979 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21980 Update block frequencies and counts.
21982 2017-03-28 Richard Biener <rguenther@suse.de>
21984 PR tree-optimization/78644
21985 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21986 of a simplification result we may not use it at all.
21988 2017-03-28 Richard Biener <rguenther@suse.de>
21991 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21992 without arguments, generate default definition of a SSA name.
21994 2017-03-28 Richard Biener <rguenther@suse.de>
21996 PR middle-end/80222
21997 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21998 TYPE_REF_CAN_ALIAS_ALL references.
21999 * fold-const.c (fold_indirect_ref_1): Likewise.
22001 2017-03-28 Martin Liska <mliska@suse.cz>
22004 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22005 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22007 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22008 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22010 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22011 (EXTRA_SPECS): Define.
22012 (SUBTARGET_EXTRA_SPECS): Likewise.
22013 (SUBTARGET_CPP_SPEC): Likewise.
22014 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22015 SUBTARGET_EXTRA_SPECS.
22016 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22018 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22020 * config/arc/simdext.md (vst64_insn): Update pattern.
22021 (vld32wh_insn): Likewise.
22022 (vld32wl_insn): Likewise.
22023 (vld64_insn): Likewise.
22024 (vld32_insn): Likewise.
22026 2017-03-28 Marek Polacek <polacek@redhat.com>
22029 * fold-const.c (fold_comparison): Use protected_set_expr_location
22030 instead of SET_EXPR_LOCATION.
22032 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22034 * tree.c (add_expr): Avoid name lookup warning.
22036 2017-03-27 Jeff Law <law@redhat.com>
22038 PR tree-optimization/80216
22039 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22040 function name. Limit recursion depth.
22041 (record_temporary_equivalences): Corresponding changes.
22043 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22045 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22048 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22051 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22053 * cfgcleanup.c (reg_note_cfa_p): New array.
22054 (insns_have_identical_cfa_notes): New function.
22055 (old_insns_match_p): Don't cross-jump in between /f
22056 and non-/f instructions. If both i1 and i2 are frame related,
22057 verify all CFA notes, their order and content.
22059 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22062 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22063 HImode and SImode with zero extend to DImode to one insn.
22064 (bswap<mode>2_extenddi): Likewise.
22065 (bswapsi2_extenddi): Likewise.
22066 (bswaphi2_extendsi): Likewise.
22067 (bswaphi2): Combine bswap HImode and SImode into one insn.
22068 Separate memory insns from swapping register.
22069 (bswapsi2): Likewise.
22070 (bswap<mode>2): Likewise.
22071 (bswaphi2_internal): Delete, no longer used.
22072 (bswapsi2_internal): Likewise.
22073 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22074 store, and gpr<-gpr swap insns.
22075 (bswap<mode>2_store): Likewise.
22076 (bswaphi2_reg): Register only splitter, combine with the splitter.
22077 (bswaphi2 splitter): Likewise.
22078 (bswapsi2_reg): Likewise.
22079 (bswapsi2 splitter): Likewise.
22080 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22081 the insns into load, store, and register/register insns.
22082 (bswapdi2_ldbrx): Likewise.
22083 (bswapdi2_load): Likewise.
22084 (bswapdi2_store): Likewise.
22085 (bswapdi2_reg): Likewise.
22087 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22089 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22090 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22092 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22095 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22097 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22098 special handling for target option conflicts between dform
22099 options (-mpower9-dform, -mpower9-dform-vector,
22100 -mpower9-dform-scalar) and -mno-direct-move.
22102 2017-03-27 Richard Biener <rguenther@suse.de>
22104 PR tree-optimization/80181
22105 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22107 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22109 * config/arc/predicates.md (move_double_src_operand): Replace the
22110 call to move_double_src_operand with a call to address_operand.
22112 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22114 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22115 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22116 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22118 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22120 * config/arc/predicates.md (long_immediate_loadstore_operand):
22121 Consider scaled addresses cases.
22123 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22125 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22126 restored when in interrupt.
22127 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22128 doesn't have delay slot.
22130 2017-03-27 Richard Biener <rguenther@suse.de>
22133 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22134 inlined thunk clones.
22136 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22139 * asan.c (instrument_derefs): Copy over last operand from
22140 original COMPONENT_REF to the new COMPONENT_REF with
22141 DECL_BIT_FIELD_REPRESENTATIVE.
22142 * ubsan.c (instrument_object_size): Likewise.
22144 2017-03-27 Richard Biener <rguenther@suse.de>
22146 PR tree-optimization/80170
22147 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22148 sure DR/SCEV didnt fold in constants we do not see when looking
22149 at the reference base alignment.
22151 2017-03-27 Richard Biener <rguenther@suse.de>
22153 PR middle-end/80171
22154 * gimple-fold.c (fold_ctor_reference): Properly guard against
22155 NULL return value from canonicalize_constructor_val.
22157 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22160 * config/i386/i386.c (ix86_expand_builtin)
22161 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22162 flags reg setting and flags reg using instructions.
22163 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22164 clobbering instructions to zero extend op2.
22166 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22168 * doc/install.texi (Configuration) <--with-aix-soname>:
22169 Update link to AIX ld.
22171 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22173 PR rtl-optimization/80160
22174 PR rtl-optimization/80159
22175 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22176 reg_alternate_class into account.
22178 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22181 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22182 to consider in curr_insn_transform.
22184 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22186 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22187 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22188 and emit_mode_inner.
22190 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22192 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22193 argument to the overloaded builtin variants. Use the new flag to
22194 deprecate certain builtin variants.
22195 * config/s390/s390-builtin-types.def: Add new builtin types.
22196 * config/s390/s390-builtins.h: Support new flags field for
22197 overloaded builtins.
22198 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22199 (s390_macro_to_expand): Enable vector float data type.
22200 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22201 builtins by incrementing the __VEC__ version number.
22202 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22204 (s390_resolve_overloaded_builtin): Emit error messages depending
22205 on the builtin flags.
22206 * config/s390/s390.c (s390_expand_builtin): Support additional
22207 flags argument. Change error message to match the messages
22208 emitted in s390-c.c.
22209 * config/s390/s390.md: New UNSPEC_* constants.
22210 (op_type): Add new instruction types.
22211 * config/s390/vecintrin.h: Add new builtins and test data class
22213 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22214 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22215 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22216 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22217 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22218 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22220 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22221 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22222 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22223 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22225 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22226 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22227 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22228 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22230 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22231 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22232 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22233 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22234 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22235 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22236 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22238 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22239 ("vec_scatter_element<V_HW_4:mode>_DI")
22240 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22241 ("vec_fpint<mode>", "vflls")
22242 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22243 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22244 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22245 ("*vec_cmphe<mode>_cc"): ... these.
22247 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22248 mode constant instead of magic value.
22250 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22252 * config/s390/s390.c (s390_expand_vec_compare): Support other
22253 vector floating point modes than just V2DF.
22254 (s390_expand_vcond): Likewise.
22255 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22256 (s390_cannot_change_mode_class): Prevent mode changes between TF
22257 and V1TF in vector registers.
22258 * config/s390/s390.md (DF, SF): New mode attributes.
22259 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22260 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22261 SFmode support for VRs.
22262 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22264 (VFT, VF_HW): New mode iterators.
22265 (vw, sdx): New mode attributes.
22266 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22267 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22268 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22269 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22270 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22271 also the new vector floating point modes. Renaming to ...
22273 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22274 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22275 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22276 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22277 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22278 ("vec_unordered<mode>"): ... these.
22280 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22281 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22282 ("*vec_extendv2df"): New insn definitions.
22284 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22286 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22287 ("mulditi3_2", "*muldi3_sign"): New patterns.
22288 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22289 rename the pattern definition.
22291 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22293 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22295 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22297 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22299 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22300 instruction if possible.
22301 * config/s390/vector.md (vec_halfnumelts): New mode
22303 ("*vec_vllezlf<mode>"): New pattern.
22305 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22307 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22308 ("popcountv4si2", "popcountv2di2"): Rename to ...
22309 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22310 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22312 ("popcount<mode>2_vxe"): New pattern.
22314 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22316 * common/config/s390/s390-common.c (processor_flags_table): Add
22318 * config.gcc: Add arch12.
22319 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22320 Default to arch12 for unknown CPU model numbers.
22321 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22322 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22323 PROCESSOR_max sanity check.
22324 * config/s390/s390-opts.h (enum processor_type): Add
22326 * config/s390/s390.c (processor_table): Add arch12.
22327 (s390_expand_builtin): Add check for B_VXE flag.
22328 (s390_issue_rate): Add PROCESSOR_ARCH12.
22329 (s390_get_sched_attrmask): Likewise.
22330 (s390_get_unit_mask): Likewise.
22331 (s390_sched_score): Enable z13 scheduling for arch12.
22332 (s390_sched_reorder): Likewise.
22333 (s390_sched_variable_issue): Likewise.
22334 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22336 (s390_tune_attr): Use z13 scheduling also for arch12.
22337 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22338 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22339 (TARGET_VXE_P): New macros.
22340 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22341 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22342 * config/s390/s390.opt: Add arch12 as processor_type.
22344 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22346 * config/s390/s390.md
22347 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22348 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22349 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22351 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22352 Rename expanders to ...
22354 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22355 ("fixuns_truncdddi2_emu"): ... these.
22357 ("fixuns_trunc<mode>si2_emu"): New expander.
22359 ("*fixuns_truncdfdi2_z13"): Rename to ...
22360 ("*fixuns_truncdfdi2_vx"): ... this.
22362 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22364 * config/s390/2964.md: Remove the single element vector compare
22365 instructions which are no longer used.
22366 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22368 (s390_canonicalize_comparison): Remove handling of DFmode
22370 (s390_expand_vec_compare_scalar): Remove function.
22371 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22372 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22374 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22376 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22378 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22380 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22381 will anyway by matched by mov<mode>_64dfp.
22383 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22385 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22386 vlef/vstef. Add missing operand to vleif.
22388 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22390 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22391 pair for all vector types with 64 bit elements.
22392 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22393 * config/s390/vector.md (V_HW_64): ... here.
22394 (V_128_NOSINGLE): New mode iterator.
22395 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22396 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22397 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22398 ("*vec_load_pairv2di"): Change to ...
22399 ("*vec_load_pair<mode>"): ... this one.
22401 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22403 * config/s390/constraints.md: Add comments.
22404 (jKK): Reject element sizes > 8 bytes.
22405 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22407 * config/s390/s390.md: Add the s_operand checks formerly in
22408 s390_split_ok_p to various splitters where they are still
22410 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22411 for 128 bit vectors. Plus two splitters.
22413 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22415 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22418 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22421 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22422 error if the boundary argument is not constant.
22424 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22426 PR rtl-optimization/80112
22427 * loop-doloop.c (doloop_condition_get): Don't check condition
22428 if cmp isn't SET with IF_THEN_ELSE src.
22430 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22432 PR tree-optimization/80158
22433 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22434 replacing a candidate statement, also replace it for the
22435 candidate's alternate interpretation.
22436 (replace_rhs_if_not_dup): Likewise.
22437 (replace_one_candidate): Likewise.
22439 2017-03-24 Richard Biener <rguenther@suse.de>
22441 PR tree-optimization/80167
22442 * graphite-isl-ast-to-gimple.c
22443 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22445 (translate_isl_ast_to_gimple::get_rename): Likewise.
22447 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22449 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22450 handling of certain combinations of target options, including the
22451 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22452 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22454 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22457 * config/arm/arm.md (*load_multiple): Add reload_completed to
22458 matching condition.
22460 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22461 Richard Biener <rguenth@suse.de>
22463 PR tree-optimization/79908
22464 PR tree-optimization/80136
22465 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22466 been cast away, gimplify_and_add suffices.
22468 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22470 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22472 2017-03-23 Richard Biener <rguenther@suse.de>
22474 PR tree-optimization/80032
22475 * gimplify.c (gimple_push_cleanup): Forced unconditional
22476 cleanups still have to go to the conditional_cleanups
22479 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22481 PR tree-optimization/80072
22482 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22484 (next_operand_entry_id): Change type to unsigned int.
22485 (sort_by_operand_rank): Make sure to return the right return value
22486 even if unsigned fields are bigger than INT_MAX.
22487 (struct oecount): Change cnt and id type to unsigned int.
22488 (oecount_hasher::equal): Formatting fix.
22489 (oecount_cmp): Make sure to return the right return value
22490 even if unsigned fields are bigger than INT_MAX.
22491 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22494 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22495 TREE_READONLY on result if writing it more than once.
22498 * doc/invoke.texi (-fsanitize=thread): Document that with
22499 -fnon-call-exceptions atomics are not able to throw
22503 * tsan.c: Include tree-eh.h.
22504 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22505 maybe_clean_or_replace_eh_stmt where needed.
22506 (instrument_memory_accesses): Add cfg_changed argument.
22507 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22508 if it returned true.
22509 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22511 PR rtl-optimization/63191
22512 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22513 wrapper function, moved the whole old content into ...
22514 (ix86_delegitimize_address_1): ... this. New inline function.
22515 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22516 true as last argument instead of ix86_delegitimize_address.
22518 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22520 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22521 cortexa57_branch_cost.
22523 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22525 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22527 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22530 * doc/md.texi (Constraints): Document wA constraint.
22531 * config/rs6000/constraints.md (wA): New.
22532 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22533 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22534 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22535 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22537 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22540 * gimplify.c (is_oacc_declared): New function.
22541 (oacc_default_clause): Use it to set default flags for acc declared
22542 variables inside parallel regions.
22543 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22544 declared variables.
22545 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22546 declare attribute to any decl as necessary.
22548 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22551 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22552 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22553 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22554 (arm_arch_lpae): This.
22555 * config/arm/arm.c (arm_arch7ve): Rename into ...
22556 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22557 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22560 2017-03-22 Martin Liska <mliska@suse.cz>
22563 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22564 error message instead of an ICE.
22566 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22568 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22570 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22572 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22574 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22577 2017-03-21 Martin Sebor <msebor@redhat.com>
22579 * doc/extend.texi: Use "cannot" instead of "can't."
22580 * doc/hostconfig.texi: Same.
22581 * doc/install.texi: Same.
22582 * doc/invoke.texi: Same.
22583 * doc/loop.texi: Same.
22584 * doc/md.texi: Same.
22585 * doc/objc.texi: Same.
22586 * doc/rtl.texi: Same.
22587 * doc/tm.texi: Same.
22588 * doc/tm.texi.in: Same.
22589 * doc/trouble.texi: Same.
22591 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22594 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22595 (collect_checksum_attributes): Set it.
22596 (die_checksum_ordered): Use it.
22598 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22600 PR tree-optimization/79908
22601 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22602 change: For a VA_ARG whose LHS has been cast away, use
22603 force_gimple_operand to construct the side effects.
22605 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22607 PR translation/80001
22608 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22609 more amenable to translation.
22610 (oacc_loop_auto_partitions): Likewise.
22612 2017-03-21 Marek Polacek <polacek@redhat.com>
22613 Martin Sebor <msebor@redhat.com>
22615 PR tree-optimization/80109
22616 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22617 on INTEGRAL_TYPE_P.
22619 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22620 Segher Boessenkool <segher@kernel.crashing.org>
22623 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22624 check reg_used_between_p between insn and one of succ or succ2
22625 depending on if succ is artificial insn not inserted into insn
22628 2017-03-21 Martin Liska <mliska@suse.cz>
22630 PR gcov-profile/80081
22631 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22632 * doc/gcc.texi: Include gcov-dump stuff.
22633 * doc/gcov-dump.texi: New file.
22635 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22637 PR rtl-optimization/79150
22638 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22639 conditional jump, if the jump is the last insn of the loop.
22641 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22642 Richard Biener <rguenth@suse.de>
22644 PR tree-optimization/79908
22645 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22646 been cast away, use force_gimple_operand to construct the side
22649 2017-03-21 Martin Liska <mliska@suse.cz>
22651 PR libfortran/79956
22652 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22655 2017-03-21 Brad Spengler <spender@grsecurity.net>
22658 * plugin.c (htab_hash_plugin): New function.
22659 (add_new_plugin): Use it and adjust.
22660 (parse_plugin_arg_opt): Adjust.
22661 (init_one_plugin): Likewise.
22663 2017-03-21 Richard Biener <rguenther@suse.de>
22665 PR tree-optimization/80032
22666 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22667 if set force the cleanup to happen unconditionally.
22668 (gimplify_target_expr): Push inserted clobbers with force_uncond
22669 to avoid them being removed by control-dependent DCE.
22671 2017-03-21 Richard Biener <rguenther@suse.de>
22673 PR tree-optimization/80122
22674 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22675 va_arg_pack_len when the inlined call stmt requires pack
22677 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22679 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22682 * tsan.c (instrument_builtin_call): If the memory model argument
22683 is not a constant, assume it is valid.
22686 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22689 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22691 PR rtl-optimization/79910
22692 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22693 if its dest is used by an insn before I2 (other than the combined
22694 insns themselves, which are properly handled already).
22696 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22699 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22701 * combine.c (record_used_regs): New static function.
22702 (try_combine): Handle situations where there is an additional
22703 instruction between I2 and I3 which needs to have a LOG_LINK
22707 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22709 * combine.c (try_combine): Delete redundant i1 test. Call
22710 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22712 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22715 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22716 alternatives 13/14.
22718 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22720 PR tree-optimization/80054
22721 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22722 the optimization if a PHI or any of its arguments is not dominated
22723 by the candidate's basis. Use gphi* rather than gimple* as
22725 (replace_profitable_candidates): Clean up a gimple* variable that
22726 should be a gphi* variable.
22728 2017-03-20 Martin Sebor <msebor@redhat.com>
22731 * doc/extend.texi (attribute constructor): Document present limitation.
22733 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22736 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22737 __POWER9_VECTOR__ #ifdef control, change template definition to
22738 use Power9-specific built-in function.
22739 (vec_any_eq): Likewise.
22740 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22741 to control outcomes from this test.
22742 (vector_ae_<mode>p): For VEC_F modes, likewise.
22744 2017-03-20 Ian Lance Taylor <iant@google.com>
22746 * config/i386/i386.c (ix86_function_regparm): Save an extra
22747 register for -fsplit-stack with DECL_STATIC_CHAIN.
22749 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22752 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22753 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22755 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22757 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22759 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22762 2017-03-20 Marek Polacek <polacek@redhat.com>
22765 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22767 2017-03-20 Richard Biener <rguenther@suse.de>
22769 PR tree-optimization/80113
22770 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22771 allocate extra SSA name for PHI def.
22772 (add_close_phis_to_outer_loops): Likewise.
22773 (add_close_phis_to_merge_points): Likewise.
22774 (copy_loop_close_phi_args): Likewise.
22775 (copy_cond_phi_nodes): Likewise.
22777 2017-03-20 Martin Liska <mliska@suse.cz>
22779 PR middle-end/79753
22780 * tree-chkp.c (chkp_build_returned_bound): Do not build
22781 returned bounds for a LHS that's not a BOUNDED_P type.
22783 2017-03-20 Martin Liska <mliska@suse.cz>
22787 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22788 COMPLEX_CST and VECTOR_CST.
22790 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22793 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22794 target operand. A new splitter adds the clobber statement in case
22795 the target operand is dead anyway.
22797 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22799 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22800 to age-old versions of binutils and glibc.
22802 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22804 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22806 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22808 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22810 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22812 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22813 requirement for binutils 2.13.
22815 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22817 * combine.c (try_combine): Delete redundant i1 test. Call
22818 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22820 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22822 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22823 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22825 <riscv64-*-elf>: Re-arrange section
22826 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22827 <riscv32-*-linux>: Likewise.
22828 <riscv64-*-elf>: Likewise
22829 <riscv64-*-linux>: Likewise.
22831 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22834 * aarch64.opt(verbose-cost-dump): Fix typo.
22836 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22839 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22840 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22842 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22844 * reload.c (find_reloads): When reloading a nonoffsettable address,
22845 use RELOAD_OTHER for it and its address reloads.
22847 PR rtl-optimization/79910
22848 * combine.c (record_used_regs): New static function.
22849 (try_combine): Handle situations where there is an additional
22850 instruction between I2 and I3 which needs to have a LOG_LINK
22853 2017-03-17 Jeff Law <law@redhat.com>
22855 PR tree-optimization/71437
22856 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22857 conditional in the hash table first.
22858 (vrp_dom_walker::before_dom_children): Extract condition from
22859 ASSERT_EXPR. Record condition, its inverion and any implied
22860 conditions as well.
22862 2017-03-17 Marek Polacek <polacek@redhat.com>
22863 Markus Trippelsdorf <markus@trippelsdorf.de>
22865 PR tree-optimization/80079
22866 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22869 2017-03-17 Richard Biener <rguenther@suse.de>
22871 PR middle-end/80075
22872 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22873 Properly verify the LHS before the RHS possibly claims to be
22875 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22878 2017-03-17 Martin Jambor <mjambor@suse.cz>
22880 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22881 (List of -O2 options): Likewise.
22882 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22885 2017-03-17 Tom de Vries <tom@codesourcery.com>
22887 * gcov-dump.c (print_usage): Print bug_report_url.
22889 2017-03-17 Richard Biener <rguenther@suse.de>
22891 PR middle-end/80050
22892 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22893 (parser::peek): Likewise.
22895 2017-03-17 Richard Biener <rguenther@suse.de>
22897 PR tree-optimization/80048
22898 * sese.c (free_sese_info): Properly release rename_map and
22899 copied_bb_map elements.
22901 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22903 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22904 Add linked-list forward and backlinks. Insert on
22905 construction, remove on destruction.
22906 (class pass_store_merging): Add m_stores_head field.
22907 (pass_store_merging::terminate_and_process_all_chains):
22908 Iterate over m_stores_head list.
22909 (pass_store_merging::terminate_all_aliasing_chains):
22911 (pass_store_merging::execute): Check for debug stmts first.
22912 Push new chains onto the m_stores_head stack.
22914 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22917 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22918 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22919 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22921 2017-03-16 Jeff Law <law@redhat.com>
22923 PR tree-optimization/71437
22924 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22925 member function. Implementation moved into after_dom_children
22926 member function and into the threader's thread_outgoing_edges
22928 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22929 some code into new thread_outgoing_edges.
22930 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22931 definition. Simplify marker handling (do it here). Assume we always
22932 have the available expression and the const/copies tables.
22933 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22935 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22936 * tree-vrp.c (equiv_stack): No longer file scoped.
22937 (vrp_dom_walker): New class.
22938 (vrp_dom_walker::before_dom_children): New member function.
22939 (vrp_dom_walker::after_dom_children): Likewise.
22940 (identify_jump_threads): Setup domwalker. Use it rather than
22941 walking edges in a random order by hand. Simplify setup/finalization.
22942 (finalize_jump_threads): Remove.
22943 (vrp_finalize): Do not call identify_jump_threads here.
22944 (execute_vrp): Do it here instead and call thread_through_all_blocks
22947 PR tree-optimization/71437
22948 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22950 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22952 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22953 (dom_opt_dom_walker::thread_across_edge): Remove
22954 handle_dominating_asserts argument. All callers changed.
22955 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22956 changes. Remove calls to lhs_of_dominating_assert. Other
22957 uses of handle_dominating_asserts turn into unconditional code
22958 (simplify_control_stmt_condition_1): Likewise.
22959 (simplify_control_stmt_condition): Likewise.
22960 (thread_through_normal_block, thread_across_edge): Likewise.
22961 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22962 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22963 object if it is not an SSA_NAME.
22964 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22965 before calling into the VRP specific simplifiers.
22966 (identify_jump_threads): Remove handle_dominating_asserts
22969 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22972 * tree-diagnostic.c (default_tree_printer): No longer static.
22973 * tree-diagnostic.h (default_tree_printer): New prototype.
22975 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22977 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22978 Change ins into fmov.
22980 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22982 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22983 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22984 Use h_con constraint for operand 1.
22985 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22986 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22988 2017-03-15 Jeff Law <law@redhat.com>
22990 PR tree-optimization/71437
22991 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22992 (record_temporary_equivalences): Use it.
22994 PR tree-optimization/71437
22995 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22996 tree-ssa-scopedtables.
22997 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22998 (record_conditions, record_cond, vuse_eq): Likewise.
22999 (record_edge_info): Adjust to API tweak of record_conditions.
23000 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23001 (record_temporary_equivalences, optimize_stmt): Likewise.
23002 (eliminate_redundant_computations): Likewise.
23003 (record_equivalences_from_stmt): Likewise.
23004 * tree-ssa-scopedtables.c: Include options.h and params.h.
23005 (vuse_eq): New function, moved from tree-ssa-dom.c
23006 (build_and_record_new_cond): Likewise.
23007 (record_conditions): Likewise. Accept vector of conditions rather
23008 than edge_equivalence structure for first argument.
23009 for the first argument.
23010 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23011 from tree-ssa-dom.c.
23012 (avail_exprs_stack::record_cond): Likewise.
23013 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23014 from tree-ssa-dom.c.
23015 (avail_exprs_stack): Add new member functions lookup_avail_expr
23017 (record_conditions): Declare.
23019 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23022 * lra-constraints.c (process_alt_operands): Increase reject for
23023 reloading an input/output operand.
23025 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23028 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23029 insns to convert from signed/unsigned char/short to IEEE 128-bit
23031 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23033 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23036 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23037 subreg of inner mode for values already in registers.
23039 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23041 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23042 iteration reg is used after the loop.
23044 2017-03-14 Martin Sebor <msebor@redhat.com>
23046 PR tree-optimization/79800
23047 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23048 precision in negative-positive range.
23049 (format_floating): Call non-const overload with adjusted precision.
23051 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23054 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23057 2017-03-14 Martin Sebor <msebor@redhat.com>
23059 PR middle-end/80020
23060 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23061 * builtins.def (aligned_alloc): Use it.
23064 * Makefile.in (GTFILES): Add calls.c.
23065 * calls.c: Include "gt-calls.h".
23067 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23069 PR rtl-optimization/79728
23070 * regs.h (struct target_regs): New field
23071 x_contains_allocatable_regs_of_mode.
23072 (contains_allocatable_regs_of_mode): New macro.
23073 * reginfo.c (init_reg_sets_1): Initialize it, and change
23074 contains_reg_of_mode so it includes global regs as well.
23075 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23076 rather than contains_regs_of_mode.
23078 2017-03-14 Martin Liska <mliska@suse.cz>
23080 * doc/invoke.texi: Document options that can't be combined with
23081 -fcheck-pointer-bounds.
23083 2017-03-14 Martin Liska <mliska@suse.cz>
23085 PR middle-end/79831
23086 * doc/invoke.texi (-Wchkp): Document the option.
23088 2017-03-14 Martin Liska <mliska@suse.cz>
23090 * Makefile.in: Install gcov-dump.
23092 2017-03-14 Martin Liska <mliska@suse.cz>
23094 * multiple_target.c (expand_target_clones): Bail out for
23095 an invalid attribute.
23097 2017-03-14 Richard Biener <rguenther@suse.de>
23099 * alias.c (struct alias_set_entry): Pack properly.
23100 * cfgloop.h (struct loop): Likewise.
23101 * cse.c (struct set): Likewise.
23102 * ipa-utils.c (struct searchc_env): Likewise.
23103 * loop-invariant.c (struct invariant): Likewise.
23104 * lra-remat.c (struct cand): Likewise.
23105 * recog.c (struct change_t): Likewise.
23106 * rtl.h (struct address_info): Likewise.
23107 * symbol-summary.h (function_summary): Likewise.
23108 * tree-loop-distribution.c (struct partition): Likewise.
23109 * tree-object-size.c (struct object_size_info): Likewise.
23110 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23111 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23112 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23113 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23114 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23115 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23116 (struct _stmt_vec_info): Likewise.
23118 2017-03-14 Martin Liska <mliska@suse.cz>
23121 * multiple_target.c (create_dispatcher_calls): Check that
23122 a target can create a function dispatcher.
23124 2017-03-14 Martin Liska <mliska@suse.cz>
23127 * multiple_target.c (expand_target_clones): Drop local.local
23128 flag for default implementation.
23130 2017-03-14 Richard Biener <rguenther@suse.de>
23132 PR tree-optimization/80030
23133 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23135 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23137 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23138 gcc_fallthrough() instead of __attribute__((fallthrough));
23140 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23142 * doc/gcc.texi: Remove "up" link to (DIR).
23143 * doc/gccint.texi: Ditto.
23145 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23147 * doc/install.texi (Specific) <avr>: Remove reference to
23150 2017-03-13 Jeff Law <law@redhat.com>
23152 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23153 attribute rather than comments.
23155 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23156 match_scratch operand is highest.
23158 2017-03-13 Martin Liska <mliska@suse.cz>
23160 PR middle-end/78339
23161 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23162 is a CHKP clone, use original declaration.
23164 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23166 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23167 (arc_conditional_register_usage): Use a different allocation order
23168 when optimizing for size.
23169 * common/config/arc/arc-common.c (arc_option_optimization_table):
23170 Section anchors default on when optimizing for size.
23172 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23174 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23176 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23178 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23179 * config/arc/arc.md (cpu_facility): Add cd variant.
23180 (*movqi_insn): Add code density variant.
23181 (*movhi_insn): Likewise.
23182 (*movqi_insn): Likewise.
23183 (*addsi3_mixed): Likewise.
23184 (subsi3_insn): Likewise.
23186 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23188 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23190 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23192 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23193 expressions with MINUS and UNARY ops.
23195 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23198 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23200 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23201 between vec_select and vector argument.
23202 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23203 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23204 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23205 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23206 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23207 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23209 2017-03-13 Richard Biener <rguenther@suse.de>
23212 * params.def (vect-max-peeling-for-alignment): Fix typo.
23214 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23216 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23217 issue that only occurred with binutils below 2.18.
23219 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23221 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23222 refer to binutils 2.11/2.12 minimum.
23224 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23226 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23227 ftp.kernel.org and simplify binutils requirement.
23229 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23231 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23233 (Optimize Options): Ditto. Also remove redundancy.
23235 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23237 PR translation/79848
23238 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23240 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23241 to G_ to avoid double translation.
23243 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23245 PR translation/79923
23246 * auto-profile.c (get_combined_location): Convert leading
23247 character of diagnostics to lower case and remove trailing period.
23248 (read_profile): Likewise for various diagnostics.
23249 * config/arm/arm.c (arm_option_override): Remove trailing period
23250 from various diagnostics.
23251 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23252 (msp430_expand_delay_cycles): Likewise.
23254 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23257 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23258 full command-line argument, rather than just "str".
23259 (aarch64_validate_march): Likewise.
23260 (aarch64_validate_mtune): Likewise.
23262 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23264 PR rtl-optimization/78911
23265 * lra-assigns.c (must_not_spill_p): New function.
23266 (spill_for): Use it.
23268 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23270 PR tree-optimization/79981
23271 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23272 ATOMIC_COMPARE_EXCHANGE ifn result.
23273 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23274 IFN_ATOMIC_COMPARE_EXCHANGE.
23276 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23279 * opts.c (parse_sanitizer_options): Add missing question mark to
23280 "did you mean" message.
23282 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23284 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23286 (VMULEUH_UNS): Likewise.
23287 (VMULOUB_UNS): Likewise.
23288 (VMULOUH_UNS): Likewise.
23289 * config/rs6000/rs6000.c (builtin_function_type): Remove
23290 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23292 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23295 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23296 x with result of extra_parsing_for_operand_code_0.
23297 (function_reader::extra_parsing_for_operand_code_0): Convert
23298 return type from void to rtx, returning x. When reading
23299 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23300 larger size containing struct block_symbol.
23302 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23304 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23305 -mfloat128-hardware without -m64.
23307 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23310 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23311 entries to the case statement that marks unsigned arguments to
23312 overloaded functions.
23314 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23316 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23317 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23319 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23322 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23323 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23325 2017-03-10 Martin Liska <mliska@suse.cz>
23329 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23330 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23333 2017-03-10 Olivier Hainque <hainque@adacore.com>
23335 * tree-switch-conversion (array_value_type): Start by resetting
23336 candidate type to it's main variant.
23338 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23340 PR rtl-optimization/79909
23341 * combine.c (try_combine): Use simplify_replace_rtx on individual
23342 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23343 of the whole CALL_INSN_FUNCTION_USAGE.
23345 PR tree-optimization/79972
23346 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23347 get_range_info on SSA_NAMEs. Formatting fixes.
23349 2017-03-10 Richard Biener <rguenther@suse.de>
23350 Jakub Jelinek <jakub@redhat.com>
23352 PR tree-optimization/77975
23353 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23354 edge to be constant.
23355 (get_val_for): For constant x return it. Formatting fix.
23356 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23357 would use the same bases as the current one.
23359 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23361 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23362 instead of vec_select for V1TImode.
23363 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23365 (VSX_LE_128): Add V1TI to this mode iterator.
23366 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23367 (*vsx_le_perm_store_<mode>): Likewise.
23368 (pre-reload splitter for VSX stores): Likewise.
23369 (post-reload splitter for VSX stores): Likewise.
23370 (*vsx_xxpermdi2_le_<mode>): Likewise.
23371 (*vsx_lxvd2x2_le_<mode>): Likewise.
23372 (*vsx_stxvd2x2_le_<mode>): Likewise.
23374 2017-03-09 Michael Eager <eager@eagercon.com>
23376 Correct failures with --enable-checking=yes,rtl.
23378 * config/microblaze/microblaze.c (microblaze_expand_shift):
23379 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23380 test for const0_rtx.
23381 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23382 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23384 2017-03-09 Richard Biener <rguenther@suse.de>
23386 PR tree-optimization/79977
23387 * graphite-scop-detection.c (scop_detection::merge_sese):
23388 Handle the case of extra exits to blocks dominating the entry.
23390 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23392 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23395 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23397 PR rtl-optimization/79949
23398 * lra-constraints.c (process_alt_operands): Check memory when
23399 trying to predict a cycle. Print about the overall increase.
23401 2017-03-09 Richard Biener <rguenther@suse.de>
23403 PR middle-end/79971
23404 * gimple-expr.c (useless_type_conversion_p): Preserve
23405 TYPE_SATURATING for fixed-point types.
23407 2017-03-09 Richard Biener <rguenther@suse.de>
23410 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23411 alignment of BLKmode params.
23413 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23416 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23417 (VALL_NO_V2Q): Likewise.
23419 * config/aarch64/aarch64-simd.md
23420 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23422 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23423 VALL_NO_V2Q mode iterator.
23424 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23426 2017-03-09 Martin Liska <mliska@suse.cz>
23428 PR tree-optimization/79631
23429 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23430 tree_int_cst_sign_bit just for INTEGER constants.
23432 2017-03-09 Martin Liska <mliska@suse.cz>
23436 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23439 2017-03-09 Marek Polacek <polacek@redhat.com>
23442 * tree.c (inchash::add_expr): Handle TREE_VEC.
23444 2017-03-09 Martin Liska <mliska@suse.cz>
23447 (chkp_narrow_size_and_offset): New function.
23448 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23449 (void chkp_parse_bit_field_ref): New function.
23450 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23451 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23453 2017-03-09 Martin Liska <mliska@suse.cz>
23456 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23457 (chkp_find_bounds_1): Remove gcc_unreachable.
23459 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23462 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23463 BUILT_IN_SYNC*, determine the access type from the size suffix and
23464 always build a MEM_REF with that type. Handle forgotten
23465 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23468 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23469 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23470 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23471 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23472 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23473 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23474 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23475 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23476 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23477 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23478 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23479 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23480 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23481 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23482 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23483 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23484 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23485 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23486 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23487 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23488 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23489 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23490 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23491 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23492 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23493 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23494 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23495 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23496 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23497 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23498 definitions outside of __OPTIMIZE__ guarded section.
23501 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23502 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23503 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23504 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23507 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23509 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23510 ("vfenez<mode>"): Add missing constraints.
23512 2017-03-08 Martin Sebor <msebor@redhat.com>
23515 * config/nds32/nds32.c (nds32_option_override):
23516 Fix misspelled diagnostic.
23518 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23521 * gimplify.c (gimplify_omp_for): Replace index var in outer
23522 taskloop statement with an artificial variable and add
23523 OMP_CLAUSE_PRIVATE clause for it.
23525 2017-03-08 Richard Biener <rguenther@suse.de>
23527 PR tree-optimization/79955
23528 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23529 for accesses that are completely outside of the variable.
23531 2017-03-08 Andrew Haley <aph@redhat.com>
23533 PR tree-optimization/79943
23534 * tree-ssa-loop-split.c (compute_new_first_bound): When
23535 calculating the new upper bound, (END-BEG) should be added, not
23538 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23540 * config/avr/avr.md (setmemhi): Make sure match_dup
23541 operand number comes before match_scratch.
23543 2017-03-08 Richard Biener <rguenther@suse.de>
23545 PR tree-optimization/79920
23546 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23547 with ncopies == 1 to ...
23548 (vect_transform_slp_perm_load): ... here. Properly compute
23549 all element loads by iterating VF times over the group. Do
23550 not handle ncopies (computed in a broken way) in
23551 vect_create_mask_and_perm.
23553 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23556 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23557 is a uniform vector, use uniform_vector_p return value instead of
23558 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23560 2017-03-07 Marek Polacek <polacek@redhat.com>
23562 PR middle-end/79809
23563 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23564 (alloca_call_type): Likewise.
23566 2017-03-07 Martin Liska <mliska@suse.cz>
23568 * gcov.c (process_args): Put comment to correct location.
23570 2017-03-07 Martin Liska <mliska@suse.cz>
23572 PR middle-end/68270
23573 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23574 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23575 (chkp_narrow_bounds_for_field): Likewise.
23576 (chkp_parse_array_and_component_ref): Pass one more argument to
23579 2017-03-07 Richard Biener <rguenther@suse.de>
23581 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23584 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23586 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23587 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23589 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23592 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23593 to end of description.
23594 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23596 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23598 PR rtl-optimization/79901
23599 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23601 (*avx512f_<code><mode>3<mask_name>): ... this.
23602 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23603 iterator instead of VI8_AVX2_AVX512BW.
23605 PR rtl-optimization/79901
23606 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23607 min/max expander, expand it using expand_vec_cond_expr.
23610 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23613 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23616 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23617 to void * for PCH reasons.
23618 * dwarf2out.c (output_loc_operands, output_die): Cast
23619 v.val_vec.array to unsigned char *.
23621 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23624 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23627 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23629 PR rtl-optimization/79571
23630 * lra-constraints.c (process_alt_operands): Calculate static
23631 reject and subtract it from overall when only addresses will be
23634 2017-03-06 Julia Koval <julia.koval@intel.com>
23637 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23638 incoming stack boundary to 128 for 64-bit targets.
23640 2017-03-06 Richard Biener <rguenther@suse.de>
23642 PR tree-optimization/79894
23643 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23644 to NULL after folding it.
23646 2017-03-06 Richard Biener <rguenther@suse.de>
23648 PR tree-optimization/79824
23649 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23650 check disabling peeling for gaps.
23652 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23654 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23655 attributes): Document gettimeofday.
23657 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23659 * config/s390/s390.c (s390_option_override_internal): Set
23660 PARAM_MIN_VECT_LOOP_BOUND
23662 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23664 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23665 * config/s390/s390.md: Likewise.
23667 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23670 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23671 (<avx2_avx512>_perm<mode>): Rename to ...
23672 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23674 (<avx512>_perm<mode>_mask): Rename to ...
23675 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23677 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23678 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23679 instead of VI8F_256_512.
23680 (avx512f_perm<mode>): New define_expand.
23681 (avx512f_perm<mode>_mask): Likewise.
23682 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23683 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23685 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23687 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23688 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23690 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23692 2017-03-06 Martin Liska <mliska@suse.cz>
23695 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23696 when having a SSA NAME w/o VAR_DECL assigned to it.
23698 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23700 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23701 msa_dpsub_<su>_d): Fix MODE for vec_select.
23703 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23705 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23707 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23709 2017-03-06 Richard Biener <rguenther@suse.de>
23711 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23712 * plugin.c (register_plugin_info): Likewise.
23713 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23715 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23717 * config/i386/sse.md (sse_storehps, sse_storelps,
23718 avx_<castmode><avxsizesuffix>_<castmode>,
23719 avx512f_<castmode><avxsizesuffix>_<castmode>,
23720 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23721 in condition that at least one operand is not a MEM.
23723 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23725 PR middle-end/79805
23726 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23727 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23729 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23730 gimple_call_nothrow_p flag based on whether original builtin can throw.
23731 If it can, emit following stmts on the fallthrough edge.
23732 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23733 don't create new bb if inserting just debug stmts on the edge, try to
23734 insert them on the fallthru bb or just reset debug stmts.
23736 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23739 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23740 restore recog_data (including the operand rtxes inside it) around
23741 the call to get_insn_template.
23743 2017-03-03 Martin Sebor <msebor@redhat.com>
23745 PR tree-optimization/79699
23746 * context.c (context::~context): Free MPFR caches to avoid
23747 a memory leak on program exit.
23749 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23751 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23752 Use wide_int::ulow () instead of .elt (0).
23754 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23756 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23757 (*pushxf): Limit oF constraint to 32bit targets and add oC
23758 constraint for 64bit targets.
23759 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23760 (*pushdf): Change rmF constraint to rmC.
23762 2017-03-03 Martin Liska <mliska@suse.cz>
23764 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23765 Remove unused variable.
23767 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23770 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23771 is a memory operand, increase num_memory.
23772 (ix86_expand_args_builtin): Likewise.
23774 2017-03-03 Jan Hubicka <jh@suse.cz>
23777 * ipa-devirt.c (maybe_record_node): Properly handle
23778 __cxa_pure_virtual visibility.
23780 2017-03-03 Martin Liska <mliska@suse.cz>
23782 PR tree-optimization/79803
23783 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23785 (pass_loop_prefetch::execute): Disabled optimization if an
23786 assumption about L1 cache size is not met.
23788 2017-03-03 Martin Liska <mliska@suse.cz>
23790 PR rtl-optimization/79574
23791 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23792 (hash_scan_set): Likewise.
23793 (dump_hash_table): Likewise.
23794 (hoist_code): Likewise.
23796 2017-03-03 Richard Biener <rguenther@suse.de>
23798 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23799 in place of uhigh ().
23800 (fixed_convert_from_real): Likewise.
23802 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23805 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23807 2017-03-03 Richard Biener <rguenther@suse.de>
23809 PR middle-end/79818
23810 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23811 TYPE_OVERFLOW_UNDEFINED check.
23813 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23815 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23817 (vector_ae_<mode>_p): Likewise.
23818 (vector_nez_<mode>_p): Likewise.
23819 (vector_ne_v2di_p): Likewise.
23820 (vector_ae_v2di_p): Likewise.
23821 (vector_ne_<mode>_p): Likewise.
23822 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23824 (vsx_tsqrt<mode>2_fe): Likewise.
23826 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23829 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23831 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23833 PR rtl-optimization/79780
23834 * cprop.c (one_cprop_pass): When second and further conditional trap
23835 in a single basic block is turned into an unconditional trap, turn it
23836 into a deleted note to avoid RTL verification failures.
23838 2017-03-02 Richard Biener <rguenther@suse.de>
23840 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23842 2017-03-02 Richard Biener <rguenther@suse.de>
23844 PR tree-optimization/79345
23846 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23847 param and abort the walk, returning -1 if it is hit.
23848 (walk_aliased_vdefs): Take a limit param and pass it on.
23849 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23850 defaulting to 0 and return a signed int.
23851 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23852 (check_defs): New helper.
23853 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23854 about uninitialized memory.
23855 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23856 bogus uninitialized warning.
23857 (fixed_convert_from_real): Likewise.
23859 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23861 PR tree-optimization/66768
23862 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23863 iv_use if base object can't be determined.
23865 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23867 PR tree-optimization/79345
23868 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23869 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23870 (get_pattern_stats): Initialize it.
23871 * genemit.c (gen_expand): Verify match_scratch numbers come after
23872 match_operand/match_dup numbers.
23873 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23874 match_scratch numbers.
23875 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23877 * config/s390/s390.md (trunctdsd2): Likewise.
23879 2017-03-02 Richard Biener <rguenther@suse.de>
23881 * wide-int.h (wide_int_storage::operator=): Implement in terms
23884 2017-03-02 Richard Biener <rguenther@suse.de>
23886 PR tree-optimization/79777
23887 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23888 the to insert expression to sth existing.
23890 2017-03-01 Martin Sebor <msebor@redhat.com>
23892 PR middle-end/79692
23893 * gimple-ssa-sprintf.c
23894 (directive::known_width_and_precision): New function.
23895 (format_integer): Use it.
23896 (get_mpfr_format_length): Consider the full range of precision
23897 when computing %g output with the # flag. Set the likely byte
23898 count to 3 rather than 1 when precision is indeterminate.
23899 (format_floating): Correct the lower bound of precision.
23901 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23903 * doc/invoke.texi: Document default code model for 64-bit Linux.
23905 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23908 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23909 udiv rather than div since input pattern is unsigned.
23911 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23913 * config/i386/i386.c (print_reg): Warn for values of
23914 unsupported size in integer register.
23916 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23919 * config/rs6000/predicates.md (current_file_function_operand): Do
23920 not allow self calls to be local if the function is replaceable.
23922 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23925 * config/rs6000/altivec.h (vec_ctz and others): Change the
23926 preprocessor macro that controls conditional compilation from
23927 _ARCH_PWR9 to __POWER9_VECTOR__.
23928 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23929 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23930 control (instead of _ARCH_PWR9 control) so that template
23931 definition uses power9-specific function.
23932 (vec_any_eq): Likewise.
23933 (vec_all_ne): Change macro definition to use a power9-specific
23934 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23935 _ARCH_PWR9 control).
23936 (vec_any_eq) Likewise.
23937 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23938 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23939 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23940 support for xvcmpnedp instruction.
23941 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23942 macro expansion so that Power9 implementation of vec_all_ne does
23943 not use the AltiVec predicate framework.
23944 (VCMPNEH_P): Likewise.
23945 (VCMPNEW_P): Likewise.
23946 (VCMPNED_P): Likewise.
23947 (VCMPNEFP_P): Likewise.
23948 (VCMPNEDP_P): Likewise.
23949 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23950 implementation of vec_any_eq to not use AltiVec predicate
23952 (VCMPAEH_P): Likewise.
23953 (VCMPAEW_P): Likewise.
23954 (VCMPAED_P): Likewise.
23955 (VCMPAEFP_P): Likewise.
23956 (VCMPAEDP_P): Likewise.
23957 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23958 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23959 not use the AltiVec predicate framework.
23960 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23961 of vec_any_eq to not use AltiVec predicate framework.
23962 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23963 support for predefined __POWER9_VECTOR__ macro to indicate that
23964 Power9 instruction selection is enabled.
23965 (altivec_overloaded_builtins): Remove extraneous
23966 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23967 function argument types RS6000_BTI_bool_V16QI and
23968 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23969 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23970 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23971 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23972 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23973 Power9 for implementations of vec_cmpne. Change the signature for
23974 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23975 (representing vec_all_ne) to remove the previously described first
23976 argument of type RS6000_BTI_INTSI, as this was an artifact of
23977 reliance on the AltiVec predicate framework, which is no longer
23978 used in the implementation of these functions. Add
23979 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23980 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23981 since, unlike the AltiVec predicate framework implementation, we
23982 do not share function descriptors between vec_alle and vec_anyeq.
23983 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23984 set of modes that receive special treatment even when
23985 TARGET_P9_VECTOR is true. The special treatment emits code that
23986 does not depend on Power9 instructions.
23987 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23988 define_expand to not rely on AltiVec predicate framework.
23989 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23991 (vector_ne_v2di_p): Change this define_expand to not rely on
23992 AltiVec predicate framework.
23993 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23995 (vector_ne_<mode>_p): Change this define_expand to not rely on
23996 AltiVec predicate framework.
23997 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23999 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24000 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24001 define_insn pattern.
24002 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24003 define_insn pattern because the xvcmpne<VSs>. instruction is not
24005 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24006 instruction is not supported.
24008 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24010 * config/nvptx/nvptx.c: Include intl.h.
24012 2017-03-01 Martin Jambor <mjambor@suse.cz>
24015 * ipa-prop.h (ipa_bits): Removed field known.
24016 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24017 to pointers. Adjusted their comments to warn about their sharing.
24018 (ipcp_transformation_summary): Change bits to a vector of pointers.
24019 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24020 (ipa_get_ipa_bits_for_value): Declare.
24021 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24022 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24023 (ipa_bits_hash_table): Likewise.
24024 (ipa_vr_ggc_hash_traits): Likewise.
24025 (ipa_vr_hash_table): Likewise.
24026 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24027 being pointers and vr_known being removed.
24028 (ipa_set_jf_unknown): Likewise.
24029 (ipa_get_ipa_bits_for_value): New function.
24030 (ipa_set_jfunc_bits): Likewise.
24031 (ipa_get_value_range): New overloaded functions.
24032 (ipa_set_jfunc_vr): Likewise.
24033 (ipa_compute_jump_functions_for_edge): Use the above functions to
24034 construct bits and vr parts of jump functions.
24035 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24036 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24038 (ipcp_grow_transformations_if_necessary): Also allocate
24039 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24041 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24042 them. Fix too long lines.
24043 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24044 vr_known being removed.
24045 (ipa_read_jump_function): Use new setter functions to construct bits
24046 and vr parts of jump functions or set them to NULL.
24047 (write_ipcp_transformation_info): Adjust for bits being pointers.
24048 (read_ipcp_transformation_info): Likewise.
24049 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24051 Include gt-ipa-prop.h.
24052 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24054 (ipcp_store_bits_results): Likewise.
24055 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24056 Do not write to existing jump functions but use a temporary instead.
24058 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24061 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24062 attempt to use its first operand as BIT_FIELD_REF base.
24064 2017-03-01 Richard Biener <rguenther@suse.de>
24066 PR middle-end/79721
24067 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24068 interpolating formula in wrapping arithmetic.
24069 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24071 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24073 PR tree-optimization/79734
24074 * tree-vect-generic.c (expand_vector_condition): Optimize
24075 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24076 Handle VEC_COND_EXPR where comparison has different inner width from
24077 type's inner width.
24079 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24081 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24082 markup, and similar issues. Remove @opindex entries for things
24083 that aren't options. Add missing -mmpy-option entries.
24085 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24087 PR tree-optimization/79737
24088 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24089 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24090 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24091 instead of byte_size. Formatting fix.
24092 (shift_bytes_in_array_right): Formatting fix.
24094 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24097 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24098 condition on optimize for the leaf function test.
24100 2017-02-28 Martin Liska <mliska@suse.cz>
24103 * read-rtl-function.c (function_reader::handle_unknown_directive):
24104 Bail out when one uses -flto.
24106 2017-02-28 Martin Liska <mliska@suse.cz>
24108 * common.opt: Replace space with tabular for options of <number>
24110 * config/i386/i386.opt: Show <number> value for
24111 -mlarge-data-threshold.
24112 * opts.c (print_filtered_help): Do not display number in hexadecimal
24115 2017-02-28 Martin Liska <mliska@suse.cz>
24117 * common.opt: Fix --help=option -Q for options which are of
24120 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24122 * config/i386/i386.c (print_reg): Error out for values
24123 of 8-bit size in invalid integer register.
24125 2017-02-28 Martin Sebor <msebor@redhat.com>
24127 PR tree-optimization/79691
24128 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24130 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24133 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24134 gcc_unreachable with output_operand_lossage.
24136 2017-02-28 Richard Biener <rguenther@suse.de>
24138 PR tree-optimization/79740
24139 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24141 (visit_nary_op): Insert the nary into the hashtable if we
24142 pattern-matched sth.
24143 * tree-ssa-pre.c (eliminate_insert): Robustify.
24145 2017-02-28 Richard Biener <rguenther@suse.de>
24147 PR middle-end/79731
24148 * fold-const.c (decode_field_reference): Reject out-of-bound
24151 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24153 * config/i386/i386.c: Include intl.h.
24154 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24155 instead of just cond ? "..." : "...".
24156 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24157 * coverage.c (read_counts_file): Likewise.
24158 * omp-offload.c: Include intl.h.
24159 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24160 of just cond ? "..." : "...".
24161 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24162 of just cond ? "..." : "...".
24164 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24167 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24169 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24170 'tune for' CPU name.
24171 * config/arm/arm-cpu-data.h: Regenerated.
24173 2017-02-28 Richard Biener <rguenther@suse.de>
24175 PR tree-optimization/79732
24176 * tree-inline.c (expand_call_inline): Do not shadow var.
24178 2017-02-28 Richard Biener <rguenther@suse.de>
24180 PR tree-optimization/79723
24181 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24182 address-space properly.
24184 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24186 * doc/optinfo.texi (Optimization groups): Fix option used for
24188 * doc/invoke.texi (-fopt-info): Document "omp".
24189 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24190 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24191 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24192 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24193 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24195 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24197 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24198 OPTGROUP_OMP with "omp".
24200 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24203 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24204 for arithmetic shift of unsigned V2DI.
24206 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24208 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24209 arc/linux.h headers.
24210 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24211 (LINK_SPEC): Likewise.
24212 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24213 (EXTRA_SPECS): Likewise.
24214 (STARTFILE_SPEC): Likewise.
24215 (ENDFILE_SPEC): Likewise.
24216 (LIB_SPEC): Likewise.
24217 (TARGET_SDATA_DEFAULT): Likewise.
24218 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24219 (MULTILIB_DEFAULTS): Likewise.
24220 (DWARF2_UNWIND_INFO): Likewise.
24221 * config/arc/big.h: New file.
24222 * config/arc/elf.h: Likewise.
24223 * config/arc/linux.h: Likewise.
24224 * config/arc/t-uClibc: Remove.
24226 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24228 PR tree-optimization/77536
24229 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24230 (tree_transform_and_unroll_loop): Use above function to compute the
24231 estimated niter of unrolled loop and use it when scaling profile.
24232 Also use count info rather than frequency if it's non-zero.
24233 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24234 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24235 (vect_transform_loop): Call above function.
24237 2017-02-27 Richard Biener <rguenther@suse.de>
24239 PR tree-optimization/45397
24240 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24241 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24242 (visit_nary_op): Add pattern matching for CSEing sign-changed
24243 or truncated operations with wider ones.
24245 2017-02-27 Richard Biener <rguenther@suse.de>
24247 PR tree-optimization/79690
24248 * tree-vect-stmts.c (vectorizable_store): Use vector type
24249 built from the DR with address-space.
24251 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24253 * doc/invoke.texi (Optimize Options): Refine the description
24254 of asan-use-after-return.
24256 2017-02-25 Alan Modra <amodra@gmail.com>
24258 PR rtl-optimization/79584
24259 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24260 base, not ad->base_term, the reg within base. Remove assertion
24261 that ad->base == ad->base_term. Replace gen_int_mode using
24262 bogus mode with const0_rtx.
24264 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24266 PR middle-end/79396
24267 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24268 FMA_EXPR like tcc_binary or tcc_unary.
24270 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24273 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24275 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24276 (output_loc_operands): Handle DW_OP_call_ref and
24277 DW_OP_GNU_variable_value.
24278 (struct variable_value_struct): New type.
24279 (struct variable_value_hasher): Likewise.
24280 (variable_value_hash): New variable.
24281 (string_types): Remove.
24282 (copy_loc_descr): New function.
24283 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24284 (prepend_loc_descr_to_each): New function.
24285 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24286 instead of add_loc_descr_to_each if the first argument is single
24287 location list and the second has multiple.
24288 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24289 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24290 when looking for variable value which doesn't have other location info.
24291 (loc_list_from_tree): Formatting fix.
24292 (gen_array_type_die): Simplify DW_AT_string_length handling.
24293 (adjust_string_types): Remove.
24294 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24295 string_types. Call resolve_variable_values.
24296 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24297 (resolve_addr_in_expr): Likewise. Add A argument.
24298 (copy_deref_exprloc): Remove deref argument. Adjust for the
24299 original expression being DW_OP_GNU_variable_value with optionally
24300 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24301 optionally after it.
24302 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24303 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24304 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24305 (variable_value_hasher::hash, variable_value_hasher::equal): New
24307 (resolve_variable_value_in_expr, resolve_variable_value,
24308 resolve_variable_values, note_variable_value_in_expr,
24309 note_variable_value): New functions.
24310 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24313 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24316 * opts.h (handle_generated_option): Add GENERATED_P argument.
24317 * opts-common.c (handle_option): Adjust function comment.
24318 (handle_generated_option): Add GENERATED_P argument, pass it to
24320 (control_warning_option): Pass false to handle_generated_option
24322 * opts.c (maybe_default_option): Pass true to handle_generated_option
24324 * optc-gen.awk: Likewise.
24326 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24328 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24329 a REG, look at the REG it is a SUBREG of.
24330 (splitter for cmpeqsi_t): Ditto.
24332 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24334 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24335 the special USEs with the pattern of the insn, not the insn itself.
24337 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24340 * doc/invoke.texi: Document -mload-store-pairs.
24342 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24343 Sandra Loosemore <sandra@codesourcery.com>
24345 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24346 argument isn't a CONST_INT.
24347 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24348 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24349 (nios2_validate_compare): Bypass alternate compare logic if *op2
24350 is not a CONST_INT.
24351 (ldstwm_operation_p): Return false if first_base is not a REG or
24352 if first_offset is not a CONST_INT.
24354 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24356 * config/cris/cris.md: Use correct operand in a define_peephole2.
24358 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24360 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24362 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24364 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24365 this_insn if it is an INSN or JUMP_INSN.
24366 (force_offsettable): Look at base, not at addr.
24367 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24368 on things that aren't necessarily CONST_INTs.
24370 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24372 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24373 -mfpmath=sse is the default also for x86-32 targets with SSE2
24374 instruction set when @option{-ffast-math} is enabled
24376 2017-02-24 Jeff Law <law@redhat.com>
24378 PR rtl-optimizatoin/79286
24379 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24382 2017-02-24 Richard Biener <rguenther@suse.de>
24384 PR tree-optimization/79389
24385 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24388 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24390 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24391 function comment to reflect reality.
24392 (loop_exits_before_overflow): Fix typo in function description.
24394 2017-02-24 Richard Biener <rguenther@suse.de>
24396 PR tree-optimization/79389
24397 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24398 properly that a threading opportunity exists. Detect conditional
24399 copy/constant propagation opportunities.
24401 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24403 * config/visium/visium.md (type): Add trap.
24404 (b): New mode attribute.
24405 (*btst): Rename into...
24406 (*btst<mode>): ...this and adjust.
24407 (*cbranchsi4_btst_insn): Rename into...
24408 (*cbranch<mode>4_btst_insn): ...this and adjust.
24409 (trap): New define_insn.
24411 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24413 PR tree-optimization/79389
24414 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24415 (noce_reversed_cond_code): New function.
24416 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24417 reversed_comparison_code. Formatting fix.
24418 (noce_try_store_flag): Test rev_cond != NULL in addition to
24419 reversed_comparison_code.
24420 (noce_try_store_flag_constants): Likewise.
24421 (noce_try_store_flag_mask): Likewise.
24422 (noce_try_addcc): Use rev_cond if non-NULL instead of
24423 reversed_comparison_code.
24424 (noce_try_cmove_arith): Likewise. Formatting fixes.
24425 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24426 (noce_find_if_block): Initialize rev_cond.
24427 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24428 instead of false as last argument never attempt to reverse it
24431 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24433 PR tree-optimization/79663
24434 * tree-predcom.c (combine_chains): Process refs in reverse order
24435 only for ZERO length chains, and add explaining comment.
24437 2017-02-23 Jeff Law <law@redhat.com>
24439 PR tree-optimization/79578
24440 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24441 in call to operand_equal_p.
24443 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24446 * config/i386/cpuid.h: Fix another undefined behavior.
24448 2017-02-23 Richard Biener <rguenther@suse.de>
24450 PR tree-optimization/79683
24451 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24452 vector types for data-refs.
24454 2017-02-23 Martin Liska <mliska@suse.cz>
24456 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24458 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24460 PR middle-end/79665
24461 * internal-fn.c (get_range_pos_neg): Moved to ...
24462 * tree.c (get_range_pos_neg): ... here. No longer static.
24463 * tree.h (get_range_pos_neg): New prototype.
24464 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24465 are known to be in between 0 and signed maximum inclusive, try to
24466 expand both unsigned and signed divmod and use the cheaper one from
24469 2017-02-22 Jeff Law <law@redhat.com>
24471 PR tree-optimization/79578
24472 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24473 to compare base operands.
24475 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24478 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24479 gpc_reg_operand instead of fpr_reg_operand.
24481 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24483 * config/mips/mips.c (mips_return_in_memory): Force FP
24484 vector types to be returned in memory for o32 ABI.
24486 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24488 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24489 instead of DW_TAG_member for static data member declarations and don't
24490 set no_linkage_name for static inline data members.
24491 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24494 2017-02-22 Martin Liska <mliska@suse.cz>
24496 * doc/invoke.texi: Replace inequality signs with square brackets
24499 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24501 PR tree-optimization/68644
24502 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24504 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24507 * lra-constraints.c (simplify_operand_subreg): Handle
24508 WORD_REGISTER_OPERATIONS targets.
24510 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24513 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24514 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24515 elimination by swapping fld*.
24517 2017-02-22 Richard Biener <rguenther@suse.de>
24519 PR tree-optimization/79673
24520 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24521 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24522 irrelevant address-space qualifiers and avoiding a
24523 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24525 2017-02-22 Richard Biener <rguenther@suse.de>
24527 PR tree-optimization/79666
24528 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24529 to not symbolically negate if that may introduce undefined
24532 2017-02-22 Martin Liska <mliska@suse.cz>
24535 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24536 * data-streamer-out.c (streamer_write_gcov_count_stream):
24538 * value-prof.c (stream_out_histogram_value): Make assert more
24539 precise based on type of counter.
24541 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24544 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24545 Use nonimmediate_operand instead of memory_operand for operand 1.
24546 (float-extend standard_x87sse_constant_load splitter): Ditto.
24548 2017-02-21 Jeff Law <law@redhat.com>
24550 PR tree-optimization/79621
24551 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24552 blocks with edges to themselves.
24554 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24557 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24558 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24559 Use gimple_call_builtin_p.
24562 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24563 on temporarily removed DEBUG_INSNs.
24565 PR tree-optimization/79649
24566 * tree-loop-distribution.c (classify_partition): Give up on
24567 non-generic address space loads/stores.
24569 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24571 * doc/loop.texi (Loop manipulation): Remove nonexistent
24572 tree_ssa_loop_version from the documentation.
24573 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24575 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24578 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24579 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24580 * config/rs6000/rs6000.c: Include except.h.
24581 (rs6000_expand_split_stack_prologue): Call
24582 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24584 2017-02-21 Martin Jambor <mjambor@suse.cz>
24587 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24588 have been analyzed.
24590 2017-02-21 Martin Jambor <mjambor@suse.cz>
24592 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24593 for backward compatibility only.
24594 * doc/invoke.texi (Option Summary): Remove all references to
24595 -fipa-cp-alignment.
24597 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24601 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24603 * lra-constraints.c (curr_insn_transform): Handle
24604 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24606 2017-02-21 Martin Liska <mliska@suse.cz>
24608 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24610 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24612 PR translation/79638
24613 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24615 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24618 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24619 (arm_function_ok_for_sibcall): Return false for an indirect call by
24620 descriptor if all the argument registers are used.
24621 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24622 alignment of the function.
24624 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24626 PR tree-optimization/61441
24627 * simplify-rtx.c (simplify_const_unary_operation): For
24628 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24629 the sNaN unmodified.
24631 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24633 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24634 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24635 instead of SYSTEM_HEADER_DIR.
24637 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24638 Martin Liška <mliska@suse.cz>
24640 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24641 Fix typos and grammar, use active voice, and clarify.
24643 2017-02-20 Marek Polacek <polacek@redhat.com>
24645 PR middle-end/79537
24646 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24649 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24651 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24654 * config/i386/i386.c (ix86_expand_builtin): Handle
24655 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24656 ix86_builtins_isa[fcode].isa as a requirement of those
24657 flags and any other flag in the bitmask.
24658 (ix86_init_mmx_sse_builtins): Use 0 instead of
24659 ~OPTION_MASK_ISA_64BIT as mask.
24660 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24661 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24662 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24663 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24665 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24668 * lra-constraints.c (split_reg): Check requested split mode
24669 is supported by the register.
24671 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24673 * lra-constraints.c (simplify_operand_subreg): Remove early
24676 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24679 * lra-constraints.c (curr_insn_transform): Tighten condition
24680 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24682 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24685 * lra-constraints.c (curr_insn_transform): Handle
24686 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24688 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24691 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24693 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24695 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24698 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24701 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24703 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24706 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24708 * final.c (last_columnnum, override_columnnum): New variables.
24709 (final_start_function): Set last_columnnum, pass it to begin_prologue
24710 hook and pass 0 to dwarf2out_begin_prologue.
24711 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24712 to source_line debug hook.
24713 (notice_source_line): Compute last_columnnum and for debug_column_info
24714 return true on column changes.
24715 * debug.h (struct gcc_debug_hooks): Add column argument to
24716 source_line and begin_prologue hooks.
24717 (debug_nothing_int_charstar_int_bool): Remove prototype.
24718 (debug_nothing_int_int_charstar,
24719 debug_nothing_int_int_charstar_int_bool): New prototypes.
24720 (dwarf2out_begin_prologue): Add column argument.
24721 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24722 begin_prologue hooks.
24723 (debug_nothing_int_charstar_int_bool): Remove.
24724 (debug_nothing_int_int_charstar,
24725 debug_nothing_int_int_charstar_int_bool): New functions.
24726 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24727 through to dwarf2out_source_line.
24728 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24729 (dwarf2out_source_line): Add column argument, emit it if requested.
24730 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24732 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24733 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24734 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24735 through to dwarf2out_begin_prologue.
24736 (vmsdbgout_source_line): Add column argument, pass it through to
24737 dwarf2out_source_line.
24738 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24739 dbxout_source_line caller.
24740 (dbxout_source_line): Add column argument.
24742 * common.opt (gno-column-info, gcolumn-info): New options.
24743 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24744 (check_die): Also test for multiple DW_AT_decl_column attributes.
24745 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24746 DW_AT_decl_column if requested.
24747 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24749 (gen_variable_die): Likewise.
24750 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24751 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24754 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24755 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24756 (ix86_handle_option): Handle OPT_m3dnowa.
24757 * doc/invoke.texi (-m3dnowa): Document.
24758 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24759 -m3dnowa instead of -m3dnow -march=athlon.
24762 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24763 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24765 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24768 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24769 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24770 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24771 generator for vsx_xxpermdi_<mode>_be.
24772 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24773 force big-endian semantics.
24774 (vsx_xxpermdi_<mode>_be): New define_expand with same
24775 implementation as previous version of vsx_xxpermdi_<mode>.
24777 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24779 PR tree-optimization/79327
24780 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24781 variable, its initialization and use.
24783 2017-02-17 Julia Koval <julia.koval@intel.com>
24785 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24786 (OPTION_MASK_ISA_PKU_UNSET): New.
24787 (ix86_handle_option): Handle -mrdpid.
24788 * config/i386/cpuid.h (bit_RDPID): New.
24789 * config/i386/driver-i386.c (host_detect_local_cpu):
24790 Detect RDPID feature.
24791 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24792 * config/i386/i386-c.c (ix86_target_macros_internal):
24794 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24795 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24796 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24797 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24798 * config/i386/i386.md (define_insn "rdpid"): New.
24799 * config/i386/i386.opt Add -mrdpid.
24800 * config/i386/immintrin.h (_rdpid_u32): New.
24802 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24804 PR rtl-optimization/79541
24805 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24806 instead of transforming it into USE.
24808 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24810 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24811 If HONOR_SNANS (SFmode) force the input to a register.
24812 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24813 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24814 an frsp or similar insn.
24816 2017-02-17 Martin Liska <mliska@suse.cz>
24818 PR rtl-optimization/79577
24819 * params.def (selsched-max-sched-times): Increase minimum to 1.
24821 2017-02-17 Martin Liska <mliska@suse.cz>
24823 PR rtl-optimization/79574
24824 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24826 2017-02-17 Martin Liska <mliska@suse.cz>
24828 PR tree-optimization/79529
24829 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24830 ssa_defined_default_def_p to handle cases which are implicitly
24832 * tree-ssa.c (ssa_defined_default_def_p): New function.
24833 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24834 which are implicitly defined.
24835 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24837 2017-02-17 Richard Biener <rguenther@suse.de>
24839 PR middle-end/79576
24840 * params.def (max-ssa-name-query-depth): Limit to 10.
24842 2017-02-17 Richard Biener <rguenther@suse.de>
24844 PR tree-optimization/79552
24845 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24848 2017-02-17 Richard Biener <rguenther@suse.de>
24851 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24853 2017-02-17 Marek Polacek <polacek@redhat.com>
24855 PR middle-end/79536
24856 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24857 (fold_negate_expr): New wrapper.
24859 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24861 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24862 Correct terminology and de-emphasize pre-standard behavior.
24864 2017-02-16 Alan Modra <amodra@gmail.com>
24866 PR rtl-optimization/79286
24867 * ira.c (def_dominates_uses): New function.
24868 (update_equiv_regs): Don't create an equivalence for insns that
24869 may trap where the register def does not dominate the use.
24871 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24873 PR rtl-optimization/78127
24874 * lra.c (lra): Call lra_eliminate before finish the loop after
24877 2017-02-16 Richard Biener <rguenther@suse.de>
24879 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24881 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24882 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24883 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24884 (isl_val_int_from_wi): New function.
24885 (extract_affine_gmp): Rename to ...
24886 (extract_affine_wi): ... this, take a widest_int.
24887 (extract_affine_int): Just wrap extract_affine_wi.
24888 (add_param_constraints): Use isl_val_int_from_wi.
24889 (add_loop_constraints): Likewise, and extract_affine_wi.
24891 2017-02-15 Jeff Law <law@redhat.com>
24893 PR middle-end/79521
24894 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24895 ira_init_register_move_cost_if_necessary.
24897 2017-02-15 Martin Sebor <msebor@redhat.com>
24899 PR middle-end/32003
24900 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24901 removed in a prior commit.
24903 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24905 PR tree-optimization/79347
24906 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24907 counters during peeling.
24909 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24911 * Makefile.in (site.exp): Remove "set ISLVER".
24913 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24916 * real.c (real_from_integer): Call real_convert even for decimal.
24918 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24921 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24923 2017-02-14 Andrew Pinski <apinski@cavium.com>
24925 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24926 cores and change the partno/implementer to be correct.
24927 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24928 the 'B" as the implementer.
24929 * config/aarch64/aarch64-tune.md: Regenerate.
24931 2017-02-14 Carl Love <cel@us.ibm.com>
24933 * config/rs6000/rs6000.c: Add case statement entry to make the
24934 xvcvuxdsp built-in argument unsigned.
24935 * config/rs6000/vsx.md: Fix the source and return operand types so they
24936 match the instruction definitions from the ISA document. Fix typo
24937 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24940 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24943 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24944 member early_clobber_alts.
24945 * lra-lives.c (reg_early_clobber_p): New.
24946 (process_bb_lives): Use it.
24947 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24948 (debug_operand_data): Initialize early_clobber_alts.
24949 (setup_operand_alternative): Set up early_clobber_alts.
24950 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24951 alternatives to new_insn_reg.
24952 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24954 (lra_update_insn_regno_info): Pass the new arg.
24956 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24958 PR middle-end/79505
24959 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24960 (new_oacc_loop_raw): Don't clear already cleared fields.
24963 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24964 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24965 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24967 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24970 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24972 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24975 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24976 the extra instruction to the right place to store 128-bit constant
24979 2017-02-14 Martin Sebor <msebor@redhat.com>
24981 PR middle-end/79448
24982 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24983 warning for strings of unknown length.
24985 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24987 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24989 2017-02-14 Jeff Law <law@redhat.com>
24992 * ira-costs.c (scan_one_insn): Initialize register move costs
24993 for pseudos seen in USE/CLOBBER insns.
24995 PR tree-optimization/79095
24996 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24997 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24998 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24999 if the operands are known to be not equal, then the resulting range
25001 (intersect_ranges): If the new range is ~[0,0] and the old range is
25002 wide, then prefer ~[0,0].
25003 * tree-vrp.c (overflow_comparison_p_1): New function.
25004 (overflow_comparison_p): New function.
25005 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25006 if NAME is used in an overflow test.
25007 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25008 overflow check that can be expressed as an equality test, then adjust
25009 ops to be that equality test.
25011 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25013 * config/s390/s390-builtin-types.def: Remove flags argument.
25014 * config/s390/s390.c (s390_init_builtins): Likewise.
25016 2017-02-14 Martin Liska <mliska@suse.cz>
25018 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25019 vector. Fix trailing white spaces.
25021 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25023 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25026 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25028 PR rtl-optimization/68664
25029 * config/arm/arm.c (arm_sched_can_speculate_insn):
25030 New function. Declare prototype.
25031 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25033 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25035 PR rtl-optimization/68664
25036 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25038 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25040 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25042 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25043 max skip bytes for function, loop and jump.
25045 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25047 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25048 ABS_EXPR for gimple dump.
25050 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25053 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25055 PR tree-optimization/79408
25056 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25057 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25058 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25059 also if rhs1 is INTEGER_CST.
25061 2017-02-14 Richard Biener <rguenther@suse.de>
25063 PR middle-end/79432
25064 * tree-into-ssa.c (insert_phi_nodes): When the function can
25065 have abnormal edges rewrite SSA names with broken use-def
25066 dominance out of SSA and register them for PHI insertion.
25068 2017-02-13 Martin Sebor <msebor@redhat.com>
25070 PR middle-end/79496
25071 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25072 clearing info.nowrite flag when snprintf size argument is a range.
25074 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25076 * cprop.c (cprop_jump): Add missing space in string literal.
25077 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25078 (get_constraint_for_component_ref): Likewise.
25079 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25080 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25081 * lra-constraints.c (process_alt_operands): Likewise.
25082 * ipa-inline.c (inline_small_functions): Likewise.
25083 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25084 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25085 * trans-mem.c (diagnose_tm_1_op): Likewise.
25086 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25087 (grid_parallel_clauses_gridifiable): Likewise.
25089 * config/nvptx/mkoffload.c (process): Add space in between
25092 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25093 "MOD4_SSE_REGS" and "ALL_REGS".
25095 * spellcheck.c (test_data): Add , in between "foo" and "food".
25097 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25100 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25101 boundary crossing check and subsequent code generation agree.
25103 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25105 * config/aarch64/aarch64.c (has_memory_op): Delete.
25106 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25109 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25111 PR rtl-optimization/79388
25112 PR rtl-optimization/79450
25113 * combine.c (distribute_notes): When removing TEM_INSN for which
25114 corresponding dest has last value recorded, invalidate that last
25117 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25119 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25120 of explicit '@'. Add missing assembly comment marker on branch costs
25123 2017-02-13 Nathan Sidwell <nathan@acm.org>
25125 * gengtype-lex.l (<in_struct>): Add '/'.
25127 2017-02-13 Martin Liska <mliska@suse.cz>
25130 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25132 2017-02-13 Richard Biener <rguenther@suse.de>
25134 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25136 * configure: Re-generate.
25137 * config.in: Likewise.
25138 * graphite-dependences.c: Simplify as if
25139 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25140 * graphite-isl-ast-to-gimple.c: Likewise.
25141 * graphite-optimize-isl.c: Likewise.
25142 * graphite-poly.c: Likewise.
25143 * graphite-sese-to-poly.c: Likewise.
25144 * graphite.h: Likewise.
25145 * toplev.c: Include isl/version.h and use isl_version () for
25146 printing the ISL version.
25147 * doc/install.texi: Update ISL requirement.
25149 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25151 * doc/standards.texi (Standards): Update reference to
25154 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25156 * doc/extend.texi (Named Address Spaces): sourceware.org now
25158 * doc/install.texi (Binaries): Ditto.
25161 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25163 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25164 standard terminology "stringize"/"stringizing" throughout.
25165 * doc/cppinternals.texi: Likewise.
25167 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25169 * doc/extend.texi: Fix some spelling mistakes and typos.
25170 * doc/invoke.texi: Likewise.
25172 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25175 * params.def (inline-min-speedup) Change from 10 to 8.
25177 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25179 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25182 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25185 * ipa-inline-analysis.c (get_minimal_bb): New function.
25186 (record_modified): Use it.
25187 (remap_edge_change_prob): Handle also ancestor functions.
25189 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25191 * doc/contrib.texi (Contributors): Remove broken link into
25192 the Mauve CVS repository.
25194 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25196 PR middle-end/79454
25197 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25198 result computation whenever lhs doesn't have vector mode, not
25199 just when it has BLKmode.
25201 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25203 * doc/makefile.texi (profiledbootstrap): Refer to the
25204 installation instructions only in textual form.
25206 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25209 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25211 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25213 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25214 (Specific): Update mingw-w64 reference.
25216 (Specific): Remove broken link to Renesas RX processor.
25218 2017-02-10 Richard Biener <rguenther@suse.de>
25220 * toplev.c (process_options): Do not mention obsolete graphite
25221 options when printing sorry message about missing graphite support.
25222 Mention -floop-nest-optimize.
25224 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25226 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25227 (vtst_p16): Likewise.
25228 (vtstq_p8): Likewise.
25229 (vtstq_p16): Likewise.
25231 (vtstq_p64): Likewise.
25232 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25233 (vset_lane_p64): New.
25234 (vsetq_lane_p64): New.
25236 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25238 PR tree-optimization/79411
25239 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25240 stmt operands are SSA_NAMEs used in abnormal phis.
25241 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25244 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25247 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25250 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25252 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25254 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25256 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25260 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25261 not arbitrary TREE_CONSTANT.
25264 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25265 "omp declare target link" attribute unless is_global_var.
25266 * omp-offload.c (find_link_var_op): Likewise.
25268 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25269 Chung-Lin Tang <cltang@codesourcery.com>
25271 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25273 (gimplify_adjust_omp_clauses): Don't delete TILE.
25274 (gimplify_omp_for): Deal with TILE.
25275 * internal-fn.c (expand_GOACC_TILE): New function.
25276 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25278 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25279 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25281 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25282 avoid DIV for outermost collapse var.
25283 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25284 Remove out of date comments, fix whitespace.
25285 * omp-general.c (omp_extract_for_data): Deal with tiling.
25286 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25287 adjust OLF_DIM_BASE value.
25288 (struct omp_for_data): Add tiling field.
25289 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25290 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25291 for auto loops. Remove default auto determining, moved to
25292 oacc_loop_fixed_partitions.
25293 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25294 stmts, add e_mask field.
25295 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25296 (oacc_thread_numbers): Use oacc_dim_call.
25297 (oacc_xform_tile): New.
25298 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25299 (finish_oacc_loop): Adjust for ifns vector.
25300 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25301 add case for GOACC_TILE fns.
25302 (oacc_loop_xform_loop): Delete.
25303 (oacc_loop_process): Iterate over call list directly, and add
25304 handling for GOACC_TILE fns.
25305 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25307 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25308 vector partitioning to outer loops. Assign 2 partitions to loops
25309 when available. Add TILE handling.
25310 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25311 (execite_oacc_device_lower): Process GOACC_TILE fns,
25312 ignore unknown specs.
25313 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25314 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25315 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25317 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25319 * configure.ac (ACX_BUGURL): Update.
25320 * configure: Regenerate.
25322 2017-02-09 Richard Biener <rguenther@suse.de>
25324 PR tree-optimization/69823
25325 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25326 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25328 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25330 * config/arc/arc-c.def: Add __NPS400__ definition.
25331 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25332 (TARGET_NPS400): Define.
25334 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25336 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25338 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25339 pointer, arch_info.
25340 (arc_cpu_types): Fill the arch_info field with a pointer into the
25341 arc_arch_types table.
25342 (arc_selected_cpu): Declare.
25343 * config/arc/arc.c (arc_selected_cpu): Make global.
25344 (arc_selected_arch): Delete.
25345 (arc_base_cpu): Delete.
25346 (arc_override_options): Remove references to deleted variables,
25347 update access to arch information.
25348 (ARC_OPT): Update access to arch information.
25349 (ARC_OPTX): Likewise.
25350 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25351 (TARGET_ARC600): Update access to arch information.
25352 (TARGET_ARC601): Likewise.
25353 (TARGET_ARC700): Likewise.
25354 (TARGET_EM): Likewise.
25355 (TARGET_HS): Likewise.
25356 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25359 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25362 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25363 condition/operands for integer GE/LE/GEU/LEU operations.
25365 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25367 PR translation/79397
25368 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25371 2017-02-08 Martin Jambor <mjambor@suse.cz>
25374 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25375 whether allocation happened.
25376 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25377 nothing was allocated.
25379 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25381 PR tree-optimization/79408
25382 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25383 constant, but SSA_NAME with a known integer range, use the minimum
25384 of that range instead of op1 to determine if modulo can be replaced
25385 with its first operand.
25387 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25389 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25391 2017-02-08 Richard Biener <rguenther@suse.de>
25393 PR tree-optimization/71824
25394 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25395 Check all loops contained in the merged region.
25397 2017-02-07 Andrew Pinski <apinski@cavium.com>
25399 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25401 2017-02-07 Andrew Pinski <apinski@cavium.com>
25403 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25404 (thunderxt88): Likewise.
25405 (thunderxt81): Disable LSE and change v8.1 to v8.
25406 (thunderxt83): Likewise.
25408 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25409 Richard Biener <rguenther@suse.de>
25411 PR middle-end/79399
25412 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25413 type from int to size_t.
25414 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25416 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25418 PR rtl-optimization/79386
25419 * cprop.c (bypass_conditional_jumps): Initialize
25420 bypass_last_basic_block already before splitting bbs after
25421 unconditional traps...
25422 (bypass_conditional_jumps): ... rather than here.
25425 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25426 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25427 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25428 fix -masm=intel patterns.
25430 2017-02-07 Richard Biener <rguenther@suse.de>
25432 PR tree-optimization/79256
25433 PR middle-end/79278
25434 * builtins.c (get_object_alignment_2): Use min_align_of_type
25435 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25436 and ADJUST_FIELD_ALIGN.
25438 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25440 * doc/tm.texi: Regenerate.
25441 * stor-layout.c (layout_decl): Adjust.
25442 (update_alignment_for_field): Likewise.
25443 (place_field): Likewise.
25444 (min_align_of_type): Likewise.
25445 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25446 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25447 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25448 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25449 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25450 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25451 * config/i386/i386.c (x86_field_alignment): Likewise.
25452 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25453 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25454 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25455 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25456 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25457 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25461 2017-01-30 Richard Biener <rguenther@suse.de>
25463 PR tree-optimization/79256
25464 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25465 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25468 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25470 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25471 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25472 builtins to SImode and emit a zero-extend, if necessary.
25474 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25476 * docs/invoke.texi (RISC-V Options): Alphabetize.
25478 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25480 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25483 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25485 * config/riscv/riscv.c: New file.
25486 * common/config/riscv/riscv-common.c: Likewise.
25487 * config.gcc: Likewise.
25488 * config/riscv/constraints.md: Likewise.
25489 * config/riscv/elf.h: Likewise.
25490 * config/riscv/generic.md: Likewise.
25491 * config/riscv/linux.h: Likewise.
25492 * config/riscv/multilib-generator: Likewise.
25493 * config/riscv/peephole.md: Likewise.
25494 * config/riscv/pic.md: Likewise.
25495 * config/riscv/predicates.md: Likewise.
25496 * config/riscv/riscv-builtins.c: Likewise.
25497 * config/riscv/riscv-c.c: Likewise.
25498 * config/riscv/riscv-ftypes.def: Likewise.
25499 * config/riscv/riscv-modes.def: Likewise.
25500 * config/riscv/riscv-opts.h: Likewise.
25501 * config/riscv/riscv-protos.h: Likewise.
25502 * config/riscv/riscv.h: Likewise.
25503 * config/riscv/riscv.md: Likewise.
25504 * config/riscv/riscv.opt: Likewise.
25505 * config/riscv/sync.md: Likewise.
25506 * config/riscv/t-elf-multilib: Likewise.
25507 * config/riscv/t-linux: Likewise.
25508 * config/riscv/t-linux-multilib: Likewise.
25509 * config/riscv/t-riscv: Likewise.
25510 * configure.ac: Likewise.
25511 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25512 Waterman as RISC-V maintainers.
25513 * doc/install.texi: Add RISC-V entries.
25514 * doc/invoke.texi: Add RISC-V options section.
25515 * doc/md.texi: Add RISC-V constraints section.
25516 * configure: Regenerated.
25518 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25521 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25522 false values to be constant vectors with all 0 or all 1 bits set.
25523 (vcondu<mode><mode>): Likewise.
25524 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25526 (fpmask_comparison_operator): Update comment.
25527 (vecint_comparison_operator): New predicate.
25528 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25529 vector conditionals when the true and false values are constant
25530 vectors with all 0 bits or all 1 bits set.
25532 2017-02-06 Martin Sebor <msebor@redhat.com>
25534 PR tree-optimization/79376
25535 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25537 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25539 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25540 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25541 to simplify split condition.
25543 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25545 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25546 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25549 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25551 PR rtl-optimization/68664
25552 * target.def (can_speculate_insn): New hook.
25553 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25554 * doc/tm.texi: Regenerate.
25555 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25556 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25557 (rs6000_sched_can_speculate_insn): New function.
25559 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25561 PR tree-optimization/79284
25562 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25563 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25564 vectorizable_mask_load_store, vectorizable_operation,
25565 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25566 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25567 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25568 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25569 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25570 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25571 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25572 is_gimple_assign (stmt). Replace another such test with
25573 is_gimple_assign (stmt).
25575 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25578 * config/avr/avr.c (rtl-iter.h): Include it.
25579 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25580 (avr_legitimate_combined_insn): ...and implementation.
25582 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25584 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25585 * config/s390/s390.c (s390_const_operand_ok)
25586 (s390_canonicalize_comparison, s390_extract_part)
25587 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25588 (s390_contiguous_bitmask_p, s390_rtx_costs)
25589 (legitimize_pic_address): Likewise.
25590 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25591 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25592 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25593 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25594 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25596 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25598 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25599 REGNO($0) == REGNO($1).
25601 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25603 * config/s390/linux.h(SIZE_TYPE): Add comment.
25605 2017-02-06 Julian Brown <julian@codesourcery.com>
25606 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25607 Virendra Pathak <virendra.pathak@broadcom.com>
25609 * config/aarch64/aarch64-cores.def: Change the scheduler
25611 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25612 * config/aarch64/thunderx2t99.md: New file.
25614 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25616 * doc/standards.texi (Go Language): Update link to language
25619 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25621 * tree-eh.c (lower_resx): Sanitize profile.
25622 (cleanup_empty_eh_move_lp): Likewise.
25624 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25627 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25629 * cfgloopmanip.h (loop_version): Update prototype.
25630 * modulo-sched.c (sms_schedule): Update call of loop_version.
25631 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25632 * tree-parloops.c (gen_parallel_loop): Likewise.
25633 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25634 * tree-ssa-loop-split.c (split_loop): Likewise.
25635 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25636 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25638 2017-02-05 Martin Liska <mliska@suse.cz>
25641 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25643 (print_operand_address): Initialize a struct to zero.
25645 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25647 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25648 garbage collector only in textual form.
25650 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25652 * doc/extend.texi (x86 specific memory model extensions for
25653 transactional memory): Simplify a phrase.
25655 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25658 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25659 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25660 (atomic_storedi_1): Likewise.
25662 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25664 PR tree-optimization/79338
25665 * tree-parloops.c (gather_scalar_reductions): Don't call
25666 vect_analyze_loop_form for loop->inner before destroying loop's
25669 2017-02-03 Martin Sebor <msebor@redhat.com>
25671 PR tree-optimization/79327
25672 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25673 when precision has resulted in leading zeros.
25674 (format_integer): Adjust the likely counter to assume an unknown
25675 argument that may be zero is non-zero.
25677 2017-02-03 Jason Merrill <jason@redhat.com>
25680 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25681 avoid copying non-taken branch.
25683 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25685 PR tree-optimization/79340
25686 * tree-vect-loop.c (vectorizable_reduction): Release
25687 vec_defs elements after safe_splicing them into other vectors.
25690 PR tree-optimization/79327
25691 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25692 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25694 (format_integer): Use wide_int_to_tree instead of build_int_cst
25695 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25696 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25697 of shortest and longest sequence.
25699 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25701 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25702 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25704 2017-02-03 Walter Lee <walt@tilera.com>
25707 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25708 after initial stackframe link reg save.
25709 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25711 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25714 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25715 wu for stxssp alternative.
25717 2017-02-03 Martin Sebor <msebor@redhat.com>
25719 PR tree-optimization/79352
25720 * gimple-fold.c (get_range_strlen): Add argument.
25721 (get_range_strlen): Change return type to bool.
25722 (get_maxval_strlen): Pass in a dummy argument.
25723 * gimple-fold.h (get_range_strlen): Change return type to bool.
25724 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25725 * tree.h (array_at_struct_end_p): Add argument.
25726 * tree.c (array_at_struct_end_p): Handle it.
25728 2017-02-03 Martin Liska <mliska@suse.cz>
25731 * multiple_target.c (create_dispatcher_calls): Redirect edge
25732 from a caller of a dispatcher.
25733 (expand_target_clones): Make the clones local.
25734 (ipa_target_clone): Do both target clones and resolvers.
25735 (ipa_dispatcher_calls): Remove the pass.
25736 (pass_dispatcher_calls::gate): Likewise.
25737 (make_pass_dispatcher_calls): Likewise.
25738 * passes.def (pass_target_clone): Put as very first IPA early
25741 2017-02-03 Martin Liska <mliska@suse.cz>
25743 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25744 in case of a function with ifunc attribute.
25746 2017-02-03 Martin Liska <mliska@suse.cz>
25748 * cgraph.c (cgraph_node::dump): Dump function version info.
25749 * symtab.c (symtab_node::dump_base): Add missing new line.
25751 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25753 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25754 (ifcombine_ifandif): Use it.
25756 2017-02-03 Martin Liska <mliska@suse.cz>
25758 * doc/invoke.texi: Document default value for
25759 use-after-scope-direct-emission-threshold.
25761 2017-02-03 Martin Liska <mliska@suse.cz>
25763 PR tree-optimization/79339
25764 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25765 (format_floating): Likewise.
25767 2017-02-03 Martin Liska <mliska@suse.cz>
25770 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25772 (ipa_node_params_t::remove): Likewise.
25773 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25774 initialization from removed ipa_node_params_t::insert.
25775 (ipa_node_params::~ipa_node_params): Move from removed
25776 ipa_node_params_t::release.
25777 * symbol-summary.h (symbol_summary::m_released): New member.
25778 Do not release a summary twice. Do not allow to call finalizer
25779 for types of a summary that live in GGC memory.
25781 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25783 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25786 2017-02-02 Martin Sebor <msebor@redhat.com>
25788 PR middle-end/79275
25789 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25790 (format_string): Tighten up the range of output for non-constant
25791 strings and correct the expected range for wide non-constant strings.
25793 2017-02-02 Martin Sebor <msebor@redhat.com>
25795 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25797 PR middle-end/32003
25798 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25800 (-fdump-tree-@var): Add to index and document how to come up
25801 with pass-specific option and dump file names.
25802 (-fdump-passes): Clarify where to look for output.
25804 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25806 PR middle-end/77445
25807 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25808 statistics of the analyzed path; allow threading for speed when
25809 any of BBs along the path are optimized for speed.
25811 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25813 PR middle-end/78468
25814 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25815 settings of the virtual registers.
25818 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25820 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25821 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25824 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25826 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25827 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25829 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25831 * config/s390/s390.md: Add missing comments with the expanded
25833 * config/s390/vector.md: Likewise.
25834 * config/s390/vx-builtins.md: Likewise.
25836 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25839 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25840 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25841 conditions on a single line.
25843 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25845 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25846 __S390_VX__ to __VX__.
25848 2017-02-01 Andrew Pinski <apinski@cavium.com>
25850 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25851 stmt_info to record_stmt_cost.
25852 (vect_get_known_peeling_cost): Pass stmt_info if known to
25854 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25855 cpu_vector_cost field into
25856 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25857 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25858 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25859 splitting of scalar_stmt_cost and vec_stmt_cost.
25860 (thunderx_vector_cost): Likewise.
25861 (cortexa57_vector_cost): LIkewise.
25862 (exynosm1_vector_cost): Likewise.
25863 (xgene1_vector_cost): Likewise.
25864 (thunderx2t99_vector_cost): Improve after the splitting of the two
25866 (aarch64_builtin_vectorization_cost): Update for the splitting of
25867 scalar_stmt_cost and vec_stmt_cost.
25869 2017-02-01 Torvald Riegel <triegel@redhat.com>
25870 Richard Henderson <rth@redhat.com>
25872 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25873 conditional on existance of a fast atomic load.
25874 * optabs-query.c (can_atomic_load_p): New function.
25875 * optabs-query.h (can_atomic_load_p): Declare it.
25876 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25877 no fast atomic load is available for the particular size of access.
25878 (expand_atomic_compare_and_swap): Likewise.
25879 (expand_atomic_load): Likewise.
25880 (expand_atomic_store): Likewise.
25881 (expand_atomic_fetch_op): Likewise.
25882 * testsuite/lib/target-supports.exp
25883 (check_effective_target_sync_int_128): Remove x86 because it provides
25884 no fast atomic load.
25885 (check_effective_target_sync_int_128_runtime): Likewise.
25887 2017-02-01 Richard Biener <rguenther@suse.de>
25889 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25890 (graphite_transform_loops): Provide opt-info for optimized nests.
25891 * tree-parloop.c (parallelize_loops): Provide opt-info for
25892 parallelized loops.
25894 2017-02-01 Richard Biener <rguenther@suse.de>
25896 PR middle-end/79315
25897 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25898 was not set before.
25900 2017-02-01 Richard Biener <rguenther@suse.de>
25902 PR tree-optimization/71824
25903 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25904 Verify the loops are valid in the merged SESE region.
25905 (scop_detection::can_represent_loop_1): Check analyzing the
25906 evolution of the number of iterations in the region succeeds.
25908 2017-01-31 Ian Lance Taylor <iant@golang.org>
25910 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25911 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25913 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25915 PR preprocessor/79210
25916 * input.c (get_substring_ranges_for_loc): Replace line_width
25917 assertion with error-handling.
25919 2017-01-31 Richard Biener <rguenther@suse.de>
25921 PR tree-optimization/77318
25922 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25923 (create_pw_aff_from_tree): Take loop parameter.
25924 (add_condition_to_pbb): Pass loop of the condition to
25925 create_pw_aff_from_tree.
25927 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25929 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25930 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25932 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25936 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25938 (convert_int_to_float128): Likewise.
25939 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25940 (convert_int_to_float128): Likewise.
25941 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25942 (UNSPEC_IEEE128_CONVERT): Likewise.
25943 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25944 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25945 Use local variables for IBM extended format.
25946 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25947 (fix_trunc<mode>si2_fprs): Likewise.
25948 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25949 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25950 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25951 to know that we can now have integers of all sizes in vector
25953 (fix<uns>_<mode>di2_hw): Likewise.
25954 (float<uns>_<mode>si2_hw): Likewise.
25955 (fix_<mode>si2_hw): Likewise.
25956 (fixuns_<mode>si2_hw): Likewise.
25957 (float<uns>_<mode>di2_hw): Likewise.
25958 (float_<mode>di2_hw): Likewise.
25959 (float_<mode>si2_hw): Likewise.
25960 (floatuns_<mode>di2_hw): Likewise.
25961 (floatuns_<mode>si2_hw): Likewise.
25962 (xscvqp<su>wz_<mode>): Delete, no longer used.
25963 (xscvqp<su>dz_<mode>): Likewise.
25964 (xscv<su>dqp_<mode>): Likewise.
25965 (ieee128_mfvsrd_64bit): Likewise.
25966 (ieee128_mfvsrd_32bit): Likewise.
25967 (ieee128_mfvsrwz): Likewise.
25968 (ieee128_mtvsrw): Likewise.
25969 (ieee128_mtvsrd_64bit): Likewise.
25970 (ieee128_mtvsrd_32bit): Likewise.
25972 2017-01-31 Martin Liska <mliska@suse.cz>
25975 * ipa-prop.c (ipa_free_all_node_params): Call release method
25976 instead of ~sumbol_summary to not to trigger double times
25979 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25981 PR tree-optimization/71691
25982 * bitmap.h (class auto_bitmap): New.
25983 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25984 is_maybe_undefined instead of ssa_undefined_value_p.
25986 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25988 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25989 __S390_ARCH_LEVEL__ to __ARCH__.
25991 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25993 PR tree-optimization/79267
25994 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25995 if should_remove_lhs_p is true.
25997 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26000 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26001 (add_alignment_attribute): New.
26002 (base_type_die): Add alignment attribute.
26003 (subrange_type_die): Likewise.
26004 (modified_type_die): Likewise.
26005 (gen_array_type_die): Likewise.
26006 (gen_descr_array_type_die: Likewise.
26007 (gen_enumeration_type_die): Likewise.
26008 (gen_subprogram_die): Likewise.
26009 (gen_variable_die): Likewise.
26010 (gen_field_die): Likewise.
26011 (gen_ptr_to_mbr_type_die): Likewise.
26012 (gen_struct_or_union_type_die): Likewise.
26013 (gen_subroutine_type_die): Likewise.
26014 (gen_typedef_die): Likewise.
26015 (base_type_cmp): Compare alignment attribute.
26017 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26020 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26021 (setb_unsigned) New pattern for setb with CCUNS.
26022 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26023 subfc./subfe sequence to avoid overflow problems. Generate a
26024 shorter sequence with cmpld/setb for power9.
26025 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26026 for generating subfc. instruction.
26027 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26028 now uses this instruction.
26030 2017-01-30 Ian Lance Taylor <iant@google.com>
26033 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26034 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26036 2017-01-30 Martin Sebor <msebor@redhat.com>
26038 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26039 Move constant to the right of a relational operator.
26040 (get_mpfr_format_length, format_character, format_string): Ditto.
26041 (should_warn_p, maybe_warn): Same.
26043 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26045 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26048 * asan.c (get_translation_unit_decl): Remove function.
26049 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26051 2017-01-30 Martin Liska <mliska@suse.cz>
26053 PR gcov-profile/79259
26054 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26055 -fprofile-generate.
26057 2017-01-30 Martin Liska <mliska@suse.cz>
26060 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26061 Initialize variables with NULL value.
26063 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26066 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26068 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26070 2017-01-30 Richard Biener <rguenther@suse.de>
26072 PR tree-optimization/79276
26073 * tree-vrp.c (process_assert_insertions): Properly adjust common
26074 when removing a duplicate.
26076 * gcc.dg/torture/pr79276.c: New testcase.
26078 2017-01-30 Richard Biener <rguenther@suse.de>
26080 PR tree-optimization/79256
26081 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26082 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26084 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26086 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26089 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26090 ("*r<noxa>sbg_<mode>_sll_bitmask")
26091 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26092 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26093 Use contiguous_bitmask_nowrap_operand.
26095 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26098 * config/rs6000/altivec.h (vec_xl): Revise #define.
26099 (vec_xst): Likewise.
26101 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26103 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26105 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26107 PR rtl-optimization/79194
26108 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26109 traps before call to bypass_conditional_jumps.
26111 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26113 PR tree-optimization/71374
26114 * lra-constraints.c (check_conflict_input_operands): New.
26115 (match_reload): Use it.
26117 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26120 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26121 account to calculate conflict_set.
26123 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26125 PR rtl-optimization/78559
26126 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26127 other_insn in combine.
26129 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26131 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26132 uint16_type_node for BT_UINT16.
26134 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26136 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26137 "RTL Tests" to menu.
26138 (GIMPLE Tests): New node.
26139 (RTL Tests): New node.
26141 2017-01-27 Richard Biener <rguenther@suse.de>
26143 PR tree-optimization/79245
26144 * tree-loop-distribution.c (distribute_loop): Apply cost
26145 modeling also to detected patterns.
26147 2017-01-27 Richard Biener <rguenther@suse.de>
26149 PR tree-optimization/71433
26150 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26151 (compare_assert_loc): New function.
26152 (process_assert_insertions): Sort and optimize assert locations
26153 to remove duplicates and push down identical assertions on
26154 edges to their destination block.
26156 2017-01-27 Richard Biener <rguenther@suse.de>
26158 PR tree-optimization/79244
26159 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26160 out SSA names even if abnormal.
26162 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26164 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26165 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26166 instead of MPFR_RNDN.
26168 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26171 * arm.c (arm_option_override): Don't call build_target_option_node
26172 until after doing all option overrides.
26173 (arm_valid_target_attribute_tree): Likewise.
26175 2017-01-27 Martin Liska <mliska@suse.cz>
26177 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26178 for {cd}tors and C++ {cd}tors.
26180 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26182 * config/s390/s390.md ("*setmem_long_and")
26183 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26185 2017-01-26 Martin Sebor <msebor@redhat.com>
26187 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26190 2017-01-26 Martin Sebor <msebor@redhat.com>
26192 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26193 HAVE_DFmode before using XFmode or DFmode.
26194 (parse_directive): Avoid using the z length modifier to avoid
26195 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26197 PR middle-end/78703
26198 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26199 to accept adjustment as an array.
26200 (get_int_range): New function.
26201 (struct directive): Make width and prec arrays.
26202 (directive::set_width, directive::set_precision): Call get_int_range.
26203 (format_integer, format_floating): Handle width and precision ranges.
26204 (format_string, parse_directive): Same.
26206 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26209 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26210 comdat_type_p set, just clone them, but keep the children in the
26214 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26215 which have direct callers with -fvar-tracking-assignments enabled
26217 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26218 inside of type units.
26220 2017-01-26 Martin Sebor <msebor@redhat.com>
26222 PR middle-end/78703
26223 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26225 (struct format_result): Replace number_chars, number_chars_min,
26226 and number_chars_max with a single member of struct result_range.
26228 (format_result::operator+=): Adjust.
26229 (struct fmtresult): Remove bounded. Handle likely and unlikely
26231 (fmtresult::adjust_for_width_or_precision): New function.
26232 (fmtresult:type_max_digits): New function.
26233 (bytes_remaining): Handle likely and unlikely counters.
26234 (min_bytes_remaining): Remove.
26235 (format_percent): Simplify.
26236 (format_integer, format_floating): Set likely and unlikely counters.
26237 (get_string_length, format_character, format_string): Same.
26238 (format_plain, should_warn_p): New function.
26239 (maybe_warn): Call should_warn_p. Update diagnostic messages
26240 and handle those for all directives, including plain strings.
26241 (format_directive): Handle likely and unlikely counters.
26242 Remove unnecessary quoting from diagnostics. Add an informational
26244 (add_bytes): Remove.
26245 (pass_sprintf_length::compute_format_length): Simplify.
26246 (try_substitute_return_value): Handle likely and unlikely counters.
26248 2017-01-26 Carl Love <cel@us.ibm.com>
26250 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26251 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26253 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26256 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26257 endianess for subregs into account.
26258 * lra-constraints.c (lra_constraints): Do risky transformations
26259 always on the first iteration.
26260 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26261 last_call_used_reg_set.
26262 (process_bb_lives): Define and use last_call_used_reg_set.
26263 * lra.c (lra): Always continue after lra_constraints on the first
26266 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26268 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26270 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26272 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26274 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26275 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26276 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26277 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26278 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26279 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26280 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26281 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26282 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26284 2017-01-26 Marek Polacek <polacek@redhat.com>
26287 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26288 for the third operand.
26290 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26292 PR middle-end/79236
26293 * omp-low.c (struct omp_context): Add simt_stmt field.
26294 (scan_omp_for): Return omp_context *.
26295 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26296 context to the _simt_ SIMD stmt.
26297 (lower_omp_for): For combined SIMD with sibling _simt_
26298 SIMD, make sure to use the same decls in _looptemp_
26299 clauses as in the sibling.
26301 2017-01-26 David Sherwood <david.sherwood@arm.com>
26303 PR middle-end/79212
26304 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26307 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26310 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26311 emit fld b; fld a; if possible.
26313 * brig-builtins.def: Update copyright years.
26314 * config/arm/arm_acle_builtins.def: Update copyright years.
26316 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26319 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26320 constraint instead of o for the stxsd instruction.
26322 2017-01-25 Carl Love <cel@us.ibm.com>
26324 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26325 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26327 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26329 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26331 2017-01-25 Richard Biener <rguenther@suse.de>
26333 PR tree-optimization/69264
26334 * target.def (vector_alignment_reachable): Improve documentation.
26335 * doc/tm.texi: Regenerate.
26336 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26338 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26339 earlier changes with respect to TYPE_USER_ALIGN.
26340 (vector_alignment_reachable_p): Likewise. Improve dumping.
26342 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26345 * config/arm/arm.md (xordi3): Force constant operand into a register
26348 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26350 * doc/invoke.texi (-fstore-merging): Correct default optimization
26351 levels at which it is enabled.
26352 (-O): Move -fstore-merging from list to...
26355 2017-01-25 Richard Biener <rguenther@suse.de>
26358 * omp-expand.c: Include debug.h.
26359 (expand_omp_taskreg): Make sure to generate early debug before
26360 outlining anything from a function.
26361 (expand_omp_target): Likewise.
26362 (grid_expand_target_grid_body): Likewise.
26364 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26367 * asan.c (get_translation_unit_decl): New function.
26368 (asan_add_global): Extract modules file name from globals
26369 TRANSLATION_UNIT_DECL name.
26371 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26374 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26375 for long calls with APCS frame and VFP.
26377 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26379 * cfg.c (original_copy_tables_initialized_p): New function.
26380 * cfg.h (original_copy_tables_initialized_p): New decl.
26381 * cfgrtl.c (relink_block_chain): Guard the call to
26382 free_original_copy_tables with a call to
26383 original_copy_tables_initialized_p.
26384 * cgraph.h (symtab_node::native_rtl_p): New decl.
26385 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26386 (symtab_node::needed_p): Don't assert for early assembly output
26387 for __RTL functions.
26388 (cgraph_node::finalize_function): Set "force_output" for __RTL
26390 (cgraph_node::analyze): Bail out early for __RTL functions.
26391 (analyze_functions): Update assertion to support __RTL functions.
26392 (cgraph_node::expand): Bail out early for __RTL functions.
26393 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26395 * function.h (struct function): Update comment for field
26397 * gimple-expr.c: Include "tree-pass.h".
26398 (gimple_has_body_p): Return false for __RTL functions.
26399 * Makefile.in (OBJS): Add run-rtl-passes.o.
26400 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26402 (gcc::pass_manager::get_clean_slate): New accessor.
26403 * passes.c: Include "insn-addr.h".
26404 (should_skip_pass_p): Add logging. Update logic for running
26405 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26406 property-provider override so it is only done for gimple passes.
26408 (skip_pass): New function.
26409 (execute_one_pass): Call skip_pass when skipping passes.
26410 * read-md.c (md_reader::read_char): Support filtering
26411 the input to a subset of line numbers.
26412 (md_reader::md_reader): Initialize fields
26413 m_first_line and m_last_line.
26414 (md_reader::read_file_fragment): New function.
26415 * read-md.h (md_reader::read_file_fragment): New decl.
26416 (md_reader::m_first_line): New field.
26417 (md_reader::m_last_line): New field.
26418 * read-rtl-function.c (function_reader::create_function): Only
26419 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26420 curr_properties. Set DECL_INITIAL to a dummy block.
26421 (read_rtl_function_body_from_file_range): New function.
26422 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26424 * run-rtl-passes.c: New file.
26425 * run-rtl-passes.h: New file.
26427 2017-01-24 Jeff Law <law@redhat.com>
26429 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26432 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26434 PR tree-optimization/79159
26435 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26436 (record_nonwrapping_iv): Improve boundary using above function if no
26437 value range information.
26439 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26440 Martin Jambor <mjambor@suse.cz>
26442 * brig-builtins.def: New file.
26443 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26444 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26445 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26446 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26447 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26448 * builtin-types.def (BT_INT8): New.
26449 (BT_INT16): Likewise.
26450 (BT_UINT8): Likewise.
26451 (BT_UINT16): Likewise.
26452 (BT_FN_ULONG): Likewise.
26453 (BT_FN_UINT_INT): Likewise.
26454 (BT_FN_UINT_ULONG): Likewise.
26455 (BT_FN_UINT_LONG): Likewise.
26456 (BT_FN_UINT_PTR): Likewise.
26457 (BT_FN_ULONG_PTR): Likewise.
26458 (BT_FN_INT8_FLOAT): Likewise.
26459 (BT_FN_INT16_FLOAT): Likewise.
26460 (BT_FN_UINT32_FLOAT): Likewise.
26461 (BT_FN_UINT16_FLOAT): Likewise.
26462 (BT_FN_UINT8_FLOAT): Likewise.
26463 (BT_FN_UINT64_FLOAT): Likewise.
26464 (BT_FN_UINT16_UINT32): Likewise.
26465 (BT_FN_UINT32_UINT16): Likewise.
26466 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26467 (BT_FN_INT_PTR_INT): Likewise.
26468 (BT_FN_UINT_PTR_UINT): Likewise.
26469 (BT_FN_LONG_PTR_LONG): Likewise.
26470 (BT_FN_ULONG_PTR_ULONG): Likewise.
26471 (BT_FN_VOID_UINT64_UINT64): Likewise.
26472 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26473 (BT_FN_INT8_INT8_INT8): Likewise.
26474 (BT_FN_INT16_INT16_INT16): Likewise.
26475 (BT_FN_INT_INT_INT): Likewise.
26476 (BT_FN_UINT_FLOAT_UINT): Likewise.
26477 (BT_FN_FLOAT_UINT_UINT): Likewise.
26478 (BT_FN_ULONG_UINT_UINT): Likewise.
26479 (BT_FN_ULONG_UINT_PTR): Likewise.
26480 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26481 (BT_FN_UINT_UINT_UINT): Likewise.
26482 (BT_FN_VOID_UINT_PTR): Likewise.
26483 (BT_FN_UINT_UINT_PTR: Likewise.
26484 (BT_FN_UINT32_UINT64_PTR): Likewise.
26485 (BT_FN_INT_INT_UINT_UINT): Likewise.
26486 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26487 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26488 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26489 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26490 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26491 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26492 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26493 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26494 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26495 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26496 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26497 * doc/frontends.texi: List BRIG FE.
26498 * doc/install.texi (Testing): Add BRIG tesring requirements.
26499 * doc/invoke.texi (Overall Options): Mention BRIG.
26500 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26502 2017-01-24 Richard Biener <rguenther@suse.de>
26504 PR translation/79208
26505 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26507 2017-01-24 Martin Jambor <mjambor@suse.cz>
26510 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26511 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26512 and known_contexts.
26514 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26516 PR middle-end/79123
26517 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26518 casts from signed to unsigned really don't have a range.
26520 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26522 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26523 GMP_RNDx for compatiblity.
26525 2017-01-24 Martin Liska <mliska@suse.cz>
26528 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26529 that would prevent us to call alloca with -1 as argument.
26531 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26533 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26534 Avoid -Wformat-security warning.
26536 2017-01-23 Andrew Pinski <apinski@cavium.com>
26538 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26541 2017-01-23 Martin Sebor <msebor@redhat.com>
26543 PR middle-end/78703
26544 * gimple-ssa-sprintf.c (warn_level): New global.
26545 (format_integer): Use it here and throughout the rest of the file.
26546 Use the same switch to compute sign as base.
26547 (maybe_warn): New function.
26548 (format_directive): Factor out warnings into maybe_warn.
26549 Add debugging output. Use warn_level.
26550 (add_bytes): Use warn_level.
26551 (pass_sprintf_length::compute_format_length): Add debugging output.
26552 (try_substitute_return_value): Same.
26553 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26555 PR middle-end/78703
26556 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26557 (struct fmtresult, format_integer, format_floating): Adjust.
26558 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26559 (get_string_length, format_string,format_directive): Same.
26560 (pass_sprintf_length::compute_format_length): Same.
26561 (try_substitute_return_value): Simplify slightly.
26563 PR middle-end/78703
26564 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26565 (fmtresult::operator+=): Outlined.
26566 (struct fmtresult): Add ctors.
26567 (struct conversion_spec): Rename...
26568 (struct directive): ...to this. Add and remove data members.
26569 (directive::set_width, directive::set_precision): New functions.
26570 (format_percent): Use fmtresult ctor.
26571 (get_width_and_precision): Remove.
26572 (format_integer): Make naming changes. Avoid computing width and
26574 (format_floating): Same. Adjust indentation.
26575 (format_character, format_none): New functions.
26576 (format_string): Moved character handling to format_character.
26577 (format_directive): Remove arguments, change return type.
26578 (parse_directive): New function.
26579 (pass_sprintf_length::compute_format_length): Move directive
26580 parsing to parse_directive.
26582 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26584 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26585 (assign_assembler_name_if_needed): ... this.
26586 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26587 (assign_assembler_name_if_needed): ... this.
26588 (free_lang_data_in_cgraph): Adjust callers.
26589 * cgraphunit.c (cgraph_node::analyze): Likewise.
26590 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26593 2017-01-23 Richard Biener <rguenther@suse.de>
26595 PR tree-optimization/79088
26596 PR tree-optimization/79188
26597 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26598 resetting loop bounds after last path deletion. Reset loop
26599 bounds of the target loop, make code match the comments.
26600 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26601 Make sure loops need no fixups.
26603 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26605 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26606 exponent support with double type for first argument.
26607 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26608 type returned by __builtin_vec_extract_sig,
26609 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26610 functions from "vector int" to "vector unsigned int" or from
26611 "vector long long int" to "vector unsigned long long int".
26612 Changed type returned by __builtin_vec_extract_exp,
26613 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26614 functions from "vector int" to "vector unsigned int" or from
26615 "vector long long int" to "vector unsigned long long int".
26616 Changed return type of __builtin_vec_test_data_class,
26617 __builtin_vec_test_data_class_sp, and
26618 __builtin_vec_test_data_class_dp from "vector int" to
26619 "vector bool int" or from "vector long long int" to "vector bool
26620 long long int" and changed second argument type from "unsigned
26621 int" to "int". Added new overloaded function forms "vector float
26622 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26623 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26624 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26625 double, vector unsigned long long int)" and "vector double
26626 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26627 long int)". Changed return type of
26628 __builtin_scalar_test_data_class and
26629 __builtin_scalar_test_data_class_sp and
26630 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26631 int" and changed second argument from "unsigned int" to "int".
26632 Changed type returned by __builtin_scalar_test_neg,
26633 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26634 from "int" to "bool int". Added new overloaded function form
26635 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26636 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26637 exponent double-precision with floating point first argument.
26638 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26639 documentation of scalar_test_data_class, scalar_test_neg,
26640 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26641 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26642 vec_test_data_class built-in functions to reflect refinements in
26643 their type signatures.
26645 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26647 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26649 (aarch64_elf_asm_destructor): Likewise.
26651 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26653 PR rtl-optimization/78634
26654 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26655 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26656 * ifcvt.c (noce_try_cmove): Add missing cost check.
26658 PR rtl-optimization/71724
26659 * combine.c (if_then_else_cond): Look for situations where it is
26660 beneficial to undo the work of one of the recursive calls.
26662 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26664 PR tree-optimization/70754
26665 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26666 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26667 combined stmt before it if not NULL.
26668 (combine_chains): Process refs reversely and compute dominance point
26671 2017-01-23 Martin Liska <mliska@suse.cz>
26673 PR tree-optimization/79196
26674 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26675 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26677 (strlen_optimize_stmt): Call the renamed function.
26679 2017-01-23 Michael Matz <matz@suse.de>
26681 PR tree-optimization/78384
26682 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26684 2017-01-23 Richard Biener <rguenther@suse.de>
26686 PR tree-optimization/79186
26687 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26688 both incoming edges before moving an assert.
26690 2017-01-23 Martin Jambor <mjambor@suse.cz>
26692 * ipa-prop.c (load_from_param_1): Removed.
26693 (load_from_unmodified_param): Bits from load_from_param_1 put back
26695 (load_from_param): Removed.
26696 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26697 with stmt. Reverted back to use of load_from_unmodified_param.
26699 2017-01-23 Martin Jambor <mjambor@suse.cz>
26702 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26703 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26704 field a pointer to garbage collected vector, mark lattices and
26705 ipcp_orig_node with GTY((skip)).
26706 (ipa_get_param_count): Adjust to descriptors being a pointer.
26707 (ipa_get_param): Likewise.
26708 (ipa_get_type): Likewise.
26709 (ipa_get_param_move_cost): Likewise.
26710 (ipa_set_param_used): Likewise.
26711 (ipa_get_controlled_uses): Likewise.
26712 (ipa_set_controlled_uses): Likewise.
26713 (ipa_is_param_used): Likewise.
26714 (ipa_node_params_t): Move into garbage collector. New methods insert
26716 (ipa_node_params_sum): Annotate wth GTY(()).
26717 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26719 (ipa_load_from_parm_agg): Adjust declaration.
26720 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26721 * ipa-profile.c (ipa_profile): Likewise.
26722 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26723 (ipa_populate_param_decls): Make descriptors parameter garbage
26725 (ipa_dump_param): Adjust to descriptors being a pointer.
26726 (ipa_alloc_node_params): Likewise.
26727 (ipa_initialize_node_params): Likewise.
26728 (load_from_param_1): Make descriptors parameter garbage collected.
26729 (load_from_unmodified_param): Likewise.
26730 (load_from_param): Likewise.
26731 (ipa_load_from_parm_agg): Likewise.
26732 (ipa_node_params::~ipa_node_params): Removed.
26733 (ipa_free_all_node_params): Remove call to delete operator.
26734 (ipa_node_params_t::insert): New.
26735 (ipa_node_params_t::remove): Likewise.
26736 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26737 copy known_csts and known_contexts vectors.
26738 (ipa_read_node_info): Adjust to descriptors being a pointer.
26739 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26741 (ipcp_transform_function): Make descriptors variable garbage
26744 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26746 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26747 * config/i386/avx512dqintrin.h: Ditto.
26748 * config/i386/avx512fintrin.h: Ditto.
26749 * config/i386/i386.c: Handle new builtins.
26750 * config/i386/i386-builtin.def: Add new builtins.
26751 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26752 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26754 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26755 Martin Liska <mliska@suse.cz>
26757 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26758 * asan.c (asan_expand_poison_ifn): Support stores and use
26759 appropriate ASAN report function.
26760 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26761 * internal-fn.def (ASAN_POISON_USE): Declare.
26762 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26763 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26764 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26765 ASAN_POISON calls w/o LHS.
26766 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26767 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26768 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26769 * gimplify.c (asan_poison_variables): Add attribute
26770 use_after_scope_memory to variables that really needs to live
26772 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26773 having the attribute.
26775 2017-01-23 Martin Liska <mliska@suse.cz>
26777 * asan.c (create_asan_shadow_var): New function.
26778 (asan_expand_poison_ifn): Likewise.
26779 * asan.h (asan_expand_poison_ifn): New declaration.
26780 * internal-fn.c (expand_ASAN_POISON): Likewise.
26781 * internal-fn.def (ASAN_POISON): New builtin.
26782 * sanopt.c (pass_sanopt::execute): Expand
26783 asan_expand_poison_ifn.
26784 * tree-inline.c (copy_decl_for_dup_finish): Make function
26786 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26787 * tree-ssa.c (is_asan_mark_p): New function.
26788 (execute_update_addresses_taken): Rewrite local variables
26789 (identified just by use-after-scope as addressable) into SSA.
26791 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26793 * doc/install.texi (Specific): opensource.apple.com uses https
26794 now. Remove trailing slash.
26796 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26798 * README.Portability: Remove note on an Irix compatibility issue.
26800 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26802 * gcov.c (INCLUDE_ALGORITHM): Define.
26803 (INCLUDE_VECTOR): Define.
26804 No longer include <vector> and <algorithm> directly.
26806 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26808 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26810 * doc/invoke.texi (Code Gen Options): Ditto.
26812 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26815 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26817 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26819 rtl-optimization/79125
26820 * cprop.c (local_cprop_pass): Handle cases where we make an
26821 unconditional trap.
26823 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26827 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26828 read from, for big endian.
26830 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26832 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26833 register pauth builtins for LP64 only.
26835 2017-01-20 Marek Polacek <polacek@redhat.com>
26838 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26841 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26843 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26845 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26846 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26847 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26849 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26852 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26853 in big-endian mode when they are not a single duplicated value.
26855 2017-01-20 Richard Biener <rguenther@suse.de>
26857 * BASE-VER: Bump to 7.0.1.
26859 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26861 * omp-low.c (omplow_simd_context): New struct. Use it...
26862 (lower_rec_simd_input_clauses): ...here and...
26863 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26864 references to idx, lane, max_vf, is_simt.
26866 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26868 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26871 2017-01-20 Martin Jambor <mjambor@suse.cz>
26873 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26874 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26876 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26877 (GTFILES): Rename hsa.c to hsa-common.c.
26878 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26879 * hsa-dump.c: Likewise.
26880 * hsa-gen.c: Likewise.
26881 * hsa-regalloc.c: Likewise.
26882 * ipa-hsa.c: Likewise.
26883 * omp-expand.c: Likewise.
26884 * omp-low.c: Likewise.
26885 * toplev.c: Likewise.
26887 2017-01-20 Marek Polacek <polacek@redhat.com>
26890 * doc/invoke.texi: Document -Wduplicated-branches.
26891 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26892 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26893 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26894 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26895 return 0 only when not OEP_LEXICOGRAPHIC.
26896 (fold_build_cleanup_point_expr): Use the expression
26897 location when building CLEANUP_POINT_EXPR.
26898 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26899 * tree.c (add_expr): Handle error_mark_node.
26901 2017-01-20 Martin Liska <mliska@suse.cz>
26904 * tree-profile.c (init_ic_make_global_vars): Do not call
26906 (gimple_init_gcov_profiler): Likewise.
26908 2017-01-20 Martin Liska <mliska@suse.cz>
26911 * cgraph.h (maybe_create_reference): Remove argument and
26913 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26915 * ipa-cp.c (create_specialized_node): Likewise.
26916 * symtab.c (symtab_node::maybe_create_reference): Handle
26917 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26919 2017-01-20 Martin Liska <mliska@suse.cz>
26921 * read-rtl-function.c (function_reader::create_function): Use
26922 build_decl instread of build_decl_stat.
26924 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26926 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26927 * config/i386/avx512dqintrin.h: Ditto.
26928 * config/i386/avx512fintrin.h: Ditto.
26929 * config/i386/i386-builtin-types.def: Add new types.
26930 * config/i386/i386.c: Handle new types.
26931 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26932 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26933 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26934 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26935 (__builtin_ia32_kshiftridi): New.
26936 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26938 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26942 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26943 define to rs6000_init_stack_protect_guard.
26944 (rs6000_init_stack_protect_guard): New function.
26946 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26947 Yunqiang Su <yunqiang.su@imgtec.com>
26949 * config.gcc (supported_defaults): Add madd4.
26950 (with_madd4): Add validation.
26951 (all_defaults): Add madd4.
26952 * config/mips/mips.opt (mmadd4): New option.
26953 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26955 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26957 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26958 (ISA_HAS_FUSED_MADD4): Likewise.
26959 * doc/invoke.texi (-mmadd4): Document the new option.
26960 * doc/install.texi (--with-madd4): Document the new option.
26962 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26964 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26965 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26966 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26967 (aarch64_init_pauth_hint_builtins): New.
26968 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26969 (aarch64_expand_builtin): Expand new builtins.
26971 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26973 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26974 * combine-stack-adj.c (no_unhandled_cfa): Handle
26975 REG_CFA_TOGGLE_RA_MANGLE.
26976 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26977 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26978 info for return address signing.
26979 (aarch64_expand_epilogue): Likewise.
26981 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26983 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26984 * config/aarch64/aarch64-protos.h
26985 (aarch64_return_address_signing_enabled): New declaration.
26986 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26988 (aarch64_expand_prologue): Sign return address before it's pushed onto
26990 (aarch64_expand_epilogue): Authenticate return address fetched from
26992 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26993 (aarch64_attributes): New function attributes for "sign-return-address".
26994 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26995 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26996 ("*do_return"): Generate combined instructions according to key index.
26997 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26998 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27000 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27001 * config/aarch64/aarch64.opt (msign-return-address=): New.
27002 * doc/extend.texi (AArch64 Function Attributes): Documents
27003 "sign-return-address=".
27004 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27006 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27008 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27009 overall option summary.
27011 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27013 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27014 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27015 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27016 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27018 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27020 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27021 -mpower9-minmax by default for -mcpu=power9.
27022 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27023 128-bit floating point.
27025 2017-01-20 Alan Modra <amodra@gmail.com>
27027 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27028 optimizing for size.
27030 2017-01-20 Alan Modra <amodra@gmail.com>
27033 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27034 for strcmp and strncmp from corresponding builtin decl.
27036 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27038 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27039 instead of i386/rtems-64.h.
27040 * config/i386/rtems-64.h: Remove.
27042 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27046 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27048 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27050 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27052 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27053 Change int to HOST_WIDE_INT.
27054 * config/aarch64/aarch64-protos.h
27055 (aarch64_simd_gen_const_vector_dup): Likewise.
27056 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27058 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27060 * langhooks-def.h (lhd_type_for_size): New decl.
27061 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27062 * langhooks.c (lhd_type_for_size): New function, taken from
27065 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27067 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27068 define_bypass for CR latency.
27069 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27070 (power9-alu2): Add define_bypass for CR latency.
27072 (power9-mul): Update insn latency.
27073 (power9-mul-compare): Update insn latency, bypass latency and remove
27076 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27078 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27080 * config/aarch64/aarch64.md
27081 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27082 aarch64_nopcrelative_literal_loads.
27083 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27085 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27087 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27088 TARGET_LOONGSON_3A.
27089 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27091 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27094 * config.gcc (supported_defaults): Add lxc1-sxc1.
27095 (with_lxc1_sxc1): Add validation.
27096 (all_defaults): Add lxc1-sxc1.
27097 * config/mips/mips.opt (mlxc1-sxc1): New option.
27098 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27100 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27101 __mips_no_lxc1_sxc1.
27102 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27103 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27104 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27106 2017-01-19 Richard Biener <rguenther@suse.de>
27108 PR tree-optimization/72488
27109 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27110 sure to restore SSA info.
27111 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27113 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27115 PR rtl-optimization/79121
27116 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27117 of the inner type when shifting an extended value.
27119 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27122 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27123 interposable aliases.
27125 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27128 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27129 Use the evmergelohi instruction.
27130 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27131 (mov_sitf_e500_subreg8_2_be): Likewise.
27132 (mov_sitf_e500_subreg12_2_le): Likewise.
27133 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27134 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27135 (mov_sitf_e500_subreg8_2_le): Likewise.
27136 (mov_sitf_e500_subreg12_2_be): Likewise.
27138 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27140 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27141 attribute from vecsimple to vecperm.
27142 (altivec_vbpermq2): Likewise.
27144 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27147 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27149 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27150 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27151 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27152 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27153 case where N arg is SIZE_MAX.
27154 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27155 (cmpstrsi): Add pattern.
27157 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27159 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27160 __builtin_vec_revb builtins.
27161 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27162 built-in functions to support generation of the ISA 3.0 XXBR<x>
27163 vector byte reverse instructions.
27164 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27165 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27166 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27167 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27168 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27169 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27170 (P9V_BUILTIN_VEC_REVB): Likewise.
27171 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27172 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27173 (p9_xxbrq_v16qi): Likewise.
27174 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27175 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27176 (p9_xxbrh_v8hi): Likewise.
27177 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27178 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27179 vec_revb built-in functions.
27181 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27183 PR rtl-optimization/78952
27184 * config/i386/i386.md (any_extract): New code iterator.
27185 (*insvqi_2): Use any_extract for source operand.
27186 (*insvqi_3): Use any_shiftrt for source operand.
27188 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27190 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27192 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27194 2017-01-18 Matthias Klose <doko@ubuntu.com>
27196 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27198 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27200 * config/rs6000/altivec.h (vec_bperm): Change #define.
27201 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27202 (altivec_vbpermq2): New define_insn.
27203 (altivec_vbpermd): Likewise.
27204 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27205 function interface.
27206 (VBPERMD): Likewise.
27207 (VBPERM): New polymorphic function interface.
27208 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27209 Add entries for P9V_BUILTIN_VEC_VBPERM.
27210 * doc/extend.texi: Add interfaces for vec_bperm.
27212 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27214 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27215 first letter of error messages.
27216 (s390_resolve_overloaded_builtin): Likewise.
27217 * config/s390/s390.c (s390_expand_builtin): Likewise.
27218 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27219 (s390_valid_target_attribute_inner_p): Likewise.
27220 * config/s390/s390.md ("tabort"): Likewise.
27222 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27224 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27225 (ISA_AVOID_DIV_HILO): New macro.
27226 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27227 (ISA_HAS_DDIV): Likewise.
27229 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27231 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27233 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27235 * doc/invoke.texi (fabi-version): Spelling fix.
27237 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27240 * doc/invoke.texi (fabi-version): Mention mangling fix for
27243 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27246 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27248 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27251 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27252 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27254 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27256 (stack_protect_test): Ditto.
27257 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27258 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27260 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27261 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27262 -mstack-protector-guard-offset=.
27263 (RS/6000 and PowerPC Options): Ditto.
27265 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27267 * config/i386/i386.h (MASK_CLASS_P): New define.
27268 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27269 there are no registers from different register sets also when
27270 mask registers are used. Update function comment.
27271 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27272 to (*k/*r) and (*k/*km) alternatives.
27274 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27276 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27277 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27278 (EH_RETURN_HANDLER_RTX): New define.
27279 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27280 Force frame pointer in EH return functions.
27281 (aarch64_expand_epilogue): Add barrier for eh_return.
27282 (aarch64_final_eh_return_addr): Remove.
27283 (aarch64_eh_return_handler_rtx): New function.
27284 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27286 (aarch64_eh_return_handler_rtx): New prototype.
27288 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27290 * config/rs6000/altivec.h (vec_rlmi): New #define.
27291 (vec_vrlnm): Likewise.
27292 (vec_rlnm): Likewise.
27293 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27294 (UNSPEC_VRLNM): Likewise.
27295 (VIlong): New mode iterator.
27296 (altivec_vrl<VI_char>mi): New define_insn.
27297 (altivec_vrl<VI_char>nm): Likewise.
27298 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27300 (VRLDNM): Likewise.
27301 (RLNM): New polymorphic function entry.
27302 (VRLWMI): New monomorphic function entry.
27303 (VRLDMI): Likewise.
27304 (RLMI): New polymorphic function entry.
27305 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27306 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27307 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27310 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27313 * dwarf2out.c (field_byte_offset): Restore the
27314 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27315 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27316 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27319 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27322 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27324 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27327 * dwarf2out.c (add_data_member_location_attribute): For constant
27328 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27329 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27330 DW_AT_byte_size attributes.
27332 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27334 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27335 after forcing to constant memory when the code model is medium.
27337 2017-01-17 Julia Koval <julia.koval@intel.com>
27340 * config/i386/avx512fintrin.h
27341 (_mm512_i32gather_ps): Change __addr type to void const*.
27342 (_mm512_mask_i32gather_ps): Ditto.
27343 (_mm512_i32gather_pd): Ditto.
27344 (_mm512_mask_i32gather_pd): Ditto.
27345 (_mm512_i64gather_ps): Ditto.
27346 (_mm512_mask_i64gather_ps): Ditto.
27347 (_mm512_i64gather_pd): Ditto.
27348 (_mm512_mask_i64gather_pd): Ditto.
27349 (_mm512_i32gather_epi32): Ditto.
27350 (_mm512_mask_i32gather_epi32): Ditto.
27351 (_mm512_i32gather_epi64): Ditto.
27352 (_mm512_mask_i32gather_epi64): Ditto.
27353 (_mm512_i64gather_epi32): Ditto.
27354 (_mm512_mask_i64gather_epi32): Ditto.
27355 (_mm512_i64gather_epi64): Ditto.
27356 (_mm512_mask_i64gather_epi64): Ditto.
27357 (_mm512_i32scatter_ps): Change __addr type to void*.
27358 (_mm512_mask_i32scatter_ps): Ditto.
27359 (_mm512_i32scatter_pd): Ditto.
27360 (_mm512_mask_i32scatter_pd): Ditto.
27361 (_mm512_i64scatter_ps): Ditto.
27362 (_mm512_mask_i64scatter_ps): Ditto.
27363 (_mm512_i64scatter_pd): Ditto.
27364 (_mm512_mask_i64scatter_pd): Ditto.
27365 (_mm512_i32scatter_epi32): Ditto.
27366 (_mm512_mask_i32scatter_epi32): Ditto.
27367 (_mm512_i32scatter_epi64): Ditto.
27368 (_mm512_mask_i32scatter_epi64): Ditto.
27369 (_mm512_i64scatter_epi32): Ditto.
27370 (_mm512_mask_i64scatter_epi32): Ditto.
27371 (_mm512_i64scatter_epi64): Ditto.
27372 (_mm512_mask_i64scatter_epi64): Ditto.
27373 * config/i386/avx512pfintrin.h
27374 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27375 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27376 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27377 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27378 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27379 (_mm512_prefetch_i32scatter_ps): Ditto.
27380 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27381 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27382 (_mm512_prefetch_i64scatter_pd): Ditto.
27383 (_mm512_prefetch_i64scatter_ps): Ditto.
27384 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27385 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27386 * config/i386/avx512vlintrin.h
27387 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27388 (_mm_mmask_i32gather_ps): Ditto.
27389 (_mm256_mmask_i32gather_pd): Ditto.
27390 (_mm_mmask_i32gather_pd): Ditto.
27391 (_mm256_mmask_i64gather_ps): Ditto.
27392 (_mm_mmask_i64gather_ps): Ditto.
27393 (_mm256_mmask_i64gather_pd): Ditto.
27394 (_mm_mmask_i64gather_pd): Ditto.
27395 (_mm256_mmask_i32gather_epi32): Ditto.
27396 (_mm_mmask_i32gather_epi32): Ditto.
27397 (_mm256_mmask_i32gather_epi64): Ditto.
27398 (_mm_mmask_i32gather_epi64): Ditto.
27399 (_mm256_mmask_i64gather_epi32): Ditto.
27400 (_mm_mmask_i64gather_epi32): Ditto.
27401 (_mm256_mmask_i64gather_epi64): Ditto.
27402 (_mm_mmask_i64gather_epi64): Ditto.
27403 (_mm256_i32scatter_ps): Change __addr type to void*.
27404 (_mm256_mask_i32scatter_ps): Ditto.
27405 (_mm_i32scatter_ps): Ditto.
27406 (_mm_mask_i32scatter_ps): Ditto.
27407 (_mm256_i32scatter_pd): Ditto.
27408 (_mm256_mask_i32scatter_pd): Ditto.
27409 (_mm_i32scatter_pd): Ditto.
27410 (_mm_mask_i32scatter_pd): Ditto.
27411 (_mm256_i64scatter_ps): Ditto.
27412 (_mm256_mask_i64scatter_ps): Ditto.
27413 (_mm_i64scatter_ps): Ditto.
27414 (_mm_mask_i64scatter_ps): Ditto.
27415 (_mm256_i64scatter_pd): Ditto.
27416 (_mm256_mask_i64scatter_pd): Ditto.
27417 (_mm_i64scatter_pd): Ditto.
27418 (_mm_mask_i64scatter_pd): Ditto.
27419 (_mm256_i32scatter_epi32): Ditto.
27420 (_mm256_mask_i32scatter_epi32): Ditto.
27421 (_mm_i32scatter_epi32): Ditto.
27422 (_mm_mask_i32scatter_epi32): Ditto.
27423 (_mm256_i32scatter_epi64): Ditto.
27424 (_mm256_mask_i32scatter_epi64): Ditto.
27425 (_mm_i32scatter_epi64): Ditto.
27426 (_mm_mask_i32scatter_epi64): Ditto.
27427 (_mm256_i64scatter_epi32): Ditto.
27428 (_mm256_mask_i64scatter_epi32): Ditto.
27429 (_mm_i64scatter_epi32): Ditto.
27430 (_mm_mask_i64scatter_epi32): Ditto.
27431 (_mm256_i64scatter_epi64): Ditto.
27432 (_mm256_mask_i64scatter_epi64): Ditto.
27433 (_mm_i64scatter_epi64): Ditto.
27434 (_mm_mask_i64scatter_epi64): Ditto.
27435 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27436 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27437 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27438 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27439 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27440 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27441 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27442 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27443 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27444 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27445 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27446 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27447 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27448 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27449 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27450 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27451 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27452 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27453 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27454 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27455 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27456 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27457 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27458 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27459 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27460 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27461 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27462 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27463 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27464 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27465 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27466 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27467 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27468 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27469 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27470 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27471 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27472 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27473 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27474 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27475 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27476 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27477 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27478 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27479 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27480 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27481 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27482 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27483 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27484 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27485 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27486 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27487 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27488 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27489 definitions accordingly.
27491 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27492 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27495 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27498 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27501 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27502 hard regs for inner regno.
27504 2017-01-17 Martin Liska <mliska@suse.cz>
27507 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27508 assumption and add comment.
27510 2017-01-17 Nathan Sidwell <nathan@acm.org>
27512 * ipa-visibility.c (localize_node): New function, broken out of ...
27513 (function_and_variable_visibility): ... here. Call it.
27515 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27517 PR middle-end/77445
27518 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27519 correctly set frequency of oudgoing edge.
27520 (duplicate_thread_path): Fix profile updating.
27522 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27525 * configure.ac: Add GCC_BASE_VER.
27526 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27527 version from BASE-VER file.
27528 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27529 (gcc.o): Depend on $(BASEVER).
27530 * common.opt (dumpfullversion): New option.
27531 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27532 * doc/invoke.texi: Document -dumpfullversion.
27533 * doc/install.texi: Document --with-gcc-major-version-only.
27534 * configure: Regenerated.
27536 2017-01-17 Richard Biener <rguenther@suse.de>
27538 PR tree-optimization/71433
27539 * tree-vrp.c (register_new_assert_for): Merge same asserts
27540 on all incoming edges.
27541 (process_assert_insertions_for): Handle insertions at the
27544 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27546 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27547 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27549 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27552 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27555 2017-01-17 Alan Modra <amodra@gmail.com>
27558 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27559 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27560 symbolic stack limit when pic.
27562 2017-01-16 Martin Sebor <msebor@redhat.com>
27564 PR tree-optimization/78608
27565 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27567 2017-01-16 Jeff Law <law@redhat.com>
27570 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27571 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27572 for several include directories that may be relative to sysroot.
27573 * config/i386/x-mingw32 (gplus_includedir): Define.
27574 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27575 (native_system_includedir): Likewise.
27576 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27577 override if TARGET_SYSTEM_ROOT is defined.
27578 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27580 PR tree-optimization/79090
27581 PR tree-optimization/33562
27582 PR tree-optimization/61912
27583 PR tree-optimization/77485
27584 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27585 and computed trims into the dump file.
27587 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27589 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27591 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27594 * gimplify.c (gimplify_init_constructor): If want_value and
27595 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27599 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27600 sequence. Formatting fixes.
27601 (doloop_optimize): Formatting fixes.
27604 * gcc.c (debug_level_greater_than_spec_func): New function.
27605 (static_spec_functions): Add debug-level-gt spec function.
27606 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27608 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27609 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27610 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27611 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27612 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27613 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27615 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27617 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27618 QImode fixups to general and mask registers only.
27620 2017-01-16 Carl Love <cel@us.ibm.com>
27622 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27623 for built-in functions
27624 vector signed char vec_nabs (vector signed char)
27625 vector signed short vec_nabs (vector signed short)
27626 vector signed int vec_nabs (vector signed int)
27627 vector signed long long vec_nabs (vector signed long long)
27628 vector float vec_nabs (vector float)
27629 vector double vec_nabs (vector double)
27630 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27632 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27633 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27634 * doc/extend.texi: Update the documentation file for the new built-in
27637 2017-01-16 Martin Sebor <msebor@redhat.com>
27639 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27642 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27644 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27645 UNSPEC_VSX__XXSPLTD to require special splat handling.
27647 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27650 * system.h: Poison strndup.
27652 2017-01-16 Alan Modra <amodra@gmail.com>
27655 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27658 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27660 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27662 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27664 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27665 call recog here. Assert that INSN_CODE (insn) is non-negative.
27667 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27670 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27672 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27673 in the currently scheduled RTL fragment.
27675 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27677 PR rtl-optimization/78751
27678 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27681 2017-01-14 Jeff Law <law@redhat.com>
27683 PR tree-optimization/79090
27684 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27685 variable length stores.
27686 (compute_trims): Delete dead assignment to *trim_tail.
27687 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27690 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27692 PR rtl-optimization/78626
27693 PR rtl-optimization/78727
27694 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27695 of a block, and split such blocks after everything else is finished.
27697 2017-01-14 Alan Modra <amodra@gmail.com>
27700 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27701 target legitimate_combined_insn.
27702 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27703 (rs6000_legitimate_combined_insn): New function.
27704 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27706 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27707 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27708 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27710 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27712 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27714 2017-01-13 Jeff Law <law@redhat.com>
27716 PR tree-optimization/33562
27717 PR tree-optimization/61912
27718 PR tree-optimization/77485
27719 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27721 (delete_dead_assignment): Likewise.
27722 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27723 statement to delete_dead_call and delete_dead_assignment.
27725 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27728 * substring-locations.c (format_warning_va): Strengthen case 1 so
27729 that both endpoints of the substring must be within the format
27730 range for just the substring to be printed.
27732 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27734 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27735 * config/i386/i386.c (ix86_target_string): Add missing options
27736 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27737 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27738 flags_other and ix86_target_other to flags2_other. Display unknown
27740 (ix86_valid_target_attribute_inner_p): Add missing options and
27741 reorder options by implied ISAs, as in ix86_target_string.
27743 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27745 * hash-table.h (hash_table::too_empty_p): New function.
27746 (hash_table::expand): Use it.
27747 (hash_table::traverse): Likewise.
27748 (hash_table::empty_slot): Use sizeof (value_type) instead of
27749 sizeof (PTR) to convert bytes to elements. Shrink the table
27750 if the current size is excessive for the current number of
27753 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27755 * ira-costs.c (record_reg_classes): Break from the inner loop
27756 early once alt_fail is known to be true. Update outer loop
27757 handling accordingly.
27759 2017-01-13 Jeff Law <law@redhat.com>
27761 * tree-ssa-dse.c (decrement_count): New function.
27762 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27763 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27764 when we know the partially dead statement is a mem* function.
27766 PR tree-optimization/61912
27767 PR tree-optimization/77485
27768 * tree-ssa-dse.c: Include expr.h.
27769 (maybe_trim_constructor_store): New function.
27770 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27772 PR tree-optimization/33562
27773 PR tree-optimization/61912
27774 PR tree-optimization/77485
27775 * doc/invoke.texi: Document new dse-max-object-size param.
27776 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27777 * tree-ssa-dse.c: Include params.h.
27778 (dse_store_status): New enum.
27779 (initialize_ao_ref_for_dse): New, partially extracted from
27781 (valid_ao_ref_for_dse, normalize_ref): New.
27782 (setup_live_bytes_from_ref, compute_trims): Likewise.
27783 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27784 (maybe_trim_partially_dead_store): Likewise.
27785 (maybe_trim_complex_store): Likewise.
27786 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27787 Track what bytes live from the original store. Return tri-state
27788 for dead, partially dead or live.
27789 (dse_dom_walker): Add constructor, destructor and new private members.
27790 (delete_dead_call, delete_dead_assignment): New extracted from
27792 (dse_optimize_stmt): Make a member of dse_dom_walker.
27793 Use initialize_ao_ref_for_dse.
27795 PR tree-optimization/33562
27796 PR tree-optimization/61912
27797 PR tree-optimization/77485
27798 * sbitmap.h (bitmap_count_bits): Prototype.
27799 (bitmap_clear_range, bitmap_set_range): Likewise.
27800 * sbitmap.c (bitmap_clear_range): New function.
27801 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27803 2017-01-13 Martin Liska <mliska@suse.cz>
27806 * function.c (set_cfun): Add new argument force.
27807 * function.h (set_cfun): Likewise.
27808 * ipa-inline-transform.c (inline_call): Use the function when
27809 strict alising from is dropped for function we inline to.
27811 2017-01-13 Richard Biener <rguenther@suse.de>
27813 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27814 for dumping GIMPLE INTEGER_CSTs.
27816 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27818 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27819 to 201112L since C++17.
27821 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27824 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27825 if -fsanitize=kernel-address is present.
27827 2017-01-13 Richard Biener <rguenther@suse.de>
27829 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27830 as _Literal ( type ) number in case usual suffixes do not
27831 preserve all information.
27833 2017-01-13 Richard Biener <rguenther@suse.de>
27835 PR tree-optimization/77283
27836 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27837 and ssa-iterators.h.
27838 (is_feasible_trace): Implement a cost model based on joiner
27841 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27844 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27845 char or short to __float128/_Float128 directly.
27847 2017-01-12 Martin Sebor <msebor@redhat.com>
27849 to -Wformat-overflow.
27850 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27851 (min_bytes_remaining): Same.
27852 (get_string_length): Same.
27853 (format_string): Same.
27854 (format_directive): Same.
27856 (pass_sprintf_length::handle_gimple_call): Same.
27858 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27860 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27861 info.nowrite calls with no lhs that can't throw. Return bool
27862 whether gsi_remove has been called or not.
27863 (pass_sprintf_length::handle_gimple_call): Return bool whether
27864 try_substitute_return_value called gsi_remove. Formatting fix.
27865 (pass_sprintf_length::execute): Don't use gsi_remove if
27866 handle_gimple_call returned true.
27869 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27870 be removed due to side-effects, don't remove following barrier nor
27871 turn the successor edge into fallthru edge.
27873 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27876 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27877 element-reversing loads and stores as not swappable.
27879 2017-01-12 Nathan Sidwell <nathan@acm.org>
27880 Nicolai Stange <nicstange@gmail.com>
27882 * combine.c (try_combine): Don't ignore result of overlap checking
27883 loop. Combine overlap & asm check into single loop.
27885 2017-01-12 Richard Biener <rguenther@suse.de>
27887 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27888 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27890 2017-01-12 Richard Biener <rguenther@suse.de>
27892 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27893 and TS_TARGET_OPTION directly derive from TS_BASE.
27894 * tree-core.h (tree_optimization_option): Derive from tree_base.
27895 (tree_target_option): Likewise.
27897 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27899 * config/i386/i386.c (memory_address_length): Increase len
27900 only when rip_relative_addr_p returns false.
27902 2017-01-11 Julia Koval <julia.koval@intel.com>
27904 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27905 (OPTION_MASK_ISA_SGX_SET): New.
27906 (ix86_handle_option): Handle OPT_msgx.
27907 * config.gcc: Added sgxintrin.h.
27908 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27909 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27910 * config/i386/i386.c (ix86_target_string): Add -msgx.
27912 (ix86_option_override_internal): Handle new options.
27913 (ix86_valid_target_attribute_inner_p): Add sgx.
27914 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27915 * config/i386/i386.opt: Add msgx.
27916 * config/i386/sgxintrin.h: New file.
27917 * config/i386/x86intrin.h: Add sgxintrin.h.
27919 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27922 * fold-const.c (maybe_nonzero_address): Return 1 for function
27924 (tree_single_nonzero_warnv_p): Don't handle function local objects
27928 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27931 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27934 * opts.c: Include "spellcheck.h"
27935 (struct string_fragment): New struct.
27936 (struct edit_distance_traits<const string_fragment &>): New
27938 (get_closest_sanitizer_option): New function.
27939 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27941 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27943 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27945 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27946 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27947 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27948 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27949 for initial die_offset if dwarf_split_debug_info.
27950 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27951 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27953 (output_skeleton_debug_sections): Formatting fix. Use
27954 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27955 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27957 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27959 * config/arm/cortex-a53.md: Add bypasses for
27960 cortex_a53_r2f_cvt.
27961 (cortex_a53_r2f): Only use for transfers.
27962 (cortex_a53_f2r): Likewise.
27963 (cortex_a53_r2f_cvt): Add reservation for conversions.
27964 (cortex_a53_f2r_cvt): Likewise.
27966 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27968 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27969 to all inlined functions, change static to extern.
27971 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27974 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27976 (arm_assemble_integer): Likewise.
27978 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27980 * config.gcc: Use new awk script to check CPU, FPU and architecture
27981 parameters for --with-... options.
27982 * config/arm/parsecpu.awk: New file
27983 * config/arm/arm-cpus.in: New file.
27984 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27986 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27988 * config/arm/t-arm: Update dependency rules.
27989 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27990 of processing .def files.
27991 * config/arm/genopt.sh: Deleted.
27992 * config/arm/gentune.sh: Deleted.
27993 * config/arm/arm-cores.def: Deleted.
27994 * config/arm/arm-arches.def: Deleted.
27995 * config/arm/arm-fpus.def: Deleted.
27996 * config/arm/arm-tune.md: Regenerated.
27997 * config/arm/arm-tables.opt: Regenerated.
27998 * config/arm/arm-cpu.h: New generated file.
27999 * config/arm/arm-cpu-data.h: New generated file.
28000 * config/arm/arm-cpu-cdata.h: New generated file.
28002 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28005 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28007 (input_varpool_node): Unpack dynamically_initialized bit.
28009 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28011 PR rtl-optimization/79032
28012 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28013 the alignment of the adjusted memory reference against that of MODE,
28014 instead of the alignment of the original memory reference.
28016 2017-01-11 Martin Jambor <mjambor@suse.cz>
28018 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28020 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28021 decorated functions.
28023 2017-01-11 Richard Biener <rguenther@suse.de>
28025 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28026 set range/nonnull info for PHI results. Do not set it on
28027 stmts marked for removal.
28029 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28031 * expr.c (store_field): In the bitfield case, fetch the return value
28032 from the registers before applying a single big-endian adjustment.
28033 Always do a final load for a BLKmode value not larger than a word.
28035 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28038 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28039 that we correctly handle column numbers greater than
28040 LINE_MAP_MAX_COLUMN_NUMBER.
28042 2017-01-10 Martin Sebor <msebor@redhat.com>
28044 PR middle-end/78245
28045 * gimple-ssa-sprintf.c (get_destination_size): Call
28046 {init,fini}object_sizes.
28047 * tree-object-size.c (addr_object_size): Adjust.
28048 (pass_through_call): Adjust.
28049 (pass_object_sizes::execute): Adjust.
28050 * tree-object-size.h (fini_object_sizes): Declare.
28052 2017-01-10 Martin Sebor <msebor@redhat.com>
28054 PR tree-optimization/78775
28055 * builtins.c (get_size_range): Move...
28056 * calls.c: ...to here.
28057 (alloc_max_size): Accept zero argument.
28058 (operand_signed_p): Remove.
28059 (maybe_warn_alloc_args_overflow): Call get_size_range.
28060 * calls.h (get_size_range): Declare.
28062 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28064 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28065 from TI's devices.csv file as of September 2016.
28066 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28068 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28070 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28071 * doc/invoke.texi: Likewise.
28072 * doc/md.texi: Likewise.
28073 * doc/objc.texi: Likewise.
28075 2017-01-10 Joshua Conner <joshconner@google.com>
28077 * config/arm/fuchsia-elf.h: New file.
28078 * config/fuchsia.h: New file.
28079 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28080 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28082 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28084 2016-01-10 Richard Biener <rguenther@suse.de>
28086 PR tree-optimization/79034
28087 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28088 Propagate out degenerate PHIs in the joiner.
28090 2017-01-10 Martin Liska <mliska@suse.cz>
28092 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28093 (sort_congruence_classes_by_decl_uid): Likewise.
28094 (sort_congruence_class_groups_by_decl_uid): Likewise.
28095 (sem_item_optimizer::merge_classes): Sort class, groups in these
28096 classes and members in the groups by DECL_UID of declarations.
28097 This would make merge operations stable.
28099 2017-01-10 Martin Liska <mliska@suse.cz>
28101 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28102 usage of m_classes_vec.
28103 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28104 (sem_item_optimizer::get_group_by_hash): Likewise.
28105 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28106 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28107 (sem_item_optimizer::verify_classes): Likewise.
28108 (sem_item_optimizer::process_cong_reduction): Likewise.
28109 (sem_item_optimizer::dump_cong_classes): Likewise.
28110 (sem_item_optimizer::merge_classes): Likewise.
28111 * ipa-icf.h (congruence_class_hash): Rename from
28112 congruence_class_group_hash. Remove declaration of m_classes_vec.
28114 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28116 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28117 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28118 * config.gcc: Add avx512vpopcntdqintrin.h.
28119 * config/i386/avx512vpopcntdqintrin.h: New.
28120 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28121 * config/i386/i386-builtin-types.def: Add new types.
28122 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28123 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28124 __builtin_ia32_vpopcountq_v8di_mask): New.
28125 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28126 __AVX512VPOPCNTDQ__.
28127 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28128 (PTA_AVX512VPOPCNTDQ): Define.
28129 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28130 TARGET_AVX512VPOPCNTDQ_P): Define.
28131 * config/i386/i386.opt: Add mavx512vpopcntdq.
28132 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28133 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28135 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28137 PR middle-end/77484
28138 * predict.def (PRED_CALL): Set to 67.
28140 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28142 * expr.c (store_field): In the bitfield case, if the value comes from
28143 a function call and is of an aggregate type returned in registers, do
28144 not modify the field mode; extract the value in all cases if the mode
28145 is BLKmode and the size is not larger than a word.
28147 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28150 * config/i386/cpuid.h: Fix undefined behavior.
28152 2017-01-04 Jeff Law <law@redhat.com>
28154 PR tree-optimization/79007
28155 PR tree-optimization/67955
28156 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28157 conservative for pt.null when flag_non_call_exceptions is on.
28159 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28161 PR translation/79019
28162 PR translation/79020
28163 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28164 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28165 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28167 * config/avr/avr.opt (maccumulate-args): Likewise.
28168 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28169 * common.opt (freport-bug): Likewise.
28170 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28171 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28172 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28173 translatable string.
28174 * config/i386/i386.c (function_value_32): Likewise.
28175 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28176 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28178 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28179 * common/config/msp430/msp430-common.c (msp430_handle_option):
28181 * symtab.c (symtab_node::verify_base): Likewise.
28182 * opts.c (set_debug_level): Likewise.
28183 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28184 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28185 missing whitespace to translatable strings.
28186 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28187 * config/sh/superh.h: Likewise.
28188 * config/i386/xopintrin.h: Likewise.
28189 * config/i386/znver1.md: Likewise.
28190 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28191 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28192 * double-int.h (struct double_int): Likewise.
28193 * double-int.c (div_and_round_double): Likewise.
28194 * wide-int.cc: Likewise.
28195 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28196 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28197 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28198 (crossjumps_occurred): ... this.
28199 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28202 PR tree-optimization/78899
28203 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28204 returning bool return struct loop *, NULL for failure and the new
28206 (versionable_outer_loop_p): Don't version outer loop if it has
28207 dont_vectorized bit set.
28208 (tree_if_conversion): When versioning outer loop, ensure
28209 tree_if_conversion is performed also on the inner loop of the
28210 non-vectorizable outer loop copy.
28211 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28212 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28213 prevent vectorization of it.
28214 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28215 the outer loop vectorization of the non-scalar version is attempted
28216 before vectorization of the inner loop in scalar version. If
28217 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28218 vectorization of its inner loop.
28219 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28220 has 2 inner loops, rename also on edges from bb whose single pred
28221 is outer_loop->header. Fix typo in function comment.
28223 2017-01-09 Martin Sebor <msebor@redhat.com>
28226 * asan.c (asan_emit_stack_protection): Increase local buffer size
28227 to avoid snprintf truncation warning.
28229 2017-01-09 Andrew Pinski <apinski@cavium.com>
28231 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28232 to reference thunderx2t99 for the tuning structure
28233 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28235 (thunderx2t99_extra_costs): This.
28236 * config/aarch64/aarch64-tune.md: Regenerate.
28237 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28238 (vulcan_addrcost_table): This.
28239 (vulcan_regmove_cost): Rename to ...
28240 (thunderx2t99_regmove_cost): This.
28241 (vulcan_vector_cost): Rename to ...
28242 (thunderx2t99_vector_cost): this.
28243 (vulcan_branch_cost): Rename to ...
28244 (thunderx2t99_branch_cost): This.
28245 (vulcan_tunings): Rename to ...
28246 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28247 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28249 2017-01-09 Martin Jambor <mjambor@suse.cz>
28253 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28254 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28255 (propagate_vr_accross_jump_function): Use the above function for all
28256 value range computations for pass-through jump functions and type
28257 converasion from explicit value range values.
28258 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28259 parameters from TYPE_ARG_TYPES.
28260 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28261 (ipa_write_node_info): Stream type of the actual argument.
28262 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28264 2017-01-09 Martin Liska <mliska@suse.cz>
28267 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28268 (lookup_compiler): Do not show error message with have_E.
28270 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28272 PR tree-optimization/78938
28273 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28274 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28275 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28276 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28279 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28281 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28284 2017-01-09 Richard Biener <rguenther@suse.de>
28286 PR tree-optimization/78997
28287 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28288 name condition properly.
28290 2017-01-09 Richard Biener <rguenther@suse.de>
28293 * dwarf2out.c (is_cxx): New overload with context.
28294 (is_naming_typedef_decl): Use it.
28296 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28298 * invoke.texi (Option Summary): Correct spacing in option lists
28299 and add line breaks to fix over-long lines.
28301 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28303 PR middle-end/17660
28305 * extend.texi (Common Variable Attributes): Add xref to GCC
28306 Internals manual to explain mode attribute keywords.
28308 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28311 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28312 and Preprocessor Options.
28313 (Options for Linking): Document -pthread here....
28314 (RS/6000 and PowerPC Options): ...not here.
28315 (Solaris 2 Options): ...or here.
28316 * doc/cppopts.texi: Document -pthread.
28318 2017-01-08 Martin Sebor <msebor@redhat.com>
28320 PR middle-end/77708
28321 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28322 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28323 New member functions.
28324 (format_directive): Used them.
28326 (pass_sprintf_length::handle_gimple_call): Same.
28327 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28328 to avoid truncation for any argument.
28329 (extract_affine_mul): Same.
28330 * tree.c (get_file_function_name): Same.
28332 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28334 PR middle-end/77484
28335 * predict.def (PRED_INDIR_CALL): Set to 86.
28337 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28339 PR preprocessor/54124
28340 * doc/cppopts.texi: Reformat -d subtable to list the full name
28341 of the options. Add cross-reference to the docs for the general
28342 compiler -d options.
28343 * doc/invoke.texi (Developer Options): Add cross-reference to the
28344 preprocessor-specific -d option documentation.
28346 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28348 PR preprocessor/13498
28349 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28350 redudant material, and reflect new command-line options.
28351 (System Headers): Likewise.
28353 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28355 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28356 -isystem, and -idirafter. Copy-edit.
28357 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28358 default for -ftrack-macro-expansion. Delete obsolete and
28359 badly-formatted implementation details about -fdebug-cpp output.
28360 * doc/cppwarnopts.texi: Copy-edit.
28362 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28365 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28366 that the transition from a max line width >= 1<<10 to narrower
28367 lines works correctly.
28369 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28371 * doc/options.texi (PerFunction): New.
28372 * opt-functions.awk (switch_flags): Map both Optimization and
28373 PerFunction to CL_OPTIMIZATION.
28374 * opth-gen.awk: Test for PerFunction flag along with
28376 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28377 it only when the latter is present. Skip those that don't in
28378 the hash function generator.
28379 * common.opt (fvar-tracking): Mark as PerFunction instead of
28381 (fvar-tracking-assignments): Likewise.
28382 (fvar-tracking-assignments-toggle): Likewise.
28383 (fvar-tracking-uninit): Likewise.
28385 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28387 PR translation/79018
28388 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28391 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28394 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28395 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28396 TARGET_LONG_JUMP_TABLE_OFFSETS.
28397 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28398 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28399 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28400 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28401 * config/m68k/m68k.md (tablejump expander): Likewise.
28402 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28403 TARGET_LONG_JUMP_TABLE_OFFSETS.
28404 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28405 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28407 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28408 David Holsgrove <david.holsgrove@xilinx.com>
28410 * common/config/microblaze/microblaze-common.c
28411 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28412 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28414 * config/microblaze/microblaze.c (microblaze_must_save_register)
28415 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28417 (microblaze_eh_return): New function.
28418 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28419 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28420 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28421 * config/microblaze/microblaze.md (eh_return): New pattern.
28423 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28425 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28426 GCC_DIAGNOSTIC_STRINGIFY): Define.
28428 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28430 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28432 * config/arm/arm.md (<mcrr>): New.
28434 * config/arm/arm.c (arm_arch5te): New.
28435 (arm_option_override): Set arm_arch5te.
28436 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28438 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28439 (arm_mcrr_qualifiers): ... this. New.
28440 (MRRC_QUALIFIERS): Define to...
28441 (arm_mrrc_qualifiers): ... this. New.
28442 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28444 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28445 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28446 (MRRCI, mrrc, MRRC): New.
28447 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28448 VUNSPEC_MRRC2): New.
28450 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28452 * config/arm/arm.md (<mcr>): New.
28454 * config/arm/arm.c (arm_coproc_builtin_available): Add
28455 support for mcr, mrc, mcr2 and mrc2.
28456 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28457 (arm_mcr_qualifiers): ... this. New.
28458 (MRC_QUALIFIERS): Define to ...
28459 (arm_mrc_qualifiers): ... this. New.
28460 (MCR_QUALIFIERS): Define to ...
28461 (arm_mcr_qualifiers): ... this. New.
28462 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28464 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28465 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28466 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28467 VUNSPEC_MRC2): New.
28469 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28471 * config/arm/arm.md (*ldc): New.
28475 * config/arm/arm.c (arm_coproc_builtin_available): Add
28476 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28477 (arm_coproc_ldc_stc_legitimate_address): New.
28478 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28479 'qualifier_const_pointer'.
28480 (LDC_QUALIFIERS): Define to...
28481 (arm_ldc_qualifiers): ... this. New.
28482 (STC_QUALIFIERS): Define to...
28483 (arm_stc_qualifiers): ... this. New.
28484 * config/arm/arm-protos.h
28485 (arm_coproc_ldc_stc_legitimate_address): New.
28486 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28487 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28488 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28489 stc2, stcl, stc2l): New.
28490 * config/arm/constraints.md (Uz): New.
28491 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28492 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28493 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28494 VUNSPEC_STC2L): New.
28496 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28498 * config/arm/arm.md (<cdp>): New.
28499 * config/arm/arm.c (neon_const_bounds): Rename this ...
28500 (arm_const_bounds): ... this.
28501 (arm_coproc_builtin_available): New.
28502 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28503 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28504 (CDP_QUALIFIERS): Define to...
28505 (arm_cdp_qualifiers): ... this. New.
28507 (arm_expand_builtin_args): Add case for 6 arguments.
28508 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28509 (arm_const_bounds): ... this.
28510 (arm_coproc_builtin_available): New.
28511 * config/arm/arm_acle.h (__arm_cdp): New.
28513 * config/arm/arm_acle_builtins.def (cdp): New.
28515 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28516 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28517 'arm_const_bounds'.
28518 * config/arm/types.md (coproc): New.
28519 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28520 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28521 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28522 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28524 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28526 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28527 (UBINOP_QUALIFIERS): New.
28529 (acle_builtin_data): New. Change comment.
28530 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28531 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28532 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28533 arm_acle_builtins.def.
28534 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28535 (arm_init_acle_builtins): New.
28536 (CRC32_BUILTIN): Remove.
28537 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28538 crc32cb, crc32ch and crc32cw.
28539 (arm_init_crc32_builtins): Remove.
28540 (arm_init_builtins): Use arm_init_acle_builtins rather
28541 than arm_init_crc32_builtins.
28542 (arm_expand_acle_builtin): New.
28543 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28544 * config/arm/arm_acle_builtins.def: New.
28546 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28548 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28549 (arm_builtin_datum): ... this.
28550 (arm_init_neon_builtin): Rename to ...
28551 (arm_init_builtin): ... this. Add a new parameters PREFIX
28552 and USE_SIG_IN_NAME.
28553 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28554 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28555 'arm_builtin_datum'.
28556 (arm_init_vfp_builtins): Likewise.
28557 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28558 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28559 (arm_expand_neon_args): Rename to ...
28560 (arm_expand_builtin_args): ... this. Rename builtin_arg
28561 enum values and differentiate between ARG_BUILTIN_MEMORY
28562 and ARG_BUILTIN_NEON_MEMORY.
28563 (arm_expand_neon_builtin_1): Rename to ...
28564 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28565 values, arm_expand_builtin_args and add bool parameter NEON.
28566 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28567 (arm_expand_vfp_builtin): Likewise.
28568 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28570 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28572 PR middle-end/77484
28573 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28574 * predict.c (tree_estimate_probability_bb): Reverse direction of
28575 polymorphic call predictor.
28577 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28579 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28580 (determine_pass_name_match): ...this new function and...
28581 (should_skip_pass_p): ...this new function.
28583 2017-01-06 Nathan Sidwell <nathan@acm.org>
28585 * ipa-visibility.c (function_and_variable_visibility): Reformat
28586 comments and long lines. Remove extrneous if.
28587 * symtab.c (symtab_node::make_decl_local): Fix code format.
28588 (symtab_node::set_section_for_node): Fix comment typo.
28590 2017-01-06 Martin Liska <mliska@suse.cz>
28593 * lra-constraints.c: Rename invariant to lra_invariant.
28594 * predict.c (set_even_probabilities): Initialize e to NULL.
28596 2017-01-05 Martin Sebor <msebor@redhat.com>
28598 PR tree-optimization/78910
28599 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28600 (format_integer): Correct off-by-one error in the handling
28601 of precision with negative numbers in signed conversions..
28603 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28605 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28607 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28609 PR tree-optimization/71016
28610 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28611 factor_out_conditional_conversion. Formatting fix.
28612 (factor_out_conditional_conversion): Add cond_stmt argument.
28613 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28614 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28617 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28619 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28620 read-rtl-function.o, and selftest-rtl.o.
28621 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28622 (selftest::aarch64_test_loading_full_dump): New function.
28623 (selftest::aarch64_run_selftests): New function.
28624 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28625 selftest::aarch64_run_selftests.
28626 * config/i386/i386.c
28627 (selftest::ix86_test_loading_dump_fragment_1): New function.
28628 (selftest::ix86_test_loading_call_insn): New function.
28629 (selftest::ix86_test_loading_full_dump): New function.
28630 (selftest::ix86_test_loading_unspec): New function.
28631 (selftest::ix86_run_selftests): Call the new functions.
28632 * emit-rtl.c (maybe_set_max_label_num): New function.
28633 * emit-rtl.h (maybe_set_max_label_num): New decl.
28634 * function.c (instantiate_decls): Guard call to
28635 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28636 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28638 * gensupport.c (gen_reader::gen_reader): Pass "false"
28639 for new "compact" param of rtx_reader.
28640 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28641 rather than an empty string for NULL strings.
28642 * read-md.c: Potentially include config.h rather than bconfig.h.
28643 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28644 (have_error): New global, copied from errors.c.
28645 (md_reader::read_name): Rename to...
28646 (md_reader::read_name_1): ...this, adding "out_loc" param,
28647 and converting "missing name or number" to returning false, rather
28649 (md_reader::read_name): Reimplement in terms of read_name_1.
28650 (md_reader::read_name_or_nil): New function.
28651 (md_reader::read_string): Handle "(nil)" by returning NULL.
28652 (md_reader::md_reader): Add new param "compact".
28653 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28654 (md_reader::read_file): New method.
28655 * read-md.h (md_reader::md_reader): Add new param "compact".
28656 (md_reader::read_file): New method.
28657 (md_reader::is_compact): New accessor.
28658 (md_reader::read_name): Convert return type from void to file_location.
28659 (md_reader::read_name_or_nil): New decl.
28660 (md_reader::read_name_1): New decl.
28661 (md_reader::m_compact): New field.
28662 (noop_reader::noop_reader): Pass "false" for new "compact" param
28664 (rtx_reader::rtx_reader): Add new "compact" param.
28665 (rtx_reader::read_rtx_operand): Make virtual and convert return
28666 type from void to rtx.
28667 (rtx_reader::read_until): New decl.
28668 (rtx_reader::handle_any_trailing_information): New virtual function.
28669 (rtx_reader::postprocess): New virtual function.
28670 (rtx_reader::finalize_string): New virtual function.
28671 (rtx_reader::m_in_call_function_usage): New field.
28672 (rtx_reader::m_reuse_rtx_by_id): New field.
28673 * read-rtl-function.c: New file.
28674 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28675 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28676 (selftest::verify_three_block_rtl_cfg): New decl.
28677 * read-rtl-function.h: New file.
28678 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28679 For host, include function.h, memmodel.h, and emit-rtl.h.
28680 (one_time_initialization): New function.
28681 (struct compact_insn_name): New struct.
28682 (compact_insn_names): New array.
28683 (find_code): Handle insn codes in compact dumps.
28684 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28685 (bind_subst_iter_and_attr): Likewise.
28686 (add_condition_to_string): Likewise.
28687 (add_condition_to_rtx): Likewise.
28688 (apply_attribute_uses): Likewise.
28689 (add_current_iterators): Likewise.
28690 (apply_iterators): Likewise.
28691 (initialize_iterators): Guard usage of apply_subst_iterator with
28692 #ifdef GENERATOR_FILE.
28693 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28694 (md_reader::read_mapping): Likewise.
28695 (add_define_attr_for_define_subst): Likewise.
28696 (add_define_subst_attr): Likewise.
28697 (read_subst_mapping): Likewise.
28698 (check_code_iterator): Likewise.
28699 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28701 (one_time_initialization): New function.
28702 (rtx_reader::read_until): New method.
28703 (read_flags): New function.
28704 (parse_reg_note_name): New function.
28705 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28706 Handle reuse_rtx ids.
28707 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28708 Add parsing support for RTL dumps, mirroring the special-cases in
28709 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28710 values, and calling handle_any_trailing_information.
28711 (rtx_reader::read_rtx_operand): Convert return type from void
28712 to rtx, returning return_rtx. Handle case 'e'. Call
28713 finalize_string on XSTR and XTMPL fields.
28714 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28715 "(nil)" values were omitted. Call the postprocess vfunc on the
28717 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28718 class ctor. Initialize m_in_call_function_usage. Call
28719 one_time_initialization.
28720 * rtl-tests.c (selftest::test_uncond_jump): Call
28721 set_new_first_and_last_insn.
28722 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28723 * selftest-rtl.c: New file.
28724 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28725 (selftest::get_insn_by_uid): New decl.
28726 * selftest-run-tests.c (selftest::run_tests): Call
28727 read_rtl_function_c_tests.
28728 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28729 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28732 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28734 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28735 operands in a special way. Assert that pos+len <= mode precision.
28737 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28739 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28740 3 argument Alias with unlimited for the negative form.
28741 (fno-vect-cost-model): Removed.
28743 2017-01-05 Martin Liska <mliska@suse.cz>
28745 * hsa-gen.c (gen_hsa_divmod): New function.
28746 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28748 2017-01-05 Martin Liska <mliska@suse.cz>
28751 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28754 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28756 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28757 small constant length operands.
28759 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28761 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28762 between loop iterations.
28764 2017-01-05 Martin Liska <mliska@suse.cz>
28767 * gimplify.c (gimplify_decl_expr): Compare to
28768 asan_poisoned_variables instread of checking flags.
28769 (gimplify_target_expr): Likewise.
28770 (gimplify_expr): Likewise.
28771 (gimplify_function_tree): Conditionally initialize
28772 asan_poisoned_variables.
28774 2017-01-04 Jeff Law <law@redhat.com>
28776 PR tree-optimizatin/78812
28777 * rtl.h (contains_mem_rtx_p): Prototype.
28778 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28779 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28780 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28781 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28782 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28784 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28786 * input.c (assert_char_at_range): Default-initialize actual_range.
28788 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28790 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28791 to match the caller.
28793 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28795 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28796 insns after final jump in test to emit dummy move.
28798 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28800 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28801 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28803 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28805 * multiple_target.c (create_dispatcher_calls): Init e_next.
28806 * tree-ssa-loop-split.c (split_loop): Init border.
28807 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28810 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28815 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28816 (altivec_register_operand): Do not return true if the operand
28817 contains a SUBREG mixing SImode and SFmode.
28818 (vsx_register_operand): Likewise.
28819 (vsx_reg_sfsubreg_ok): New predicate.
28820 (vfloat_operand): Do not return true if the operand contains a
28821 SUBREG mixing SImode and SFmode.
28822 (vint_operand): Likewise.
28823 (vlogical_operand): Likewise.
28824 (gpc_reg_operand): Likewise.
28825 (int_reg_operand): Likewise.
28826 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28827 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28828 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28830 (rs6000_emit_move_si_sf_subreg): New helper function.
28831 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28832 fixup SUBREGs involving SImode and SFmode.
28833 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28834 numbers for the new peephole2 optimization.
28835 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28836 the GLIBC math library that do AND/IOR/XOR operations on single
28837 precision floating point.
28838 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28839 target macros to say whether we need to avoid SUBREGs mixing
28841 (TARGET_ALLOW_SF_SUBREG): Likewise.
28842 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28843 (UNSPEC_SI_FROM_SF): Likewise.
28844 (iorxor): Change spacing.
28845 (and_ior_xor): New iterator for AND, IOR, and XOR.
28846 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28847 (movdi_from_sf_zero_ext): Likewise.
28848 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28849 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28850 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28851 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28852 (fms<mode>4): Likewise.
28853 (fnma<mode>4): Likewise.
28854 (fnms<mode>4): Likewise.
28855 (nfma<mode>4): Likewise.
28856 (nfms<mode>4): Likewise.
28858 2017-01-04 Marek Polacek <polacek@redhat.com>
28861 * doc/invoke.texi: Document -Wpointer-compare.
28863 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28865 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28868 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28869 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28872 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28875 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28876 documentation of the powerpc_popcntb_ok attribute.
28877 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28878 code to issue warning messages if a requested CPU configuration is
28879 not supported by the binary (assembler and loader) toolchain.
28880 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28881 made to define a built-in function that has been disabled.
28882 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28883 made to define a built-in function that has been disabled.
28884 (altivec_init_builtins): Add comment explaining why definition
28885 of the DST built-in functions is not preceded by an assertion
28886 check. Add assertions to prevent ICE if attempts are made to
28887 define an altivec predicate or an abs* built-in function that has
28889 (htm_init_builtins): Add comment explaining why definition of the
28890 htm built-in functions is not preceded by an assertion check.
28892 2017-01-04 Jeff Law <law@redhat.com>
28894 PR tree-optimizatin/67955
28895 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28896 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28897 the points-to solution does not include pt_null. Use DECL_PT_UID
28900 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28902 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28903 Use gen_int_mode instead of gen_lopwart for const_int operands.
28905 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28907 PR tree-optimization/71563
28908 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28909 out of range value - has low bits known to be zero.
28911 2017-01-04 Alan Modra <amodra@gmail.com>
28913 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28914 * configure: Regenerate.
28915 * config.in: Regenerate.
28917 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28920 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28921 a substring of the message, but strcmp with the whole message. Ifdef
28922 ENABLE_NLS, translate the message first using dgettext.
28924 2017-01-03 Jeff Law <law@redhat.com>
28926 PR tree-optimizatin/78856
28927 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28928 (mark_threaded_blocks): Remove code to truncate thread paths that
28929 cross multiple loop headers. Instead invalidate the cached loop
28930 iteration information and handle case of a thread path walking
28931 into an irreducible region.
28933 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28936 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28937 assertions. Add support for doing the signbit if the IEEE 128-bit
28938 floating point value is in a GPR.
28939 * config/rs6000/rs6000.md (Fsignbit): Delete.
28940 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28941 Update the length attribute if the value is in a GPR.
28942 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28943 the sign or zero extension instruction, since the value is always 0/1.
28944 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28947 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28948 extracting SImode to a GPR register so that we can generate a
28949 store, limit the vector to be in a traditional Altivec register
28950 for the vextuwrx instruction.
28952 2017-01-03 Ian Lance Taylor <iant@google.com>
28954 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28956 2017-01-03 Martin Sebor <msebor@redhat.com>
28958 PR tree-optimization/78696
28959 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28960 precision. Use MPFR for %f for greater fidelity. Correct handling
28962 (pass_sprintf_length::compute_format_length): Set width and precision
28963 specified by asrerisk to void_node for vararg functions.
28964 (try_substitute_return_value): Adjust dump output.
28966 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28968 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28970 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28972 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28973 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28974 -mlra/-mno-lra was passed to the compiler.
28976 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28978 PR rtl-optimization/65618
28979 * emit-rtl.c (try_split): Move initialization of "before" and
28980 "after" to just before the call to emit_insn_after_setloc.
28982 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28984 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28986 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28988 * dwarf2out.c (gen_enumeration_type_die): When
28989 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28991 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28993 PR tree-optimization/78965
28994 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28995 Change first argument from const call_info & to call_info &. For %n
28996 set info.nowrite to false.
28998 PR middle-end/78901
28999 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29000 possibly throwing calls.
29002 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29003 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29004 and fns handling, rather than in a separate case SSA_NAME.
29006 2017-01-02 Jeff Law <law@redhat.com>
29008 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29009 fixes for first_period and second_period variables.
29011 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29014 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29015 (*insvqi_1): New insn pattern.
29016 (*insvqi_1_mem_rex64): Ditto.
29017 (*insvqi_2): Ditto.
29018 (*insvqi_3): Rename from *insvqi.
29020 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29022 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29024 * doc/cfg.texi (Edges): Remove reference to Java.
29025 (Maintaining the CFG): Ditto.
29027 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29029 PR middle-end/77674
29030 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29031 transparent aliases.
29033 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29035 PR middle-end/77484
29036 * predict.def (PRED_CALL): Update hitrate.
29037 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29038 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29039 into direct/indirect/polymorphic variants.
29041 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29043 Update copyright years.
29045 * gcc.c (process_command): Update copyright notice dates.
29046 * gcov-dump.c (print_version): Ditto.
29047 * gcov.c (print_version): Ditto.
29048 * gcov-tool.c (print_version): Ditto.
29049 * gengtype.c (create_file): Ditto.
29050 * doc/cpp.texi: Bump @copying's copyright year.
29051 * doc/cppinternals.texi: Ditto.
29052 * doc/gcc.texi: Ditto.
29053 * doc/gccint.texi: Ditto.
29054 * doc/gcov.texi: Ditto.
29055 * doc/install.texi: Ditto.
29056 * doc/invoke.texi: Ditto.
29058 Copyright (C) 2017 Free Software Foundation, Inc.
29060 Copying and distribution of this file, with or without modification,
29061 are permitted in any medium without royalty provided the copyright
29062 notice and this notice are preserved.